lectures on groups, entropies and number theorycomplex/files/slidesptempesta.pdf · lectures on...

433
Introduction: A Foundational Perspective Formal Group Theory Strong and weak composability Groups, Difference Operators and Entropies LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de Madrid and Instituto de Ciencias Matem´ aticas (ICMAT), Madrid, Spain. TEMPLETON SCHOOL ON FOUNDATIONS OF COMPLEXITY Rio de Janeiro, October 14 - 15, 2015 Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Upload: others

Post on 19-Aug-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

LECTURES ONGROUPS, ENTROPIES AND NUMBER

THEORY

Piergiulio Tempesta

Universidad Complutense de Madridand

Instituto de Ciencias Matematicas (ICMAT),Madrid, Spain.

TEMPLETON SCHOOL ON FOUNDATIONS OF COMPLEXITY

Rio de Janeiro, October 14 - 15, 2015

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 2: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

LECTURE I

On the Mathematical Foundations of theTheory of Generalized Entropies

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 3: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

OutlineIntroduction: A Foundational Perspective

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

Formal Group TheoryDefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Strong and weak composabilityStrong and weak composabilityGeneral properties of the trace-form class

Groups, Difference Operators and EntropiesDelta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operatorsPiergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 4: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

SK and

Composability

axioms Generalized Entropies

Group-theoretical

Structure

Delta Operators Formal group

theory

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 5: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

A Foundational Perspective

Statement of the problem

I What are the mathematical foundations of the notion ofentropy?

I The main idea is to propose a group-theoretical approach to thetheory of generalized entropies

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 6: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

A Foundational Perspective

Statement of the problem

I What are the mathematical foundations of the notion ofentropy?

I The main idea is to propose a group-theoretical approach to thetheory of generalized entropies

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 7: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

A Foundational Perspective

Statement of the problem

I What are the mathematical foundations of the notion ofentropy?

I The main idea is to propose a group-theoretical approach to thetheory of generalized entropies

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 8: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

Different foundational approaches

This problem has been addressed from many different perspectives:

I Large deviation theory

I Maximum entropy principle

I Superstatistics

I Information Theory:

Divergences (Kullback-Leibler, Renyi, Bregman, Csiczar, α-,β-, etc.)

Exponential families

Information Geometry (Riemannian manifolds, conformal geometry),

etc.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 9: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

Different foundational approaches

This problem has been addressed from many different perspectives:

I Large deviation theory

I Maximum entropy principle

I Superstatistics

I Information Theory:

Divergences (Kullback-Leibler, Renyi, Bregman, Csiczar, α-,β-, etc.)

Exponential families

Information Geometry (Riemannian manifolds, conformal geometry),

etc.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 10: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

Different foundational approaches

This problem has been addressed from many different perspectives:

I Large deviation theory

I Maximum entropy principle

I Superstatistics

I Information Theory:

Divergences (Kullback-Leibler, Renyi, Bregman, Csiczar, α-,β-, etc.)

Exponential families

Information Geometry (Riemannian manifolds, conformal geometry),

etc.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 11: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

Different foundational approaches

This problem has been addressed from many different perspectives:

I Large deviation theory

I Maximum entropy principle

I Superstatistics

I Information Theory:

Divergences (Kullback-Leibler, Renyi, Bregman, Csiczar, α-,β-, etc.)

Exponential families

Information Geometry (Riemannian manifolds, conformal geometry),

etc.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 12: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

Characterization theorems

Let W ∈ N/0. Let piWi=1, with pi ≥ 0 for all i = 1, . . . ,W ,∑Wi=1 pi = 1 denote a discrete probability distribution, where PW

represents the set of all discrete probability distributions with W entries.The Boltzmann-Gibbs entropy

SBG : PW −→ R+ ∪ 0

SBG [p] :=W∑i=1

pi ln1

p1, piWi=1 ∈ PW

has been characterized by several existence and uniqueness theorems,proved in the last century.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 13: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

Khinchin’s theorem

(A. I. Khinchin, Mathematical foundations of information theory, 1957).Given two systems A and B, let pij(A ∪ B) denote its joint probabilitydistribution, i = 1, . . . ,WA, j = 1, . . . ,WB . Also,

pi (A) :=

WB∑j=1

pij(A ∪ B).

Let S(B | A) denote the conditional entropy associated with theconditional probability distribution pij(B | A) = pij(A ∪ B)/pi (A).Theorem 1.1.i) Let S be a continuous function of all its arguments (p1, . . . , pW ).i) Assume that it takes its maximum for the equiprobability distribution,i.e. when pi = 1/W , for i = 1, . . . ,W .iii) Also, assume that S(p1, . . . , pw , 0) = S(p1, . . . , pw ).iv) Given two systems A and B, S(A ∪ B) = S(A) + S(B | A).Then the unique entropy satisfying (i)-(iv) is Boltzmann’s entropy.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 14: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

Abe’s theorem

S. Abe, Phys. Lett. A, 271, 74-79 (2000).Theorem 1.2.i) Let S be a continuous function of all its arguments (p1, . . . , pW ).i) Assume that it takes its maximum for the equiprobability distribution,pi = 1/W , for i = 1, . . . ,W .iii) Also, assume that S(p1, . . . , pw , 0) = S(p1, . . . , pw ).iv) Given two systems A and B,S(A ∪ B) = S(A) + S(B | A) + (1− q)S(A)S(B | A).Then the unique entropy satisfying the axioms (i)-(iv) is Tsallis’entropy:

Sq[p] :=

∑Wi=1 pq

i − 1

1− q, q > 0.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 15: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

dos Santos’ theorem

R. J. V. dos Santos, J. M. Phys. 38, 4104 (1997)This theorem generalizes the original Shannon’s uniqueness theorem forthe Boltzmann entropy.

Theorem 1.3i) Let S be a continuous function of all its arguments (p1, . . . , pW ).ii) For the equiprobability distribution, pi = 1/W , for i = 1, . . . ,W , S isa monotonic increasing function of W , i.e.

S =W 1−q − 1

1− q.

iii) Given two statistically independent systems A and B,S(A ∪ B) = S(A) + S(B) + (1− q)S(A)S(B).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 16: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

iv) If W = WL + WM , with

PL =

WL∑i=1

pi , (WL terms)

PM =W∑

i=WL+1

pi , (WM terms)

(hence PL + PM = 1)we have:

S [p] = S(pL, pM) + (pL)qS

(pi

pL

)+ (pM)qS

(pi

pM

).

Then, the unique entropy satisfying the axioms (i)-(iv) is Tsallis’entropy Sq.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 17: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

The modern Shannon-Khinchin formulation

I (SK1) (Continuity). The function S(p1, . . . , pW ) is continuous withrespect to all its arguments

I (SK2) (Maximum principle). The function S(p1, . . . , pW ) takes itsmaximum value over the uniform distribution pi = 1/W ,i = 1, . . . ,W .

I (SK3) (Expansibility). Adding an impossible event to a probabilitydistribution does not change its entropy:S(p1, . . . , pW , 0) = S(p1, . . . , pW ).

I (SK4) (Additivity). Given two subsystems A, B of a statisticalsystem,

S(A ∪ B) = S(A) + S(B | A).

In particular, if they are statistically independent,

S(A ∪ B) = S(A) + S(B).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 18: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

The modern Shannon-Khinchin formulation

I (SK1) (Continuity). The function S(p1, . . . , pW ) is continuous withrespect to all its arguments

I (SK2) (Maximum principle). The function S(p1, . . . , pW ) takes itsmaximum value over the uniform distribution pi = 1/W ,i = 1, . . . ,W .

I (SK3) (Expansibility). Adding an impossible event to a probabilitydistribution does not change its entropy:S(p1, . . . , pW , 0) = S(p1, . . . , pW ).

I (SK4) (Additivity). Given two subsystems A, B of a statisticalsystem,

S(A ∪ B) = S(A) + S(B | A).

In particular, if they are statistically independent,

S(A ∪ B) = S(A) + S(B).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 19: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

The modern Shannon-Khinchin formulation

I (SK1) (Continuity). The function S(p1, . . . , pW ) is continuous withrespect to all its arguments

I (SK2) (Maximum principle). The function S(p1, . . . , pW ) takes itsmaximum value over the uniform distribution pi = 1/W ,i = 1, . . . ,W .

I (SK3) (Expansibility). Adding an impossible event to a probabilitydistribution does not change its entropy:S(p1, . . . , pW , 0) = S(p1, . . . , pW ).

I (SK4) (Additivity). Given two subsystems A, B of a statisticalsystem,

S(A ∪ B) = S(A) + S(B | A).

In particular, if they are statistically independent,

S(A ∪ B) = S(A) + S(B).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 20: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

The modern Shannon-Khinchin formulation

I (SK1) (Continuity). The function S(p1, . . . , pW ) is continuous withrespect to all its arguments

I (SK2) (Maximum principle). The function S(p1, . . . , pW ) takes itsmaximum value over the uniform distribution pi = 1/W ,i = 1, . . . ,W .

I (SK3) (Expansibility). Adding an impossible event to a probabilitydistribution does not change its entropy:S(p1, . . . , pW , 0) = S(p1, . . . , pW ).

I (SK4) (Additivity). Given two subsystems A, B of a statisticalsystem,

S(A ∪ B) = S(A) + S(B | A).

In particular, if they are statistically independent,

S(A ∪ B) = S(A) + S(B).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 21: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

A possible unifying principle: The group-theoreticalapproach

We wish to find, possibly, a unifying principle, that would be responsibleof the many properties of a generalized entropy:

Thermodynamics: Legendre structure, Extensivity, SK axioms, etc.

Classical and Quantum Information Theory: Generalized divergences,etc.This principle should allow to classify all the known entropies and allowto design new ones.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 22: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

Beyond the SK formulation: the Composability Axiom

P. T., Beyond the SK formulation: the Composability axiom and theUniversal-group entropy, arXiv: 1407.3807 Annals of Physics (in press)

I The notion of entropy should be properly defined just in terms ofmacroscopic configurations of a system (without the need for amicroscopic description of the associated dynamics).

I Foundational property: an entropy should be a coarse-grainedquantity (Gell-Mann); if not so, the concept of entropy would besimply empty.

I Even the second law of thermodynamics loses any meaning if notreferred to the evolution of macroscopic subsystems: the entropywould stay invariant if defined on microscopic configurations.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 23: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

Beyond the SK formulation: the Composability Axiom

P. T., Beyond the SK formulation: the Composability axiom and theUniversal-group entropy, arXiv: 1407.3807 Annals of Physics (in press)

I The notion of entropy should be properly defined just in terms ofmacroscopic configurations of a system (without the need for amicroscopic description of the associated dynamics).

I Foundational property: an entropy should be a coarse-grainedquantity (Gell-Mann); if not so, the concept of entropy would besimply empty.

I Even the second law of thermodynamics loses any meaning if notreferred to the evolution of macroscopic subsystems: the entropywould stay invariant if defined on microscopic configurations.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 24: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

Beyond the SK formulation: the Composability Axiom

P. T., Beyond the SK formulation: the Composability axiom and theUniversal-group entropy, arXiv: 1407.3807 Annals of Physics (in press)

I The notion of entropy should be properly defined just in terms ofmacroscopic configurations of a system (without the need for amicroscopic description of the associated dynamics).

I Foundational property: an entropy should be a coarse-grainedquantity (Gell-Mann); if not so, the concept of entropy would besimply empty.

I Even the second law of thermodynamics loses any meaning if notreferred to the evolution of macroscopic subsystems: the entropywould stay invariant if defined on microscopic configurations.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 25: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

I Therefore, when composing two different systems, defined overany probability distribution piWi=1, we should have:

S(A ∪ B) = Φ(S(A),S(B); η) (1)

where η is a possible set of parameters.

I In additionS(A ∪ B) = S(B ∪ A).

I Another crucial property:

S(A ∪ (B ∪ C )) = S((A ∪ B) ∪ C )

I Finally, when we compose a system A with another system B, suchthat S(B) = 0, we should have that

S(A ∪ B) = S(A).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 26: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

I Therefore, when composing two different systems, defined overany probability distribution piWi=1, we should have:

S(A ∪ B) = Φ(S(A),S(B); η) (1)

where η is a possible set of parameters.

I In additionS(A ∪ B) = S(B ∪ A).

I Another crucial property:

S(A ∪ (B ∪ C )) = S((A ∪ B) ∪ C )

I Finally, when we compose a system A with another system B, suchthat S(B) = 0, we should have that

S(A ∪ B) = S(A).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 27: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

I Therefore, when composing two different systems, defined overany probability distribution piWi=1, we should have:

S(A ∪ B) = Φ(S(A),S(B); η) (1)

where η is a possible set of parameters.

I In additionS(A ∪ B) = S(B ∪ A).

I Another crucial property:

S(A ∪ (B ∪ C )) = S((A ∪ B) ∪ C )

I Finally, when we compose a system A with another system B, suchthat S(B) = 0, we should have that

S(A ∪ B) = S(A).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 28: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

I Therefore, when composing two different systems, defined overany probability distribution piWi=1, we should have:

S(A ∪ B) = Φ(S(A),S(B); η) (1)

where η is a possible set of parameters.

I In additionS(A ∪ B) = S(B ∪ A).

I Another crucial property:

S(A ∪ (B ∪ C )) = S((A ∪ B) ∪ C )

I Finally, when we compose a system A with another system B, suchthat S(B) = 0, we should have that

S(A ∪ B) = S(A).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 29: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

The Shannon-Khinchin axioms (SK1)-(SK3) are basic properties, nonnegotiable.

However, they are not sufficient for thermodynamical purposes.

We need a further requirement, that replaces the axiom (SK4) in fullgenerality: the Composability Axiom.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 30: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

The Composability Axiom

Definition 1.4. An entropy S is (strictly) composable if there exists asmooth function of two real variables Φ(x , y) such that (C1)

S(A ∪ B) = Φ(S(A),S(B); η) (2)

where A ⊂ X and B ⊂ X are two statistically independent systems,defined over any probability distribution piWi=1, with the furtherproperties(C2) Symmetry:

Φ(x , y) = Φ(y , x). (3)

(C3) Associativity:

Φ(x ,Φ(y , z)) = Φ(Φ(x , y), z). (4)

(C4) Null-composability:

Φ(x , 0) = x . (5)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 31: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

Composition of two arbitrary systems

Remark. Condition (2) can be formulated in more general terms.Consider the case of a composite system A ∪ B arising from two systemsnot statistically independent, with a conditional probability distributionpij(B | A) := pij(A ∪ B)/pi (A).Here pij(A ∪ B), i = 1, . . . ,WA, j = 1, . . . ,WB denotes the jointprobability distribution for the composite system A ∪ B, and pi (A) is

the marginal probability distribution pi (A) =∑WB

j=1 pij(A ∪ B).In this general context we postulate the relation

S(A ∪ B) = Φ(S(A),S(B | A); η), (6)

where S(B | A) denotes the conditional entropy associated with theconditional distribution pij(B | A). Equation (6) reduces to the relation(2) in the case of statistically independent subsystems. The relation (6)generalizes the original axiom (SK4) for the case of systems notindependent.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 32: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Different foundational approachesThe uniqueness theorems for the Boltzmann-Gibbs entropyThe SK approachBeyond the SK formulation: the Composability Axiom

How to realize the composability axiom?

According to the Composability Axiom, we are interested to the generalclass of functions

Φ : R2 −→ R

such that the properties (C 1)-(C 4) apply. Surprisingly enough, there is afull mathematical theory that deals with such functions: the FormalGroup Theory. It was developed in the context of algebraic topologystarting from the seminal work of Bochner in 1946, and is still an activeresearch area.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 33: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Formal Group Theory: a crash course

The theory of formal groups offers a natural language for formulatingthe theory of generalized entropies.S. Bochner, Formal Lie groups, Annals of Mathematics 47, 192–201(1946).M. Hazewinkel, Formal Groups and Applications, Academic Press, NewYork, 1978.Many important mathematicians have been working in this area: S. P.Novikov, D. Quillen, J. P. Serre, G. Faltings, etc.Nowadays, it has a prominent role in fields as

I Algebraic topology,

I The theory of elliptic curves

I Arithmetic number theory.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 34: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Formal Group Theory: a crash course

The theory of formal groups offers a natural language for formulatingthe theory of generalized entropies.S. Bochner, Formal Lie groups, Annals of Mathematics 47, 192–201(1946).M. Hazewinkel, Formal Groups and Applications, Academic Press, NewYork, 1978.Many important mathematicians have been working in this area: S. P.Novikov, D. Quillen, J. P. Serre, G. Faltings, etc.Nowadays, it has a prominent role in fields as

I Algebraic topology,

I The theory of elliptic curves

I Arithmetic number theory.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 35: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Formal Group Theory: a crash course

The theory of formal groups offers a natural language for formulatingthe theory of generalized entropies.S. Bochner, Formal Lie groups, Annals of Mathematics 47, 192–201(1946).M. Hazewinkel, Formal Groups and Applications, Academic Press, NewYork, 1978.Many important mathematicians have been working in this area: S. P.Novikov, D. Quillen, J. P. Serre, G. Faltings, etc.Nowadays, it has a prominent role in fields as

I Algebraic topology,

I The theory of elliptic curves

I Arithmetic number theory.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 36: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Definitions

I Definition 1.5. Let R be a commutative ring with identity, andR x1, x2, .. be the ring of formal power series in the variables x1,x2, ... with coefficients in R.

I A commutative one-dimensional formal group law over R is aformal power series in two variables Φ (x , y) ∈ R x , y of the formΦ (x , y) = x + y + terms of higher degree, such that

i) Φ (x , 0) = Φ (0, x) = x

ii) Φ (Φ (x , y) , z) = Φ (x ,Φ (y , z)).

I When Φ (x , y) = Φ (y , x), the formal group law is said to becommutative.

I Proposition 1.6. Form any formal group law Φ(x , t), there existsan inverse formal series ϕ (x) ∈ R x such that Φ (x , ϕ (x)) = 0.

I This is the reason why we talk about formal group laws.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 37: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Definitions

I Definition 1.5. Let R be a commutative ring with identity, andR x1, x2, .. be the ring of formal power series in the variables x1,x2, ... with coefficients in R.

I A commutative one-dimensional formal group law over R is aformal power series in two variables Φ (x , y) ∈ R x , y of the formΦ (x , y) = x + y + terms of higher degree, such that

i) Φ (x , 0) = Φ (0, x) = x

ii) Φ (Φ (x , y) , z) = Φ (x ,Φ (y , z)).

I When Φ (x , y) = Φ (y , x), the formal group law is said to becommutative.

I Proposition 1.6. Form any formal group law Φ(x , t), there existsan inverse formal series ϕ (x) ∈ R x such that Φ (x , ϕ (x)) = 0.

I This is the reason why we talk about formal group laws.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 38: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Definitions

I Definition 1.5. Let R be a commutative ring with identity, andR x1, x2, .. be the ring of formal power series in the variables x1,x2, ... with coefficients in R.

I A commutative one-dimensional formal group law over R is aformal power series in two variables Φ (x , y) ∈ R x , y of the formΦ (x , y) = x + y + terms of higher degree, such that

i) Φ (x , 0) = Φ (0, x) = x

ii) Φ (Φ (x , y) , z) = Φ (x ,Φ (y , z)).

I When Φ (x , y) = Φ (y , x), the formal group law is said to becommutative.

I Proposition 1.6. Form any formal group law Φ(x , t), there existsan inverse formal series ϕ (x) ∈ R x such that Φ (x , ϕ (x)) = 0.

I This is the reason why we talk about formal group laws.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 39: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Definitions

I Definition 1.5. Let R be a commutative ring with identity, andR x1, x2, .. be the ring of formal power series in the variables x1,x2, ... with coefficients in R.

I A commutative one-dimensional formal group law over R is aformal power series in two variables Φ (x , y) ∈ R x , y of the formΦ (x , y) = x + y + terms of higher degree, such that

i) Φ (x , 0) = Φ (0, x) = x

ii) Φ (Φ (x , y) , z) = Φ (x ,Φ (y , z)).

I When Φ (x , y) = Φ (y , x), the formal group law is said to becommutative.

I Proposition 1.6. Form any formal group law Φ(x , t), there existsan inverse formal series ϕ (x) ∈ R x such that Φ (x , ϕ (x)) = 0.

I This is the reason why we talk about formal group laws.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 40: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Definitions

I Definition 1.5. Let R be a commutative ring with identity, andR x1, x2, .. be the ring of formal power series in the variables x1,x2, ... with coefficients in R.

I A commutative one-dimensional formal group law over R is aformal power series in two variables Φ (x , y) ∈ R x , y of the formΦ (x , y) = x + y + terms of higher degree, such that

i) Φ (x , 0) = Φ (0, x) = x

ii) Φ (Φ (x , y) , z) = Φ (x ,Φ (y , z)).

I When Φ (x , y) = Φ (y , x), the formal group law is said to becommutative.

I Proposition 1.6. Form any formal group law Φ(x , t), there existsan inverse formal series ϕ (x) ∈ R x such that Φ (x , ϕ (x)) = 0.

I This is the reason why we talk about formal group laws.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 41: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

The previous definition can be naturally extended to the case ofn-dimensional formal group laws.

Definition 1.7.Let R be a commutative ring with identity, and R x1, x2, .. be the ringof formal power series in the variables x1, x2, ... with coefficients in R.An n-dimensional formal group law over R is an n-tuple

Φ = (Φ1, . . . ,Φn)

of power series Φi (x,y), i = 1, . . . , n, where x= (x1, . . . , xn), withcoefficients in R, such that

Φ(x, 0) = x

Φ(x,Φ(y, z)) = Φ(Φ(x, y), z)

If Φ(x, y) = Φ(y, x), the n- dimensional formal group law is said to becommutative.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 42: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Formal groups and formal group laws

I The relevance of formal group laws relies first of all on their closeconnection with classical group theory.

I A formal group law Ψ(x , y) defines a functor F : AlgR −→ Group,where AlgR denotes the category of commutative unitary algebrasover R and Group denotes the category of groups. The functor F isby definition the formal group (sometimes called the formal groupscheme) associated to the formal group law Ψ.

I We can construct a formal group law of dimension n from anyalgebraic group or Lie group of the same dimension n, by takingcoordinates at the identity and writing down the formal power seriesexpansion of the product map. An important special case of this isthe formal group law of an elliptic curve (or abelian variety).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 43: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Formal groups and formal group laws

I The relevance of formal group laws relies first of all on their closeconnection with classical group theory.

I A formal group law Ψ(x , y) defines a functor F : AlgR −→ Group,where AlgR denotes the category of commutative unitary algebrasover R and Group denotes the category of groups. The functor F isby definition the formal group (sometimes called the formal groupscheme) associated to the formal group law Ψ.

I We can construct a formal group law of dimension n from anyalgebraic group or Lie group of the same dimension n, by takingcoordinates at the identity and writing down the formal power seriesexpansion of the product map. An important special case of this isthe formal group law of an elliptic curve (or abelian variety).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 44: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Formal groups and formal group laws

I The relevance of formal group laws relies first of all on their closeconnection with classical group theory.

I A formal group law Ψ(x , y) defines a functor F : AlgR −→ Group,where AlgR denotes the category of commutative unitary algebrasover R and Group denotes the category of groups. The functor F isby definition the formal group (sometimes called the formal groupscheme) associated to the formal group law Ψ.

I We can construct a formal group law of dimension n from anyalgebraic group or Lie group of the same dimension n, by takingcoordinates at the identity and writing down the formal power seriesexpansion of the product map. An important special case of this isthe formal group law of an elliptic curve (or abelian variety).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 45: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

I Viceversa, given a formal group we can construct a Lie algebra.

I Construction. Let us write

Φ(x, y) = x + y + Φ2(x, y) + Φ3(x, y) + . . .

where Φ2(x , y) denotes the quadratic part of the formal group law,Φ3(x , y) the cubic part, and so on.Any n-dimensional formal group law induces an n-dimensional Liealgebra over the same ring R, by means of the formula

[x, y] := Φ2(x, y)−Φ2(y, x).

I Lie Groups −→ Formal Group Laws ←→ Lie Algebras

I This correspondence is true over fields F of characteristic zero, andis lost if char(F) 6= 0. (See J.-P. Serre, Lie Algebras and Lie Groups,LNM 1500, Springer, 1992).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 46: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

I Viceversa, given a formal group we can construct a Lie algebra.

I Construction. Let us write

Φ(x, y) = x + y + Φ2(x, y) + Φ3(x, y) + . . .

where Φ2(x , y) denotes the quadratic part of the formal group law,Φ3(x , y) the cubic part, and so on.Any n-dimensional formal group law induces an n-dimensional Liealgebra over the same ring R, by means of the formula

[x, y] := Φ2(x, y)−Φ2(y, x).

I Lie Groups −→ Formal Group Laws ←→ Lie Algebras

I This correspondence is true over fields F of characteristic zero, andis lost if char(F) 6= 0. (See J.-P. Serre, Lie Algebras and Lie Groups,LNM 1500, Springer, 1992).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 47: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

I Viceversa, given a formal group we can construct a Lie algebra.

I Construction. Let us write

Φ(x, y) = x + y + Φ2(x, y) + Φ3(x, y) + . . .

where Φ2(x , y) denotes the quadratic part of the formal group law,Φ3(x , y) the cubic part, and so on.Any n-dimensional formal group law induces an n-dimensional Liealgebra over the same ring R, by means of the formula

[x, y] := Φ2(x, y)−Φ2(y, x).

I Lie Groups −→ Formal Group Laws ←→ Lie Algebras

I This correspondence is true over fields F of characteristic zero, andis lost if char(F) 6= 0. (See J.-P. Serre, Lie Algebras and Lie Groups,LNM 1500, Springer, 1992).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 48: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

I Viceversa, given a formal group we can construct a Lie algebra.

I Construction. Let us write

Φ(x, y) = x + y + Φ2(x, y) + Φ3(x, y) + . . .

where Φ2(x , y) denotes the quadratic part of the formal group law,Φ3(x , y) the cubic part, and so on.Any n-dimensional formal group law induces an n-dimensional Liealgebra over the same ring R, by means of the formula

[x, y] := Φ2(x, y)−Φ2(y, x).

I Lie Groups −→ Formal Group Laws ←→ Lie Algebras

I This correspondence is true over fields F of characteristic zero, andis lost if char(F) 6= 0. (See J.-P. Serre, Lie Algebras and Lie Groups,LNM 1500, Springer, 1992).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 49: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Examples

I The additive group law

Φ(x , y) = x + y

I The multiplicative group law

Φ(x , y) = x + y + axy

I The hyperbolic group law (addition of velocities in special relativity)

Φ(x , y) =x + y

1 + xy

I The Euler group law for elliptic integrals.

Φ(x , y) = (x√

1− y 4 + y√

1− x4)/(1 + x2y 2)

I ∫ x

0

dt√1− t4

+

∫ y

0

dt√1− t4

=

∫ Φ(x,y)

0

dt√1− t4

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 50: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Examples

I The additive group law

Φ(x , y) = x + y

I The multiplicative group law

Φ(x , y) = x + y + axy

I The hyperbolic group law (addition of velocities in special relativity)

Φ(x , y) =x + y

1 + xy

I The Euler group law for elliptic integrals.

Φ(x , y) = (x√

1− y 4 + y√

1− x4)/(1 + x2y 2)

I ∫ x

0

dt√1− t4

+

∫ y

0

dt√1− t4

=

∫ Φ(x,y)

0

dt√1− t4

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 51: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Examples

I The additive group law

Φ(x , y) = x + y

I The multiplicative group law

Φ(x , y) = x + y + axy

I The hyperbolic group law (addition of velocities in special relativity)

Φ(x , y) =x + y

1 + xy

I The Euler group law for elliptic integrals.

Φ(x , y) = (x√

1− y 4 + y√

1− x4)/(1 + x2y 2)

I ∫ x

0

dt√1− t4

+

∫ y

0

dt√1− t4

=

∫ Φ(x,y)

0

dt√1− t4

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 52: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Examples

I The additive group law

Φ(x , y) = x + y

I The multiplicative group law

Φ(x , y) = x + y + axy

I The hyperbolic group law (addition of velocities in special relativity)

Φ(x , y) =x + y

1 + xy

I The Euler group law for elliptic integrals.

Φ(x , y) = (x√

1− y 4 + y√

1− x4)/(1 + x2y 2)

I ∫ x

0

dt√1− t4

+

∫ y

0

dt√1− t4

=

∫ Φ(x,y)

0

dt√1− t4

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 53: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Examples

I The additive group law

Φ(x , y) = x + y

I The multiplicative group law

Φ(x , y) = x + y + axy

I The hyperbolic group law (addition of velocities in special relativity)

Φ(x , y) =x + y

1 + xy

I The Euler group law for elliptic integrals.

Φ(x , y) = (x√

1− y 4 + y√

1− x4)/(1 + x2y 2)

I ∫ x

0

dt√1− t4

+

∫ y

0

dt√1− t4

=

∫ Φ(x,y)

0

dt√1− t4

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 54: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Motivation: Thermodynamics

I The additive group law ⇐⇒ The composition law for the SBentropy:

Φ(x , y) = x + y

SBG (A ∪ B) = SBG (A) + SBG (B)

I The multiplicative group law ⇐⇒ The composition law for theTsallis entropy:

Φ(x , y) = x + y + axy

Sq(A ∪ B) = Sq(A) + Sq(B) + (1− q)Sq(A)Sq(B)

I These two results clearly support the existence of a directconnection among group laws and composition laws of generalizedentropies.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 55: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Motivation: Thermodynamics

I The additive group law ⇐⇒ The composition law for the SBentropy:

Φ(x , y) = x + y

SBG (A ∪ B) = SBG (A) + SBG (B)

I The multiplicative group law ⇐⇒ The composition law for theTsallis entropy:

Φ(x , y) = x + y + axy

Sq(A ∪ B) = Sq(A) + Sq(B) + (1− q)Sq(A)Sq(B)

I These two results clearly support the existence of a directconnection among group laws and composition laws of generalizedentropies.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 56: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Motivation: Thermodynamics

I The additive group law ⇐⇒ The composition law for the SBentropy:

Φ(x , y) = x + y

SBG (A ∪ B) = SBG (A) + SBG (B)

I The multiplicative group law ⇐⇒ The composition law for theTsallis entropy:

Φ(x , y) = x + y + axy

Sq(A ∪ B) = Sq(A) + Sq(B) + (1− q)Sq(A)Sq(B)

I These two results clearly support the existence of a directconnection among group laws and composition laws of generalizedentropies.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 57: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

A generalized rational group law (V. M. Bukhshtaber, A. N.Khodolov, Math. USSR Sbornik, 69, 77–97 (1991)).

I

Φ(x , y) =x + y + axy

1 + bxy

I A particular case of this group law (a = α− 1), b = α alreadyappeared in the context of the Hirzebruch theory of genera inalgebraic topology:

Φ(x , y) =x + y + (α− 1)xy

1 + αxy

(V. M. Bukhshtaber, A. S. Mishchenko, S. P. Novikov, Russ.Math. Surv. 26, 63–90 (1971)).

I When a = b = 0, we obtain the additive group law.I When b = 0, we obtain the multiplicative group law.I Also, when a = 0, b = 1, we obtain the hyperbolic group law.

(See also E. Curado, P. Tempesta, C. Tsallis, preprint arxiv:1507.05058, (2015)).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 58: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

A generalized rational group law (V. M. Bukhshtaber, A. N.Khodolov, Math. USSR Sbornik, 69, 77–97 (1991)).

I

Φ(x , y) =x + y + axy

1 + bxy

I A particular case of this group law (a = α− 1), b = α alreadyappeared in the context of the Hirzebruch theory of genera inalgebraic topology:

Φ(x , y) =x + y + (α− 1)xy

1 + αxy

(V. M. Bukhshtaber, A. S. Mishchenko, S. P. Novikov, Russ.Math. Surv. 26, 63–90 (1971)).

I When a = b = 0, we obtain the additive group law.I When b = 0, we obtain the multiplicative group law.I Also, when a = 0, b = 1, we obtain the hyperbolic group law.

(See also E. Curado, P. Tempesta, C. Tsallis, preprint arxiv:1507.05058, (2015)).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 59: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

A generalized rational group law (V. M. Bukhshtaber, A. N.Khodolov, Math. USSR Sbornik, 69, 77–97 (1991)).

I

Φ(x , y) =x + y + axy

1 + bxy

I A particular case of this group law (a = α− 1), b = α alreadyappeared in the context of the Hirzebruch theory of genera inalgebraic topology:

Φ(x , y) =x + y + (α− 1)xy

1 + αxy

(V. M. Bukhshtaber, A. S. Mishchenko, S. P. Novikov, Russ.Math. Surv. 26, 63–90 (1971)).

I When a = b = 0, we obtain the additive group law.

I When b = 0, we obtain the multiplicative group law.I Also, when a = 0, b = 1, we obtain the hyperbolic group law.

(See also E. Curado, P. Tempesta, C. Tsallis, preprint arxiv:1507.05058, (2015)).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 60: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

A generalized rational group law (V. M. Bukhshtaber, A. N.Khodolov, Math. USSR Sbornik, 69, 77–97 (1991)).

I

Φ(x , y) =x + y + axy

1 + bxy

I A particular case of this group law (a = α− 1), b = α alreadyappeared in the context of the Hirzebruch theory of genera inalgebraic topology:

Φ(x , y) =x + y + (α− 1)xy

1 + αxy

(V. M. Bukhshtaber, A. S. Mishchenko, S. P. Novikov, Russ.Math. Surv. 26, 63–90 (1971)).

I When a = b = 0, we obtain the additive group law.I When b = 0, we obtain the multiplicative group law.

I Also, when a = 0, b = 1, we obtain the hyperbolic group law.(See also E. Curado, P. Tempesta, C. Tsallis, preprint arxiv:1507.05058, (2015)).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 61: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

A generalized rational group law (V. M. Bukhshtaber, A. N.Khodolov, Math. USSR Sbornik, 69, 77–97 (1991)).

I

Φ(x , y) =x + y + axy

1 + bxy

I A particular case of this group law (a = α− 1), b = α alreadyappeared in the context of the Hirzebruch theory of genera inalgebraic topology:

Φ(x , y) =x + y + (α− 1)xy

1 + αxy

(V. M. Bukhshtaber, A. S. Mishchenko, S. P. Novikov, Russ.Math. Surv. 26, 63–90 (1971)).

I When a = b = 0, we obtain the additive group law.I When b = 0, we obtain the multiplicative group law.I Also, when a = 0, b = 1, we obtain the hyperbolic group law.

(See also E. Curado, P. Tempesta, C. Tsallis, preprint arxiv:1507.05058, (2015)).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 62: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

The Lazard Universal Formal Group

Let us consider the series (the so called formal group exponential)

G (t) =∞∑k=0

aktk+1

k + 1(7)

over the ring B = Z[a1, a2, ...] of integral polynomials in infinitely manyvariables a1, a2, . . ., with a0 = 1. Let G−1 (s) be its compositional inverse:

G−1 (s) =∞∑i=0

bis i+1

i + 1, (8)

so that G−1 (G (t)) = t. We have b0 = 1, b1 = −a1, b2 = 32 a2

1 − a2, . . .(Lagrange theorem). The Lazard universal formal group law is definedto be the formal power series

Φ (x, y) = G(G−1 (x) + G−1 (y)

).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 63: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Properties

I The following two important results, due to Lazard, hold.

I Proposition 1.8. For any commutative one-dimensional formalgroup law over any ring R, there exists a unique homomorphismL→ R under which the Lazard group law is mapped into the givengroup law (the so called universal property of the Lazard group).

I Proposition 1.9. For any commutative one-dimensional formalgroup law Ψ(x , y) over any ring R, there exists a seriesψ(x) ∈ R[[x ]]⊗Q such that

ψ(x) = x + O(x2), and Ψ(x , y) = ψ−1 (ψ(x) + ψ(y)) ∈ R[[x , y ]]⊗Q.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 64: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Properties

I The following two important results, due to Lazard, hold.

I Proposition 1.8. For any commutative one-dimensional formalgroup law over any ring R, there exists a unique homomorphismL→ R under which the Lazard group law is mapped into the givengroup law (the so called universal property of the Lazard group).

I Proposition 1.9. For any commutative one-dimensional formalgroup law Ψ(x , y) over any ring R, there exists a seriesψ(x) ∈ R[[x ]]⊗Q such that

ψ(x) = x + O(x2), and Ψ(x , y) = ψ−1 (ψ(x) + ψ(y)) ∈ R[[x , y ]]⊗Q.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 65: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

Properties

I The following two important results, due to Lazard, hold.

I Proposition 1.8. For any commutative one-dimensional formalgroup law over any ring R, there exists a unique homomorphismL→ R under which the Lazard group law is mapped into the givengroup law (the so called universal property of the Lazard group).

I Proposition 1.9. For any commutative one-dimensional formalgroup law Ψ(x , y) over any ring R, there exists a seriesψ(x) ∈ R[[x ]]⊗Q such that

ψ(x) = x + O(x2), and Ψ(x , y) = ψ−1 (ψ(x) + ψ(y)) ∈ R[[x , y ]]⊗Q.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 66: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

The cohomological interpretation of formal group theory(U can sleep here)

(Novikov’s school, Quillen, Ray, etc.)There is a clear relationship between cobordism theory and formalgroup laws.Definition 1.10. Suppose we are assigned with some class of manifolds,closed and with boundary, and possibly with additional structure. Werequire thata) The boundary of a manifold in the class belongs to the same class;b) The direct product of manifolds of the class belong to the class(“multiplicative property”);c) a closed domain with smooth boundary in a manifold of the classbelongs to the class (“excision axiom”)We say that such a class defines a cobordism (and a bordism) theory,denoted by C.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 67: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

U can sleep here. Part II

Let M be a closed manifold in C. The cycles (bordisms in C) for anycomplex K are the pairs (M, f ) where f : M → K is a continuousmapping.The “films” are the pairs (N, g) where N ∈ C has boundary andg : N → K .Definition. The bordism group of K relative to the class C, denoted byΩCn (K ) is

ΩCn (K ) =group of n-dimensional cycles

group of boundaries of films in CFor finite complexes K , the Cobordism groups Ωn

C associated can beobtained by means of the Alexander-Pontryagin duality law. Let N ∈ Nlarge. If K ⊂ SN , where SN is a sphere, by definition we put

ΩnC(K ) = ΩCN−n

(SN ,SN \ K

)and this definition is independent of N and the embedding K ⊂ SN .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 68: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

U can sleep here. Part III

There are various correlated aspects of the theory:

I Unitary cobordisms

I Geometric bordisms

I Chern characters and Hirzebruch genera

I All the fundamental concepts and facts of the theory of unitarycobordisms, both modern and classical, can be expressed by means ofLazard’s formal group (S. P. Novikov).

I Main reason : if [CPn] are cobordism classes of complex projective spaces,the formal group exponential

G(u) =∑k≥0

[CPn]

n + 1un+1

generates the formal groups associated with the previous theories (byintroducing a suitable multiplication in each theory). At the same time,

G(G−1(u) + G−1(v))

generates the Lazard formal group.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 69: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

U can sleep here. Part III

There are various correlated aspects of the theory:

I Unitary cobordisms

I Geometric bordisms

I Chern characters and Hirzebruch genera

I All the fundamental concepts and facts of the theory of unitarycobordisms, both modern and classical, can be expressed by means ofLazard’s formal group (S. P. Novikov).

I Main reason : if [CPn] are cobordism classes of complex projective spaces,the formal group exponential

G(u) =∑k≥0

[CPn]

n + 1un+1

generates the formal groups associated with the previous theories (byintroducing a suitable multiplication in each theory). At the same time,

G(G−1(u) + G−1(v))

generates the Lazard formal group.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 70: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

U can sleep here. Part III

There are various correlated aspects of the theory:

I Unitary cobordisms

I Geometric bordisms

I Chern characters and Hirzebruch genera

I All the fundamental concepts and facts of the theory of unitarycobordisms, both modern and classical, can be expressed by means ofLazard’s formal group (S. P. Novikov).

I Main reason : if [CPn] are cobordism classes of complex projective spaces,the formal group exponential

G(u) =∑k≥0

[CPn]

n + 1un+1

generates the formal groups associated with the previous theories (byintroducing a suitable multiplication in each theory). At the same time,

G(G−1(u) + G−1(v))

generates the Lazard formal group.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 71: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

U can sleep here. Part III

There are various correlated aspects of the theory:

I Unitary cobordisms

I Geometric bordisms

I Chern characters and Hirzebruch genera

I All the fundamental concepts and facts of the theory of unitarycobordisms, both modern and classical, can be expressed by means ofLazard’s formal group (S. P. Novikov).

I Main reason : if [CPn] are cobordism classes of complex projective spaces,the formal group exponential

G(u) =∑k≥0

[CPn]

n + 1un+1

generates the formal groups associated with the previous theories (byintroducing a suitable multiplication in each theory). At the same time,

G(G−1(u) + G−1(v))

generates the Lazard formal group.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 72: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

DefinitionsExamplesA rational group lawThe Lazard Universal Formal GroupFormal Groups and Cobordism Theory

U can sleep here. Part III

There are various correlated aspects of the theory:

I Unitary cobordisms

I Geometric bordisms

I Chern characters and Hirzebruch genera

I All the fundamental concepts and facts of the theory of unitarycobordisms, both modern and classical, can be expressed by means ofLazard’s formal group (S. P. Novikov).

I Main reason : if [CPn] are cobordism classes of complex projective spaces,the formal group exponential

G(u) =∑k≥0

[CPn]

n + 1un+1

generates the formal groups associated with the previous theories (byintroducing a suitable multiplication in each theory). At the same time,

G(G−1(u) + G−1(v))

generates the Lazard formal group.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 73: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

Resume

The first three Shannon-Khinchin axioms, jointly with thecomposability axiom, are a necessary set of axioms to doThermodynamics.In particular, the composability axiom replaces and generalizes thefourth axiom, and provides a group-theoretical structure to the notion ofentropy.

I Definition 1.11. We shall say that an entropy isthermodynamically admissible if satisfies the first three SK axiomsand is composable.

I Every “decent” entropy has associated a group structure.

I Vice-versa, given a group structure, is there a suitable entropyassociated with it? (Lectures II and III).

I In particular, is there any mathematical obstruction to thiscorrespondence?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 74: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

Resume

The first three Shannon-Khinchin axioms, jointly with thecomposability axiom, are a necessary set of axioms to doThermodynamics.In particular, the composability axiom replaces and generalizes thefourth axiom, and provides a group-theoretical structure to the notion ofentropy.

I Definition 1.11. We shall say that an entropy isthermodynamically admissible if satisfies the first three SK axiomsand is composable.

I Every “decent” entropy has associated a group structure.

I Vice-versa, given a group structure, is there a suitable entropyassociated with it? (Lectures II and III).

I In particular, is there any mathematical obstruction to thiscorrespondence?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 75: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

Resume

The first three Shannon-Khinchin axioms, jointly with thecomposability axiom, are a necessary set of axioms to doThermodynamics.In particular, the composability axiom replaces and generalizes thefourth axiom, and provides a group-theoretical structure to the notion ofentropy.

I Definition 1.11. We shall say that an entropy isthermodynamically admissible if satisfies the first three SK axiomsand is composable.

I Every “decent” entropy has associated a group structure.

I Vice-versa, given a group structure, is there a suitable entropyassociated with it? (Lectures II and III).

I In particular, is there any mathematical obstruction to thiscorrespondence?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 76: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

Resume

The first three Shannon-Khinchin axioms, jointly with thecomposability axiom, are a necessary set of axioms to doThermodynamics.In particular, the composability axiom replaces and generalizes thefourth axiom, and provides a group-theoretical structure to the notion ofentropy.

I Definition 1.11. We shall say that an entropy isthermodynamically admissible if satisfies the first three SK axiomsand is composable.

I Every “decent” entropy has associated a group structure.

I Vice-versa, given a group structure, is there a suitable entropyassociated with it? (Lectures II and III).

I In particular, is there any mathematical obstruction to thiscorrespondence?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 77: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

Another uniqueness theorem

P. T., Formal groups and Z-entropies, arxiv: 1507.07436 (2015).Let us denote by N the number of particles of a complex system.Definition 1.12. An entropy S is said to be extensive if there exists a function(“occupation law”) W = W (N) such that, over the uniform distributionpi = 1/W , i = 1, . . . ,W , we have, for large values of N:

S [W (N)] ∝ N

Observation. No entropy is extensive in all possible regimes. For instance, theBoltzmann entropy is extensive when the occupation law of the phase space isW (N) = kN ; the Tsallis entropy is extensive for W (N) = Nρ, and so on.Definition 1.13. An entropy belongs to the trace-form class if it can bewritten in the general form S [p] =

∑Wi=1 F (pi ), where F : [0, 1] −→ R+ ∪ 0 is

assumed to be a function at least twice differentiable in (0, 1), strictly concave,with the further properties F (0) = F (1) = 0. We shall also assume that S beextensive. The previous definition ensures that S [p] satisfies the axioms(SK1)-(SK3).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 78: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

I Theorem 1.11 The Boltzmann entropy and the Tsallis entropy arethe only trace-form and C∞(0, 1) entropies which are strictlycomposable.

I Consequence. The trace-form class has a serious “drawback”.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 79: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

I Theorem 1.11 The Boltzmann entropy and the Tsallis entropy arethe only trace-form and C∞(0, 1) entropies which are strictlycomposable.

I Consequence. The trace-form class has a serious “drawback”.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 80: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

Strong and weak composability

I The composability axiom is very demanding: for trace-formentropies

∑Wi=1 f (pi ), only two entropies are strictly composable:

The entropy SBG =∑

i pi ln 1p1

, the entropy Sq =∑

i pqi −1

1−q .

I We can consider a weaker formulation of this notion.

I Definition 2. An entropy is weakly composable if the properties(C1)–(C3) of the composability axiom are satisfied at least when theprobability distributions of the two independent subsystems A and Bare both uniform, and property (C4) holds in general.

I The uniform distribution emerges when considering isolated physicalsystems at the equilibrium (microcanonical ensemble), or incontact with a thermostat at very high temperature (canonicalensemble) : for instance, for high-temperature astrophysical objects.A large set of entropies recently considered are weakly composable,although even not all of them.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 81: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

Strong and weak composability

I The composability axiom is very demanding: for trace-formentropies

∑Wi=1 f (pi ), only two entropies are strictly composable:

The entropy SBG =∑

i pi ln 1p1

, the entropy Sq =∑

i pqi −1

1−q .I We can consider a weaker formulation of this notion.

I Definition 2. An entropy is weakly composable if the properties(C1)–(C3) of the composability axiom are satisfied at least when theprobability distributions of the two independent subsystems A and Bare both uniform, and property (C4) holds in general.

I The uniform distribution emerges when considering isolated physicalsystems at the equilibrium (microcanonical ensemble), or incontact with a thermostat at very high temperature (canonicalensemble) : for instance, for high-temperature astrophysical objects.A large set of entropies recently considered are weakly composable,although even not all of them.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 82: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

Strong and weak composability

I The composability axiom is very demanding: for trace-formentropies

∑Wi=1 f (pi ), only two entropies are strictly composable:

The entropy SBG =∑

i pi ln 1p1

, the entropy Sq =∑

i pqi −1

1−q .I We can consider a weaker formulation of this notion.

I Definition 2. An entropy is weakly composable if the properties(C1)–(C3) of the composability axiom are satisfied at least when theprobability distributions of the two independent subsystems A and Bare both uniform, and property (C4) holds in general.

I The uniform distribution emerges when considering isolated physicalsystems at the equilibrium (microcanonical ensemble), or incontact with a thermostat at very high temperature (canonicalensemble) : for instance, for high-temperature astrophysical objects.A large set of entropies recently considered are weakly composable,although even not all of them.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 83: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

Strong and weak composability

I The composability axiom is very demanding: for trace-formentropies

∑Wi=1 f (pi ), only two entropies are strictly composable:

The entropy SBG =∑

i pi ln 1p1

, the entropy Sq =∑

i pqi −1

1−q .I We can consider a weaker formulation of this notion.

I Definition 2. An entropy is weakly composable if the properties(C1)–(C3) of the composability axiom are satisfied at least when theprobability distributions of the two independent subsystems A and Bare both uniform, and property (C4) holds in general.

I The uniform distribution emerges when considering isolated physicalsystems at the equilibrium (microcanonical ensemble), or incontact with a thermostat at very high temperature (canonicalensemble) : for instance, for high-temperature astrophysical objects.A large set of entropies recently considered are weakly composable,although even not all of them.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 84: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

Examples of weakly composable entropies

Typical examples of this (huge) class are the most well-known entropies.

I The Kaniadakis entropy

Sκ[p] = kB

W∑i=1

pip−κi − pκi

2κ(9)

I The Anteneodo-Plastino entropy

Sη =W∑i=1

(η + 1

η,− ln pi

)− piΓ

(η + 1

η

)).

I The HT Sc,d entropy

Sc,d =e

1− c + cd

W∑i=1

Γ(1 + d , 1− c ln pi )−c

1− c + cd. (10)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 85: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

Examples of weakly composable entropies

Typical examples of this (huge) class are the most well-known entropies.

I The Kaniadakis entropy

Sκ[p] = kB

W∑i=1

pip−κi − pκi

2κ(9)

I The Anteneodo-Plastino entropy

Sη =W∑i=1

(η + 1

η,− ln pi

)− piΓ

(η + 1

η

)).

I The HT Sc,d entropy

Sc,d =e

1− c + cd

W∑i=1

Γ(1 + d , 1− c ln pi )−c

1− c + cd. (10)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 86: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

Examples of weakly composable entropies

Typical examples of this (huge) class are the most well-known entropies.

I The Kaniadakis entropy

Sκ[p] = kB

W∑i=1

pip−κi − pκi

2κ(9)

I The Anteneodo-Plastino entropy

Sη =W∑i=1

(η + 1

η,− ln pi

)− piΓ

(η + 1

η

)).

I The HT Sc,d entropy

Sc,d =e

1− c + cd

W∑i=1

Γ(1 + d , 1− c ln pi )−c

1− c + cd. (10)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 87: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

I The Abe entropy

SAbe =W∑i=1

pq−1

i − pqi

q − q−1

I The Borges-Roditi entropy

SBR =W∑i=1

pq′

i − pqi

q − q′

I The Tsallis-Cirto entropy

Sδ = kB

W∑i=1

pi

(ln

1

pi

)δ, 0 < δ ≤ (1 + ln W ). (11)

I The Shafee entropy

Sδ = kB

W∑i=1

pci

(ln

1

pi

). (12)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 88: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

II The Abe entropy

SAbe =W∑i=1

pq−1

i − pqi

q − q−1

I The Borges-Roditi entropy

SBR =W∑i=1

pq′

i − pqi

q − q′

I The Tsallis-Cirto entropy

Sδ = kB

W∑i=1

pi

(ln

1

pi

)δ, 0 < δ ≤ (1 + ln W ). (11)

I The Shafee entropy

Sδ = kB

W∑i=1

pci

(ln

1

pi

). (12)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 89: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

II The Abe entropy

SAbe =W∑i=1

pq−1

i − pqi

q − q−1

I The Borges-Roditi entropy

SBR =W∑i=1

pq′

i − pqi

q − q′

I The Tsallis-Cirto entropy

Sδ = kB

W∑i=1

pi

(ln

1

pi

)δ, 0 < δ ≤ (1 + ln W ). (11)

I The Shafee entropy

Sδ = kB

W∑i=1

pci

(ln

1

pi

). (12)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 90: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

General properties of the trace-form class

II The Abe entropy

SAbe =W∑i=1

pq−1

i − pqi

q − q−1

I The Borges-Roditi entropy

SBR =W∑i=1

pq′

i − pqi

q − q′

I The Tsallis-Cirto entropy

Sδ = kB

W∑i=1

pi

(ln

1

pi

)δ, 0 < δ ≤ (1 + ln W ). (11)

I The Shafee entropy

Sδ = kB

W∑i=1

pci

(ln

1

pi

). (12)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 91: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

Entropies and the discrete world

II There is a quite interesting and unexpected relationship between thetheory of difference equations and a large class of generalizedentropies.

Main idea: the nonstandard logarithm appearing at the heart of thenotion of generalized entropy can be seen (in most cases) as arealization of a specific difference operator.

I The modern algebraic theory of difference operators is intimatelyrelated with the theory of polynomial sequences of Sheffer andAppell type.

It is due to G. C. Rota and collaborators (finite operator theory,umbral calculus).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 92: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

Entropies and the discrete world

I There is a quite interesting and unexpected relationship between thetheory of difference equations and a large class of generalizedentropies.

Main idea: the nonstandard logarithm appearing at the heart of thenotion of generalized entropy can be seen (in most cases) as arealization of a specific difference operator.

I The modern algebraic theory of difference operators is intimatelyrelated with the theory of polynomial sequences of Sheffer andAppell type.

It is due to G. C. Rota and collaborators (finite operator theory,umbral calculus).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 93: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

I In the following, we shall consider a regular equally spaced latticeof points, denoted by L and indexed by x = nσ, with n ∈ N, σ ∈ R.

Also, we shall deal with operators acting on the algebra of(sufficiently regular) functions FL defined on L.

I The crucial notion of the theory is that of Delta Operator. Todefine it, we shall first introduce shift operators.

I The shift operator, denoted by T , is the operator on L, whoseaction on a function f ∈ FL is given by

Tf (x) = f (x + σ) .

As usual, the operator T can also be represented in terms ofdifferential operators as T = eσD ≡ eσt .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 94: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

I In the following, we shall consider a regular equally spaced latticeof points, denoted by L and indexed by x = nσ, with n ∈ N, σ ∈ R.

Also, we shall deal with operators acting on the algebra of(sufficiently regular) functions FL defined on L.

I The crucial notion of the theory is that of Delta Operator. Todefine it, we shall first introduce shift operators.

I The shift operator, denoted by T , is the operator on L, whoseaction on a function f ∈ FL is given by

Tf (x) = f (x + σ) .

As usual, the operator T can also be represented in terms ofdifferential operators as T = eσD ≡ eσt .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 95: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

I In the following, we shall consider a regular equally spaced latticeof points, denoted by L and indexed by x = nσ, with n ∈ N, σ ∈ R.

Also, we shall deal with operators acting on the algebra of(sufficiently regular) functions FL defined on L.

I The crucial notion of the theory is that of Delta Operator. Todefine it, we shall first introduce shift operators.

I The shift operator, denoted by T , is the operator on L, whoseaction on a function f ∈ FL is given by

Tf (x) = f (x + σ) .

As usual, the operator T can also be represented in terms ofdifferential operators as T = eσD ≡ eσt .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 96: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

I Definition 1.14. An operator S is said to be shift–invariant if commuteswith T : [S ,T ] = 0.

I Definition 1.15. A delta operator Q is a shift–invariant operator suchthat Qx = const 6= 0.

Directly from this definition we deduce the following property.

I Corollary For every constant c ∈ R, Qc = 0.

I Examples. The most common example of delta operator are provided by

a) the standard derivative: Q = ∂x .

b) The forward discrete derivative

Q = ∆+ :=T − 1

σ

c) The backward discrete derivative

Q = ∆− :=1− T−1

σ

d) The symmetric discrete derivative

Q = ∆s :=T − T−1

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 97: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

I Definition 1.14. An operator S is said to be shift–invariant if commuteswith T : [S ,T ] = 0.

I Definition 1.15. A delta operator Q is a shift–invariant operator suchthat Qx = const 6= 0.

Directly from this definition we deduce the following property.

I Corollary For every constant c ∈ R, Qc = 0.

I Examples. The most common example of delta operator are provided by

a) the standard derivative: Q = ∂x .

b) The forward discrete derivative

Q = ∆+ :=T − 1

σ

c) The backward discrete derivative

Q = ∆− :=1− T−1

σ

d) The symmetric discrete derivative

Q = ∆s :=T − T−1

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 98: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

I Definition 1.14. An operator S is said to be shift–invariant if commuteswith T : [S ,T ] = 0.

I Definition 1.15. A delta operator Q is a shift–invariant operator suchthat Qx = const 6= 0.

Directly from this definition we deduce the following property.

I Corollary For every constant c ∈ R, Qc = 0.

I Examples. The most common example of delta operator are provided by

a) the standard derivative: Q = ∂x .

b) The forward discrete derivative

Q = ∆+ :=T − 1

σ

c) The backward discrete derivative

Q = ∆− :=1− T−1

σ

d) The symmetric discrete derivative

Q = ∆s :=T − T−1

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 99: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

I Definition 1.14. An operator S is said to be shift–invariant if commuteswith T : [S ,T ] = 0.

I Definition 1.15. A delta operator Q is a shift–invariant operator suchthat Qx = const 6= 0.

Directly from this definition we deduce the following property.

I Corollary For every constant c ∈ R, Qc = 0.

I Examples. The most common example of delta operator are provided by

a) the standard derivative: Q = ∂x .

b) The forward discrete derivative

Q = ∆+ :=T − 1

σ

c) The backward discrete derivative

Q = ∆− :=1− T−1

σ

d) The symmetric discrete derivative

Q = ∆s :=T − T−1

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 100: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

I Definition 1.16.Given a delta operator Q, a polynomial sequence pn (x)n∈N willbe said to be the sequence of basic polynomials for Q if thefollowing conditions are satisfied:

1) p0 (x) = 1;

2) pn (0) = 0 for all n > 0;

3) Qpn (x) = npn−1 (x) .

Notice that, for a given delta operator Q the sequence of associatedbasic polynomials is unique.

I Examples.Let Q = ∂x . We have pn(x) = xn

Qxn = nxn−1

and the other properties are obvious.Let Q = ∆+. The basic polynomials for Q are the lower factorialpolynomials (x)n := x(x − 1) · . . . · (x − n + 1). We have

∆+(x)n = n(x)n−1.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 101: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

I Definition 1.16.Given a delta operator Q, a polynomial sequence pn (x)n∈N willbe said to be the sequence of basic polynomials for Q if thefollowing conditions are satisfied:

1) p0 (x) = 1;

2) pn (0) = 0 for all n > 0;

3) Qpn (x) = npn−1 (x) .

Notice that, for a given delta operator Q the sequence of associatedbasic polynomials is unique.

I Examples.Let Q = ∂x . We have pn(x) = xn

Qxn = nxn−1

and the other properties are obvious.Let Q = ∆+. The basic polynomials for Q are the lower factorialpolynomials (x)n := x(x − 1) · . . . · (x − n + 1). We have

∆+(x)n = n(x)n−1.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 102: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

I As has been proved by G. C. Rota, there is an isomorphism between thering of formal power series in a variable t and the ring of shift–invariantoperators, carrying

f (t) =∞∑k=0

aktk

k!into

∞∑k=0

akQk

k!. (13)

Definition 1.17. Given a delta operator Q, the function (or a formalpower series) G (t), associated with S under Rota’s isomorphism will becalled the indicator of S .

I With a slight abuse of notation, in the following we often identify a deltaoperator with its indicator. For instance, for ∆+ we have:

∆+ =T − 1

σ=

eσ∂x − 1

σ=∞∑k=0

(σ)k−1∂kx

k!⇐⇒ δ+(t) =

eσt − 1

σ.

For ∆s , we deduce

∆s =T − T−1

2σ=⇐⇒ δs(t) =

eσt − s−σt

2σ.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 103: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

I As has been proved by G. C. Rota, there is an isomorphism between thering of formal power series in a variable t and the ring of shift–invariantoperators, carrying

f (t) =∞∑k=0

aktk

k!into

∞∑k=0

akQk

k!. (13)

Definition 1.17. Given a delta operator Q, the function (or a formalpower series) G (t), associated with S under Rota’s isomorphism will becalled the indicator of S .

I With a slight abuse of notation, in the following we often identify a deltaoperator with its indicator. For instance, for ∆+ we have:

∆+ =T − 1

σ=

eσ∂x − 1

σ=∞∑k=0

(σ)k−1∂kx

k!⇐⇒ δ+(t) =

eσt − 1

σ.

For ∆s , we deduce

∆s =T − T−1

2σ=⇐⇒ δs(t) =

eσt − s−σt

2σ.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 104: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

Definition 1.17 An Appell sequence of polynomials is a Sheffersequence for the delta operator ∂x .

Consequence. For an Appell of polynomials pn (x), we have

∂xpn (x) = npn−1 (x) .

Among the most important examples of Appell sequences, apart fromxnn∈N, are the classical Bernoulli and Euler polynomials and theirgeneralizations.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 105: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

Difference operators and generalized entropies

In order to define the class of entropic functionals of interest for thiswork, we will consider operators expressed as finite Laurent series in shiftoperators

∆r =1

σ

m∑n=l

knT n, l , m ∈ Z, m − l = r > 0, (14)

where kn are real constants such that

m∑n=l

kn = 0,m∑n=l

nkn = c . (15)

and km 6= 0, kl 6= 0. We choose c = 1, to reproduce the derivative ∂x inthe continuum limit, when the lattice spacing σ → 0. The main ideaunderlying our construction is to represent delta operators in logarithmicform, in terms of a suitable function.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 106: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

Generalized logarithms from delta operators

I Definition 1.18. We call logarithmic representation of the delta operator(14) the correspondence T ↔ eσ ln x ≡ xσ, that defines an isomorphism Ibetween the space of shift–invariant operators and the space of functionsf ∈ H.

Definition 1.19. We call generalized logarithm the function

LogG (x) := G(ln x) =1

σ

m∑n=l

knxσn, l ,m ∈ Z,

l < m, m − l = r , x > 0 (16)

with the constraints (15), i.e. the image of the operator (14) under theisomorphism I.

I Lemma 1.20 The following property holds:

limσ→0

LogG (x) = ln x . (17)

Tsallis’ logarithm is

logq(x) =x1−q − 1

1− q= G(ln x), G(t) = δ+(t) =

e(1−q)t − 1

1− q

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 107: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

Generalized logarithms from delta operators

I Definition 1.18. We call logarithmic representation of the delta operator(14) the correspondence T ↔ eσ ln x ≡ xσ, that defines an isomorphism Ibetween the space of shift–invariant operators and the space of functionsf ∈ H.

Definition 1.19. We call generalized logarithm the function

LogG (x) := G(ln x) =1

σ

m∑n=l

knxσn, l ,m ∈ Z,

l < m, m − l = r , x > 0 (16)

with the constraints (15), i.e. the image of the operator (14) under theisomorphism I.

I Lemma 1.20 The following property holds:

limσ→0

LogG (x) = ln x . (17)

Tsallis’ logarithm is

logq(x) =x1−q − 1

1− q= G(ln x), G(t) = δ+(t) =

e(1−q)t − 1

1− qPiergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 108: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

Other examples

I Kaniadakis’ logarithm is

logk(x) =xk − x−k

2k= G(ln x), G(t) = δs(t) =

ekt − e−kt

2k.

Let us now construct some new examples. Consider for instance thedifference operators

∆III = T−2T−1+T−2

σ, ∆IV =

T 2− 32T+ 3

2T−1−T−2

σ,

∆V = T 3−2T 2+2T−2T−1+T−2

−σ ,

and so on.

I The corresponding generalized logarithms are

LogGIII (x) =1

σ

(xσ − 2x−σ + x−2σ

),

LogGIV (x) =1

σ

(x2σ − 3

2xσ +

3

2x−σ − x−2σ

),

LogGV (x) =1

σ

(x3σ − 2x2σ + 2xσ − 2x−σ + x−2σ

).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 109: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

Other examples

I Kaniadakis’ logarithm is

logk(x) =xk − x−k

2k= G(ln x), G(t) = δs(t) =

ekt − e−kt

2k.

Let us now construct some new examples. Consider for instance thedifference operators

∆III = T−2T−1+T−2

σ, ∆IV =

T 2− 32T+ 3

2T−1−T−2

σ,

∆V = T 3−2T 2+2T−2T−1+T−2

−σ ,

and so on.

I The corresponding generalized logarithms are

LogGIII (x) =1

σ

(xσ − 2x−σ + x−2σ

),

LogGIV (x) =1

σ

(x2σ − 3

2xσ +

3

2x−σ − x−2σ

),

LogGV (x) =1

σ

(x3σ − 2x2σ + 2xσ − 2x−σ + x−2σ

).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 110: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

State of Art

I A generalization of the Shannon-Khinchin axiomatic formulation,originally proposed to characterize uniquely the Boltzmann entropy,has been proposed.

I Although the first three SK axioms are mantained, the fourth one isreplaced by the new composability axiom.

I This axiom encodes a group theoretical structure, which isresponsible of essentially all the properties of a generalized entropy.

I Formal group theory, borrowed from algebraic topology, has beenused to characterize the composability process, and consequently,the class of thermodynamically admissible entropies.

I A connection between discrete mathematics and generalizedlogarithms has been sketched, via the finite operator theory by C.G. Rota and coll.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 111: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

State of Art

I A generalization of the Shannon-Khinchin axiomatic formulation,originally proposed to characterize uniquely the Boltzmann entropy,has been proposed.

I Although the first three SK axioms are mantained, the fourth one isreplaced by the new composability axiom.

I This axiom encodes a group theoretical structure, which isresponsible of essentially all the properties of a generalized entropy.

I Formal group theory, borrowed from algebraic topology, has beenused to characterize the composability process, and consequently,the class of thermodynamically admissible entropies.

I A connection between discrete mathematics and generalizedlogarithms has been sketched, via the finite operator theory by C.G. Rota and coll.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 112: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

State of Art

I A generalization of the Shannon-Khinchin axiomatic formulation,originally proposed to characterize uniquely the Boltzmann entropy,has been proposed.

I Although the first three SK axioms are mantained, the fourth one isreplaced by the new composability axiom.

I This axiom encodes a group theoretical structure, which isresponsible of essentially all the properties of a generalized entropy.

I Formal group theory, borrowed from algebraic topology, has beenused to characterize the composability process, and consequently,the class of thermodynamically admissible entropies.

I A connection between discrete mathematics and generalizedlogarithms has been sketched, via the finite operator theory by C.G. Rota and coll.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 113: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

State of Art

I A generalization of the Shannon-Khinchin axiomatic formulation,originally proposed to characterize uniquely the Boltzmann entropy,has been proposed.

I Although the first three SK axioms are mantained, the fourth one isreplaced by the new composability axiom.

I This axiom encodes a group theoretical structure, which isresponsible of essentially all the properties of a generalized entropy.

I Formal group theory, borrowed from algebraic topology, has beenused to characterize the composability process, and consequently,the class of thermodynamically admissible entropies.

I A connection between discrete mathematics and generalizedlogarithms has been sketched, via the finite operator theory by C.G. Rota and coll.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 114: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

State of Art

I A generalization of the Shannon-Khinchin axiomatic formulation,originally proposed to characterize uniquely the Boltzmann entropy,has been proposed.

I Although the first three SK axioms are mantained, the fourth one isreplaced by the new composability axiom.

I This axiom encodes a group theoretical structure, which isresponsible of essentially all the properties of a generalized entropy.

I Formal group theory, borrowed from algebraic topology, has beenused to characterize the composability process, and consequently,the class of thermodynamically admissible entropies.

I A connection between discrete mathematics and generalizedlogarithms has been sketched, via the finite operator theory by C.G. Rota and coll.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 115: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

SK and

Composability

axioms Generalized Entropies

Group-theoretical

Structure

Delta Operators Formal group

theory

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 116: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Introduction: A Foundational PerspectiveFormal Group Theory

Strong and weak composabilityGroups, Difference Operators and Entropies

Delta operators: a short reviewBasic sequences, Sheffer and Appell sequencesGeneralized logarithms from delta operators

Thank you!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY

Page 117: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

GROUPS, ENTROPIES AND NUMBERTHEORY

Piergiulio Tempesta

Universidad Complutense de Madridand

Instituto de Ciencias Matematicas (ICMAT),Madrid, Spain.

TEMPLETON SCHOOL ON FOUNDATIONS OF COMPLEXITY

October 14 - 15, 2015

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 118: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

LECTURE II

From Groups to Entropies: The composabilityaxiom and the Universal Group Entropy

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 119: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

Outline

The Universal Group EntropyThe universal-group entropyMain properties of SU entropyRelation with other entropies

Distribution functions and thermodynamic properties

An entropy for the rational group lawA new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

A “magical recipe” to generate multiparametric entropiesThe microcanonical description and associated thermodynamicsMultiparametric entropies

On the asymptotic behaviour of generalized entropies

State of Art

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 120: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A little reminder: The composability axiom

An entropy S is (strictly) composable if there exists a continuous function oftwo real variables Φ(x , y) such that (C1)

S(A ∪ B) = Φ(S(A), S(B); η) (1)

where A ⊂ X and B ⊂ X are two statistically independent subsystems of agiven system X , defined over any probability distribution piWi=1, with thefurther properties(C2) Symmetry:

Φ(x , y) = Φ(y , x). (2)

(C3) Associativity:Φ(x ,Φ(y , z)) = Φ(Φ(x , y), z). (3)

(C4) Null-composability:Φ(x , 0) = x . (4)

An entropy is weakly composable if the properties (C1)–(C3) are satisfied atleast when the probability distributions of the two independent systems A andB are both uniform, and property (C4) holds in general.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 121: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The universal-group entropy

I Is there an entropy associated with the Lazard universal formalgroup?

I Let us denote by P the set of all probability distributions.Definition 2.1. Let pii=1,··· ,W , W ≥ 1, with

∑Wi=1 pi = 1, be a

discrete probability distribution. Let

G (t) =∞∑k=0

aktk+1

k + 1, (5)

be a real analytic function, where akk∈N is a sequence ofparameters, with a0 6= 0, such that the function SU : P → R+ ∪ 0,

SU(p1, . . . , pW ) := kB

W∑i=1

pi G

(ln

1

pi

), (6)

is a concave one. This function is the universal-group entropy.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 122: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The universal-group entropy

I Is there an entropy associated with the Lazard universal formalgroup?

I Let us denote by P the set of all probability distributions.Definition 2.1. Let pii=1,··· ,W , W ≥ 1, with

∑Wi=1 pi = 1, be a

discrete probability distribution. Let

G (t) =∞∑k=0

aktk+1

k + 1, (5)

be a real analytic function, where akk∈N is a sequence ofparameters, with a0 6= 0, such that the function SU : P → R+ ∪ 0,

SU(p1, . . . , pW ) := kB

W∑i=1

pi G

(ln

1

pi

), (6)

is a concave one. This function is the universal-group entropy.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 123: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

First properties

I Remark 2.2. The condition

ak > (k + 1)ak+1 ∀k ∈ N with akk∈N ≥ 0 (7)

is sufficient to ensure that the series G (t) is absolutely anduniformly convergent with a radius r =∞ and that SU [p] is astrictly concave functional. Although certainly restrictive, condition(19) is satisfied by many of the entropies known in the literature.

I Remark 2.3. The universal-group entropy depends on the infiniteset of parameters ak , which are a priori independent. To recoverknown cases of one-parametric or two-parametric entropies,depending for instance on the parameters q1 and q2, we shall havethat ak = fk(q1, q2).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 124: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

First properties

I Remark 2.2. The condition

ak > (k + 1)ak+1 ∀k ∈ N with akk∈N ≥ 0 (7)

is sufficient to ensure that the series G (t) is absolutely anduniformly convergent with a radius r =∞ and that SU [p] is astrictly concave functional. Although certainly restrictive, condition(19) is satisfied by many of the entropies known in the literature.

I Remark 2.3. The universal-group entropy depends on the infiniteset of parameters ak , which are a priori independent. To recoverknown cases of one-parametric or two-parametric entropies,depending for instance on the parameters q1 and q2, we shall havethat ak = fk(q1, q2).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 125: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

Main properties of SU entropy

I Theorem 2.4 The entropy SU [p] is at least weakly composable.

I Proof. Let pAi WA

i=1 and pBj WB

j=1 two sets of probabilities associatedwith two statistically independent systems A and B. The jointprobability is given by

pA∪Bij = pAi · pBjThe total number of states of the composed system (including stateswith possibly null probability), is WAB = WAWB . We have

SU(A ∪ B) := kB

WA∑i=1

WB∑j=1

pA∪Bij G

(ln

1

pA∪Bij

)=

= kB

WA∑i=1

WB∑j=1

pAi ·pBj G

(ln

1

pAi+ ln

1

pBj

)= kB

WA∑i=1

WB∑j=1

pAi ·pBj G (t1 + t2) =

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 126: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

Main properties of SU entropy

I Theorem 2.4 The entropy SU [p] is at least weakly composable.

I Proof. Let pAi WA

i=1 and pBj WB

j=1 two sets of probabilities associatedwith two statistically independent systems A and B. The jointprobability is given by

pA∪Bij = pAi · pBjThe total number of states of the composed system (including stateswith possibly null probability), is WAB = WAWB . We have

SU(A ∪ B) := kB

WA∑i=1

WB∑j=1

pA∪Bij G

(ln

1

pA∪Bij

)=

= kB

WA∑i=1

WB∑j=1

pAi ·pBj G

(ln

1

pAi+ ln

1

pBj

)= kB

WA∑i=1

WB∑j=1

pAi ·pBj G (t1 + t2) =

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 127: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

= kB

WA∑i=1

WB∑j=1

pAi · pB

j G(

G−1(s1) + G−1(s2))

= kB

WA∑i=1

WB∑j=1

pAi · pB

j Φ (s1, s2) =

= kB

WA∑i=1

WB∑j=1

pAi · pB

j

s1 + s2 +∞∑

k,m=1

ckmsk1 sm2

=

= kB

WA∑i=1

WB∑j=1

pAi · pB

j

[G

(ln

1

pAi

)+ G

(ln

1

pBj

)+ c11G

(ln

1

pAi

)· G

(ln

1

pBj

)+ . . .

]

The Boltzmann entropy corresponds to the case (i.e. ckm = 0 ∀k,m). TheTsallis entropy corresponds to the case c11 6= 0 and ckm = 0 for(k,m) 6= (1, 1). In both cases, we get immediately

SU(A + B) = Φ (SU(A), SU(B)) , (8)

with Φ(x , y) = x + y for ckm = 0, and Φ(x , y) = x + y + c11xy for c11 6= 0 andckm = 0. However, whenever ckm 6= 0 for (k,m) 6= (1, 1), a priori formula (8)does not hold in general (i.e. for any possible choice of pA

i WAi=1 and pB

j WBj=1).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 128: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

This because

Yk

(pA1 , . . . , p

Aw

)=

WA∑i=1

pAi Gk

(ln

1

pAi

)−

[WA∑i=1

pAi G

(ln

1

pAi

)]k6= 0!!.

Nevertheless, in the case that pAi and pBj are both the uniformdistribution, formula (8) holds for the whole family of entropiesrepresented by (6).

The formal power series φ(x , y) = G (G−1(x) + G−1(y)) for any choiceof G defines a formal group law. It verifies automatically the conditionsof symmetry, null composability and transitivity. We deduce that SU isweakly composable.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 129: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

Theorem 2.5

I The entropy SU [p] satisfies the Shannon-Khinchin axioms(SK1)-(SK3).

I (SK1). The group exponential G(t) is a real analytic function of t.Consequently, the universal group entropy is (at least) a continuousfunction of its arguments (p1, . . . , pw ).

I (SK2). The entropy SU [p] is supposed to be concave by definition.Nevertheless, we wish to prove here that condition (19), which allows toconstruct a large subclass of entropies of the form (6), is sufficient toguarantee concavity. To this aim, consider the quantity∑∞

k=0akk+1

x(ln 1

x

)k+1. By imposing strict concavity we get

− 1

xa0−a1+(a1−2a2) ln

1

x+(a2−3a3)

(ln

1

x

)2

+(a3−4a4)

(ln

1

x

)3

+. . . < 0.

(9)Condition ak > (k + 1)ak+1 ensures that the inequality (55) is satisfied.Consequently, the associated entropy (6) is strictly concave in its space ofparameters. Other choices of the sequence akk∈N are clearly possible.

I (SK3). Since by construction G(0) = 0, and limx→0 x(ln 1

x

)k= 0, it

follows that SU(0) = 0. Similarly, SU(1) = 0.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 130: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

Theorem 2.5

I The entropy SU [p] satisfies the Shannon-Khinchin axioms(SK1)-(SK3).

I (SK1). The group exponential G(t) is a real analytic function of t.Consequently, the universal group entropy is (at least) a continuousfunction of its arguments (p1, . . . , pw ).

I (SK2). The entropy SU [p] is supposed to be concave by definition.Nevertheless, we wish to prove here that condition (19), which allows toconstruct a large subclass of entropies of the form (6), is sufficient toguarantee concavity. To this aim, consider the quantity∑∞

k=0akk+1

x(ln 1

x

)k+1. By imposing strict concavity we get

− 1

xa0−a1+(a1−2a2) ln

1

x+(a2−3a3)

(ln

1

x

)2

+(a3−4a4)

(ln

1

x

)3

+. . . < 0.

(9)Condition ak > (k + 1)ak+1 ensures that the inequality (55) is satisfied.Consequently, the associated entropy (6) is strictly concave in its space ofparameters. Other choices of the sequence akk∈N are clearly possible.

I (SK3). Since by construction G(0) = 0, and limx→0 x(ln 1

x

)k= 0, it

follows that SU(0) = 0. Similarly, SU(1) = 0.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 131: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

Theorem 2.5

I The entropy SU [p] satisfies the Shannon-Khinchin axioms(SK1)-(SK3).

I (SK1). The group exponential G(t) is a real analytic function of t.Consequently, the universal group entropy is (at least) a continuousfunction of its arguments (p1, . . . , pw ).

I (SK2). The entropy SU [p] is supposed to be concave by definition.Nevertheless, we wish to prove here that condition (19), which allows toconstruct a large subclass of entropies of the form (6), is sufficient toguarantee concavity. To this aim, consider the quantity∑∞

k=0akk+1

x(ln 1

x

)k+1. By imposing strict concavity we get

− 1

xa0−a1+(a1−2a2) ln

1

x+(a2−3a3)

(ln

1

x

)2

+(a3−4a4)

(ln

1

x

)3

+. . . < 0.

(9)Condition ak > (k + 1)ak+1 ensures that the inequality (55) is satisfied.Consequently, the associated entropy (6) is strictly concave in its space ofparameters. Other choices of the sequence akk∈N are clearly possible.

I (SK3). Since by construction G(0) = 0, and limx→0 x(ln 1

x

)k= 0, it

follows that SU(0) = 0. Similarly, SU(1) = 0.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 132: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

Theorem 2.5

I The entropy SU [p] satisfies the Shannon-Khinchin axioms(SK1)-(SK3).

I (SK1). The group exponential G(t) is a real analytic function of t.Consequently, the universal group entropy is (at least) a continuousfunction of its arguments (p1, . . . , pw ).

I (SK2). The entropy SU [p] is supposed to be concave by definition.Nevertheless, we wish to prove here that condition (19), which allows toconstruct a large subclass of entropies of the form (6), is sufficient toguarantee concavity. To this aim, consider the quantity∑∞

k=0akk+1

x(ln 1

x

)k+1. By imposing strict concavity we get

− 1

xa0−a1+(a1−2a2) ln

1

x+(a2−3a3)

(ln

1

x

)2

+(a3−4a4)

(ln

1

x

)3

+. . . < 0.

(9)Condition ak > (k + 1)ak+1 ensures that the inequality (55) is satisfied.Consequently, the associated entropy (6) is strictly concave in its space ofparameters. Other choices of the sequence akk∈N are clearly possible.

I (SK3). Since by construction G(0) = 0, and limx→0 x(ln 1

x

)k= 0, it

follows that SU(0) = 0. Similarly, SU(1) = 0.Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 133: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

Theorem 2.6 The entropy SU [p] is extensive: on the uniformprobability distribution (i.e. pi = 1/W for all i = 1, . . . ,W ), we have

SU [W ] = kBG (lnW ) ∼ N ⇐⇒W (N) ∼ exp(G−1(N)

). (10)

We should also require that W (N) ∼ exp(G−1(N)

), to guarantee that

W (N) is interpretable as an occupation law.Viceversa, given an occupation law W (N), we get G−1(x) = lnW (x);the inverse of this expression gives us G (x). In this way, we can designan entropy extensive in a given regime.These properties are all necessary for the applicability of the entropy (6)in thermodynamical contexts.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 134: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

Relation with other entropies

I General approach. An useful property of the entropy SU [p] is thatit admits the following decomposition

SU [p] =∞∑k=1

γkSk [p], γk = ak−1/k, (11)

in terms of a set of elementary functionals

Sk [p] := kB∑W

i=1 pi(

ln 1pi

)k, with k ∈ N.

I Therefore we have the general expression

SU [p] = kB

W∑i=1

pi

∞∑k=1

ak−1k

(ln

1

pi

)k

I This analytic form is shared by essentially all the entropies known inthe literature, and allows a quick comparison among them.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 135: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

Relation with other entropies

I General approach. An useful property of the entropy SU [p] is thatit admits the following decomposition

SU [p] =∞∑k=1

γkSk [p], γk = ak−1/k, (11)

in terms of a set of elementary functionals

Sk [p] := kB∑W

i=1 pi(

ln 1pi

)k, with k ∈ N.

I Therefore we have the general expression

SU [p] = kB

W∑i=1

pi

∞∑k=1

ak−1k

(ln

1

pi

)k

I This analytic form is shared by essentially all the entropies known inthe literature, and allows a quick comparison among them.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 136: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

Relation with other entropies

I General approach. An useful property of the entropy SU [p] is thatit admits the following decomposition

SU [p] =∞∑k=1

γkSk [p], γk = ak−1/k, (11)

in terms of a set of elementary functionals

Sk [p] := kB∑W

i=1 pi(

ln 1pi

)k, with k ∈ N.

I Therefore we have the general expression

SU [p] = kB

W∑i=1

pi

∞∑k=1

ak−1k

(ln

1

pi

)k

I This analytic form is shared by essentially all the entropies known inthe literature, and allows a quick comparison among them.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 137: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The Boltzmann-Gibbs entropy

I a) The Boltzmann-Gibbs entropy

SB[p] = kB

W∑i=1

pi ln1

pi. (12)

It has the general form∑

i piG(

ln 1pi

), with

G (t) = t

(i.e. for a1 = 1, ai = 0, ∀i = 2, 3, . . .). The delta operatorassociated with G (t) according to Rota’s isomorphism is t = ∂

I The group structure is simply the additive group law

Φ(x , y) = G (G−1(x) + G−1(y)) = x + y .

Therefore:

S(A ∪ B) = Φ(S(A),S(B)) = S(A) + S(B).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 138: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The Boltzmann-Gibbs entropy

I a) The Boltzmann-Gibbs entropy

SB[p] = kB

W∑i=1

pi ln1

pi. (12)

It has the general form∑

i piG(

ln 1pi

), with

G (t) = t

(i.e. for a1 = 1, ai = 0, ∀i = 2, 3, . . .). The delta operatorassociated with G (t) according to Rota’s isomorphism is t = ∂

I The group structure is simply the additive group law

Φ(x , y) = G (G−1(x) + G−1(y)) = x + y .

Therefore:

S(A ∪ B) = Φ(S(A),S(B)) = S(A) + S(B).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 139: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The group-theoretical approach for the Tsallis entropy

C. Tsallis, J. Stat. Phys. 52, 479-487 (1988).I b) The Tsallis entropy

Sq[p] = kB

∑Wi=1 p

qi − 1

1− q=

W∑i=1

pi lnq1

pi, 0 ≤ q < 1.

I It has the general form∑

i piG(

ln 1pi

), with

G (t) =exp[(1− q)t]− 1

1− q= t +

1

2(1− q)t2 +

1

6(1− q)2t3 + . . .

I Group logarithm: lnq x = x1−q−11−q .

I The associated finite difference delta operator is δ+ = Tσ−1σ , with

σ = 1− q.I When q → 1, G (t)→ t and Tsallis’ entropy reduces to Boltzmann’s

entropy

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 140: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The group-theoretical approach for the Tsallis entropy

C. Tsallis, J. Stat. Phys. 52, 479-487 (1988).I b) The Tsallis entropy

Sq[p] = kB

∑Wi=1 p

qi − 1

1− q=

W∑i=1

pi lnq1

pi, 0 ≤ q < 1.

I It has the general form∑

i piG(

ln 1pi

), with

G (t) =exp[(1− q)t]− 1

1− q= t +

1

2(1− q)t2 +

1

6(1− q)2t3 + . . .

I Group logarithm: lnq x = x1−q−11−q .

I The associated finite difference delta operator is δ+ = Tσ−1σ , with

σ = 1− q.I When q → 1, G (t)→ t and Tsallis’ entropy reduces to Boltzmann’s

entropy

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 141: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The group-theoretical approach for the Tsallis entropy

C. Tsallis, J. Stat. Phys. 52, 479-487 (1988).I b) The Tsallis entropy

Sq[p] = kB

∑Wi=1 p

qi − 1

1− q=

W∑i=1

pi lnq1

pi, 0 ≤ q < 1.

I It has the general form∑

i piG(

ln 1pi

), with

G (t) =exp[(1− q)t]− 1

1− q= t +

1

2(1− q)t2 +

1

6(1− q)2t3 + . . .

I Group logarithm: lnq x = x1−q−11−q .

I The associated finite difference delta operator is δ+ = Tσ−1σ , with

σ = 1− q.I When q → 1, G (t)→ t and Tsallis’ entropy reduces to Boltzmann’s

entropy

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 142: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The group-theoretical approach for the Tsallis entropy

C. Tsallis, J. Stat. Phys. 52, 479-487 (1988).I b) The Tsallis entropy

Sq[p] = kB

∑Wi=1 p

qi − 1

1− q=

W∑i=1

pi lnq1

pi, 0 ≤ q < 1.

I It has the general form∑

i piG(

ln 1pi

), with

G (t) =exp[(1− q)t]− 1

1− q= t +

1

2(1− q)t2 +

1

6(1− q)2t3 + . . .

I Group logarithm: lnq x = x1−q−11−q .

I The associated finite difference delta operator is δ+ = Tσ−1σ , with

σ = 1− q.

I When q → 1, G (t)→ t and Tsallis’ entropy reduces to Boltzmann’sentropy

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 143: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The group-theoretical approach for the Tsallis entropy

C. Tsallis, J. Stat. Phys. 52, 479-487 (1988).I b) The Tsallis entropy

Sq[p] = kB

∑Wi=1 p

qi − 1

1− q=

W∑i=1

pi lnq1

pi, 0 ≤ q < 1.

I It has the general form∑

i piG(

ln 1pi

), with

G (t) =exp[(1− q)t]− 1

1− q= t +

1

2(1− q)t2 +

1

6(1− q)2t3 + . . .

I Group logarithm: lnq x = x1−q−11−q .

I The associated finite difference delta operator is δ+ = Tσ−1σ , with

σ = 1− q.I When q → 1, G (t)→ t and Tsallis’ entropy reduces to Boltzmann’s

entropy

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 144: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

I Its decomposition is provided by:

Sq[p] = kB

W∑i=1

piln1

pi+

1

2(1− q)

(ln

1

pi

)2

+1

6(1− q)2

(ln

1

pi

)3

+ . . .

I Let us determine the associated group law:

G−1(s) =1

1− qln(1 + (1− q)s)

G−1(x) + G−1(y) =1

1− qln(

1 + (1− q)x + (1− q)y + (1− q)2xy)

Consequently we get

Φ(x , y) = G(G−1(x) + G−1(y)) = x + y + (1− q)xy .

which is nothing but the multiplicative group law. Once again, wededuce

S(A ∪ B) = Φ(S(A),S(B)) = S(A) + S(B) + (1− q)S(A)S(B).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 145: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

I Its decomposition is provided by:

Sq[p] = kB

W∑i=1

piln1

pi+

1

2(1− q)

(ln

1

pi

)2

+1

6(1− q)2

(ln

1

pi

)3

+ . . .

I Let us determine the associated group law:

G−1(s) =1

1− qln(1 + (1− q)s)

G−1(x) + G−1(y) =1

1− qln(

1 + (1− q)x + (1− q)y + (1− q)2xy)

Consequently we get

Φ(x , y) = G(G−1(x) + G−1(y)) = x + y + (1− q)xy .

which is nothing but the multiplicative group law. Once again, wededuce

S(A ∪ B) = Φ(S(A), S(B)) = S(A) + S(B) + (1− q)S(A)S(B).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 146: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The Kaniadakis entropy

G. Kaniadakis, Phys. Rev. E, 66, 056125 (2002).I c) The Kaniadakis entropy

Sκ[p] = kB

W∑i=1

pip−κi − pκi

2κ. (13)

I It has the general form∑

i piG(

ln 1pi

), with

G (t) =exp(κt)− exp(−κt)

I The group logarithm is lnk x = xk−x−k

2k . The associated finite

difference operator is δ+ = Tσ−T−σ

2σ . Its decomposition is

Sκ[p] = kB

W∑i=1

pi

ln

1

pi+

1

3!κ2(

ln1

pi

)3

+1

5!κ4(

ln1

pi

)5

+ . . .

(14)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 147: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The Kaniadakis entropy

G. Kaniadakis, Phys. Rev. E, 66, 056125 (2002).I c) The Kaniadakis entropy

Sκ[p] = kB

W∑i=1

pip−κi − pκi

2κ. (13)

I It has the general form∑

i piG(

ln 1pi

), with

G (t) =exp(κt)− exp(−κt)

I The group logarithm is lnk x = xk−x−k

2k . The associated finite

difference operator is δ+ = Tσ−T−σ

2σ . Its decomposition is

Sκ[p] = kB

W∑i=1

pi

ln

1

pi+

1

3!κ2(

ln1

pi

)3

+1

5!κ4(

ln1

pi

)5

+ . . .

(14)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 148: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The Kaniadakis entropy

G. Kaniadakis, Phys. Rev. E, 66, 056125 (2002).I c) The Kaniadakis entropy

Sκ[p] = kB

W∑i=1

pip−κi − pκi

2κ. (13)

I It has the general form∑

i piG(

ln 1pi

), with

G (t) =exp(κt)− exp(−κt)

I The group logarithm is lnk x = xk−x−k

2k . The associated finite

difference operator is δ+ = Tσ−T−σ

2σ . Its decomposition is

Sκ[p] = kB

W∑i=1

pi

ln

1

pi+

1

3!κ2(

ln1

pi

)3

+1

5!κ4(

ln1

pi

)5

+ . . .

(14)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 149: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

Weak composability of Sk

Nota Bene. Kaniadakis’ entropy is not composable in full generality.

I Indeed,Sk(A ∪ B) 6= F (Sk(A),Sk(B), κ)

for any function F (x , y), if we wish that the previous property holdsfor any possible probability distribution piWi=1.

I However, if we choose the uniform distribution, then

Sk [p] = kB lnk W ,

andSk(Au ∪ Bu) = Φ(Sk(Au),Sk(Bu))

I whereΦ(x , y) = x

√1 + k2y2 + y

√1 + k2x2.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 150: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

Weak composability of Sk

Nota Bene. Kaniadakis’ entropy is not composable in full generality.

I Indeed,Sk(A ∪ B) 6= F (Sk(A),Sk(B), κ)

for any function F (x , y), if we wish that the previous property holdsfor any possible probability distribution piWi=1.

I However, if we choose the uniform distribution, then

Sk [p] = kB lnk W ,

andSk(Au ∪ Bu) = Φ(Sk(Au),Sk(Bu))

I whereΦ(x , y) = x

√1 + k2y2 + y

√1 + k2x2.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 151: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

Weak composability of Sk

Nota Bene. Kaniadakis’ entropy is not composable in full generality.

I Indeed,Sk(A ∪ B) 6= F (Sk(A),Sk(B), κ)

for any function F (x , y), if we wish that the previous property holdsfor any possible probability distribution piWi=1.

I However, if we choose the uniform distribution, then

Sk [p] = kB lnk W ,

andSk(Au ∪ Bu) = Φ(Sk(Au),Sk(Bu))

I whereΦ(x , y) = x

√1 + k2y2 + y

√1 + k2x2.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 152: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The S(c , d) entropy

R. Hanel and S. Thurner, Europhys. Lett. 93 20006 (2011).

I d) The Sc,d entropy reads

Sc,d =e

1− c + cd

W∑i=1

Γ(1+d , 1−c ln pi )−c

1− c + cd, c ∈ (0, 1], d ∈ R.

(15)

I We remind that the upper incomplete gamma function Γ(s, x) and theupper incomplete gamma functions are, respectively :

Γ(s, x) =

∫ ∞x

ts−1e−tdt, γ(s, x) =

∫ x

0

ts−1e−tdt, s ∈ C, Re s > 0.

I To study the group-theoretical structure of the Sc,d entropy, observe that

Γ(1 + d , 1− c ln pi ) = Γ(1 + d) + td∞∑k=0

δktk+1

k + 1, (16)

δk =(−1)k+1(k + 1)

k!(k + d + 1), t =

[ln

e

pic

]. (17)

In the subsequent considerations, we shall restrict to the case d ∈ N.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 153: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The S(c , d) entropy

R. Hanel and S. Thurner, Europhys. Lett. 93 20006 (2011).

I d) The Sc,d entropy reads

Sc,d =e

1− c + cd

W∑i=1

Γ(1+d , 1−c ln pi )−c

1− c + cd, c ∈ (0, 1], d ∈ R.

(15)

I We remind that the upper incomplete gamma function Γ(s, x) and theupper incomplete gamma functions are, respectively :

Γ(s, x) =

∫ ∞x

ts−1e−tdt, γ(s, x) =

∫ x

0

ts−1e−tdt, s ∈ C, Re s > 0.

I To study the group-theoretical structure of the Sc,d entropy, observe that

Γ(1 + d , 1− c ln pi ) = Γ(1 + d) + td∞∑k=0

δktk+1

k + 1, (16)

δk =(−1)k+1(k + 1)

k!(k + d + 1), t =

[ln

e

pic

]. (17)

In the subsequent considerations, we shall restrict to the case d ∈ N.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 154: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The S(c , d) entropy

R. Hanel and S. Thurner, Europhys. Lett. 93 20006 (2011).

I d) The Sc,d entropy reads

Sc,d =e

1− c + cd

W∑i=1

Γ(1+d , 1−c ln pi )−c

1− c + cd, c ∈ (0, 1], d ∈ R.

(15)

I We remind that the upper incomplete gamma function Γ(s, x) and theupper incomplete gamma functions are, respectively :

Γ(s, x) =

∫ ∞x

ts−1e−tdt, γ(s, x) =

∫ x

0

ts−1e−tdt, s ∈ C, Re s > 0.

I To study the group-theoretical structure of the Sc,d entropy, observe that

Γ(1 + d , 1− c ln pi ) = Γ(1 + d) + td∞∑k=0

δktk+1

k + 1, (16)

δk =(−1)k+1(k + 1)

k!(k + d + 1), t =

[ln

e

pic

]. (17)

In the subsequent considerations, we shall restrict to the case d ∈ N.Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 155: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

I To perform our analysis we shall use the identity, valid for d ∈ N:∫ ∞K

tde−tdt = e−Kd∑

n=0

∏nj=0(d − j + 1)

d + 1K d−n. (18)

This identity can be proven by a direct computation. It allows to expandthe entropy in terms of the set Sk [p].

I In general, we have

Sc,d =1

1− c + cd

W∑i=1

pic

d∑

k=0

1

d + 1

k∏j=0

(d − j + 1)d−k∑n=0

(d − k

n

)(ln

1

pic

)n

− c

1− c + cd, d ∈ N.

I Particular cases of the previous formula:

S1,1[p] = 1+∑i

pi ln1

pi, S1,2[p] = 2

(1 +

∑i

pi ln1

pi

)+

1

2

∑i

pi

(ln

1

pi

)2

.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 156: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

I To perform our analysis we shall use the identity, valid for d ∈ N:∫ ∞K

tde−tdt = e−Kd∑

n=0

∏nj=0(d − j + 1)

d + 1K d−n. (18)

This identity can be proven by a direct computation. It allows to expandthe entropy in terms of the set Sk [p].

I In general, we have

Sc,d =1

1− c + cd

W∑i=1

pic

d∑

k=0

1

d + 1

k∏j=0

(d − j + 1)d−k∑n=0

(d − k

n

)(ln

1

pic

)n

− c

1− c + cd, d ∈ N.

I Particular cases of the previous formula:

S1,1[p] = 1+∑i

pi ln1

pi, S1,2[p] = 2

(1 +

∑i

pi ln1

pi

)+

1

2

∑i

pi

(ln

1

pi

)2

.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 157: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

I To perform our analysis we shall use the identity, valid for d ∈ N:∫ ∞K

tde−tdt = e−Kd∑

n=0

∏nj=0(d − j + 1)

d + 1K d−n. (18)

This identity can be proven by a direct computation. It allows to expandthe entropy in terms of the set Sk [p].

I In general, we have

Sc,d =1

1− c + cd

W∑i=1

pic

d∑

k=0

1

d + 1

k∏j=0

(d − j + 1)d−k∑n=0

(d − k

n

)(ln

1

pic

)n

− c

1− c + cd, d ∈ N.

I Particular cases of the previous formula:

S1,1[p] = 1+∑i

pi ln1

pi, S1,2[p] = 2

(1 +

∑i

pi ln1

pi

)+

1

2

∑i

pi

(ln

1

pi

)2

.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 158: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

I For c arbitrary, let us write explicitly, for instance, the functionalscorresponding to d = 3:

Sc,3[p] =1

1 + 2c

W∑i=1

pic

16 + 15 ln

1

pic

+ 6

(ln

1

pic

)2

+

(ln

1

pic

)3− c

1 + 2c,

We have:G(t) = 16 + 15t + 6t2 + t3

I and d = 5:

Sc,5[p] =1

1 + 4c

W∑i=1

pic

326 + 325 ln

1

pic

+ 160

(ln

1

pic

)2

+ 50

(ln

1

pic

)3

+ 10

(ln

1

pic

)4

+

(ln

1

pic

)5− c

1 + 4c.

withG(t) = 326 + 325t + 160t2 + 50t3 + 10t4 + t5

I In both cases:

ak > (k + 1)ak+1 ∀k ∈ N with akk∈N ≥ 0 (19)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 159: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

I For c arbitrary, let us write explicitly, for instance, the functionalscorresponding to d = 3:

Sc,3[p] =1

1 + 2c

W∑i=1

pic

16 + 15 ln

1

pic

+ 6

(ln

1

pic

)2

+

(ln

1

pic

)3− c

1 + 2c,

We have:G(t) = 16 + 15t + 6t2 + t3

I and d = 5:

Sc,5[p] =1

1 + 4c

W∑i=1

pic

326 + 325 ln

1

pic

+ 160

(ln

1

pic

)2

+ 50

(ln

1

pic

)3

+ 10

(ln

1

pic

)4

+

(ln

1

pic

)5− c

1 + 4c.

withG(t) = 326 + 325t + 160t2 + 50t3 + 10t4 + t5

I In both cases:

ak > (k + 1)ak+1 ∀k ∈ N with akk∈N ≥ 0 (19)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 160: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

I For c arbitrary, let us write explicitly, for instance, the functionalscorresponding to d = 3:

Sc,3[p] =1

1 + 2c

W∑i=1

pic

16 + 15 ln

1

pic

+ 6

(ln

1

pic

)2

+

(ln

1

pic

)3− c

1 + 2c,

We have:G(t) = 16 + 15t + 6t2 + t3

I and d = 5:

Sc,5[p] =1

1 + 4c

W∑i=1

pic

326 + 325 ln

1

pic

+ 160

(ln

1

pic

)2

+ 50

(ln

1

pic

)3

+ 10

(ln

1

pic

)4

+

(ln

1

pic

)5− c

1 + 4c.

withG(t) = 326 + 325t + 160t2 + 50t3 + 10t4 + t5

I In both cases:

ak > (k + 1)ak+1 ∀k ∈ N with akk∈N ≥ 0 (19)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 161: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

General group properties

From the previous analysis, it emerges that the S(c , d) entropy fits intothe class SU in the two cases

I a) (c = 1, d ∈ N) (we can get rid of the constant term in theexpansion)

I b) (c > 0, d = 0).

I In the first case, we get that the S(c , d) is weakly composable.The group structure Φ(x , y) is generated by functions G (t) that arepolynomials. The explicit expression of G (t) can be explicitlycomputed, but it is usually cumbersome.

I For other cases, S(c , d) does not seem to be weakly composable.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 162: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

General group properties

From the previous analysis, it emerges that the S(c , d) entropy fits intothe class SU in the two cases

I a) (c = 1, d ∈ N) (we can get rid of the constant term in theexpansion)

I b) (c > 0, d = 0).

I In the first case, we get that the S(c , d) is weakly composable.The group structure Φ(x , y) is generated by functions G (t) that arepolynomials. The explicit expression of G (t) can be explicitlycomputed, but it is usually cumbersome.

I For other cases, S(c , d) does not seem to be weakly composable.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 163: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

General group properties

From the previous analysis, it emerges that the S(c , d) entropy fits intothe class SU in the two cases

I a) (c = 1, d ∈ N) (we can get rid of the constant term in theexpansion)

I b) (c > 0, d = 0).

I In the first case, we get that the S(c , d) is weakly composable.The group structure Φ(x , y) is generated by functions G (t) that arepolynomials. The explicit expression of G (t) can be explicitlycomputed, but it is usually cumbersome.

I For other cases, S(c , d) does not seem to be weakly composable.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 164: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

General group properties

From the previous analysis, it emerges that the S(c , d) entropy fits intothe class SU in the two cases

I a) (c = 1, d ∈ N) (we can get rid of the constant term in theexpansion)

I b) (c > 0, d = 0).

I In the first case, we get that the S(c , d) is weakly composable.The group structure Φ(x , y) is generated by functions G (t) that arepolynomials. The explicit expression of G (t) can be explicitlycomputed, but it is usually cumbersome.

I For other cases, S(c , d) does not seem to be weakly composable.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 165: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

E. P. Borges and I. Roditi, Phys. Lett. A 246 399–402 (1998).

I e) The Borges-Roditi entropy is a two-parametric entropy. Theassociated generalized logarithm reads

Loga,b(x) =xa − xb

a− b, (20)

which reproduces Abe’s entropy for a = σ − 1, b = σ−1 − 1. The related

group exponential is GA(t) = eat−ebt

a−b. The associated formal group is

known in the literature as the Abel formal group:

ΦA(x , y) = x + y + β1xy +∑j>i

βi(

xy i − x iy). (21)

The coefficients βn in (21) can be expressed as

βn =(−1)n−1

n!(n − 1)

∏i+j=n−1, i,j≥0

(ia + jb).

SBR [p] = kB

W∑i=1

pi

ln

1

pi+

1

2(a + b)

(ln

1

pi

)2

+1

6

(a2 + ab + b2

)(ln

1

pi

)3

. . .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 166: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

The Tsallis-Cirto entropy

C. Tsallis, Introduction to nonextensive statistical mechanics, Springer, 2009 ;C. Tsallis, L. Cirto, Eur. Phys. J. C 73, 2487 (2013); M. R. Ubriaco, Phys.Lett. A 373, 2516–2519 2009.f) The Sδ entropy reads

Sδ = kB

W∑i=1

pi

(ln

1

pi

)δ, 0 < δ ≤ (1 + ln W ). (22)

The underlying algebraic structure can be analyzed on the uniform distribution:Sδ is weakly composable. We get The case δ = 1 is the only strictlycomposable case. Also, Φ(x , y) does not define a group law over the reals,but simply a monoid, except for δ ∈ N, δ odd. A similar analysis can beperformed for the case of the entropic functional

Sq,δ = kB

W∑i=1

pi

(lnq

1

pi

)δ. (23)

It reduces to the Tsallis entropy for q ∈ R and δ = 1.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 167: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

g) Entropies from delta operators

As we have seen, from a class of discrete derivatives we can define, via theRota isomorphism, the generalized logarithm

LogG (x) =1

σ

m∑n=l

knxσn, l ,m ∈ Z, m − l = r > 0, x > 0. (24)

We can associate with these logarithms the class of entropies

SG (p) := kB

W∑i=1

piLogG

(1

pi

). (25)

where kn are real constants such thatm∑n=l

kn = 0,m∑n=l

nkn = 1, (26)

and km 6= 0, kl 6= 0. Conditions (26) ensure that limσ→0 LogG (x) = ln x .The class (25) is a representation of SU entropy, corresponding to the choice

G(t) =1

σ

m∑n=l

kn exp(nσt), l , m ∈ Z, m − l = r > 0, (27)

with the constraints (26).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 168: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

I Let us construct some new examples. Consider the delta operators

∆III = T−2T−1+T−2

σ, ∆IV =

T 2− 32T+ 3

2T−1−T−2

σ,

∆V = T 3−2T 2+2T−2T−1+T−2

−σ , . . .

I The corresponding logarithms are

LogGIII (x) =1

σ

(xσ − 2x−σ + x−2σ

),

LogGIV (x) =1

σ

(x2σ − 3

2xσ +

3

2x−σ − x−2σ

),

LogGV (x) =1

σ

(x3σ − 2x2σ + 2xσ − 2x−σ + x−2σ

).

I Consequently, we introduce the new entropic forms

SGIII (p) :=k

σ

N∑i=1

pi

(p2σi − 2pσi + p−σi

), (28)

SGIV (p) :=k

σ

N∑i=1

pi

(p−2σi − 3

2p−σi +

3

2pσi − p2σ

i

), (29)

SGV (p) :=k

σ

N∑i=1

pi

(p−3σi − 2p−2σ

i + 2p−σi − 2pσi + p2σi

), . . .

(30)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 169: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

I Let us construct some new examples. Consider the delta operators

∆III = T−2T−1+T−2

σ, ∆IV =

T 2− 32T+ 3

2T−1−T−2

σ,

∆V = T 3−2T 2+2T−2T−1+T−2

−σ , . . .

I The corresponding logarithms are

LogGIII (x) =1

σ

(xσ − 2x−σ + x−2σ

),

LogGIV (x) =1

σ

(x2σ − 3

2xσ +

3

2x−σ − x−2σ

),

LogGV (x) =1

σ

(x3σ − 2x2σ + 2xσ − 2x−σ + x−2σ

).

I Consequently, we introduce the new entropic forms

SGIII (p) :=k

σ

N∑i=1

pi

(p2σi − 2pσi + p−σi

), (28)

SGIV (p) :=k

σ

N∑i=1

pi

(p−2σi − 3

2p−σi +

3

2pσi − p2σ

i

), (29)

SGV (p) :=k

σ

N∑i=1

pi

(p−3σi − 2p−2σ

i + 2p−σi − 2pσi + p2σi

), . . .

(30)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 170: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The universal-group entropyMain properties of SU entropyRelation with other entropies

I Let us construct some new examples. Consider the delta operators

∆III = T−2T−1+T−2

σ, ∆IV =

T 2− 32T+ 3

2T−1−T−2

σ,

∆V = T 3−2T 2+2T−2T−1+T−2

−σ , . . .

I The corresponding logarithms are

LogGIII (x) =1

σ

(xσ − 2x−σ + x−2σ

),

LogGIV (x) =1

σ

(x2σ − 3

2xσ +

3

2x−σ − x−2σ

),

LogGV (x) =1

σ

(x3σ − 2x2σ + 2xσ − 2x−σ + x−2σ

).

I Consequently, we introduce the new entropic forms

SGIII (p) :=k

σ

N∑i=1

pi

(p2σi − 2pσi + p−σi

), (28)

SGIV (p) :=k

σ

N∑i=1

pi

(p−2σi − 3

2p−σi +

3

2pσi − p2σ

i

), (29)

SGV (p) :=k

σ

N∑i=1

pi

(p−3σi − 2p−2σ

i + 2p−σi − 2pσi + p2σi

), . . .

(30)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 171: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

Thermodynamics from the group structure

We shall discuss the maximization of group entropies under appropriateconstraints: we adopt a generalized maximum entropy principle. We willsee that the Legendre structure of classical thermodynamics ispreserved in the group-theoretical framework.Precisely, let

LogU [ε] = G (ln ε)

where G (t) is the power series (5), with the constraint (19). Consider anisolated system in a stationary state (microcanonical ensemble). Theoptimization of SU leads to the equal probability case, i.e. pi = 1/W ,∀i .Therefore, we have

SU [p] = kBLogUW , (31)

which reduces to the celebrated Boltzmann formula SBG = kB lnW inthe case of uncorrelated particles.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 172: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

On the Legendre structure

I Let us consider a system in thermal contact with a reservoir (canonicalensemble). We introduce the numbers εi , interpreted as the values of aphysically relevant observable, typically the value of the energy of thesystem in its ith state. Assume that pi (εi ) is a normalized andmonotonically decreasing distribution function of εi . The internal energyV in a given state is defined as V =

∑Wi=1 εipi (εi ).

I As usual, we shall study the variational problem of the existence of astationary distribution pi (ε). To this aim, we introduce the functional

L = SG [p]− α

[∑i

p(εi )− 1

]− β

[W∑i=1

εipi (εi )− V

], (32)

where α and β are Lagrange multipliers.

I The vanishing of the variational derivative of this functional with respectto the distribution pi provides the stationary solution

pi = γE

(−α− β(εi )

λ

), (33)

with γ a priori a normalization constant, and E(·) is an invertible function.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 173: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

On the Legendre structure

I Let us consider a system in thermal contact with a reservoir (canonicalensemble). We introduce the numbers εi , interpreted as the values of aphysically relevant observable, typically the value of the energy of thesystem in its ith state. Assume that pi (εi ) is a normalized andmonotonically decreasing distribution function of εi . The internal energyV in a given state is defined as V =

∑Wi=1 εipi (εi ).

I As usual, we shall study the variational problem of the existence of astationary distribution pi (ε). To this aim, we introduce the functional

L = SG [p]− α

[∑i

p(εi )− 1

]− β

[W∑i=1

εipi (εi )− V

], (32)

where α and β are Lagrange multipliers.

I The vanishing of the variational derivative of this functional with respectto the distribution pi provides the stationary solution

pi = γE

(−α− β(εi )

λ

), (33)

with γ a priori a normalization constant, and E(·) is an invertible function.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 174: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

On the Legendre structure

I Let us consider a system in thermal contact with a reservoir (canonicalensemble). We introduce the numbers εi , interpreted as the values of aphysically relevant observable, typically the value of the energy of thesystem in its ith state. Assume that pi (εi ) is a normalized andmonotonically decreasing distribution function of εi . The internal energyV in a given state is defined as V =

∑Wi=1 εipi (εi ).

I As usual, we shall study the variational problem of the existence of astationary distribution pi (ε). To this aim, we introduce the functional

L = SG [p]− α

[∑i

p(εi )− 1

]− β

[W∑i=1

εipi (εi )− V

], (32)

where α and β are Lagrange multipliers.

I The vanishing of the variational derivative of this functional with respectto the distribution pi provides the stationary solution

pi = γE

(−α− β(εi )

λ

), (33)

with γ a priori a normalization constant, and E(·) is an invertible function.Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 175: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

I G. Kaniadakis, M. Lissia and A. M. Scarfone, Physica A 340 (2004)41–49 ; Phys. Rev. E 71 (2005) 046128 ; S. Abe and S. Thurner,Europhys. Lett 81 (2008) 1004 ; G. Kaniadakis, Eur. Phys. J. B 70 3–13(2009).

Let SG = −∑W

i=1 piLogG (pi ). The condition for identifying E with ageneralized logarithm is

d

dpj[pjLogG (pj)] = λLogG

(pj

γ

).

I For a specific class of entropies (Kaniadakis, Sharma-Mittal, etc.), onecan construct an ad-hoc Legendre structure (by using Legendre andMassieu potentials):

LogG (Z) + βVG = SG , Z =W∑i=1

E(−α− βεi )). (34)

I The previous equation can be used to introduce a thermodynamicobservable T , which has the interpretation of a local temperature for anon-equilibrium metastable state. Precisely, we can define ∂SU

∂V = 1T.

I Usually, a more accurate selection of the constraints is necessary (see C.Tsallis, R. Mendes, A. R. Plastino Physica A, 261 534-554 (1998)).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 176: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

I G. Kaniadakis, M. Lissia and A. M. Scarfone, Physica A 340 (2004)41–49 ; Phys. Rev. E 71 (2005) 046128 ; S. Abe and S. Thurner,Europhys. Lett 81 (2008) 1004 ; G. Kaniadakis, Eur. Phys. J. B 70 3–13(2009).

Let SG = −∑W

i=1 piLogG (pi ). The condition for identifying E with ageneralized logarithm is

d

dpj[pjLogG (pj)] = λLogG

(pj

γ

).

I For a specific class of entropies (Kaniadakis, Sharma-Mittal, etc.), onecan construct an ad-hoc Legendre structure (by using Legendre andMassieu potentials):

LogG (Z) + βVG = SG , Z =W∑i=1

E(−α− βεi )). (34)

I The previous equation can be used to introduce a thermodynamicobservable T , which has the interpretation of a local temperature for anon-equilibrium metastable state. Precisely, we can define ∂SU

∂V = 1T.

I Usually, a more accurate selection of the constraints is necessary (see C.Tsallis, R. Mendes, A. R. Plastino Physica A, 261 534-554 (1998)).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 177: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

I G. Kaniadakis, M. Lissia and A. M. Scarfone, Physica A 340 (2004)41–49 ; Phys. Rev. E 71 (2005) 046128 ; S. Abe and S. Thurner,Europhys. Lett 81 (2008) 1004 ; G. Kaniadakis, Eur. Phys. J. B 70 3–13(2009).

Let SG = −∑W

i=1 piLogG (pi ). The condition for identifying E with ageneralized logarithm is

d

dpj[pjLogG (pj)] = λLogG

(pj

γ

).

I For a specific class of entropies (Kaniadakis, Sharma-Mittal, etc.), onecan construct an ad-hoc Legendre structure (by using Legendre andMassieu potentials):

LogG (Z) + βVG = SG , Z =W∑i=1

E(−α− βεi )). (34)

I The previous equation can be used to introduce a thermodynamicobservable T , which has the interpretation of a local temperature for anon-equilibrium metastable state. Precisely, we can define ∂SU

∂V = 1T.

I Usually, a more accurate selection of the constraints is necessary (see C.Tsallis, R. Mendes, A. R. Plastino Physica A, 261 534-554 (1998)).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 178: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

A new class of trace-form multidimensional entropies

How to design a trace-form entropy by using the notion ofuniversal-group entropy?

I I) Given a group law Φ(x , y) (a priori function of a set ofparameters), determine the function G such that the group law canbe represented in the form

Φ(x , y) = G (G−1(x) + G−1(y)).

I II) The corresponding entropy will be a specific representation ofSU [p], and is provided by

W∑i=1

piG

(ln

1

pi

)I III) Under mild hypotheses (concerning the set of parametersakn∈N), this entropy is weakly composable, satisfies the firstthree SK axioms, is extensive, possesses a Legendre structure,etc.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 179: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

A new class of trace-form multidimensional entropies

How to design a trace-form entropy by using the notion ofuniversal-group entropy?

I I) Given a group law Φ(x , y) (a priori function of a set ofparameters), determine the function G such that the group law canbe represented in the form

Φ(x , y) = G (G−1(x) + G−1(y)).

I II) The corresponding entropy will be a specific representation ofSU [p], and is provided by

W∑i=1

piG

(ln

1

pi

)

I III) Under mild hypotheses (concerning the set of parametersakn∈N), this entropy is weakly composable, satisfies the firstthree SK axioms, is extensive, possesses a Legendre structure,etc.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 180: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

A new class of trace-form multidimensional entropies

How to design a trace-form entropy by using the notion ofuniversal-group entropy?

I I) Given a group law Φ(x , y) (a priori function of a set ofparameters), determine the function G such that the group law canbe represented in the form

Φ(x , y) = G (G−1(x) + G−1(y)).

I II) The corresponding entropy will be a specific representation ofSU [p], and is provided by

W∑i=1

piG

(ln

1

pi

)I III) Under mild hypotheses (concerning the set of parametersakn∈N), this entropy is weakly composable, satisfies the firstthree SK axioms, is extensive, possesses a Legendre structure,etc.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 181: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

A new entropy associated with the rational group law

E. Curado, P. Tempesta, C. Tsallis, arxiv: 1507.05058 (2015)

I There is a remarkable example of rational group law:

S(A ∪ B) =S(A) + S(B) + aS(A)S(B)

1 + bS(A)S(B), (35)

where a, b ∈ R.

I The corresponding formal group is given by

ΦR(x , y) =x + y + axy

1 + bxy(36)

When a = b = 0, we recover the additive law; for b = 0, we recover themultiplicative case. Whenever b 6= 0, we have a genuinely new case.

I A specific one-parametric realization of it is

Φ(x , y) =x + y + (α− 1)xy

1 + αxy(37)

For α = −1, 0, 1, we obtain group laws respectively associated with theEuler characteristic, the Todd genus and the Hirzebruch L-genus.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 182: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

A new entropy associated with the rational group law

E. Curado, P. Tempesta, C. Tsallis, arxiv: 1507.05058 (2015)

I There is a remarkable example of rational group law:

S(A ∪ B) =S(A) + S(B) + aS(A)S(B)

1 + bS(A)S(B), (35)

where a, b ∈ R.

I The corresponding formal group is given by

ΦR(x , y) =x + y + axy

1 + bxy(36)

When a = b = 0, we recover the additive law; for b = 0, we recover themultiplicative case. Whenever b 6= 0, we have a genuinely new case.

I A specific one-parametric realization of it is

Φ(x , y) =x + y + (α− 1)xy

1 + αxy(37)

For α = −1, 0, 1, we obtain group laws respectively associated with theEuler characteristic, the Todd genus and the Hirzebruch L-genus.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 183: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

A new entropy associated with the rational group law

E. Curado, P. Tempesta, C. Tsallis, arxiv: 1507.05058 (2015)

I There is a remarkable example of rational group law:

S(A ∪ B) =S(A) + S(B) + aS(A)S(B)

1 + bS(A)S(B), (35)

where a, b ∈ R.

I The corresponding formal group is given by

ΦR(x , y) =x + y + axy

1 + bxy(36)

When a = b = 0, we recover the additive law; for b = 0, we recover themultiplicative case. Whenever b 6= 0, we have a genuinely new case.

I A specific one-parametric realization of it is

Φ(x , y) =x + y + (α− 1)xy

1 + αxy(37)

For α = −1, 0, 1, we obtain group laws respectively associated with theEuler characteristic, the Todd genus and the Hirzebruch L-genus.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 184: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

I It is easy to verify that ΦR(x , y) satisfies the group axioms. Also, itadmits an inverse, i.e. there exists a real function φ(x) such that

ΦR(x , φ(x)) = 0. (38)

For instance, for the case of the additive law, φ(x) = −x . For the rationalgroup law(35), we have φ(x) = −x/(1 + ax)

I Problem: how to construct the entropic form associated with therational group law (35)?

I We look for a function G(t), which a priori is a formal power series, suchthat

ΦR(x , y) = G(G−1(x) + G−1(y)). (39)

I The most general form of G(t) is

G(t) =∞∑k=0

Aktk+1

k + 1= A0t + A1

t2

2+ A2

t3

3+ . . . . (40)

The inverse G−1(s) can be computed by means of the Lagrange inversiontheorem. We get:

G−1(s) =s

A0− A1

2(A0)3s2+ (41)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 185: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

I It is easy to verify that ΦR(x , y) satisfies the group axioms. Also, itadmits an inverse, i.e. there exists a real function φ(x) such that

ΦR(x , φ(x)) = 0. (38)

For instance, for the case of the additive law, φ(x) = −x . For the rationalgroup law(35), we have φ(x) = −x/(1 + ax)

I Problem: how to construct the entropic form associated with therational group law (35)?

I We look for a function G(t), which a priori is a formal power series, suchthat

ΦR(x , y) = G(G−1(x) + G−1(y)). (39)

I The most general form of G(t) is

G(t) =∞∑k=0

Aktk+1

k + 1= A0t + A1

t2

2+ A2

t3

3+ . . . . (40)

The inverse G−1(s) can be computed by means of the Lagrange inversiontheorem. We get:

G−1(s) =s

A0− A1

2(A0)3s2+ (41)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 186: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

I It is easy to verify that ΦR(x , y) satisfies the group axioms. Also, itadmits an inverse, i.e. there exists a real function φ(x) such that

ΦR(x , φ(x)) = 0. (38)

For instance, for the case of the additive law, φ(x) = −x . For the rationalgroup law(35), we have φ(x) = −x/(1 + ax)

I Problem: how to construct the entropic form associated with therational group law (35)?

I We look for a function G(t), which a priori is a formal power series, suchthat

ΦR(x , y) = G(G−1(x) + G−1(y)). (39)

I The most general form of G(t) is

G(t) =∞∑k=0

Aktk+1

k + 1= A0t + A1

t2

2+ A2

t3

3+ . . . . (40)

The inverse G−1(s) can be computed by means of the Lagrange inversiontheorem. We get:

G−1(s) =s

A0− A1

2(A0)3s2+ (41)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 187: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

I It is easy to verify that ΦR(x , y) satisfies the group axioms. Also, itadmits an inverse, i.e. there exists a real function φ(x) such that

ΦR(x , φ(x)) = 0. (38)

For instance, for the case of the additive law, φ(x) = −x . For the rationalgroup law(35), we have φ(x) = −x/(1 + ax)

I Problem: how to construct the entropic form associated with therational group law (35)?

I We look for a function G(t), which a priori is a formal power series, suchthat

ΦR(x , y) = G(G−1(x) + G−1(y)). (39)

I The most general form of G(t) is

G(t) =∞∑k=0

Aktk+1

k + 1= A0t + A1

t2

2+ A2

t3

3+ . . . . (40)

The inverse G−1(s) can be computed by means of the Lagrange inversiontheorem. We get:

G−1(s) =s

A0− A1

2(A0)3s2+ (41)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 188: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

The expansion of the rational group law ΦR(x , y) is

ΦR(x , y) = x + y + axy − b(xy 2 + yx2)− abx2y 2

+ b2(x2y 3 + x3y 2) + ab2x3y 3 + higher order terms

Procedure. By computing the expression ΦR(x , y) = G(G−1(x) + G−1(y))

with the form G(t) = A0t + A1t2

2+ A2

t3

3+ . . . for the expansion of the formal

exponential, and identifying the terms appearing in this expansion with thosecoming from (42), we get an infinite set of relations for the coefficients Ak :

A0 ∈ RA1 = aA2

0

A2 =1

2

(a2 − 2b

)A3

0

A3 =1

3!

(a3 − 8ab

)A4

0

A4 =1

4!

(a4 − 22 a2b + 16 b2

)A5

0

A5 =1

5!

(a5 − 52 a3b + 136 ab2

)A6

0

... (42)

The coefficients Ak provide the most general solution to our problem.Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 189: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

Before proceeding further, let us consider the particular case b = 0. Then

Φ(x , y) = x + y + axy (43)

If we put b = 0, A0 = 1 in the previous coefficients (42), we getimmediately

ak =1

k!ak (44)

i.e.

G (t) =eat − 1

a, (45)

which is the correct form we were looking for! Indeed, using the

prescription S =∑W

i=1 piG(

ln 1pi

)and putting a = 1− q we get back the

Sq entropy

S =W∑i=1

pipq−1i − 1

1− q=

1−∑W

i=1 pqi

q − 1. (46)

The general case provides us with a series solution: indeed, wereconstruct G (t) and hence the entropy, term by term. However, thisseries can be re-summed up!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 190: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

I We obtain the closed form solution

G(t) =2(ert − 1)

−a(ert − 1)±√

a2 + 4b(ert + 1)(47)

In particular, the coefficient A0 is fixed to be

A0 = ± r√a2 + 4b

. (48)

Consequently, we have deduced the following expression for the entropyassociated with ΦR(x , y).

I Definition 2.7. The S(+)a,b,r entropy, for r > 0, is the function

S(+)a,b,r [p] =

W∑i=1

s(+)[pi ] :=W∑i=1

piLog(+)a,b,r

(1

pi

), (49)

where the generalized plus logarithm is defined as

Log(+)a,b,r (x) :=

2(x r − 1)

−a(x r − 1) +√

a2 + 4b (x r + 1). (50)

A similar expression holds for the S(−)a,b,r entropy.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 191: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

I We obtain the closed form solution

G(t) =2(ert − 1)

−a(ert − 1)±√

a2 + 4b(ert + 1)(47)

In particular, the coefficient A0 is fixed to be

A0 = ± r√a2 + 4b

. (48)

Consequently, we have deduced the following expression for the entropyassociated with ΦR(x , y).

I Definition 2.7. The S(+)a,b,r entropy, for r > 0, is the function

S(+)a,b,r [p] =

W∑i=1

s(+)[pi ] :=W∑i=1

piLog(+)a,b,r

(1

pi

), (49)

where the generalized plus logarithm is defined as

Log(+)a,b,r (x) :=

2(x r − 1)

−a(x r − 1) +√

a2 + 4b (x r + 1). (50)

A similar expression holds for the S(−)a,b,r entropy.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 192: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

Properties of the rational entropy S(+)a,b,r [p]

I Proposition. The entropies S(±)a,b,r reproduce the standard SBG entropy for

b = 0, a = r , in the limit r → 0:

limr→0

S(±)r,0,r [p] = SBG [p]. (51)

Proposition 2.8. The entropy S(+)a,b,r [p] satisfies the first three Khinchin

axioms.

I The extensivity problem

One of the main reasons to consider generalized entropies is the fact thatthey can be useful, or even mandatory, to describe systems with unusualbehavior. If an entropy is extensive, it essentially means that, for anoccupation law W = W (N) of the phase space associated with a givensystem, it is asymptotically proportional to N, the number of particles ofthe system. Precisely, SBG is extensive whenever W (N) ∼ kN , wherek ∈ R+ is a suitable constant. However, for substantially different choicesof W = W (N), this property is no longer true for SBG .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 193: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

Properties of the rational entropy S(+)a,b,r [p]

I Proposition. The entropies S(±)a,b,r reproduce the standard SBG entropy for

b = 0, a = r , in the limit r → 0:

limr→0

S(±)r,0,r [p] = SBG [p]. (51)

Proposition 2.8. The entropy S(+)a,b,r [p] satisfies the first three Khinchin

axioms.

I The extensivity problem

One of the main reasons to consider generalized entropies is the fact thatthey can be useful, or even mandatory, to describe systems with unusualbehavior. If an entropy is extensive, it essentially means that, for anoccupation law W = W (N) of the phase space associated with a givensystem, it is asymptotically proportional to N, the number of particles ofthe system. Precisely, SBG is extensive whenever W (N) ∼ kN , wherek ∈ R+ is a suitable constant. However, for substantially different choicesof W = W (N), this property is no longer true for SBG .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 194: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

The extensivity problem

I A natural question is to ascertain whether the new entropy we propose inthis paper is extensive. Its group structure, once again, ensures that thisproperty holds for a suitable asymptotic occupation law W = W (N) ofphase space. A general result of the group-theoretical approach is that a

sufficient condition for an entropy of the form S =∑W

i=1 piG(

ln 1pi

)to

be extensive is thatln W (N) ∼ G−1(N), (52)

provided that W (N) be real and defined for all N ∈ N, with W (N)→∞for N →∞. These requirements usually restrict the space of parameters.

I In our case, we observe that when pi = 1/W for all i = 1, · · · ,W , both

entropies S(±)a,b,r tend to the limit value

S(±)a,b,r [1/W ]→ 2√

a2 + 4b − a, (53)

if limN→∞W (N) =∞. In particular, for b → 0, the entropy diverges; asa consequence of the previous discussion, there exists a regime ofextensivity, for W (N) ∼ Nγ , with γ = 1/a.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 195: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

The extensivity problem

I A natural question is to ascertain whether the new entropy we propose inthis paper is extensive. Its group structure, once again, ensures that thisproperty holds for a suitable asymptotic occupation law W = W (N) ofphase space. A general result of the group-theoretical approach is that a

sufficient condition for an entropy of the form S =∑W

i=1 piG(

ln 1pi

)to

be extensive is thatln W (N) ∼ G−1(N), (52)

provided that W (N) be real and defined for all N ∈ N, with W (N)→∞for N →∞. These requirements usually restrict the space of parameters.

I In our case, we observe that when pi = 1/W for all i = 1, · · · ,W , both

entropies S(±)a,b,r tend to the limit value

S(±)a,b,r [1/W ]→ 2√

a2 + 4b − a, (53)

if limN→∞W (N) =∞. In particular, for b → 0, the entropy diverges; asa consequence of the previous discussion, there exists a regime ofextensivity, for W (N) ∼ Nγ , with γ = 1/a.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 196: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

I If limN→∞W (N) = c ∈ R+, with c > 1, these entropies tend to the value

S(±)a,b,r [1/c]→ 2 (c r − 1)

±√

a2 + 4b (c r + 1)− a (c r − 1). (54)

I In both cases, if b 6= 0 the limiting value is finite, independently onwhether W diverges or tends to a constant for N →∞. Consequently, forb 6= 0 these entropies, albeit monotonically increasing functions of W , cannot be extensive.

I A natural question emerges, concerning the kind of systems such type ofentropies could be useful for.

I A possible answer is that the present formalism could be relevantwhenever treating systems highly correlated, where the addition of anew degree of freedom essentially does not change the value of theentropy, for a large number of degrees of freedom.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 197: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

I If limN→∞W (N) = c ∈ R+, with c > 1, these entropies tend to the value

S(±)a,b,r [1/c]→ 2 (c r − 1)

±√

a2 + 4b (c r + 1)− a (c r − 1). (54)

I In both cases, if b 6= 0 the limiting value is finite, independently onwhether W diverges or tends to a constant for N →∞. Consequently, forb 6= 0 these entropies, albeit monotonically increasing functions of W , cannot be extensive.

I A natural question emerges, concerning the kind of systems such type ofentropies could be useful for.

I A possible answer is that the present formalism could be relevantwhenever treating systems highly correlated, where the addition of anew degree of freedom essentially does not change the value of theentropy, for a large number of degrees of freedom.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 198: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

I If limN→∞W (N) = c ∈ R+, with c > 1, these entropies tend to the value

S(±)a,b,r [1/c]→ 2 (c r − 1)

±√

a2 + 4b (c r + 1)− a (c r − 1). (54)

I In both cases, if b 6= 0 the limiting value is finite, independently onwhether W diverges or tends to a constant for N →∞. Consequently, forb 6= 0 these entropies, albeit monotonically increasing functions of W , cannot be extensive.

I A natural question emerges, concerning the kind of systems such type ofentropies could be useful for.

I A possible answer is that the present formalism could be relevantwhenever treating systems highly correlated, where the addition of anew degree of freedom essentially does not change the value of theentropy, for a large number of degrees of freedom.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 199: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

I If limN→∞W (N) = c ∈ R+, with c > 1, these entropies tend to the value

S(±)a,b,r [1/c]→ 2 (c r − 1)

±√

a2 + 4b (c r + 1)− a (c r − 1). (54)

I In both cases, if b 6= 0 the limiting value is finite, independently onwhether W diverges or tends to a constant for N →∞. Consequently, forb 6= 0 these entropies, albeit monotonically increasing functions of W , cannot be extensive.

I A natural question emerges, concerning the kind of systems such type ofentropies could be useful for.

I A possible answer is that the present formalism could be relevantwhenever treating systems highly correlated, where the addition of anew degree of freedom essentially does not change the value of theentropy, for a large number of degrees of freedom.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 200: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

I One can also consider different scenarios, borrowed from socialsciences, where no thermodynamical or energetical aspects areinvolved, and extensivity is a priori not required. Again suchentropies, that increase very little with the addition of new degreesof freedom, could be play a relevant role in describing situationswhere the amount of information tends to stabilize,irrespectively of the increase of new agents involved in theinformation exchange.

I Consequently, the multiparametric entropy Sa,b,r is compatible withboth scenarios: the standard one, where an increase of the numbersof degrees of freedom converts into an increase of the entropy, andthe “anomalous” one, where an increase of the number of particlesessentially freezes the system, by confining it in the phase space.Excepting for Sq for q > 1, this flexibility in the limit W →∞ isseemingly not shared by the entropies typically used in the literature.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 201: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

A new class of trace-form multidimensional entropiesThe reconstruction procedureThe general expressionMain properties: the extensivity problem

I One can also consider different scenarios, borrowed from socialsciences, where no thermodynamical or energetical aspects areinvolved, and extensivity is a priori not required. Again suchentropies, that increase very little with the addition of new degreesof freedom, could be play a relevant role in describing situationswhere the amount of information tends to stabilize,irrespectively of the increase of new agents involved in theinformation exchange.

I Consequently, the multiparametric entropy Sa,b,r is compatible withboth scenarios: the standard one, where an increase of the numbersof degrees of freedom converts into an increase of the entropy, andthe “anomalous” one, where an increase of the number of particlesessentially freezes the system, by confining it in the phase space.Excepting for Sq for q > 1, this flexibility in the limit W →∞ isseemingly not shared by the entropies typically used in the literature.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 202: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The microcanonical description and associated thermodynamicsMultiparametric entropies

A magical recipe to generate multiparametric entropies

P. T., Proc. Royal Society A, to appear (2015).

I Definition 2.9. Let f : [0, 1)→ R be a C∞(0, 1) function,continuous in [0, 1), such that

1− af ′(axσ)− a2xσ σ1+σ f

′′(axσ)

1− af ′(a)> 0 (55)

for x ∈ [0, 1], a ∈ (0, la), σ ∈ (0, lσ), with la, lσ ∈ (0, 1). Then f willbe said to be an admissible function.We shall denote by A the set of admissible functions. The values ofla, lσ usually will depend on the choice of f .

I Definition 5 makes sense for functions f at least of class C2(0, 1).The regularity C∞(0, 1) is required for later convenience.

I Notice that the condition 1− af ′(a) 6= 0 is implicit in Definition 5.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 203: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The microcanonical description and associated thermodynamicsMultiparametric entropies

A magical recipe to generate multiparametric entropies

P. T., Proc. Royal Society A, to appear (2015).

I Definition 2.9. Let f : [0, 1)→ R be a C∞(0, 1) function,continuous in [0, 1), such that

1− af ′(axσ)− a2xσ σ1+σ f

′′(axσ)

1− af ′(a)> 0 (55)

for x ∈ [0, 1], a ∈ (0, la), σ ∈ (0, lσ), with la, lσ ∈ (0, 1). Then f willbe said to be an admissible function.We shall denote by A the set of admissible functions. The values ofla, lσ usually will depend on the choice of f .

I Definition 5 makes sense for functions f at least of class C2(0, 1).The regularity C∞(0, 1) is required for later convenience.

I Notice that the condition 1− af ′(a) 6= 0 is implicit in Definition 5.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 204: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The microcanonical description and associated thermodynamicsMultiparametric entropies

A magical recipe to generate multiparametric entropies

P. T., Proc. Royal Society A, to appear (2015).

I Definition 2.9. Let f : [0, 1)→ R be a C∞(0, 1) function,continuous in [0, 1), such that

1− af ′(axσ)− a2xσ σ1+σ f

′′(axσ)

1− af ′(a)> 0 (55)

for x ∈ [0, 1], a ∈ (0, la), σ ∈ (0, lσ), with la, lσ ∈ (0, 1). Then f willbe said to be an admissible function.We shall denote by A the set of admissible functions. The values ofla, lσ usually will depend on the choice of f .

I Definition 5 makes sense for functions f at least of class C2(0, 1).The regularity C∞(0, 1) is required for later convenience.

I Notice that the condition 1− af ′(a) 6= 0 is implicit in Definition 5.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 205: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The microcanonical description and associated thermodynamicsMultiparametric entropies

I Definition 2.10 Let pii=1,··· ,W , with W ≥ 1,∑W

i=1 pi = 1, be a discreteprobability distribution. Let f be an admissible function. The functional

Sf [p] =kB

1− af ′(a)

W∑i=1

pi

(f (apσi )− pσi

σ+

1− f (a)

σ

)(56)

for a ∈ (0, la), σ ∈ (0, lσ), with la, lσ ∈ (0, 1), will be called the entropyassociated with f .

I Theorem 2.11 For any choice of f ∈ A, Sf [p] satisfies the firstthree SK axioms.

I Theorem 2.12 For any f ∈ A, Sf is weakly composable. In

particular, the function G (t) such that S =∑

i piG(

ln 1pi

)is

G(t) = t − σ a2f ′′(a) + af ′(a)− 1

2(−1 + af ′(a))t2 +

σ2 a3f ′′′(a) + 3a2f ′′(a) + af ′(a)− 1

6(−1 + af ′(a))t3 + ... =

∞∑k=1

γktk

k + 1,

with γ0 = 1, γ1 = σ a2f ′′(a)+af ′(a)−12(1−af ′(a)) , etc.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 206: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The microcanonical description and associated thermodynamicsMultiparametric entropies

I Definition 2.10 Let pii=1,··· ,W , with W ≥ 1,∑W

i=1 pi = 1, be a discreteprobability distribution. Let f be an admissible function. The functional

Sf [p] =kB

1− af ′(a)

W∑i=1

pi

(f (apσi )− pσi

σ+

1− f (a)

σ

)(56)

for a ∈ (0, la), σ ∈ (0, lσ), with la, lσ ∈ (0, 1), will be called the entropyassociated with f .

I Theorem 2.11 For any choice of f ∈ A, Sf [p] satisfies the firstthree SK axioms.

I Theorem 2.12 For any f ∈ A, Sf is weakly composable. In

particular, the function G (t) such that S =∑

i piG(

ln 1pi

)is

G(t) = t − σ a2f ′′(a) + af ′(a)− 1

2(−1 + af ′(a))t2 +

σ2 a3f ′′′(a) + 3a2f ′′(a) + af ′(a)− 1

6(−1 + af ′(a))t3 + ... =

∞∑k=1

γktk

k + 1,

with γ0 = 1, γ1 = σ a2f ′′(a)+af ′(a)−12(1−af ′(a)) , etc.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 207: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The microcanonical description and associated thermodynamicsMultiparametric entropies

I Definition 2.10 Let pii=1,··· ,W , with W ≥ 1,∑W

i=1 pi = 1, be a discreteprobability distribution. Let f be an admissible function. The functional

Sf [p] =kB

1− af ′(a)

W∑i=1

pi

(f (apσi )− pσi

σ+

1− f (a)

σ

)(56)

for a ∈ (0, la), σ ∈ (0, lσ), with la, lσ ∈ (0, 1), will be called the entropyassociated with f .

I Theorem 2.11 For any choice of f ∈ A, Sf [p] satisfies the firstthree SK axioms.

I Theorem 2.12 For any f ∈ A, Sf is weakly composable. In

particular, the function G (t) such that S =∑

i piG(

ln 1pi

)is

G(t) = t − σ a2f ′′(a) + af ′(a)− 1

2(−1 + af ′(a))t2 +

σ2 a3f ′′′(a) + 3a2f ′′(a) + af ′(a)− 1

6(−1 + af ′(a))t3 + ... =

∞∑k=1

γktk

k + 1,

with γ0 = 1, γ1 = σ a2f ′′(a)+af ′(a)−12(1−af ′(a)) , etc.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 208: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The microcanonical description and associated thermodynamicsMultiparametric entropies

We shall briefly describe the thermodynamic properties of the entropiesdiscussed above. Despite of their different functional form, they shareseveral relevant features. Let us first introduce the generalizedlogarithm

Logf (x) :=1

1− af ′(a)

(f (ax−σ)− x−σ

σ+

1− f (a)

σ

). (57)

Equation (56) takes the usual form

Sf [p] = kB

W∑i=1

piLogf

(1

pi

).

In the micro-canonical ensemble, all microstates have equal probabilitypi = 1/W . In this case, Eq. (56) becomes

Sf [W ] := kBLog[f ][W ].

It reduces to the celebrated Boltzmann formula SBG = k lnW whenσ → 0. All the discussion valid for the UGE concerning the existence of aLegendre structure applies also to the class Sf .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 209: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The microcanonical description and associated thermodynamicsMultiparametric entropies

New entropic forms from the recipe

Definition 2.13. The generalized Kaniadakis entropy is the function

S1/x [p] =kB2

W∑i=1

pi

(αp−σi − pσi

σ

). (58)

Indeed, it can be considered as a two-parametric version of Kaniadakisentropy, due to the presence of the extra parameter α = 1/a. It coincideswith the standard Kaniadakis entropy for α = 1 and σ = κ, κ ∈ (−1, 1).The exponential entropy

Sexp[p] :=kB

1− aea

W∑i=1

pi

(eap

σi − pσiσ

+1− ea

σ

)(59)

is defined for a ∈ (0, 1/2), σ ∈ (0, 1/2).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 210: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The microcanonical description and associated thermodynamicsMultiparametric entropies

The sin-entropy is the functional

Ssin[p] :=kB

1− a cos a

W∑i=1

pi

(sin apσi − pσi

σ+

1− sin a

σ

)(60)

with a ∈ (0, 1), σ ∈ (0, 1).

Is it possible to “compose” old entropies to get new ones?For instance, in the theory of modular forms, there is a procedure to passfrom

“old” modular forms → newforms

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 211: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The microcanonical description and associated thermodynamicsMultiparametric entropies

Composition theorem

I Theorem 2.14 Assume that f ∈ A. Let

Sf [p] :=kB

1− a1f ′(a1)

W∑i=1

pi

(f (a1pσ1i )− pσ1i

σ1+

1− f (a1)

σ1

),

with a1 ∈ (0, la1), σ1 ∈ (0, lσ1) be the entropy associated with f .

I Then the function

s(x) :=x

1− a1f ′(a1)

(f (a1xσ1)− xσ1

σ1+

1− f (a1)

σ1

)is an admissible function (it belongs to ∈ A) and

I

Ss [p] :=kB

1− a2s ′(a2)

W∑i=1

pi

(s(a2pσ2i )− pσ2i

σ2+

1− s(a2)

σ2

)(61)

with a2 ∈ (0, la2), σ2 ∈ (0, lσ2), la2 , lσ2 ∈ (0, 1) is also an entropy,associated with s(x), and depending on the four parameters(a1, a2, σ1, σ2).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 212: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The microcanonical description and associated thermodynamicsMultiparametric entropies

Composition theorem

I Theorem 2.14 Assume that f ∈ A. Let

Sf [p] :=kB

1− a1f ′(a1)

W∑i=1

pi

(f (a1pσ1i )− pσ1i

σ1+

1− f (a1)

σ1

),

with a1 ∈ (0, la1), σ1 ∈ (0, lσ1) be the entropy associated with f .

I Then the function

s(x) :=x

1− a1f ′(a1)

(f (a1xσ1)− xσ1

σ1+

1− f (a1)

σ1

)is an admissible function (it belongs to ∈ A) and

I

Ss [p] :=kB

1− a2s ′(a2)

W∑i=1

pi

(s(a2pσ2i )− pσ2i

σ2+

1− s(a2)

σ2

)(61)

with a2 ∈ (0, la2), σ2 ∈ (0, lσ2), la2 , lσ2 ∈ (0, 1) is also an entropy,associated with s(x), and depending on the four parameters(a1, a2, σ1, σ2).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 213: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

The microcanonical description and associated thermodynamicsMultiparametric entropies

Composition theorem

I Theorem 2.14 Assume that f ∈ A. Let

Sf [p] :=kB

1− a1f ′(a1)

W∑i=1

pi

(f (a1pσ1i )− pσ1i

σ1+

1− f (a1)

σ1

),

with a1 ∈ (0, la1), σ1 ∈ (0, lσ1) be the entropy associated with f .

I Then the function

s(x) :=x

1− a1f ′(a1)

(f (a1xσ1)− xσ1

σ1+

1− f (a1)

σ1

)is an admissible function (it belongs to ∈ A) and

I

Ss [p] :=kB

1− a2s ′(a2)

W∑i=1

pi

(s(a2pσ2i )− pσ2i

σ2+

1− s(a2)

σ2

)(61)

with a2 ∈ (0, la2), σ2 ∈ (0, lσ2), la2 , lσ2 ∈ (0, 1) is also an entropy,associated with s(x), and depending on the four parameters(a1, a2, σ1, σ2).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 214: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

On the asymptotic behaviour of generalized entropies

I From a mathematical point of view, the study of the asymptoticbehaviour of a given entropy, in the limit of large size systems, is not awell defined task, even for the standard Boltzmann-Gibbs entropy.

I Let W be the number of states admitted by a system. We shall focus onthe case of large size systems (which implies W →∞). An entropy isthen a functional S = S [p] defined on the space P∞.

I A simple argument proves that, depending on the choice of thedistribution in P∞, we can get infinitely many different limits even for theSBG case.

Indeed, consider the probability distribution p = (1, 0, 0, . . .) withinfinitely many entries. Then SBG [p] = 0. If p = (1/2, 1/2, 0, 0, . . .),then SBG [p] = ln 2 (in units of kB). Let N = 1080 (i.e., the estimatednumber of atoms in the observable Universe), andp = (1/N , 1/N , . . . , 1/N︸ ︷︷ ︸

N−times

, 0, 0, . . .), then SBG [p] = 80 · ln 10 ' 184.2.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 215: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

On the asymptotic behaviour of generalized entropies

I From a mathematical point of view, the study of the asymptoticbehaviour of a given entropy, in the limit of large size systems, is not awell defined task, even for the standard Boltzmann-Gibbs entropy.

I Let W be the number of states admitted by a system. We shall focus onthe case of large size systems (which implies W →∞). An entropy isthen a functional S = S [p] defined on the space P∞.

I A simple argument proves that, depending on the choice of thedistribution in P∞, we can get infinitely many different limits even for theSBG case.

Indeed, consider the probability distribution p = (1, 0, 0, . . .) withinfinitely many entries. Then SBG [p] = 0. If p = (1/2, 1/2, 0, 0, . . .),then SBG [p] = ln 2 (in units of kB). Let N = 1080 (i.e., the estimatednumber of atoms in the observable Universe), andp = (1/N , 1/N , . . . , 1/N︸ ︷︷ ︸

N−times

, 0, 0, . . .), then SBG [p] = 80 · ln 10 ' 184.2.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 216: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

On the asymptotic behaviour of generalized entropies

I From a mathematical point of view, the study of the asymptoticbehaviour of a given entropy, in the limit of large size systems, is not awell defined task, even for the standard Boltzmann-Gibbs entropy.

I Let W be the number of states admitted by a system. We shall focus onthe case of large size systems (which implies W →∞). An entropy isthen a functional S = S [p] defined on the space P∞.

I A simple argument proves that, depending on the choice of thedistribution in P∞, we can get infinitely many different limits even for theSBG case.

Indeed, consider the probability distribution p = (1, 0, 0, . . .) withinfinitely many entries. Then SBG [p] = 0. If p = (1/2, 1/2, 0, 0, . . .),then SBG [p] = ln 2 (in units of kB). Let N = 1080 (i.e., the estimatednumber of atoms in the observable Universe), andp = (1/N , 1/N , . . . , 1/N︸ ︷︷ ︸

N−times

, 0, 0, . . .), then SBG [p] = 80 · ln 10 ' 184.2.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 217: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

I From the point of view of probability and information theory, there is noway to have uniqueness of the limit on the full space P∞.

I In the domain of classical thermodynamics, a priori the same objectionapplies. However, if we accept to restrict to the important case of theuniform distribution (which is not the only physically interesting case),then one can define properly a thermodynamic limit of an entropy onthe uniform distribution. From a physical perspective, this wouldcorrespond to the case of a double limit, both of large system size and oflarge times.

I The only meaningful way to compare the behaviour of different entropiesis to compute them all in the same regime. Once we accept to restrict touniform probabilities, then the known entropies assume the form of thefollowing asymptotic functions: either (lnW )a or W b (possibly multipliedby parameters), or the product of these forms. Presently, we cannotexclude that other forms could also be possible.

This is a consequence of the analysis of scaling laws performed in R.Hanel and S. Thurner, Europhys. Lett. 93 20006 (2011). .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 218: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

I From the point of view of probability and information theory, there is noway to have uniqueness of the limit on the full space P∞.

I In the domain of classical thermodynamics, a priori the same objectionapplies. However, if we accept to restrict to the important case of theuniform distribution (which is not the only physically interesting case),then one can define properly a thermodynamic limit of an entropy onthe uniform distribution. From a physical perspective, this wouldcorrespond to the case of a double limit, both of large system size and oflarge times.

I The only meaningful way to compare the behaviour of different entropiesis to compute them all in the same regime. Once we accept to restrict touniform probabilities, then the known entropies assume the form of thefollowing asymptotic functions: either (lnW )a or W b (possibly multipliedby parameters), or the product of these forms. Presently, we cannotexclude that other forms could also be possible.

This is a consequence of the analysis of scaling laws performed in R.Hanel and S. Thurner, Europhys. Lett. 93 20006 (2011). .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 219: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

I From the point of view of probability and information theory, there is noway to have uniqueness of the limit on the full space P∞.

I In the domain of classical thermodynamics, a priori the same objectionapplies. However, if we accept to restrict to the important case of theuniform distribution (which is not the only physically interesting case),then one can define properly a thermodynamic limit of an entropy onthe uniform distribution. From a physical perspective, this wouldcorrespond to the case of a double limit, both of large system size and oflarge times.

I The only meaningful way to compare the behaviour of different entropiesis to compute them all in the same regime. Once we accept to restrict touniform probabilities, then the known entropies assume the form of thefollowing asymptotic functions: either (lnW )a or W b (possibly multipliedby parameters), or the product of these forms. Presently, we cannotexclude that other forms could also be possible.

This is a consequence of the analysis of scaling laws performed in R.Hanel and S. Thurner, Europhys. Lett. 93 20006 (2011). .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 220: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

I The problem of “comparing” entropies in the large size limit finds its mostsimple answer in the regime, if exists, where they are extensive. Inclassical thermodynamics, the only reason to consider generalizedentropies is the fact that they can be extensive in regimes where the SBG

entropy is not. This is the truly important limiting property.

I What is crucial is that all admissible entropies have the same behaviorin the regime where they are extensive, i.e. proportional to the number Nof particles of the system.

I Therefore, it is not surprising that different entropies could share the sameasymptotic behavior.

When we are interested in more general contexts as probability theory andinformation theory, then all of the infinitely many possible distributions area priori relevant. Therefore, according to the previous discussion, thecomparison among asymptotic behaviors of different entropies loses itsmeaning.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 221: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

I The problem of “comparing” entropies in the large size limit finds its mostsimple answer in the regime, if exists, where they are extensive. Inclassical thermodynamics, the only reason to consider generalizedentropies is the fact that they can be extensive in regimes where the SBG

entropy is not. This is the truly important limiting property.

I What is crucial is that all admissible entropies have the same behaviorin the regime where they are extensive, i.e. proportional to the number Nof particles of the system.

I Therefore, it is not surprising that different entropies could share the sameasymptotic behavior.

When we are interested in more general contexts as probability theory andinformation theory, then all of the infinitely many possible distributions area priori relevant. Therefore, according to the previous discussion, thecomparison among asymptotic behaviors of different entropies loses itsmeaning.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 222: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

I The problem of “comparing” entropies in the large size limit finds its mostsimple answer in the regime, if exists, where they are extensive. Inclassical thermodynamics, the only reason to consider generalizedentropies is the fact that they can be extensive in regimes where the SBG

entropy is not. This is the truly important limiting property.

I What is crucial is that all admissible entropies have the same behaviorin the regime where they are extensive, i.e. proportional to the number Nof particles of the system.

I Therefore, it is not surprising that different entropies could share the sameasymptotic behavior.

When we are interested in more general contexts as probability theory andinformation theory, then all of the infinitely many possible distributions area priori relevant. Therefore, according to the previous discussion, thecomparison among asymptotic behaviors of different entropies loses itsmeaning.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 223: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

State of Art

I The correspondence among groups and entropies leads naturally to thenotion of universal-group entropy SU , as the entropy associated with theLazard universal formal group.

I It includes a huge class of entropies, satisfying the first three SK axioms,and the composability axiom at least in the weak sense.

I The Legendre structure of the group entropies is not recovered fromclassical constraints; non standard ones should be introduced (openproblem).

I A simple reconstruction procedure can be defined, allowing to define thetrace-form entropy associated with a group law.

I Multiparametric entropies can be introduced from a “recipe”; newentropies can be generated from old ones.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 224: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

State of Art

I The correspondence among groups and entropies leads naturally to thenotion of universal-group entropy SU , as the entropy associated with theLazard universal formal group.

I It includes a huge class of entropies, satisfying the first three SK axioms,and the composability axiom at least in the weak sense.

I The Legendre structure of the group entropies is not recovered fromclassical constraints; non standard ones should be introduced (openproblem).

I A simple reconstruction procedure can be defined, allowing to define thetrace-form entropy associated with a group law.

I Multiparametric entropies can be introduced from a “recipe”; newentropies can be generated from old ones.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 225: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

State of Art

I The correspondence among groups and entropies leads naturally to thenotion of universal-group entropy SU , as the entropy associated with theLazard universal formal group.

I It includes a huge class of entropies, satisfying the first three SK axioms,and the composability axiom at least in the weak sense.

I The Legendre structure of the group entropies is not recovered fromclassical constraints; non standard ones should be introduced (openproblem).

I A simple reconstruction procedure can be defined, allowing to define thetrace-form entropy associated with a group law.

I Multiparametric entropies can be introduced from a “recipe”; newentropies can be generated from old ones.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 226: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

State of Art

I The correspondence among groups and entropies leads naturally to thenotion of universal-group entropy SU , as the entropy associated with theLazard universal formal group.

I It includes a huge class of entropies, satisfying the first three SK axioms,and the composability axiom at least in the weak sense.

I The Legendre structure of the group entropies is not recovered fromclassical constraints; non standard ones should be introduced (openproblem).

I A simple reconstruction procedure can be defined, allowing to define thetrace-form entropy associated with a group law.

I Multiparametric entropies can be introduced from a “recipe”; newentropies can be generated from old ones.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 227: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

State of Art

I The correspondence among groups and entropies leads naturally to thenotion of universal-group entropy SU , as the entropy associated with theLazard universal formal group.

I It includes a huge class of entropies, satisfying the first three SK axioms,and the composability axiom at least in the weak sense.

I The Legendre structure of the group entropies is not recovered fromclassical constraints; non standard ones should be introduced (openproblem).

I A simple reconstruction procedure can be defined, allowing to define thetrace-form entropy associated with a group law.

I Multiparametric entropies can be introduced from a “recipe”; newentropies can be generated from old ones.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 228: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

Generalized Entropy

Lazard’s universal

formal group

and

universal-group

Entropy

Generalized Entropy Group law

(weak or strong sense)

Direct Approach Inverse Approach

Group law

Composability

axiom

Reconstruction

procedure

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 229: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

The Universal Group EntropyDistribution functions and thermodynamic properties

An entropy for the rational group lawA “magical recipe” to generate multiparametric entropies

On the asymptotic behaviour of generalized entropiesState of Art

Thank you!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 230: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

GROUPS, ENTROPIES AND NUMBERTHEORY

Piergiulio Tempesta

Universidad Complutense de Madridand

Instituto de Ciencias Matematicas (ICMAT),Madrid, Spain.

TEMPLETON SCHOOL ON FOUNDATIONS OF COMPLEXITY

October 14 - 15, 2015

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 231: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

LECTURE III

Information Theory of group entropies:Renyi’s entropy and

the Z-class of Entropies

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 232: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

OutlineInformation-theoretical aspects of the notion of entropy

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

The Z-class of EntropiesThe composability problem revisitedThe Z-entropies: main definitions

Main properties of the Z-entropiesConcavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-MittalentropiesZ-delta entropiesInformation-theoretical content: Z-divergences

State of ArtPiergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 233: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Information Theory and Entropy

I H. Nyquist, Bell Syst. Tech. J., 3, 324 (1924).R. V. Hartley, Bell Syst. Tech. J. 7, 535 (1928).

I A. Kolmogorov, Atti della R. Acc. Naz. Lincei, 12, 388 (1930)M. Nagumo, Jap. J. Math. 7, 71 (1930)

I C. Shannon, Bell Syst. Tech. J. 27, 379 (1948); 27, 623 (1948),Shannon, W. Weaver, The mathematical Theory of Communication,University of Illinois Press, Urbana, IL, 1949.

I A. I. Khinchin, Mathematical Foundations of Information Theory,Dover, New York, 1957.

I E. T. Jaynes, Information theory and Statistical Mechanics. Part I,Phys. Rev. 106, 620 (1957); Part II, Phys. Rev. 108, 171 (1957).

I S. Kullback, Information Theory and Statistics, Wiley, 1959.I A. Renyi, Probability Theory, North–Holland, Amsterdam, 1970.I B. B. Mandelbrot, Fractal-form. Chance and Dimension, Freeman,

1977.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 234: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Information Theory and Entropy

I H. Nyquist, Bell Syst. Tech. J., 3, 324 (1924).R. V. Hartley, Bell Syst. Tech. J. 7, 535 (1928).

I A. Kolmogorov, Atti della R. Acc. Naz. Lincei, 12, 388 (1930)M. Nagumo, Jap. J. Math. 7, 71 (1930)

I C. Shannon, Bell Syst. Tech. J. 27, 379 (1948); 27, 623 (1948),Shannon, W. Weaver, The mathematical Theory of Communication,University of Illinois Press, Urbana, IL, 1949.

I A. I. Khinchin, Mathematical Foundations of Information Theory,Dover, New York, 1957.

I E. T. Jaynes, Information theory and Statistical Mechanics. Part I,Phys. Rev. 106, 620 (1957); Part II, Phys. Rev. 108, 171 (1957).

I S. Kullback, Information Theory and Statistics, Wiley, 1959.I A. Renyi, Probability Theory, North–Holland, Amsterdam, 1970.I B. B. Mandelbrot, Fractal-form. Chance and Dimension, Freeman,

1977.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 235: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Information Theory and Entropy

I H. Nyquist, Bell Syst. Tech. J., 3, 324 (1924).R. V. Hartley, Bell Syst. Tech. J. 7, 535 (1928).

I A. Kolmogorov, Atti della R. Acc. Naz. Lincei, 12, 388 (1930)M. Nagumo, Jap. J. Math. 7, 71 (1930)

I C. Shannon, Bell Syst. Tech. J. 27, 379 (1948); 27, 623 (1948),Shannon, W. Weaver, The mathematical Theory of Communication,University of Illinois Press, Urbana, IL, 1949.

I A. I. Khinchin, Mathematical Foundations of Information Theory,Dover, New York, 1957.

I E. T. Jaynes, Information theory and Statistical Mechanics. Part I,Phys. Rev. 106, 620 (1957); Part II, Phys. Rev. 108, 171 (1957).

I S. Kullback, Information Theory and Statistics, Wiley, 1959.I A. Renyi, Probability Theory, North–Holland, Amsterdam, 1970.I B. B. Mandelbrot, Fractal-form. Chance and Dimension, Freeman,

1977.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 236: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Information Theory and Entropy

I H. Nyquist, Bell Syst. Tech. J., 3, 324 (1924).R. V. Hartley, Bell Syst. Tech. J. 7, 535 (1928).

I A. Kolmogorov, Atti della R. Acc. Naz. Lincei, 12, 388 (1930)M. Nagumo, Jap. J. Math. 7, 71 (1930)

I C. Shannon, Bell Syst. Tech. J. 27, 379 (1948); 27, 623 (1948),Shannon, W. Weaver, The mathematical Theory of Communication,University of Illinois Press, Urbana, IL, 1949.

I A. I. Khinchin, Mathematical Foundations of Information Theory,Dover, New York, 1957.

I E. T. Jaynes, Information theory and Statistical Mechanics. Part I,Phys. Rev. 106, 620 (1957); Part II, Phys. Rev. 108, 171 (1957).

I S. Kullback, Information Theory and Statistics, Wiley, 1959.I A. Renyi, Probability Theory, North–Holland, Amsterdam, 1970.I B. B. Mandelbrot, Fractal-form. Chance and Dimension, Freeman,

1977.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 237: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Information Theory and Entropy

I H. Nyquist, Bell Syst. Tech. J., 3, 324 (1924).R. V. Hartley, Bell Syst. Tech. J. 7, 535 (1928).

I A. Kolmogorov, Atti della R. Acc. Naz. Lincei, 12, 388 (1930)M. Nagumo, Jap. J. Math. 7, 71 (1930)

I C. Shannon, Bell Syst. Tech. J. 27, 379 (1948); 27, 623 (1948),Shannon, W. Weaver, The mathematical Theory of Communication,University of Illinois Press, Urbana, IL, 1949.

I A. I. Khinchin, Mathematical Foundations of Information Theory,Dover, New York, 1957.

I E. T. Jaynes, Information theory and Statistical Mechanics. Part I,Phys. Rev. 106, 620 (1957); Part II, Phys. Rev. 108, 171 (1957).

I S. Kullback, Information Theory and Statistics, Wiley, 1959.I A. Renyi, Probability Theory, North–Holland, Amsterdam, 1970.I B. B. Mandelbrot, Fractal-form. Chance and Dimension, Freeman,

1977.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 238: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Information Theory and Entropy

I H. Nyquist, Bell Syst. Tech. J., 3, 324 (1924).R. V. Hartley, Bell Syst. Tech. J. 7, 535 (1928).

I A. Kolmogorov, Atti della R. Acc. Naz. Lincei, 12, 388 (1930)M. Nagumo, Jap. J. Math. 7, 71 (1930)

I C. Shannon, Bell Syst. Tech. J. 27, 379 (1948); 27, 623 (1948),Shannon, W. Weaver, The mathematical Theory of Communication,University of Illinois Press, Urbana, IL, 1949.

I A. I. Khinchin, Mathematical Foundations of Information Theory,Dover, New York, 1957.

I E. T. Jaynes, Information theory and Statistical Mechanics. Part I,Phys. Rev. 106, 620 (1957); Part II, Phys. Rev. 108, 171 (1957).

I S. Kullback, Information Theory and Statistics, Wiley, 1959.

I A. Renyi, Probability Theory, North–Holland, Amsterdam, 1970.I B. B. Mandelbrot, Fractal-form. Chance and Dimension, Freeman,

1977.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 239: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Information Theory and Entropy

I H. Nyquist, Bell Syst. Tech. J., 3, 324 (1924).R. V. Hartley, Bell Syst. Tech. J. 7, 535 (1928).

I A. Kolmogorov, Atti della R. Acc. Naz. Lincei, 12, 388 (1930)M. Nagumo, Jap. J. Math. 7, 71 (1930)

I C. Shannon, Bell Syst. Tech. J. 27, 379 (1948); 27, 623 (1948),Shannon, W. Weaver, The mathematical Theory of Communication,University of Illinois Press, Urbana, IL, 1949.

I A. I. Khinchin, Mathematical Foundations of Information Theory,Dover, New York, 1957.

I E. T. Jaynes, Information theory and Statistical Mechanics. Part I,Phys. Rev. 106, 620 (1957); Part II, Phys. Rev. 108, 171 (1957).

I S. Kullback, Information Theory and Statistics, Wiley, 1959.I A. Renyi, Probability Theory, North–Holland, Amsterdam, 1970.

I B. B. Mandelbrot, Fractal-form. Chance and Dimension, Freeman,1977.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 240: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Information Theory and Entropy

I H. Nyquist, Bell Syst. Tech. J., 3, 324 (1924).R. V. Hartley, Bell Syst. Tech. J. 7, 535 (1928).

I A. Kolmogorov, Atti della R. Acc. Naz. Lincei, 12, 388 (1930)M. Nagumo, Jap. J. Math. 7, 71 (1930)

I C. Shannon, Bell Syst. Tech. J. 27, 379 (1948); 27, 623 (1948),Shannon, W. Weaver, The mathematical Theory of Communication,University of Illinois Press, Urbana, IL, 1949.

I A. I. Khinchin, Mathematical Foundations of Information Theory,Dover, New York, 1957.

I E. T. Jaynes, Information theory and Statistical Mechanics. Part I,Phys. Rev. 106, 620 (1957); Part II, Phys. Rev. 108, 171 (1957).

I S. Kullback, Information Theory and Statistics, Wiley, 1959.I A. Renyi, Probability Theory, North–Holland, Amsterdam, 1970.I B. B. Mandelbrot, Fractal-form. Chance and Dimension, Freeman,

1977.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 241: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Modern Theory: Applicability

I From the original vision of Shannon, concerning the limits on signalprocessing operations, i.e. storing, compressing andcommunicating data (e.g. the “noisy channel coding theorem”),Information Theory has broadened to cover many different areas:

I Cryptography

I Statistical Inference

I Quantum Computing

I Pattern Recognition

I Theoretical Linguistics

I Ecology, etc.

I A crucial notion is that of Information Entropy.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 242: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Modern Theory: Applicability

I From the original vision of Shannon, concerning the limits on signalprocessing operations, i.e. storing, compressing andcommunicating data (e.g. the “noisy channel coding theorem”),Information Theory has broadened to cover many different areas:

I Cryptography

I Statistical Inference

I Quantum Computing

I Pattern Recognition

I Theoretical Linguistics

I Ecology, etc.

I A crucial notion is that of Information Entropy.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 243: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Modern Theory: Applicability

I From the original vision of Shannon, concerning the limits on signalprocessing operations, i.e. storing, compressing andcommunicating data (e.g. the “noisy channel coding theorem”),Information Theory has broadened to cover many different areas:

I Cryptography

I Statistical Inference

I Quantum Computing

I Pattern Recognition

I Theoretical Linguistics

I Ecology, etc.

I A crucial notion is that of Information Entropy.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 244: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Modern Theory: Applicability

I From the original vision of Shannon, concerning the limits on signalprocessing operations, i.e. storing, compressing andcommunicating data (e.g. the “noisy channel coding theorem”),Information Theory has broadened to cover many different areas:

I Cryptography

I Statistical Inference

I Quantum Computing

I Pattern Recognition

I Theoretical Linguistics

I Ecology, etc.

I A crucial notion is that of Information Entropy.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 245: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Modern Theory: Applicability

I From the original vision of Shannon, concerning the limits on signalprocessing operations, i.e. storing, compressing andcommunicating data (e.g. the “noisy channel coding theorem”),Information Theory has broadened to cover many different areas:

I Cryptography

I Statistical Inference

I Quantum Computing

I Pattern Recognition

I Theoretical Linguistics

I Ecology, etc.

I A crucial notion is that of Information Entropy.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 246: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Modern Theory: Applicability

I From the original vision of Shannon, concerning the limits on signalprocessing operations, i.e. storing, compressing andcommunicating data (e.g. the “noisy channel coding theorem”),Information Theory has broadened to cover many different areas:

I Cryptography

I Statistical Inference

I Quantum Computing

I Pattern Recognition

I Theoretical Linguistics

I Ecology, etc.

I A crucial notion is that of Information Entropy.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 247: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Modern Theory: Applicability

I From the original vision of Shannon, concerning the limits on signalprocessing operations, i.e. storing, compressing andcommunicating data (e.g. the “noisy channel coding theorem”),Information Theory has broadened to cover many different areas:

I Cryptography

I Statistical Inference

I Quantum Computing

I Pattern Recognition

I Theoretical Linguistics

I Ecology, etc.

I A crucial notion is that of Information Entropy.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 248: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Modern Theory: Applicability

I From the original vision of Shannon, concerning the limits on signalprocessing operations, i.e. storing, compressing andcommunicating data (e.g. the “noisy channel coding theorem”),Information Theory has broadened to cover many different areas:

I Cryptography

I Statistical Inference

I Quantum Computing

I Pattern Recognition

I Theoretical Linguistics

I Ecology, etc.

I A crucial notion is that of Information Entropy.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 249: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Some foundational aspects

A. Renyi, Probability Theory, North–Holland, Amsterdam, 1970.

Let I be the amount of information (expressed in bits) associated withan event, described by a random variable X , whose probabilitydistribution is p := piWi=1.We shall make three fundamental assumptions.

I 1) Information should depends purely on p.

I 2) Information should be additive for two independent events.It means that if we observe the outcome of two independent eventswith probabilities p and q, the total information associated is thesum of the two. This implies the Cauchy functional equation

I(p · q) = I(p) + I(q).

I Under very mild assumptions, this equation admits a unique class ofsolutions: I(p) = −k ln2(p). For instance, by putting I (1/2) = 1,we get the so called Hartley measure of information: I = − ln2(p)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 250: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Some foundational aspects

A. Renyi, Probability Theory, North–Holland, Amsterdam, 1970.

Let I be the amount of information (expressed in bits) associated withan event, described by a random variable X , whose probabilitydistribution is p := piWi=1.We shall make three fundamental assumptions.

I 1) Information should depends purely on p.I 2) Information should be additive for two independent events.

It means that if we observe the outcome of two independent eventswith probabilities p and q, the total information associated is thesum of the two. This implies the Cauchy functional equation

I(p · q) = I(p) + I(q).

I Under very mild assumptions, this equation admits a unique class ofsolutions: I(p) = −k ln2(p). For instance, by putting I (1/2) = 1,we get the so called Hartley measure of information: I = − ln2(p)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 251: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Some foundational aspects

A. Renyi, Probability Theory, North–Holland, Amsterdam, 1970.

Let I be the amount of information (expressed in bits) associated withan event, described by a random variable X , whose probabilitydistribution is p := piWi=1.We shall make three fundamental assumptions.

I 1) Information should depends purely on p.I 2) Information should be additive for two independent events.

It means that if we observe the outcome of two independent eventswith probabilities p and q, the total information associated is thesum of the two. This implies the Cauchy functional equation

I(p · q) = I(p) + I(q).

I Under very mild assumptions, this equation admits a unique class ofsolutions: I(p) = −k ln2(p). For instance, by putting I (1/2) = 1,we get the so called Hartley measure of information: I = − ln2(p)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 252: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

P. Jizba, T. Arimitsu, Ann. of Phys. 312, 17 (2004).

I 3) If different amounts of information occur with differentprobabilities, the total amount of information is the average ofthe individual information, weighted by the probabilities of theiroccurrences.

Let A1, . . . ,An be n independent random variables, interpreted aspossible outcomes of an experiment, with probabilities p1, . . . , pn.Assume that Ak conveys Ik bits of information. Then, a natural wayto express the total amount of informationI = I(p1, . . . , pn; I1, . . . , In) is

I =n∑

k=1

pkIk .

I Notice that this linear averaging is just a particular case of a moregeneral mean, introduced by Kolmogorov and Nagumo in 1930!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 253: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

P. Jizba, T. Arimitsu, Ann. of Phys. 312, 17 (2004).

I 3) If different amounts of information occur with differentprobabilities, the total amount of information is the average ofthe individual information, weighted by the probabilities of theiroccurrences.

Let A1, . . . ,An be n independent random variables, interpreted aspossible outcomes of an experiment, with probabilities p1, . . . , pn.Assume that Ak conveys Ik bits of information. Then, a natural wayto express the total amount of informationI = I(p1, . . . , pn; I1, . . . , In) is

I =n∑

k=1

pkIk .

I Notice that this linear averaging is just a particular case of a moregeneral mean, introduced by Kolmogorov and Nagumo in 1930!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 254: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

The Kolmogorov-Nagumo approach

Kolmogorov and Nagumo independently introduced the notion ofKolmogorov-Nagumo (KN) average.Definition 3.1. Let f be an arbitrary real function, and piWi=1 be a probabilitydistribution. Let fk := f (k), where k ∈ N. Assume that φ is a monotonicallyincreasing real function, called the KN function. The KN average of f is

〈f 〉 = φ−1

(W∑k=0

pkφ(fk)

)Now, fix an additional monotonically increasing function ω(x). We shall callinformation content Ik the quantity

Ik = ω

(1

pk

).

Therefore, the average information content is

〈I 〉 = φ−1

(W∑k=0

pkφ

(1

pk

)))Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 255: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Without loss of generality, we shall define the information content by therelation ω(x) = ln2(x). We are led to the most general measure of theamount of transmitted information:

〈I 〉 = φ−1

(W∑k=0

pkφ

(ln2

1

pk

))

Problem. What are the mathematical forms allowed for the KN functionφ(x)? Once again, the composition process plays a crucial role.Consider an experiment E which comes from the union of two

independent experiments E1 and E2. Suppose that I(1)k bits of

information are received with probability pk , k = 1, . . . , n from the

experiment E1 and I(2)l bits of information are received with probability

ql from the experiment E2, l = 1, . . . ,m.From the postulate of additivity, we get

φ−1

(n∑

k=1

m∑l=1

pkqlφ(I(1)k + I(2)

l

))= φ−1

(n∑

k=1

pkφ(I(1)k

))+φ−1

(m∑l=1

qlφ(I(2)l

))

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 256: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Shannon’s and Renyi entropies

I Result. The additivity postulate allows only for two classes: f linearand f exponential.a) φ(x) = cx (the most elementary KN function).

I We obtain

〈I〉 = −n∑

k=1

pk log2(pk)

i.e., the celebrated Shannon entropyI b) φ(x) = 2(1−α)x−1

γ .I We obtain

〈I〉 = Hα(p) :=1

1− αLog2

(n∑

k=1

pαk

).

It is called the Renyi entropy or information measure of order α.I Renyi entropy has been axiomatized by several authors: Renyi

himself, Daroczy (1964), Jizba and Arimitsu (2004).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 257: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Shannon’s and Renyi entropies

I Result. The additivity postulate allows only for two classes: f linearand f exponential.a) φ(x) = cx (the most elementary KN function).

I We obtain

〈I〉 = −n∑

k=1

pk log2(pk)

i.e., the celebrated Shannon entropy

I b) φ(x) = 2(1−α)x−1γ .

I We obtain

〈I〉 = Hα(p) :=1

1− αLog2

(n∑

k=1

pαk

).

It is called the Renyi entropy or information measure of order α.I Renyi entropy has been axiomatized by several authors: Renyi

himself, Daroczy (1964), Jizba and Arimitsu (2004).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 258: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Shannon’s and Renyi entropies

I Result. The additivity postulate allows only for two classes: f linearand f exponential.a) φ(x) = cx (the most elementary KN function).

I We obtain

〈I〉 = −n∑

k=1

pk log2(pk)

i.e., the celebrated Shannon entropyI b) φ(x) = 2(1−α)x−1

γ .

I We obtain

〈I〉 = Hα(p) :=1

1− αLog2

(n∑

k=1

pαk

).

It is called the Renyi entropy or information measure of order α.I Renyi entropy has been axiomatized by several authors: Renyi

himself, Daroczy (1964), Jizba and Arimitsu (2004).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 259: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Shannon’s and Renyi entropies

I Result. The additivity postulate allows only for two classes: f linearand f exponential.a) φ(x) = cx (the most elementary KN function).

I We obtain

〈I〉 = −n∑

k=1

pk log2(pk)

i.e., the celebrated Shannon entropyI b) φ(x) = 2(1−α)x−1

γ .I We obtain

〈I〉 = Hα(p) :=1

1− αLog2

(n∑

k=1

pαk

).

It is called the Renyi entropy or information measure of order α.

I Renyi entropy has been axiomatized by several authors: Renyihimself, Daroczy (1964), Jizba and Arimitsu (2004).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 260: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Shannon’s and Renyi entropies

I Result. The additivity postulate allows only for two classes: f linearand f exponential.a) φ(x) = cx (the most elementary KN function).

I We obtain

〈I〉 = −n∑

k=1

pk log2(pk)

i.e., the celebrated Shannon entropyI b) φ(x) = 2(1−α)x−1

γ .I We obtain

〈I〉 = Hα(p) :=1

1− αLog2

(n∑

k=1

pαk

).

It is called the Renyi entropy or information measure of order α.I Renyi entropy has been axiomatized by several authors: Renyi

himself, Daroczy (1964), Jizba and Arimitsu (2004).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 261: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

On the notion of Information Entropy

I Consequently, Information Entropy is the expected value(average) of the information contained in an experiment (message,flow of information, etc.).

I Given a random variable X , with associated a certain distributionof outcomes x1, . . . , xn, information entropy can be interpreted asa measure of the amount of uncertainty associated with the value ofX if only the distribution is known.

I If the distribution associated with X is constant, i.e. equal to someknown value with probability 1, then the information entropy isminimal, and equal to zero.

I If X takes a (finite) number of values, each of them equally likely,(i.e. they are uniformly distributed), the information entropy ismaximal.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 262: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

On the notion of Information Entropy

I Consequently, Information Entropy is the expected value(average) of the information contained in an experiment (message,flow of information, etc.).

I Given a random variable X , with associated a certain distributionof outcomes x1, . . . , xn, information entropy can be interpreted asa measure of the amount of uncertainty associated with the value ofX if only the distribution is known.

I If the distribution associated with X is constant, i.e. equal to someknown value with probability 1, then the information entropy isminimal, and equal to zero.

I If X takes a (finite) number of values, each of them equally likely,(i.e. they are uniformly distributed), the information entropy ismaximal.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 263: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

On the notion of Information Entropy

I Consequently, Information Entropy is the expected value(average) of the information contained in an experiment (message,flow of information, etc.).

I Given a random variable X , with associated a certain distributionof outcomes x1, . . . , xn, information entropy can be interpreted asa measure of the amount of uncertainty associated with the value ofX if only the distribution is known.

I If the distribution associated with X is constant, i.e. equal to someknown value with probability 1, then the information entropy isminimal, and equal to zero.

I If X takes a (finite) number of values, each of them equally likely,(i.e. they are uniformly distributed), the information entropy ismaximal.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 264: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

On the notion of Information Entropy

I Consequently, Information Entropy is the expected value(average) of the information contained in an experiment (message,flow of information, etc.).

I Given a random variable X , with associated a certain distributionof outcomes x1, . . . , xn, information entropy can be interpreted asa measure of the amount of uncertainty associated with the value ofX if only the distribution is known.

I If the distribution associated with X is constant, i.e. equal to someknown value with probability 1, then the information entropy isminimal, and equal to zero.

I If X takes a (finite) number of values, each of them equally likely,(i.e. they are uniformly distributed), the information entropy ismaximal.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 265: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

On the notion of Information Entropy

I Examples. Suppose to transmit a string M of bits (0s and 1s).

I If the receiver already knows the value of each bit of M, (e.g.p(0) = 1, p(1) = 0, no information is transmitted. In this case,S = 0

I If each bit is independently equally likely transmitted (i.e.p(0) = 1/2, p(1) = 1/2), then we have transmitted M shannons ofinformation. In this case, S = log2M.

I In the intermediate cases, the Shannon information entropy takesvalues between 0 and log2M.

I More generally, if a random variable X takes values x1, . . . , xn,where xi ∈ X, the Shannon information entropy takes values

S(M) = −∑xi∈X

p(xi )log2(p(xi )).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 266: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

On the notion of Information Entropy

I Examples. Suppose to transmit a string M of bits (0s and 1s).

I If the receiver already knows the value of each bit of M, (e.g.p(0) = 1, p(1) = 0, no information is transmitted. In this case,S = 0

I If each bit is independently equally likely transmitted (i.e.p(0) = 1/2, p(1) = 1/2), then we have transmitted M shannons ofinformation. In this case, S = log2M.

I In the intermediate cases, the Shannon information entropy takesvalues between 0 and log2M.

I More generally, if a random variable X takes values x1, . . . , xn,where xi ∈ X, the Shannon information entropy takes values

S(M) = −∑xi∈X

p(xi )log2(p(xi )).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 267: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

On the notion of Information Entropy

I Examples. Suppose to transmit a string M of bits (0s and 1s).

I If the receiver already knows the value of each bit of M, (e.g.p(0) = 1, p(1) = 0, no information is transmitted. In this case,S = 0

I If each bit is independently equally likely transmitted (i.e.p(0) = 1/2, p(1) = 1/2), then we have transmitted M shannons ofinformation. In this case, S = log2M.

I In the intermediate cases, the Shannon information entropy takesvalues between 0 and log2M.

I More generally, if a random variable X takes values x1, . . . , xn,where xi ∈ X, the Shannon information entropy takes values

S(M) = −∑xi∈X

p(xi )log2(p(xi )).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 268: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

On the notion of Information Entropy

I Examples. Suppose to transmit a string M of bits (0s and 1s).

I If the receiver already knows the value of each bit of M, (e.g.p(0) = 1, p(1) = 0, no information is transmitted. In this case,S = 0

I If each bit is independently equally likely transmitted (i.e.p(0) = 1/2, p(1) = 1/2), then we have transmitted M shannons ofinformation. In this case, S = log2M.

I In the intermediate cases, the Shannon information entropy takesvalues between 0 and log2M.

I More generally, if a random variable X takes values x1, . . . , xn,where xi ∈ X, the Shannon information entropy takes values

S(M) = −∑xi∈X

p(xi )log2(p(xi )).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 269: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

On the notion of Information Entropy

I Examples. Suppose to transmit a string M of bits (0s and 1s).

I If the receiver already knows the value of each bit of M, (e.g.p(0) = 1, p(1) = 0, no information is transmitted. In this case,S = 0

I If each bit is independently equally likely transmitted (i.e.p(0) = 1/2, p(1) = 1/2), then we have transmitted M shannons ofinformation. In this case, S = log2M.

I In the intermediate cases, the Shannon information entropy takesvalues between 0 and log2M.

I More generally, if a random variable X takes values x1, . . . , xn,where xi ∈ X, the Shannon information entropy takes values

S(M) = −∑xi∈X

p(xi )log2(p(xi )).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 270: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

The Renyi entropy, revisited

A. Renyi, On measures of information and entropy, Proc. of the 4th BerkeleySymposium on Mathematics, Statistics and Probability, 547, 1960. Thecelebrated Renyi entropy is one of most important measure entropies.

Hα(p1, . . . , pW ) :=ln(∑W

i=1 pαi

)1− α .

It is additive:Hα(A ∪ B) = Hα(A) + Hα(B).

It is concave for 0 < α < 1. In the limit α→ 1, it reduces to Shannon’sentropy. For α = 0, it reduces to Hartley’s entropy:

H0 = ln W .

The differential Renyi entropy of a distribution P with density p is given by

hα(P) =1

1− α ln

∫(p(x))αdx .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 271: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Renyi divergence

T. van Erven, P. Harremoes, arXiv: 1206.2459v2 (2014).To this aim, let (X ,Σ, µ) be a measure space, where X is a set, Σ a σ–algebraover X , and µ : Σ→ R+ ∪ 0 a measure. A Σ–measurable functionp : X → R+ ∪ 0 will be called a probability distribution function (p.d.f.) if∫

X

pdµ = 1. (1)

The probability measure induced by a p.d.f. p is defined by

P(E) =

∫E

p(x)dµ(x), ∀E ∈ Σ. (2)

We shall assume that (X ,Σ, µ) is a σ–finite measure space.A measure P is called absolutely continuous with respect to a measure Q ifP(A) = 0 whenever Q(A) = 0 for all events A ∈ Σ. We shall write P Q if Pis absolutely continuous with respect to Q.We shall assume that our probability measures are absolutely continuous withrespect to µ. We will also identify functions differing on a µ–null set only.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 272: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Definition 3.1. The Renyi divergence of order α of P from Q is defined to be

Dα(P‖Q) :=1

α− 1ln

(∫pαi

qα−1

)dµ,

where we adopt the conventions that 0/0 = 0 and x/0 =∞ for x > 0.Properties.

D1(P‖Q) = DKL(P‖Q),

i.e. it coincides with the celebrated Kullback-Leibler divergence.Kullback-Leibler divergence:

D1(P‖Q) :=

∫p ln

p

qdµ.

Property 3.2 For any order α ∈ [0,∞],

Dα(P‖Q) ≥ 0.

For α > 0, we have

Dα(P‖Q) = 0 if and only if P = Q.

For α = 0,Dα(P‖Q) = 0 if and only if Q P.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 273: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

I Property 3.3 (joint convexity). For any order α ∈ [0, 1], and for any twopairs of probability measures (P0,Q0) and (P1,Q1), and any 0 < λ < 1,we have

Dα [(1− λ)P0 + λP1 ‖ (1− λ)Q0 + λQ1] ≤ (1−λ)Dα (P0 ‖ Q0)+λDα (P1 ‖ Q1)

I Property 3.4 (convexity in the second argument).

for any order α ∈ [0,∞] Renyi divergence is convex in its secondargument: for any probability distributions P, Q0 and Q1 we have

Dα [P ‖ (1− λ)Q0 + λQ1] ≤ (1− λ)Dα (P ‖ Q0) + λDα (P ‖ Q1) .

I Property 3.5 For any 0 < α ≤ β < 1,

α

β

1− β1− αDβ(P ‖ Q) ≤ Dα(P ‖ Q) ≤ Dβ(P ‖ Q).

This means that the topologies induced by Renyi divergences of orderα ∈ (0, 1) are all equivalent.

I Property 3.6 (Skew symmetry) For any 0 < α < 1,

Dα(P ‖ Q) =α

1− αD1−α(Q ‖ P).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 274: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

I Property 3.3 (joint convexity). For any order α ∈ [0, 1], and for any twopairs of probability measures (P0,Q0) and (P1,Q1), and any 0 < λ < 1,we have

Dα [(1− λ)P0 + λP1 ‖ (1− λ)Q0 + λQ1] ≤ (1−λ)Dα (P0 ‖ Q0)+λDα (P1 ‖ Q1)

I Property 3.4 (convexity in the second argument).

for any order α ∈ [0,∞] Renyi divergence is convex in its secondargument: for any probability distributions P, Q0 and Q1 we have

Dα [P ‖ (1− λ)Q0 + λQ1] ≤ (1− λ)Dα (P ‖ Q0) + λDα (P ‖ Q1) .

I Property 3.5 For any 0 < α ≤ β < 1,

α

β

1− β1− αDβ(P ‖ Q) ≤ Dα(P ‖ Q) ≤ Dβ(P ‖ Q).

This means that the topologies induced by Renyi divergences of orderα ∈ (0, 1) are all equivalent.

I Property 3.6 (Skew symmetry) For any 0 < α < 1,

Dα(P ‖ Q) =α

1− αD1−α(Q ‖ P).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 275: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

I Property 3.3 (joint convexity). For any order α ∈ [0, 1], and for any twopairs of probability measures (P0,Q0) and (P1,Q1), and any 0 < λ < 1,we have

Dα [(1− λ)P0 + λP1 ‖ (1− λ)Q0 + λQ1] ≤ (1−λ)Dα (P0 ‖ Q0)+λDα (P1 ‖ Q1)

I Property 3.4 (convexity in the second argument).

for any order α ∈ [0,∞] Renyi divergence is convex in its secondargument: for any probability distributions P, Q0 and Q1 we have

Dα [P ‖ (1− λ)Q0 + λQ1] ≤ (1− λ)Dα (P ‖ Q0) + λDα (P ‖ Q1) .

I Property 3.5 For any 0 < α ≤ β < 1,

α

β

1− β1− αDβ(P ‖ Q) ≤ Dα(P ‖ Q) ≤ Dβ(P ‖ Q).

This means that the topologies induced by Renyi divergences of orderα ∈ (0, 1) are all equivalent.

I Property 3.6 (Skew symmetry) For any 0 < α < 1,

Dα(P ‖ Q) =α

1− αD1−α(Q ‖ P).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 276: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

I Property 3.3 (joint convexity). For any order α ∈ [0, 1], and for any twopairs of probability measures (P0,Q0) and (P1,Q1), and any 0 < λ < 1,we have

Dα [(1− λ)P0 + λP1 ‖ (1− λ)Q0 + λQ1] ≤ (1−λ)Dα (P0 ‖ Q0)+λDα (P1 ‖ Q1)

I Property 3.4 (convexity in the second argument).

for any order α ∈ [0,∞] Renyi divergence is convex in its secondargument: for any probability distributions P, Q0 and Q1 we have

Dα [P ‖ (1− λ)Q0 + λQ1] ≤ (1− λ)Dα (P ‖ Q0) + λDα (P ‖ Q1) .

I Property 3.5 For any 0 < α ≤ β < 1,

α

β

1− β1− αDβ(P ‖ Q) ≤ Dα(P ‖ Q) ≤ Dβ(P ‖ Q).

This means that the topologies induced by Renyi divergences of orderα ∈ (0, 1) are all equivalent.

I Property 3.6 (Skew symmetry) For any 0 < α < 1,

Dα(P ‖ Q) =α

1− αD1−α(Q ‖ P).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 277: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Lesche stability of Renyi entropy

B. Lesche, J. Stat. Phys. 27, 419 (1982).

I Lesche’s stability. Let f (x) be a scalar quantity, x ∈ X ⊂ Rn be astate variable. Lesche’s criterion provides a necessary condition forf (x) to be observable.

I Let

‖ x− x′ ‖1=n∑k

| xk − x′

k |

be the Holder l1 metric on Rn. Then, ∀ε > 0, there exists δε > 0such that, for any pair x, x′, one has

‖ x− x‘ ‖≤ δε =⇒ | f (x)− f (x′) |fmax

< ε

I This property is nothing but the uniform continuity of f (x) on thestate space X .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 278: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Lesche stability of Renyi entropy

B. Lesche, J. Stat. Phys. 27, 419 (1982).

I Lesche’s stability. Let f (x) be a scalar quantity, x ∈ X ⊂ Rn be astate variable. Lesche’s criterion provides a necessary condition forf (x) to be observable.

I Let

‖ x− x′ ‖1=n∑k

| xk − x′

k |

be the Holder l1 metric on Rn. Then, ∀ε > 0, there exists δε > 0such that, for any pair x, x′, one has

‖ x− x‘ ‖≤ δε =⇒ | f (x)− f (x′) |fmax

< ε

I This property is nothing but the uniform continuity of f (x) on thestate space X .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 279: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Lesche stability of Renyi entropy

B. Lesche, J. Stat. Phys. 27, 419 (1982).

I Lesche’s stability. Let f (x) be a scalar quantity, x ∈ X ⊂ Rn be astate variable. Lesche’s criterion provides a necessary condition forf (x) to be observable.

I Let

‖ x− x′ ‖1=n∑k

| xk − x′

k |

be the Holder l1 metric on Rn. Then, ∀ε > 0, there exists δε > 0such that, for any pair x, x′, one has

‖ x− x‘ ‖≤ δε =⇒ | f (x)− f (x′) |fmax

< ε

I This property is nothing but the uniform continuity of f (x) on thestate space X .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 280: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

I Many of the trace-form entropies are Lesche stable. In particular,Tsallis’ entropy and Kaniadakis’ entropy are.There are results stating that, under mild conditions, a very largeclass of trace-form entropies are indeed Lesche stable.For long time, Renyi entropy was considered to be not Lesche stable.However, this is not really the case.

I P. Jizba and T. Arimitsu, PRE 69, 026128 (2004).Results.a) Renyi’s entropy is Lesche stable for systems with a finite numberof microstates.b) For systems with an infinite number of microstates, the domainof instability has zero Bhattacharyya measure. The possibleinstabilities can be emended by introducing a coarse graining into anactual measurement.c) In case of systems with continuous probability distributions ormultifractal systems, again the lesche condition applies.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 281: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

I Many of the trace-form entropies are Lesche stable. In particular,Tsallis’ entropy and Kaniadakis’ entropy are.There are results stating that, under mild conditions, a very largeclass of trace-form entropies are indeed Lesche stable.For long time, Renyi entropy was considered to be not Lesche stable.However, this is not really the case.

I P. Jizba and T. Arimitsu, PRE 69, 026128 (2004).Results.a) Renyi’s entropy is Lesche stable for systems with a finite numberof microstates.b) For systems with an infinite number of microstates, the domainof instability has zero Bhattacharyya measure. The possibleinstabilities can be emended by introducing a coarse graining into anactual measurement.c) In case of systems with continuous probability distributions ormultifractal systems, again the lesche condition applies.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 282: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Einstein’s likelihood principle

A. Einstein, The Theory of the Opalescence of Homogeneous Fluids andLiquid Mixtures near the Critical State, Annalen der Physik 33,1275-1298 (1910).E. G. D. Cohen, Pramana J. of Physics, 64, 635 (2005).C. Tsallis and H. J. Haubold, EPL, 110, 30005 (2015).

A longstanding unsolved problem. The foundations of statisticalmechanics should be based on a probabilistic approach or on a dynamicalone?Boltzmann’s approach at the equilibrium: SBG = k ln W , where W is thetotal amount of microscopic possibilities, assumed equally probable.

Einstein’s point of view was a dynamical one: “one cannot fix theweights of the various regions in phase space without using the dynamics,i.e. the equations of motion of the system” (Cohen).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 283: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

I Therefore, Einstein never used Boltzmann’s principle. Instead, heproposed the equation (the reversal of Boltzmann’s formula)

W ∝ eSBG /k .

The meaning of this formula is that W is determined from theentropy S , which, as any other physical quantity, is determinedfrom the dynamics. We shall call W the likelihood function.

I Now, assume that we have two systems A and B which arestatistically (and dynamically) independent. Then we postulate that

W(A ∪ B) =W(A)W(B).

Following Tsallis and Haubold, we shall call it the Einsteinlikelihood principle.This principle has very profound consequences. It implies that thephysics of the system A ∪ B depends on the systems A and B only,without any need for the knowledge of the rest of the Universe.Needless to say, A and B are allowed to stay in an arbitrary (a prioriunknown) state.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 284: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

I Therefore, Einstein never used Boltzmann’s principle. Instead, heproposed the equation (the reversal of Boltzmann’s formula)

W ∝ eSBG /k .

The meaning of this formula is that W is determined from theentropy S , which, as any other physical quantity, is determinedfrom the dynamics. We shall call W the likelihood function.

I Now, assume that we have two systems A and B which arestatistically (and dynamically) independent. Then we postulate that

W(A ∪ B) =W(A)W(B).

Following Tsallis and Haubold, we shall call it the Einsteinlikelihood principle.This principle has very profound consequences. It implies that thephysics of the system A ∪ B depends on the systems A and B only,without any need for the knowledge of the rest of the Universe.Needless to say, A and B are allowed to stay in an arbitrary (a prioriunknown) state.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 285: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Groups and Information Theory

G. Sicuro and P.T, preprint (2015).

Problem. How can we put together Information Theory, Einstein’slikelihood principle and Generalized Entropies?A possible merge of these ideas is again offered by the group-theoreticalapproach.

I Main result: All entropies satisfying the first three SK axiomsand the composability axiom do have:

I a) An information-theoretical content.

I b) They satisfy the Einstein’s likelihood principle.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 286: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Groups and Information Theory

G. Sicuro and P.T, preprint (2015).

Problem. How can we put together Information Theory, Einstein’slikelihood principle and Generalized Entropies?A possible merge of these ideas is again offered by the group-theoreticalapproach.

I Main result: All entropies satisfying the first three SK axiomsand the composability axiom do have:

I a) An information-theoretical content.

I b) They satisfy the Einstein’s likelihood principle.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 287: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Groups and Information Theory

G. Sicuro and P.T, preprint (2015).

Problem. How can we put together Information Theory, Einstein’slikelihood principle and Generalized Entropies?A possible merge of these ideas is again offered by the group-theoreticalapproach.

I Main result: All entropies satisfying the first three SK axiomsand the composability axiom do have:

I a) An information-theoretical content.

I b) They satisfy the Einstein’s likelihood principle.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 288: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

The group theoretical information content

Definition 3.7. Assume that S is an entropy satisfying the first three SKaxioms, and that is composable, with

S(A ∪ B) = Φ(S(A),S(B)),

where G is a strictly monotone function such thatΦ(x , y) = G (G−1(x) + G−1(y)). We call the quantity

IG (A) := G−1 (S(A)) . (3)

the information functional associated with the generalized entropy S .Properties.Continuity. IG is continuous respect to its arguments;Maximum principle. IG is maximized on the uniform distribution;Expansibility. The addition of a zero–probability event do not changethe value of I;Additivity. Given two independent systems A and B, in arbitrary states,

IG (A ∪ B) = IG (A) + IG (B). (4)Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 289: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

I Proof of additivity. It is based on the universal property of theLazard formal group: There exists a unique application that mapsthe Lazard group into any other group, in particular into theadditive one. We have:

IG (A ∪ B) = G−1(S(A ∪ B)) = G−1(Φ(S(A),S(B))) (5)

= G−1(G (G−1(S(A) + G−1(S(B))))

= G−1 (S(A)) + G−1 (S(B))

= IG (A) + IG (B),

where A and B are two statistically independent systems, in thestates described by the probability distributions p and q respectively.

I In this context, we shall call G−1(t) the linearization map.I For instance, in the case of Tsallis entropy,

G (t) =e(1−q)t − 1

1− q, G−1(s) =

1

1− qln(1 + (1− q)s)

The linearization map G−1 is exactly the map which transformsTsallis entropy into Reny’s entropy!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 290: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

I Proof of additivity. It is based on the universal property of theLazard formal group: There exists a unique application that mapsthe Lazard group into any other group, in particular into theadditive one. We have:

IG (A ∪ B) = G−1(S(A ∪ B)) = G−1(Φ(S(A),S(B))) (5)

= G−1(G (G−1(S(A) + G−1(S(B))))

= G−1 (S(A)) + G−1 (S(B))

= IG (A) + IG (B),

where A and B are two statistically independent systems, in thestates described by the probability distributions p and q respectively.

I In this context, we shall call G−1(t) the linearization map.

I For instance, in the case of Tsallis entropy,

G (t) =e(1−q)t − 1

1− q, G−1(s) =

1

1− qln(1 + (1− q)s)

The linearization map G−1 is exactly the map which transformsTsallis entropy into Reny’s entropy!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 291: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

I Proof of additivity. It is based on the universal property of theLazard formal group: There exists a unique application that mapsthe Lazard group into any other group, in particular into theadditive one. We have:

IG (A ∪ B) = G−1(S(A ∪ B)) = G−1(Φ(S(A),S(B))) (5)

= G−1(G (G−1(S(A) + G−1(S(B))))

= G−1 (S(A)) + G−1 (S(B))

= IG (A) + IG (B),

where A and B are two statistically independent systems, in thestates described by the probability distributions p and q respectively.

I In this context, we shall call G−1(t) the linearization map.I For instance, in the case of Tsallis entropy,

G (t) =e(1−q)t − 1

1− q, G−1(s) =

1

1− qln(1 + (1− q)s)

The linearization map G−1 is exactly the map which transformsTsallis entropy into Reny’s entropy!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 292: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

Einstein’s likelihood function

Definition 3.8. Assume that S is an entropy satisfying the first three SKaxioms, and that is composable, with

S(A ∪ B) = Φ(S(A),S(B)),

where G is a strictly monotone function such thatΦ(x , y) = G (G−1(x) + G−1(y)). The Einstein’s likelihood functionassociated with the entropy S is

W(A) = eG−1(S(A)) (6)

Theorem 3.9.Let S be a composable entropy. Then

W(A ∪ B) =W(A) · W(B).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 293: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

I Proof of Einstein’s principle. We have:

W(A ∪ B) = eG−1(A∪B) = eG−1(Φ(S(A),S(B)))

= eG−1(G(G−1(S(A))+G−1(S(B)))

= eG−1(S(A))+G−1(S(B)) =W(A) · W(B). (7)

I Example. For the Tsallis entropy case, we have the formula:

eG−1(Sq) = eSqq , where eq(x) := [1 + (1− q)x ]

11−q is the

q-exponential. It recovers the formula by Tsallis and Haubold.

I What happens if S is an entropy which is not composable?

I For entropies at least weakly composable, both definitions ofinformation functional IG (A) := G−1 (S(A)) and likekihood

W(A) = eG−1(S(A)) are formally valid, but the additivity andEinstein’s principle are valid only on the uniform distribution. Thisin turn implies that this class of entropies is not very satisfactory.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 294: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

I Proof of Einstein’s principle. We have:

W(A ∪ B) = eG−1(A∪B) = eG−1(Φ(S(A),S(B)))

= eG−1(G(G−1(S(A))+G−1(S(B)))

= eG−1(S(A))+G−1(S(B)) =W(A) · W(B). (7)

I Example. For the Tsallis entropy case, we have the formula:

eG−1(Sq) = eSqq , where eq(x) := [1 + (1− q)x ]

11−q is the

q-exponential. It recovers the formula by Tsallis and Haubold.

I What happens if S is an entropy which is not composable?

I For entropies at least weakly composable, both definitions ofinformation functional IG (A) := G−1 (S(A)) and likekihood

W(A) = eG−1(S(A)) are formally valid, but the additivity andEinstein’s principle are valid only on the uniform distribution. Thisin turn implies that this class of entropies is not very satisfactory.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 295: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Information EntropyRenyi’s entropyEinstein’s likelihood principleGroups and Information Theory

I Proof of Einstein’s principle. We have:

W(A ∪ B) = eG−1(A∪B) = eG−1(Φ(S(A),S(B)))

= eG−1(G(G−1(S(A))+G−1(S(B)))

= eG−1(S(A))+G−1(S(B)) =W(A) · W(B). (7)

I Example. For the Tsallis entropy case, we have the formula:

eG−1(Sq) = eSqq , where eq(x) := [1 + (1− q)x ]

11−q is the

q-exponential. It recovers the formula by Tsallis and Haubold.

I What happens if S is an entropy which is not composable?

I For entropies at least weakly composable, both definitions ofinformation functional IG (A) := G−1 (S(A)) and likekihood

W(A) = eG−1(S(A)) are formally valid, but the additivity andEinstein’s principle are valid only on the uniform distribution. Thisin turn implies that this class of entropies is not very satisfactory.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 296: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

The composability problem revisitedThe Z-entropies: main definitions

The composability problem revisited

The trace-form class of entropies has a serious drawback:

I Only the Boltzmann and Tsallis entropies are composable.

I This property seems crucial, for thermodynamic andinformation-theoretical purposes.

I The whole plethora of entropies proposed since 1989, is at mostweakly composable. From the previous analysis, this property isnecessary, but probably not sufficient.

I Is there any way out ?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 297: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

The composability problem revisitedThe Z-entropies: main definitions

The composability problem revisited

The trace-form class of entropies has a serious drawback:

I Only the Boltzmann and Tsallis entropies are composable.

I This property seems crucial, for thermodynamic andinformation-theoretical purposes.

I The whole plethora of entropies proposed since 1989, is at mostweakly composable. From the previous analysis, this property isnecessary, but probably not sufficient.

I Is there any way out ?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 298: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

The composability problem revisitedThe Z-entropies: main definitions

The composability problem revisited

The trace-form class of entropies has a serious drawback:

I Only the Boltzmann and Tsallis entropies are composable.

I This property seems crucial, for thermodynamic andinformation-theoretical purposes.

I The whole plethora of entropies proposed since 1989, is at mostweakly composable. From the previous analysis, this property isnecessary, but probably not sufficient.

I Is there any way out ?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 299: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

The composability problem revisitedThe Z-entropies: main definitions

The composability problem revisited

The trace-form class of entropies has a serious drawback:

I Only the Boltzmann and Tsallis entropies are composable.

I This property seems crucial, for thermodynamic andinformation-theoretical purposes.

I The whole plethora of entropies proposed since 1989, is at mostweakly composable. From the previous analysis, this property isnecessary, but probably not sufficient.

I Is there any way out ?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 300: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

The composability problem revisitedThe Z-entropies: main definitions

Main result

Main result: There is a new class of non trace-form type of strictlycomposable entropies, coming from formal group theory.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 301: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

The composability problem revisitedThe Z-entropies: main definitions

The Z-entropies

P. T., A New Class of Composable Entropies from Group Theory: TheZ-entropies, arxiv: 1507.07436Definition 3.10. Let pii=1,··· ,W , W ≥ 1, with

∑Wi=1 pi = 1, be a

discrete probability distribution. The Z -entropy associated with G isdefined to be the function

ZG ,α(p1, . . . , pW ) :=lnG

(∑Wi=1 pαi

)1− α

, (8)

with 0 < α < 1. Here lnG (x) denotes the generalized group logarithmassociated with G .The Z -entropies are the non trace-form equivalent of the trace-form SU

entropy.They generalized at the same time Boltzmann’s andRenyi’s entropies.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 302: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

Group logarithms

Let’s come back to the construction of Z -entropies.Definition 3.11. A group logarithm is a continuous, concave andmonotonically increasing function lnG (x) : [1,∞]→ R+ ∪ 0, possiblydepending on a set of real parameters, such that lnG (w) solves thefunctional equation for the group law corresponding to G , i.e

lnG (xy) = Φ(lnG (x), lnG (y)) (9)

where Φ(x , y) = G (G−1(x) + G−1(y)).For instance, when Φ(x , y) = x + y , we have directly that lnG (x) = ln x .If Φ(x , y) = x + y + (1− q)xy , we get for the group logarithm the Tsallislogarithm

lnq =x1−q − 1

q − 1. (10)

Remark 3.12. Any known trace-form entropy provides a grouplogarithm.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 303: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

Schur-concavity

I Majorization. Given two vectors a, b ∈Rn, we shall say that aweakly majorizes b from below (a w b) if

k∑i=1

a↓i ≥k∑

i=1

b↓i , k = 1, . . . , n,

where a↓i and b↓i are the elements of a and b sorted in decreasingorder. If a w b and also

∑ni=1 ai =

∑ni=1 bi , we say that a

majorizes b (a b).

I An entropy S [p] is said to be Schur-concave if for all probabilitydistributions p = pii∈N and q = qjj∈N such that p ≺ q, we have

S [p] ≥ S [q].

This property is typical of Renyi entropy and is sufficient forguaranteeing that the axiom (SK2) is fulfilled.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 304: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

Schur-concavity

I Majorization. Given two vectors a, b ∈Rn, we shall say that aweakly majorizes b from below (a w b) if

k∑i=1

a↓i ≥k∑

i=1

b↓i , k = 1, . . . , n,

where a↓i and b↓i are the elements of a and b sorted in decreasingorder. If a w b and also

∑ni=1 ai =

∑ni=1 bi , we say that a

majorizes b (a b).I An entropy S [p] is said to be Schur-concave if for all probability

distributions p = pii∈N and q = qjj∈N such that p ≺ q, we have

S [p] ≥ S [q].

This property is typical of Renyi entropy and is sufficient forguaranteeing that the axiom (SK2) is fulfilled.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 305: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

I There is a simple criterion to ascertain the Schur-concavity (orconvexity) of a function.Theorem 3.13 (Schur-Ostrowski criterion). Let f be a symmetricfunction, and assume that all first order partial derivatives exist.Then F is Schur-convace if and only if

(xi − xj)

(∂f

∂xi− ∂f

∂xj

)≤ 0.

I Theorem 3.14 (concavity of the Z -family). All the ZG ,α entropiesare concave for 0 < α < 1 and Schur-concave for α > 1.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 306: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

I There is a simple criterion to ascertain the Schur-concavity (orconvexity) of a function.Theorem 3.13 (Schur-Ostrowski criterion). Let f be a symmetricfunction, and assume that all first order partial derivatives exist.Then F is Schur-convace if and only if

(xi − xj)

(∂f

∂xi− ∂f

∂xj

)≤ 0.

I Theorem 3.14 (concavity of the Z -family). All the ZG ,α entropiesare concave for 0 < α < 1 and Schur-concave for α > 1.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 307: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

Theorem 3.15. The ZG ,α entropies are all strictly composable, i.e.given any two statistically independent subsystems A, B, defined on anarbitrary probability distribution pii=1,··· ,W , they satisfy thecomposition rule

ZG ,α(A ∪ B) = Ψ(ZG ,α(A),ZG ,α(B)) (11)

where

Ψ(x , y) =1

1− αΦ ((1− α)x , (1− α)y)

with Φ (x , y) being the group law satisfied by the generalized grouplogarithm lnG associated to ZG ,α.

Proposition 3.16. The function Ψ(x , y) is also a group law.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 308: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

Limiting properties

The class of entropies (1) possesses a simple relation with the most celebratedentropies. In particular, each entropic functional of the class generalizesBoltzmann’s and Renyi’s entropies at the same time, as stated by the followingresults.Proposition 3.17 In the limit α→ 1, the ZG ,α-entropy reduces to theBoltzmann-Gibbs entropy.Proof.

limα→1

lnG

(∑Wi=1 pαi

)1− α = lim

α→1

G[ln(∑W

i=1 pαi

)]1− α =

= limα→1

G ′[ln(∑W

i=1 pαi

)]· 1∑W

i=1 pαi·∑W

i=1 pαi ln pi

−1=

W∑i=1

pi ln1

pi.

where we took into account that G ′(0) = 1.Proposition 3.18 Each of the Z -entropies generalizes the Renyi entropy.Proof. It suffices to observe that the function lnG (x) tends to ln x whenai → 0, i = 1, 2, . . . in the expression of G(t).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 309: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

Theorem 3.19 The ZG ,α-entropy satisfies the first three SK axioms.(SK1). By definition, the function (8) is continuous with respect to all of itsarguments.(SK2). The entropy (8) is concave for 0 < α < 1. Indeed, given two probability

distributions p(1) =(

p(1)1 , . . . , p

(1)W

)and p(2) =

(p

(21 , . . . , p

(2)W

), we have

ZG ,α(λp1 + (1− λ)p2) =lnG

(∑Wi=1

(λp

(1)i + (1− λ)p

(2)i

)α)1− α ≥

≥lnG

(λ∑W

i=1(p(1)i )α + (1− λ)

∑Wi=1(p

(2)i )α

)1− α ≥

≥ λlnG

(∑Wi=1(p

(1)i )α

)1− α + (1− λ)

lnG

(∑Wi=1(p

(2)i )α

)1− α ,

where we took into account that lnG is by construction monotonically increasingand concave, and that the function

∑Wi=1 pαi is concave for 0 < α < 1.

(SK3). By adding an event of zero probability, we have

ZG ,α(p1, . . . , pW+1) =lnG

(∑W+1i=1 pαi

)1− α =

lnG

(∑Wi=1 pαi

)1− α . (12)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 310: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

A Tower of infinitely many new composable entropies

I 1) Let Φ(x , y) = x + y . We get lnG (x) = ln x . The Renyi entropy

Hα(p1, . . . , pW ) :=ln(∑W

i=1 pαi

)1− α

. (13)

Composition law:

hα(A ∪ B) = Hα(A) + Hα(B)

I 2) Let Φ(x , y) = x + y + qxy . We have lnG (x) = xq−1q .

The associated Zq-entropy is

Zq,α(p1, . . . , pW ) :=

(∑Wi=1 pαi

)q− 1

q(1− α). (14)

Composition law:

Zq,α(A ∪ B) = Zq,α(A) + Zq,α(B) + q(1− α)Zq,α(A)Zq,α(B)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 311: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

A Tower of infinitely many new composable entropies

I 1) Let Φ(x , y) = x + y . We get lnG (x) = ln x . The Renyi entropy

Hα(p1, . . . , pW ) :=ln(∑W

i=1 pαi

)1− α

. (13)

Composition law:

hα(A ∪ B) = Hα(A) + Hα(B)

I 2) Let Φ(x , y) = x + y + qxy . We have lnG (x) = xq−1q .

The associated Zq-entropy is

Zq,α(p1, . . . , pW ) :=

(∑Wi=1 pαi

)q− 1

q(1− α). (14)

Composition law:

Zq,α(A ∪ B) = Zq,α(A) + Zq,α(B) + q(1− α)Zq,α(A)Zq,α(B)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 312: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

Nani gigantum humeris insidentes:

We stay on the shoulders of giants!

(I. Newton).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 313: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

A generalization of the Boltzmann, Tsallis, Renyi,Sharma-Mittal entropies

Definition 3.20. The Za,b-entropy is defined to be

Za,b(p1, . . . , pW ) :=

(∑Wi=1 pαi

)a−(∑W

i=1 pαi

)b(a− b)(1− α)

, . (15)

with 0 < α < 1, a < 0, 0 < b < 1.Limiting properties

I Proposition 3.21. The Za,b entropy reduces to the Boltzmannentropy in the limit α→ 1.

I Proposition 3.22. The Za,b entropy reduces to the Renyi entropyin the limit a→ 0, b → 0.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 314: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

A generalization of the Boltzmann, Tsallis, Renyi,Sharma-Mittal entropies

Definition 3.20. The Za,b-entropy is defined to be

Za,b(p1, . . . , pW ) :=

(∑Wi=1 pαi

)a−(∑W

i=1 pαi

)b(a− b)(1− α)

, . (15)

with 0 < α < 1, a < 0, 0 < b < 1.Limiting properties

I Proposition 3.21. The Za,b entropy reduces to the Boltzmannentropy in the limit α→ 1.

I Proposition 3.22. The Za,b entropy reduces to the Renyi entropyin the limit a→ 0, b → 0.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 315: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

I Proposition 3.23. The Za,b entropy reduces to the Tsallis entropyin the limit b → 0, a→ 1.

I Proof.

limb→0

lima→1

(∑Wi=1 pαi

)a−(∑W

i=1 pαi

)b(a− b)(1− α)

= lima→1

(∑Wi=1 pαi

)a− 1

a(1− α)=

=1−

∑Wi=1 pαi

α− 1,

which is the expression of the Tsallis entropy, as a function of theparameter α.

I Proposition 3.24. The Za,b entropy reduces to the Sharma-Mittalentropy in the limit b → 0.Proof. Obvious.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 316: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

I Proposition 3.23. The Za,b entropy reduces to the Tsallis entropyin the limit b → 0, a→ 1.

I Proof.

limb→0

lima→1

(∑Wi=1 pαi

)a−(∑W

i=1 pαi

)b(a− b)(1− α)

= lima→1

(∑Wi=1 pαi

)a− 1

a(1− α)=

=1−

∑Wi=1 pαi

α− 1,

which is the expression of the Tsallis entropy, as a function of theparameter α.

I Proposition 3.24. The Za,b entropy reduces to the Sharma-Mittalentropy in the limit b → 0.Proof. Obvious.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 317: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

I Proposition 3.23. The Za,b entropy reduces to the Tsallis entropyin the limit b → 0, a→ 1.

I Proof.

limb→0

lima→1

(∑Wi=1 pαi

)a−(∑W

i=1 pαi

)b(a− b)(1− α)

= lima→1

(∑Wi=1 pαi

)a− 1

a(1− α)=

=1−

∑Wi=1 pαi

α− 1,

which is the expression of the Tsallis entropy, as a function of theparameter α.

I Proposition 3.24. The Za,b entropy reduces to the Sharma-Mittalentropy in the limit b → 0.Proof. Obvious.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 318: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

I The Za,b entropy is related to the Borges-Roditi logarithm:

lnBR(x) :=xa − xb

a− b.

I If we put x = eu (Rota’s isomorphism), we recover what in themathematical literature is called the Abel exponential, given by

expAbel =eau − ebu

a− b=

eγu√δ

sinh(√

δu), (16)

where γ = (a + b)/2,

δ = (a− b)2/4.

I The Abel exponential (16) satisties the Abel functional equation, as wasproven in N. H. Abel, Magazin for Naturvidenskaberne 1 (1823) 216-229.reprinted in: L. Sylow, S. Lie (Eds.), Oeuvres Completes, vol. 1,Christiania, 1881.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 319: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

I The Za,b entropy is related to the Borges-Roditi logarithm:

lnBR(x) :=xa − xb

a− b.

I If we put x = eu (Rota’s isomorphism), we recover what in themathematical literature is called the Abel exponential, given by

expAbel =eau − ebu

a− b=

eγu√δ

sinh(√

δu), (16)

where γ = (a + b)/2,

δ = (a− b)2/4.

I The Abel exponential (16) satisties the Abel functional equation, as wasproven in N. H. Abel, Magazin for Naturvidenskaberne 1 (1823) 216-229.reprinted in: L. Sylow, S. Lie (Eds.), Oeuvres Completes, vol. 1,Christiania, 1881.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 320: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

I The Za,b entropy is related to the Borges-Roditi logarithm:

lnBR(x) :=xa − xb

a− b.

I If we put x = eu (Rota’s isomorphism), we recover what in themathematical literature is called the Abel exponential, given by

expAbel =eau − ebu

a− b=

eγu√δ

sinh(√

δu), (16)

where γ = (a + b)/2,

δ = (a− b)2/4.

I The Abel exponential (16) satisties the Abel functional equation, as wasproven in N. H. Abel, Magazin for Naturvidenskaberne 1 (1823) 216-229.reprinted in: L. Sylow, S. Lie (Eds.), Oeuvres Completes, vol. 1,Christiania, 1881.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 321: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

I The Composition law is the Abel group law:

ΦA(x , y) = x + y + β1xy +∑j>i

βi(

xy i − x iy). (17)

The coefficients βn in (17) can be expressed as

βn =(−1)n−1

n!(n − 1)

∏i+j=n−1, i,j≥0

(ia + jb).

I In particular, if a = −b = k, we have the Z -analog of Kaniadakisentropy:

Zk,α(p1, . . . , pW ) :=

(∑Wi=1 pαi

)k−(∑W

i=1 pαi

)−k

(2k)(1− α). (18)

Zk,α(A ∪ B) = Zk,α(A)√

1 + k2(1− α)2Zk,α(B)2 (19)

+ Zk,α(B)√

1 + k2(1− α)2Zk,α(A)2

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 322: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

I The Composition law is the Abel group law:

ΦA(x , y) = x + y + β1xy +∑j>i

βi(

xy i − x iy). (17)

The coefficients βn in (17) can be expressed as

βn =(−1)n−1

n!(n − 1)

∏i+j=n−1, i,j≥0

(ia + jb).

I In particular, if a = −b = k, we have the Z -analog of Kaniadakisentropy:

Zk,α(p1, . . . , pW ) :=

(∑Wi=1 pαi

)k−(∑W

i=1 pαi

)−k

(2k)(1− α). (18)

Zk,α(A ∪ B) = Zk,α(A)√

1 + k2(1− α)2Zk,α(B)2 (19)

+ Zk,α(B)√

1 + k2(1− α)2Zk,α(A)2

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 323: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

I Definition 3.25. Let pii=1,··· ,W , W ≥ 1, with∑W

i=1 pi = 1, be adiscrete probability distribution. Let G be a formal groupexponential, and lnG (x) denotes the associated generalized grouplogarithm (according to Definition 9). The Z δ-entropy associatedwith G is defined to be the function

Z δG ,α(p1, . . . , pW ) :=

lnδG

(∑Wi=1 pαi

)(1− α)δ

. (20)

where 0 < α < 1 if 0 < δ < 1 and α > 1 if δ > 1.

I This class of entropies is concave for 0 < α < 1, 0 < δ < 1 andSchur-concave for α > 1, δ > 1 (according to theSchur-Ostrowski criterion).

I All the class Zδ satisfies the first three SK axiom and in particular isstrictly composable!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 324: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

I Definition 3.25. Let pii=1,··· ,W , W ≥ 1, with∑W

i=1 pi = 1, be adiscrete probability distribution. Let G be a formal groupexponential, and lnG (x) denotes the associated generalized grouplogarithm (according to Definition 9). The Z δ-entropy associatedwith G is defined to be the function

Z δG ,α(p1, . . . , pW ) :=

lnδG

(∑Wi=1 pαi

)(1− α)δ

. (20)

where 0 < α < 1 if 0 < δ < 1 and α > 1 if δ > 1.

I This class of entropies is concave for 0 < α < 1, 0 < δ < 1 andSchur-concave for α > 1, δ > 1 (according to theSchur-Ostrowski criterion).

I All the class Zδ satisfies the first three SK axiom and in particular isstrictly composable!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 325: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

I Definition 3.25. Let pii=1,··· ,W , W ≥ 1, with∑W

i=1 pi = 1, be adiscrete probability distribution. Let G be a formal groupexponential, and lnG (x) denotes the associated generalized grouplogarithm (according to Definition 9). The Z δ-entropy associatedwith G is defined to be the function

Z δG ,α(p1, . . . , pW ) :=

lnδG

(∑Wi=1 pαi

)(1− α)δ

. (20)

where 0 < α < 1 if 0 < δ < 1 and α > 1 if δ > 1.

I This class of entropies is concave for 0 < α < 1, 0 < δ < 1 andSchur-concave for α > 1, δ > 1 (according to theSchur-Ostrowski criterion).

I All the class Zδ satisfies the first three SK axiom and in particular isstrictly composable!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 326: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

Special case:

Z δG ,α(p1, . . . , pW ) :=

lnδq

(∑Wi=1 pαi

)(1− α)δ

=

(∑Wi=1 pαi

)1−q

− 1

(1− q)(1− α)δ.

This entropy possesses all the good properties of the Tsallis-Cirto entropy(TC) Sq,δ. Precisely, it is extensive when

W (N) = ANµBNν

, A > 0, µ ≥ 0,B > 1, 0 ≤ ν ≤ 1.

In particular, consider the case q = 1. Given a d-dimensional system, theentropy

Z δ(p1, . . . , pW ) :=lnδ(∑W

i=1 pαi

)(1− α)δ

for δ = d/d − 1 is extensive. For d = 3, we have the same beautifulinterpretation already found by Tsallis and Cirto concerning thethermodynamics of black-holes:

Z3/2 ∝ (SBH)3/2.

This entropy does not reduce to the TC entropy; however, it is extensive in thesame regimes, and is strictly composable.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 327: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

Information-theoretical content: Z-divergences

I Definition 3.26. Let P = pi and R = ri two discrete probabilitydistributions. The Z -divergence associated to G of order α of thedistribution P from the distribution R, denoted by Dα

G , is defined to be

DαG (P||R) :=

lnG

(∑Wi=1

pαirα−1i

)(α− 1)

.

where 0 < α < 1 (here we adopt the standard conventions that 0/0 = 0and x/0 =∞ for x > 0).

I Theorem 3.27 The following inequality holds

DαG (P||R) ≥ 0.

In particular,

DαG (P||R) = 0 if and only if P = R.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 328: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

Information-theoretical content: Z-divergences

I Definition 3.26. Let P = pi and R = ri two discrete probabilitydistributions. The Z -divergence associated to G of order α of thedistribution P from the distribution R, denoted by Dα

G , is defined to be

DαG (P||R) :=

lnG

(∑Wi=1

pαirα−1i

)(α− 1)

.

where 0 < α < 1 (here we adopt the standard conventions that 0/0 = 0and x/0 =∞ for x > 0).

I Theorem 3.27 The following inequality holds

DαG (P||R) ≥ 0.

In particular,

DαG (P||R) = 0 if and only if P = R.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 329: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

Specific example: the Za,b-divergence of order (α, a, b) of the distribution Pfrom the distribution R, denoted by Dα

a,b, is defined to be

Dαa,b (P||R) :=

(∑Wi=1

pαirα−1i

)a

−(∑W

i=1

pαirα−1i

)b

(a− b)(α− 1).

where 0 < α < 1.Observe that

D10,0 (P||R) := lim

α→1Dα

0,0 (P||R) =W∑i=1

pi

(ln

pi

qi

)≥ 0,

i.e. it reduces to the standard Kullback-Leibler divergence. At the same time,

Dα0,0 (P||R) := lim

a→0limb→0

Dαa,b =

1

α− 1ln

(W∑i=1

pαirα−1i

),

which gives the Renyi divergence.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 330: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

In short summary:

I Group Law Φ(x , y)←→ Entropy

I Φ(x , y) = G(G−1(x) + G−1(y)⇐⇒ G(t) =∑∞

k=0 aktk+1/(k + 1)

I G(t)←→∑W

i=1 piG(

ln 1pi

)(UGE construction)

I G(t)←→ 11−α lnG

(∑Wi=1 pαi

)(Z-entropy construction)

I where lnG (xy) = Φ(lnG (x), lnG (y))

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 331: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

In short summary:

I Group Law Φ(x , y)←→ Entropy

I Φ(x , y) = G(G−1(x) + G−1(y)⇐⇒ G(t) =∑∞

k=0 aktk+1/(k + 1)

I G(t)←→∑W

i=1 piG(

ln 1pi

)(UGE construction)

I G(t)←→ 11−α lnG

(∑Wi=1 pαi

)(Z-entropy construction)

I where lnG (xy) = Φ(lnG (x), lnG (y))

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 332: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

In short summary:

I Group Law Φ(x , y)←→ Entropy

I Φ(x , y) = G(G−1(x) + G−1(y)⇐⇒ G(t) =∑∞

k=0 aktk+1/(k + 1)

I G(t)←→∑W

i=1 piG(

ln 1pi

)(UGE construction)

I G(t)←→ 11−α lnG

(∑Wi=1 pαi

)(Z-entropy construction)

I where lnG (xy) = Φ(lnG (x), lnG (y))

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 333: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

In short summary:

I Group Law Φ(x , y)←→ Entropy

I Φ(x , y) = G(G−1(x) + G−1(y)⇐⇒ G(t) =∑∞

k=0 aktk+1/(k + 1)

I G(t)←→∑W

i=1 piG(

ln 1pi

)(UGE construction)

I G(t)←→ 11−α lnG

(∑Wi=1 pαi

)(Z-entropy construction)

I where lnG (xy) = Φ(lnG (x), lnG (y))

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 334: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Concavity and Schur-concavityLimiting propertiesA Tower of infinitely many new composable entropiesA generalization of the Boltzmann, Tsallis, Renyi, Sharma-Mittal entropiesZ-delta entropiesInformation-theoretical content: Z-divergences

In short summary:

I Group Law Φ(x , y)←→ Entropy

I Φ(x , y) = G(G−1(x) + G−1(y)⇐⇒ G(t) =∑∞

k=0 aktk+1/(k + 1)

I G(t)←→∑W

i=1 piG(

ln 1pi

)(UGE construction)

I G(t)←→ 11−α lnG

(∑Wi=1 pαi

)(Z-entropy construction)

I where lnG (xy) = Φ(lnG (x), lnG (y))

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 335: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Conclusions and Future Perspectives

I The Composability Axiom, jointly with the axioms (SK1)-(SK3), allowsto classify the known entropies and to generate infinitely many new ones.

I Under mild hypotheses, a group law Φ(x , y) corresponds to an entropyand vice-versa.

I There are two classes of group entropies.

I The trace-form entropies can be regarded as representation of a uniquegeneral entropy: the universal-group entropy.

I The group entropies of non trace-form type belong to a new family: theZ -class.

I All these entropies are extensive; the trace-form ones are (at least) weaklycomposable; the Z -family is strictly composable.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 336: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Conclusions and Future Perspectives

I The Composability Axiom, jointly with the axioms (SK1)-(SK3), allowsto classify the known entropies and to generate infinitely many new ones.

I Under mild hypotheses, a group law Φ(x , y) corresponds to an entropyand vice-versa.

I There are two classes of group entropies.

I The trace-form entropies can be regarded as representation of a uniquegeneral entropy: the universal-group entropy.

I The group entropies of non trace-form type belong to a new family: theZ -class.

I All these entropies are extensive; the trace-form ones are (at least) weaklycomposable; the Z -family is strictly composable.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 337: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Conclusions and Future Perspectives

I The Composability Axiom, jointly with the axioms (SK1)-(SK3), allowsto classify the known entropies and to generate infinitely many new ones.

I Under mild hypotheses, a group law Φ(x , y) corresponds to an entropyand vice-versa.

I There are two classes of group entropies.

I The trace-form entropies can be regarded as representation of a uniquegeneral entropy: the universal-group entropy.

I The group entropies of non trace-form type belong to a new family: theZ -class.

I All these entropies are extensive; the trace-form ones are (at least) weaklycomposable; the Z -family is strictly composable.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 338: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Conclusions and Future Perspectives

I The Composability Axiom, jointly with the axioms (SK1)-(SK3), allowsto classify the known entropies and to generate infinitely many new ones.

I Under mild hypotheses, a group law Φ(x , y) corresponds to an entropyand vice-versa.

I There are two classes of group entropies.

I The trace-form entropies can be regarded as representation of a uniquegeneral entropy: the universal-group entropy.

I The group entropies of non trace-form type belong to a new family: theZ -class.

I All these entropies are extensive; the trace-form ones are (at least) weaklycomposable; the Z -family is strictly composable.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 339: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Conclusions and Future Perspectives

I The Composability Axiom, jointly with the axioms (SK1)-(SK3), allowsto classify the known entropies and to generate infinitely many new ones.

I Under mild hypotheses, a group law Φ(x , y) corresponds to an entropyand vice-versa.

I There are two classes of group entropies.

I The trace-form entropies can be regarded as representation of a uniquegeneral entropy: the universal-group entropy.

I The group entropies of non trace-form type belong to a new family: theZ -class.

I All these entropies are extensive; the trace-form ones are (at least) weaklycomposable; the Z -family is strictly composable.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 340: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Conclusions and Future Perspectives

I The Composability Axiom, jointly with the axioms (SK1)-(SK3), allowsto classify the known entropies and to generate infinitely many new ones.

I Under mild hypotheses, a group law Φ(x , y) corresponds to an entropyand vice-versa.

I There are two classes of group entropies.

I The trace-form entropies can be regarded as representation of a uniquegeneral entropy: the universal-group entropy.

I The group entropies of non trace-form type belong to a new family: theZ -class.

I All these entropies are extensive; the trace-form ones are (at least) weaklycomposable; the Z -family is strictly composable.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 341: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

I In particular, the Z(a,b) entropy is presented as a new composablegeneralization of many known entropies.

I The group-entropic forms possess many remarkable properties: Legendrestructure, information-theoretical content, etc. (another seminar), allcoming from the group structure.

I A general study of the role of the Z -class in the theory of complexsystems is in order (long-term project).

I The extension of the correspondence among entropies and zetafunctions to the case of multi-parametric entropies and multiple zetavalues and polylogarithms is an open problem.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 342: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

I In particular, the Z(a,b) entropy is presented as a new composablegeneralization of many known entropies.

I The group-entropic forms possess many remarkable properties: Legendrestructure, information-theoretical content, etc. (another seminar), allcoming from the group structure.

I A general study of the role of the Z -class in the theory of complexsystems is in order (long-term project).

I The extension of the correspondence among entropies and zetafunctions to the case of multi-parametric entropies and multiple zetavalues and polylogarithms is an open problem.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 343: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

I In particular, the Z(a,b) entropy is presented as a new composablegeneralization of many known entropies.

I The group-entropic forms possess many remarkable properties: Legendrestructure, information-theoretical content, etc. (another seminar), allcoming from the group structure.

I A general study of the role of the Z -class in the theory of complexsystems is in order (long-term project).

I The extension of the correspondence among entropies and zetafunctions to the case of multi-parametric entropies and multiple zetavalues and polylogarithms is an open problem.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 344: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

I In particular, the Z(a,b) entropy is presented as a new composablegeneralization of many known entropies.

I The group-entropic forms possess many remarkable properties: Legendrestructure, information-theoretical content, etc. (another seminar), allcoming from the group structure.

I A general study of the role of the Z -class in the theory of complexsystems is in order (long-term project).

I The extension of the correspondence among entropies and zetafunctions to the case of multi-parametric entropies and multiple zetavalues and polylogarithms is an open problem.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 345: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

Formal groups Z-Entropies The universal-group

entropy

Information Theory

Shannon-Khinchin axioms

and composability

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 346: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Information-theoretical aspects of the notion of entropyThe Z-class of Entropies

Main properties of the Z-entropiesState of Art

THANK YOU!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 347: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

GROUPS, ENTROPIES AND NUMBERTHEORY

Piergiulio Tempesta

Universidad Complutense de Madridand

Instituto de Ciencias Matematicas (ICMAT),Madrid, Spain.

TEMPLETON SCHOOL ON FOUNDATIONS OF COMPLEXITY

October 14 - 15, 2015

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 348: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

LECTURE IV

Formal Groups and Zeta Functions:A Possible Root from Number Theory to

Statistical Mechanics

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 349: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

OutlineNumber Theory and Statistical Mechanics

The multiple connections between Number Theory and PhysicsEntropies and Number Theory

Preliminaries: The Theory of L-functionsThe Riemann zeta functionL-series and L-functionsThe Hurwitz zeta function

Bernoulli numbers and polynomialsSpecial values of zeta functionsFamous CongruencesUniversal structures

Formal Groups and Number TheoryUniversal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 350: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The multiple connections between Number Theory and PhysicsEntropies and Number Theory

Number Theory and Physics

I Perhaps the first connection was established by Montgomery(1970’s): the sequence of zeros of the Riemann zeta function is veryclose to the distributions of energy levels of heavy nuclei!

I Zeta function regularization in quantum field theory (see e.g. theCasimir effect)

I This connection has been extended by considering zeroes ofL-functions and energy levels of heavy nuclei.

I Modern Cryptography over elliptic and hyperelluptic curves

I Polylogarithms, multiple zeta values and Feynman diagrams.

I Are there connections between nonadditive entropies and NumberTheory?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 351: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The multiple connections between Number Theory and PhysicsEntropies and Number Theory

Number Theory and Physics

I Perhaps the first connection was established by Montgomery(1970’s): the sequence of zeros of the Riemann zeta function is veryclose to the distributions of energy levels of heavy nuclei!

I Zeta function regularization in quantum field theory (see e.g. theCasimir effect)

I This connection has been extended by considering zeroes ofL-functions and energy levels of heavy nuclei.

I Modern Cryptography over elliptic and hyperelluptic curves

I Polylogarithms, multiple zeta values and Feynman diagrams.

I Are there connections between nonadditive entropies and NumberTheory?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 352: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The multiple connections between Number Theory and PhysicsEntropies and Number Theory

Number Theory and Physics

I Perhaps the first connection was established by Montgomery(1970’s): the sequence of zeros of the Riemann zeta function is veryclose to the distributions of energy levels of heavy nuclei!

I Zeta function regularization in quantum field theory (see e.g. theCasimir effect)

I This connection has been extended by considering zeroes ofL-functions and energy levels of heavy nuclei.

I Modern Cryptography over elliptic and hyperelluptic curves

I Polylogarithms, multiple zeta values and Feynman diagrams.

I Are there connections between nonadditive entropies and NumberTheory?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 353: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The multiple connections between Number Theory and PhysicsEntropies and Number Theory

Number Theory and Physics

I Perhaps the first connection was established by Montgomery(1970’s): the sequence of zeros of the Riemann zeta function is veryclose to the distributions of energy levels of heavy nuclei!

I Zeta function regularization in quantum field theory (see e.g. theCasimir effect)

I This connection has been extended by considering zeroes ofL-functions and energy levels of heavy nuclei.

I Modern Cryptography over elliptic and hyperelluptic curves

I Polylogarithms, multiple zeta values and Feynman diagrams.

I Are there connections between nonadditive entropies and NumberTheory?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 354: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The multiple connections between Number Theory and PhysicsEntropies and Number Theory

Number Theory and Physics

I Perhaps the first connection was established by Montgomery(1970’s): the sequence of zeros of the Riemann zeta function is veryclose to the distributions of energy levels of heavy nuclei!

I Zeta function regularization in quantum field theory (see e.g. theCasimir effect)

I This connection has been extended by considering zeroes ofL-functions and energy levels of heavy nuclei.

I Modern Cryptography over elliptic and hyperelluptic curves

I Polylogarithms, multiple zeta values and Feynman diagrams.

I Are there connections between nonadditive entropies and NumberTheory?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 355: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The multiple connections between Number Theory and PhysicsEntropies and Number Theory

Number Theory and Physics

I Perhaps the first connection was established by Montgomery(1970’s): the sequence of zeros of the Riemann zeta function is veryclose to the distributions of energy levels of heavy nuclei!

I Zeta function regularization in quantum field theory (see e.g. theCasimir effect)

I This connection has been extended by considering zeroes ofL-functions and energy levels of heavy nuclei.

I Modern Cryptography over elliptic and hyperelluptic curves

I Polylogarithms, multiple zeta values and Feynman diagrams.

I Are there connections between nonadditive entropies and NumberTheory?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 356: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The multiple connections between Number Theory and PhysicsEntropies and Number Theory

Entropies and Number Theory

Main idea. Under suitable regularity hypotheses, we can associate with agroup entropy a class of number-theoretical structures:

Generalized Bernoulli polynomials,

L-series (generalized Riemann-zeta functions),

Hurwitz-type zeta functions,

Polylogarithms, etc.

These generalized objects preserve the same relations possessed by thestandard objects, i.e.:

ζ(s) −→ Bernoulli numbers

Hurwitz zeta function −→ Bernoulli polynomials

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 357: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The multiple connections between Number Theory and PhysicsEntropies and Number Theory

Classical Bernoulli

numbers

and polynomials

Riemann zeta function

Hurwitz zeta function

Formal groups

L-functions Generalized Bernoulli

structures

Hyperfunctions

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 358: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The Riemann zeta functionL-series and L-functionsThe Hurwitz zeta function

The Riemann zeta function

E. C. Titchmarsh The Theory of the Riemann Zeta Function, (1986); H.Iwaniec, E. Kowalski, Analytic Number Theory, AMS, 53 (2004).The Riemann Zeta Function is defined to be the series

ζ (s) =∞∑n=0

1

nss ∈ C, Re s > 1.

It was introduced by Euler for s ∈ R and by Riemann (1860) for s ∈ C. It isabsolutely and uniformly convergent for Re s > 1. It can be represented as aMellin transform:

ζ (s) =1

Γ (s)

∫ ∞0

1

ex − 1x s−1dx .

Euler product:∞∑n=0

1

ns=∏p

1

(1− p−s).

Functional equation:

ξ(s) = ξ(1− s), ξ(s) =1

2π−s/2s(s − 1)Γ

( s2

)ζ(s).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 359: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The Riemann zeta functionL-series and L-functionsThe Hurwitz zeta function

The Riemann hypothesis: all the non-trivial zeros of the Riemann zetafunction lie in the axis Re s = 1/2: The most important unresolved problemin pure mathematics (Bombieri, 2000).

Figure: ζ(s) doesn’t have any zero on the right of Re s = 1 and on the left ofRe s = 0. Furthermore, the non-trivial zeros are symmetric about the real axisand the line Re s = 1/2. According to the Riemann Hypothesis, they all lie onthe line Re s = 1/2.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 360: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The Riemann zeta functionL-series and L-functionsThe Hurwitz zeta function

Dirichlet series

T. Apostol, Introduction to Analytic Number Theory, Springer, 1976.A Dirichlet series is a series of the form

L(s) =∞∑n=1

anns, Re s > σ0

These series have great relevance in analytic number theory.Examples.

∞∑n=1

µ(n)

ns=

1

ζ(s),

where µ(n) is the Mobius function, equal to the sum of the primitive n-throots of unity.

∞∑n=1

φ(n)

ns=ζ(s − 1)

ζ(s)

where φ(n) = n∏

p|n (1− 1/p) is Euler’s totient function : it is defined as the

number of integers k ∈ N, 1 ≤ k ≤ n, s.t. gcd(n, k) = 1.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 361: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The Riemann zeta functionL-series and L-functionsThe Hurwitz zeta function

L-series and L-functions

I Generally speaking an L-series is a Dirichlet series, which has ameromorphic prolongation to the whole complex plane. Thisprolongation defines a complex function, which is the L-functionassociated with the L-series. Usually, one also asks for the existenceof an Euler product.

I In particular, there is a famous axiomatization, due to A. Selberg,of a class of L-series, known as the Selberg class. It is based on fouraxioms (meromorphicity, existence of an Euluer product expansion,existence of a functional equation, Ramanujan’s growth condition.

I We also have other classes: Artin’s L-functions, automorphicL-functions, Dirichlet L-functions, etc.

I A Generalized Riemann hypothesis has been conjectured for theDirichlet L-functions (1884).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 362: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The Riemann zeta functionL-series and L-functionsThe Hurwitz zeta function

L-series and L-functions

I Generally speaking an L-series is a Dirichlet series, which has ameromorphic prolongation to the whole complex plane. Thisprolongation defines a complex function, which is the L-functionassociated with the L-series. Usually, one also asks for the existenceof an Euler product.

I In particular, there is a famous axiomatization, due to A. Selberg,of a class of L-series, known as the Selberg class. It is based on fouraxioms (meromorphicity, existence of an Euluer product expansion,existence of a functional equation, Ramanujan’s growth condition.

I We also have other classes: Artin’s L-functions, automorphicL-functions, Dirichlet L-functions, etc.

I A Generalized Riemann hypothesis has been conjectured for theDirichlet L-functions (1884).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 363: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The Riemann zeta functionL-series and L-functionsThe Hurwitz zeta function

L-series and L-functions

I Generally speaking an L-series is a Dirichlet series, which has ameromorphic prolongation to the whole complex plane. Thisprolongation defines a complex function, which is the L-functionassociated with the L-series. Usually, one also asks for the existenceof an Euler product.

I In particular, there is a famous axiomatization, due to A. Selberg,of a class of L-series, known as the Selberg class. It is based on fouraxioms (meromorphicity, existence of an Euluer product expansion,existence of a functional equation, Ramanujan’s growth condition.

I We also have other classes: Artin’s L-functions, automorphicL-functions, Dirichlet L-functions, etc.

I A Generalized Riemann hypothesis has been conjectured for theDirichlet L-functions (1884).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 364: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The Riemann zeta functionL-series and L-functionsThe Hurwitz zeta function

L-series and L-functions

I Generally speaking an L-series is a Dirichlet series, which has ameromorphic prolongation to the whole complex plane. Thisprolongation defines a complex function, which is the L-functionassociated with the L-series. Usually, one also asks for the existenceof an Euler product.

I In particular, there is a famous axiomatization, due to A. Selberg,of a class of L-series, known as the Selberg class. It is based on fouraxioms (meromorphicity, existence of an Euluer product expansion,existence of a functional equation, Ramanujan’s growth condition.

I We also have other classes: Artin’s L-functions, automorphicL-functions, Dirichlet L-functions, etc.

I A Generalized Riemann hypothesis has been conjectured for theDirichlet L-functions (1884).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 365: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The Riemann zeta functionL-series and L-functionsThe Hurwitz zeta function

The Hurwitz zeta function

I The Hurwitz zeta function is defined to be the series

ζ(a, s) =∞∑n=1

1

(n + a)s, Re s > 1, a > 0.

This series is absolutely convergent for the given values of s and qand can be extended to a meromorphic function for all s 6= 1. Noticethat ζ(s, 1) = ζ(s).

I Its Mellin transform representation is

ζ (s, a) =∞∑n=0

(n + a)−s =1

Γ (s)

∫ ∞0

e−ax

1− e−xx s−1dx .

I Notice that

ζ(s, a) =1

s − 1

∞∑n=0

(−1)n

n + 1∆n

+a1−s

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 366: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The Riemann zeta functionL-series and L-functionsThe Hurwitz zeta function

The Hurwitz zeta function

I The Hurwitz zeta function is defined to be the series

ζ(a, s) =∞∑n=1

1

(n + a)s, Re s > 1, a > 0.

This series is absolutely convergent for the given values of s and qand can be extended to a meromorphic function for all s 6= 1. Noticethat ζ(s, 1) = ζ(s).

I Its Mellin transform representation is

ζ (s, a) =∞∑n=0

(n + a)−s =1

Γ (s)

∫ ∞0

e−ax

1− e−xx s−1dx .

I Notice that

ζ(s, a) =1

s − 1

∞∑n=0

(−1)n

n + 1∆n

+a1−s

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 367: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The Riemann zeta functionL-series and L-functionsThe Hurwitz zeta function

The Hurwitz zeta function

I The Hurwitz zeta function is defined to be the series

ζ(a, s) =∞∑n=1

1

(n + a)s, Re s > 1, a > 0.

This series is absolutely convergent for the given values of s and qand can be extended to a meromorphic function for all s 6= 1. Noticethat ζ(s, 1) = ζ(s).

I Its Mellin transform representation is

ζ (s, a) =∞∑n=0

(n + a)−s =1

Γ (s)

∫ ∞0

e−ax

1− e−xx s−1dx .

I Notice that

ζ(s, a) =1

s − 1

∞∑n=0

(−1)n

n + 1∆n

+a1−s

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 368: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The Riemann zeta functionL-series and L-functionsThe Hurwitz zeta function

I) NEXT and Hurwitz

I Is there any connection between Nonextensive Statistical Mechanicsand the Hurwitz zeta function?

I Consider the Legendre constraints for the maximization of the Sq

entropy. We know that, under suitable constraints, the canonicaldistributions takes the form

pi = (Zq)−1[1− (1− q)βεi ]1/(1−q).

whereZq =

∑i

[1− (1− q)βεi ]1/(1−q).

I We assume that εk = γ0k, where γ0 is a suitable constant. Then weintroduce

s =1

q − 1, α =

1

β(q − 1)k0We have:

pk =1

(k + α)s/ζ(k, α).

The canonical distribution takes now the form of a Hurwitz distribution.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 369: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The Riemann zeta functionL-series and L-functionsThe Hurwitz zeta function

I) NEXT and Hurwitz

I Is there any connection between Nonextensive Statistical Mechanicsand the Hurwitz zeta function?

I Consider the Legendre constraints for the maximization of the Sq

entropy. We know that, under suitable constraints, the canonicaldistributions takes the form

pi = (Zq)−1[1− (1− q)βεi ]1/(1−q).

whereZq =

∑i

[1− (1− q)βεi ]1/(1−q).

I We assume that εk = γ0k, where γ0 is a suitable constant. Then weintroduce

s =1

q − 1, α =

1

β(q − 1)k0We have:

pk =1

(k + α)s/ζ(k, α).

The canonical distribution takes now the form of a Hurwitz distribution.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 370: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

The Riemann zeta functionL-series and L-functionsThe Hurwitz zeta function

I) NEXT and Hurwitz

I Is there any connection between Nonextensive Statistical Mechanicsand the Hurwitz zeta function?

I Consider the Legendre constraints for the maximization of the Sq

entropy. We know that, under suitable constraints, the canonicaldistributions takes the form

pi = (Zq)−1[1− (1− q)βεi ]1/(1−q).

whereZq =

∑i

[1− (1− q)βεi ]1/(1−q).

I We assume that εk = γ0k, where γ0 is a suitable constant. Then weintroduce

s =1

q − 1, α =

1

β(q − 1)k0We have:

pk =1

(k + α)s/ζ(k, α).

The canonical distribution takes now the form of a Hurwitz distribution.Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 371: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The classical Bernoulli numbers and polynomials

I Fermat’s Last Theorem and class field theory (Kummer)

I Theory of Riemann and Hurwitz zeta functions

I Measure theory in p-adic analysis (Mazur)

I Interpolation Theory (Boas and Buck)

I Combinatorics of groups (V. I. Arnold)

I Congruences and Theory of Algebraic Equations

I Ramanujan identities: QFT and Feynman diagrams

I GW invariants, soliton theory (Pandharipande, Veselov)

I More than 1500 papers!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 372: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The classical Bernoulli numbers and polynomials

I Fermat’s Last Theorem and class field theory (Kummer)

I Theory of Riemann and Hurwitz zeta functions

I Measure theory in p-adic analysis (Mazur)

I Interpolation Theory (Boas and Buck)

I Combinatorics of groups (V. I. Arnold)

I Congruences and Theory of Algebraic Equations

I Ramanujan identities: QFT and Feynman diagrams

I GW invariants, soliton theory (Pandharipande, Veselov)

I More than 1500 papers!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 373: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The classical Bernoulli numbers and polynomials

I Fermat’s Last Theorem and class field theory (Kummer)

I Theory of Riemann and Hurwitz zeta functions

I Measure theory in p-adic analysis (Mazur)

I Interpolation Theory (Boas and Buck)

I Combinatorics of groups (V. I. Arnold)

I Congruences and Theory of Algebraic Equations

I Ramanujan identities: QFT and Feynman diagrams

I GW invariants, soliton theory (Pandharipande, Veselov)

I More than 1500 papers!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 374: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The classical Bernoulli numbers and polynomials

I Fermat’s Last Theorem and class field theory (Kummer)

I Theory of Riemann and Hurwitz zeta functions

I Measure theory in p-adic analysis (Mazur)

I Interpolation Theory (Boas and Buck)

I Combinatorics of groups (V. I. Arnold)

I Congruences and Theory of Algebraic Equations

I Ramanujan identities: QFT and Feynman diagrams

I GW invariants, soliton theory (Pandharipande, Veselov)

I More than 1500 papers!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 375: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The classical Bernoulli numbers and polynomials

I Fermat’s Last Theorem and class field theory (Kummer)

I Theory of Riemann and Hurwitz zeta functions

I Measure theory in p-adic analysis (Mazur)

I Interpolation Theory (Boas and Buck)

I Combinatorics of groups (V. I. Arnold)

I Congruences and Theory of Algebraic Equations

I Ramanujan identities: QFT and Feynman diagrams

I GW invariants, soliton theory (Pandharipande, Veselov)

I More than 1500 papers!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 376: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The classical Bernoulli numbers and polynomials

I Fermat’s Last Theorem and class field theory (Kummer)

I Theory of Riemann and Hurwitz zeta functions

I Measure theory in p-adic analysis (Mazur)

I Interpolation Theory (Boas and Buck)

I Combinatorics of groups (V. I. Arnold)

I Congruences and Theory of Algebraic Equations

I Ramanujan identities: QFT and Feynman diagrams

I GW invariants, soliton theory (Pandharipande, Veselov)

I More than 1500 papers!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 377: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The classical Bernoulli numbers and polynomials

I Fermat’s Last Theorem and class field theory (Kummer)

I Theory of Riemann and Hurwitz zeta functions

I Measure theory in p-adic analysis (Mazur)

I Interpolation Theory (Boas and Buck)

I Combinatorics of groups (V. I. Arnold)

I Congruences and Theory of Algebraic Equations

I Ramanujan identities: QFT and Feynman diagrams

I GW invariants, soliton theory (Pandharipande, Veselov)

I More than 1500 papers!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 378: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The classical Bernoulli numbers and polynomials

I Fermat’s Last Theorem and class field theory (Kummer)

I Theory of Riemann and Hurwitz zeta functions

I Measure theory in p-adic analysis (Mazur)

I Interpolation Theory (Boas and Buck)

I Combinatorics of groups (V. I. Arnold)

I Congruences and Theory of Algebraic Equations

I Ramanujan identities: QFT and Feynman diagrams

I GW invariants, soliton theory (Pandharipande, Veselov)

I More than 1500 papers!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 379: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The classical Bernoulli numbers and polynomials

I Fermat’s Last Theorem and class field theory (Kummer)

I Theory of Riemann and Hurwitz zeta functions

I Measure theory in p-adic analysis (Mazur)

I Interpolation Theory (Boas and Buck)

I Combinatorics of groups (V. I. Arnold)

I Congruences and Theory of Algebraic Equations

I Ramanujan identities: QFT and Feynman diagrams

I GW invariants, soliton theory (Pandharipande, Veselov)

I More than 1500 papers!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 380: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Generating functions

Definition. The Bernoulli numbers are the sequence of rational numbersdefined by the generating function

t

et − 1=∞∑k=0

Bk

k!tk .

I The first Bernoulli numbers are

I

B0 = 1, B1 = −1

2,

I

B2 = −1

6, B3 = 0, B4 = − 1

30, B5 = 0,

I

B6 =1

42, B7 = 0, B8 = − 1

30, . . .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 381: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Generating functions

Definition. The Bernoulli numbers are the sequence of rational numbersdefined by the generating function

t

et − 1=∞∑k=0

Bk

k!tk .

I The first Bernoulli numbers are

I

B0 = 1, B1 = −1

2,

I

B2 = −1

6, B3 = 0, B4 = − 1

30, B5 = 0,

I

B6 =1

42, B7 = 0, B8 = − 1

30, . . .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 382: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Generating functions

Definition. The Bernoulli numbers are the sequence of rational numbersdefined by the generating function

t

et − 1=∞∑k=0

Bk

k!tk .

I The first Bernoulli numbers are

I

B0 = 1, B1 = −1

2,

I

B2 = −1

6, B3 = 0, B4 = − 1

30, B5 = 0,

I

B6 =1

42, B7 = 0, B8 = − 1

30, . . .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 383: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Generating functions

Definition. The Bernoulli numbers are the sequence of rational numbersdefined by the generating function

t

et − 1=∞∑k=0

Bk

k!tk .

I The first Bernoulli numbers are

I

B0 = 1, B1 = −1

2,

I

B2 = −1

6, B3 = 0, B4 = − 1

30, B5 = 0,

I

B6 =1

42, B7 = 0, B8 = − 1

30, . . .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 384: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Classical Bernoulli polynomials

Definition. The Bernoulli polynomials are the sequence of polynomialsdefined by the generating function

t

et − 1ext =

∞∑k=0

Bk(x)

k!tk .

I The first Bernoulli polynomials are

I

B0 = 1, B1 = x − 1

2,

I

B2 = x2−x− 1

6, B3 = x3− 3

2x2+

1

2x , B4 = x4−2x3+x2− 1

30,

I

B5 = x5− 5

2x4 +

5

3x3− 1

6x , B6 = x6−3x5 +

5

2x4− 1

2x2 +

1

42, . . .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 385: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Classical Bernoulli polynomials

Definition. The Bernoulli polynomials are the sequence of polynomialsdefined by the generating function

t

et − 1ext =

∞∑k=0

Bk(x)

k!tk .

I The first Bernoulli polynomials areI

B0 = 1, B1 = x − 1

2,

I

B2 = x2−x− 1

6, B3 = x3− 3

2x2+

1

2x , B4 = x4−2x3+x2− 1

30,

I

B5 = x5− 5

2x4 +

5

3x3− 1

6x , B6 = x6−3x5 +

5

2x4− 1

2x2 +

1

42, . . .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 386: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Classical Bernoulli polynomials

Definition. The Bernoulli polynomials are the sequence of polynomialsdefined by the generating function

t

et − 1ext =

∞∑k=0

Bk(x)

k!tk .

I The first Bernoulli polynomials areI

B0 = 1, B1 = x − 1

2,

I

B2 = x2−x− 1

6, B3 = x3− 3

2x2+

1

2x , B4 = x4−2x3+x2− 1

30,

I

B5 = x5− 5

2x4 +

5

3x3− 1

6x , B6 = x6−3x5 +

5

2x4− 1

2x2 +

1

42, . . .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 387: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Classical Bernoulli polynomials

Definition. The Bernoulli polynomials are the sequence of polynomialsdefined by the generating function

t

et − 1ext =

∞∑k=0

Bk(x)

k!tk .

I The first Bernoulli polynomials areI

B0 = 1, B1 = x − 1

2,

I

B2 = x2−x− 1

6, B3 = x3− 3

2x2+

1

2x , B4 = x4−2x3+x2− 1

30,

I

B5 = x5− 5

2x4 +

5

3x3− 1

6x , B6 = x6−3x5 +

5

2x4− 1

2x2 +

1

42, . . .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 388: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

“Magic” properties of Bernoulli numbers and polynomials

Sums of n-th powers

m∑k=0

kn =Bn+1(m + 1)− Bn+1(0)

n + 1

Appell’s property:∂Bn(x) = nBn−1(x)

Finite differences:∆Bn(x) = nxn−1

Raabe’s theorem (p-adic analysis)

Bn(mx) = mn−1m−1∑k=0

Bn

(x +

k

m

)Fourier series representation:

Bn(x) = − n!

(2πi)n

∑k 6=0

e2πikx

kn

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 389: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Special values of zeta functions

There is a close, general connection among special values of zetafunctions and suitable classes of polynomials.

Bernoulli numbers ←→ Riemann’s zeta function

Bernoulli polynomials ←→ Hurwitz’ zeta function

The most basic is the following one:

ζ(1− n) = −Bn

n, n ∈ N

ζ(1− n, a) = −Bn(a)

nn ∈ N, a > 0

Other relations are also considered in the technical literature.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 390: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The Clausen-von Staudt congruence

I The study of congruences in mathematics was started by C. F.Gauss.

I One of the most beautiful congruence of mathematics is that due toClausen and von Staudt, independently. It relates the primenumbers with the Bernoulli numbers.Let p be a prime number, such that p − 1 divides 2k. Then

B2k +∑

p−1|2k

1

p∈ Z

I Prime numbers are both related to the Riemann zeta function andthe Bernoulli numbers

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 391: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The Clausen-von Staudt congruence

I The study of congruences in mathematics was started by C. F.Gauss.

I One of the most beautiful congruence of mathematics is that due toClausen and von Staudt, independently. It relates the primenumbers with the Bernoulli numbers.Let p be a prime number, such that p − 1 divides 2k. Then

B2k +∑

p−1|2k

1

p∈ Z

I Prime numbers are both related to the Riemann zeta function andthe Bernoulli numbers

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 392: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The Clausen-von Staudt congruence

I The study of congruences in mathematics was started by C. F.Gauss.

I One of the most beautiful congruence of mathematics is that due toClausen and von Staudt, independently. It relates the primenumbers with the Bernoulli numbers.Let p be a prime number, such that p − 1 divides 2k. Then

B2k +∑

p−1|2k

1

p∈ Z

I Prime numbers are both related to the Riemann zeta function andthe Bernoulli numbers

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 393: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Prime numbers

Formal groups

Special values

Euler product Clausen-von Staudt

congruence

𝜁(𝑠) 𝐵𝑛

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 394: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

B2k +∑

p−1|2k

1

p∈ Z

I Example 4.1. B12 = −691/2730. The prime numbers p such thatp − 1 divides 12 are p = 2, 3, 5, 7, 13.

I We have1

2+

1

3+

1

5+

1

7+

1

13− 691

2730= 1 ∈ Z

I Example 4.2. B20 = −174611/330. The prime numbers p suchthat p − 1 divides 20 are p = 2, 3, 5, 11.

I We have1

2+

1

3+

1

5+

1

11− 174611

330= −528 ∈ Z

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 395: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

B2k +∑

p−1|2k

1

p∈ Z

I Example 4.1. B12 = −691/2730. The prime numbers p such thatp − 1 divides 12 are p = 2, 3, 5, 7, 13.

I We have1

2+

1

3+

1

5+

1

7+

1

13− 691

2730= 1 ∈ Z

I Example 4.2. B20 = −174611/330. The prime numbers p suchthat p − 1 divides 20 are p = 2, 3, 5, 11.

I We have1

2+

1

3+

1

5+

1

11− 174611

330= −528 ∈ Z

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 396: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

B2k +∑

p−1|2k

1

p∈ Z

I Example 4.1. B12 = −691/2730. The prime numbers p such thatp − 1 divides 12 are p = 2, 3, 5, 7, 13.

I We have1

2+

1

3+

1

5+

1

7+

1

13− 691

2730= 1 ∈ Z

I Example 4.2. B20 = −174611/330. The prime numbers p suchthat p − 1 divides 20 are p = 2, 3, 5, 11.

I We have1

2+

1

3+

1

5+

1

11− 174611

330= −528 ∈ Z

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 397: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

B2k +∑

p−1|2k

1

p∈ Z

I Example 4.1. B12 = −691/2730. The prime numbers p such thatp − 1 divides 12 are p = 2, 3, 5, 7, 13.

I We have1

2+

1

3+

1

5+

1

7+

1

13− 691

2730= 1 ∈ Z

I Example 4.2. B20 = −174611/330. The prime numbers p suchthat p − 1 divides 20 are p = 2, 3, 5, 11.

I We have1

2+

1

3+

1

5+

1

11− 174611

330= −528 ∈ Z

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 398: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The Almkvist-Meurman congruence

There is a famous congruence for the Bernoulli polynomials, due toAlmkvist and Meurman (1991).Theorem. (Almkvist-Meurman) Let h, k be positive integers. Let

Bn(x) := Bn(x)− Bn(0)

Then

knBn

(h

k

)∈ Z

Example 4.3. Choose B5(x) = x5 − 52x4 + 5

3x3 − x6 . Then

AM(5, h, k) = h5 − 5

2h4k +

5

3h3k2 − hk4

6

Example 4.4. Choose B8(x) = x8 − 4x7 + 143 x6 − 7

3x4 + 23x2 − 1

30 . Then

AM(8, h, k) = h8 − 4h7k +14

3h6k2 − 7

3h4k4 +

2

3h2k6

AM(5, h, k) and AM(8, h, k) are integer numbers for any h, k ∈ N/0.Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 399: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Generalized Bernoulli structure

I Are there generalizations of the Bernoulli numbers and polynomials?

I Surprisingly enough, there are very many: dozens, perhaps hundreds!

I Natural questions:

I A) is there a general structure behind all these generalizations?

I B) In particular, are there fundamental properties which arepreserved?

I C) Are there generalized zeta functions in correspondence?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 400: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Generalized Bernoulli structure

I Are there generalizations of the Bernoulli numbers and polynomials?

I Surprisingly enough, there are very many: dozens, perhaps hundreds!

I Natural questions:

I A) is there a general structure behind all these generalizations?

I B) In particular, are there fundamental properties which arepreserved?

I C) Are there generalized zeta functions in correspondence?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 401: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Generalized Bernoulli structure

I Are there generalizations of the Bernoulli numbers and polynomials?

I Surprisingly enough, there are very many: dozens, perhaps hundreds!

I Natural questions:

I A) is there a general structure behind all these generalizations?

I B) In particular, are there fundamental properties which arepreserved?

I C) Are there generalized zeta functions in correspondence?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 402: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Generalized Bernoulli structure

I Are there generalizations of the Bernoulli numbers and polynomials?

I Surprisingly enough, there are very many: dozens, perhaps hundreds!

I Natural questions:

I A) is there a general structure behind all these generalizations?

I B) In particular, are there fundamental properties which arepreserved?

I C) Are there generalized zeta functions in correspondence?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 403: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Generalized Bernoulli structure

I Are there generalizations of the Bernoulli numbers and polynomials?

I Surprisingly enough, there are very many: dozens, perhaps hundreds!

I Natural questions:

I A) is there a general structure behind all these generalizations?

I B) In particular, are there fundamental properties which arepreserved?

I C) Are there generalized zeta functions in correspondence?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 404: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Generalized Bernoulli structure

I Are there generalizations of the Bernoulli numbers and polynomials?

I Surprisingly enough, there are very many: dozens, perhaps hundreds!

I Natural questions:

I A) is there a general structure behind all these generalizations?

I B) In particular, are there fundamental properties which arepreserved?

I C) Are there generalized zeta functions in correspondence?

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 405: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The Universal Bernoulli Polynomials

P. T., Transactions of the AMS, 367, 7015-7028 (2015). Definition4.1. Let

G (t) =∞∑k=0

aktk+1

k + 1, (1)

be a formal group exponential, where akk∈N a real sequence, witha0 6= 0.The universal higher–order Bernoulli polynomials

B(α)k (x , c1, . . . , cn) ≡ B

(α)k (x) are defined by the relation(

t

G (t)

)αext =

∑k≥0

B(α)k (x)

tk

k!, x , α ∈ R. (2)

Particular cases. If ci = (−1)i , α = 1, then G (t) = et − 1, and theuniversal Bernoulli polynomials and numbers reduce to the standard ones.If ci = (−1)i , α > 1, we get back the higher-order Bernoulli polynomials.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 406: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

Universal Bernoulli numbers

The quantities Bk ∈ Q [c1, c2, . . .] coincide with Clarke’s universalBernoulli numbers. They satisfy a Universal Clausen–von Staudtcongruence. Let B0 = 1, and if n > 0 is even, then

Bn ≡ −∑p−1|n

p prime

cn/(p−1)p−1

pmod Z [c1, c2, ...] ; (3)

Let B1 = c1/2 and if n > 1 is odd, then

Bn ≡cn1 + cn−3

1 c32

mod Z [c1, c2, ...] . (4)

When cn = (−1)n, the celebrated Clausen–Von Staudt congruence forBernoulli numbers is obtained.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 407: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Special values of zeta functionsFamous CongruencesUniversal structures

The universal AM Theorem

Theorem 4.2. Let h > 0, k > 0, n ≥ 0 be integers. Consider thepolynomials defined by

t

G (t)ext =

∑m≥0

BGm (x)

tm

m!,

where

G (t) =∞∑k=0

aktk+1

k + 1, (5)

is a formal group exponential, such that cm ∈ Z for all m. Assume thatcp−1 ≡ 0, 1 mod p for all odd primes p, and either c1 ≡ c3 mod 2, or c1is odd and c3 even. Then

knBGn

(h

k

)∈ Z, (6)

where BGn (x) = BG

n (x)− BGn .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 408: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Formal Groups and Number Theory

Universal Bernoulli

Polynomials

and Congruences

Formal groups Integer sequences

Rota’s theory of

Delta Operators

Hyperfunctions

Theory of Functional

Equations

L-series and

L-functions

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 409: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Sequences of integer numbers

P. T., Transactions of the AMS, 367, 7015-7028 (2015).Sequences of integer numbers and Appell sequences of polynomials withinteger coefficients are constructed as a byproduct of the previous theory.Lemma 4.3. Consider a sequence of the form

t

G1 (t)− t

G2 (t)=∞∑k=0

Nk

2 k!tk , (7)

where G1 (t) and G2 (t) are formal group exponentials, defined as in formula

(5). Assume that cGj

i ∈ Z for all i ∈ N, j = 1, 2 and cG1p−1 ≡ cG2

p−1 mod p for all

p ≥ 2 (here cGjn denotes the nth coefficient of the expansion for the logarithm

associated to Gj). Then Nkk∈N is a sequence of integers.

Proof. The Bernoulli–type numbers BG1k and BG2

k associated with the formalgroup exponentials G1(t) and G2(t), under the previous assumptions mustsatisfy Clarke’s universal congruence. It follows that the difference BG1

k − BG2k

for k even is an integer and for k odd is a half–integer or an integer.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 410: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

a) The characteristic power series obtained as the difference between thepower series associated with the L–genus and that one associated withthe Todd genus, i.e.

Q(t) = t/ tanh t − t/(1− e−t)

is the generating function of the sequence

−1, 1, 0,−1, 0, 3, 0,−17, 0, 155, . . . (8)

b) Realizations of the universal Bernoulli polynomials can be constructedby using delta operators. Here we quote two generating functions ofthese classes of polynomials, related to certain difference delta operators:

∞∑k=0

BVk (x)

k!tk =

text

e3t − 2e2t + 2et − 2e−t + e−2t,

∞∑k=0

BVIIk (x)

k!tk =

−text

e4t − e3t + e2t − 2et + e−t − e−2t + e−3t. (9)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 411: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Integer sequences

By combining these delta operators (and different ones), one gets easily apriori infinitely many generating functions of integer sequences.b.1)

t(1 + et)

2(1 + et − e2t).

The sequence generated is 2, 6, 39, 324, 3365, 41958, . . .b.2)

−2t (1 + 2 cosh t + 4 cosh 2t − 6 sinh t)

(−6 + 8 cosh t)(2 + cosh t − cosh 2t − sinh t + sinh 2t + 2 sinh 3t).

The sequence generated is −7, 61,−642, 10127,−207110, 5001663, ...As an immediate consequence of the previous results, we can alsoconstruct new sequences of Appell polynomials possessing integercoefficients.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 412: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Polynomial sequences with integer coefficients

Lemma 4.4Assume the hypotheses of the previous Lemma. Then any sequence ofpolynomials of the form[

t

G1 (t)− t

G2 (t)

]ext =

∞∑k=0

Nk (x)

2k!tk (10)

is an Appell sequence with integer coefficients. As an example, consider thesequence of polynomials pn(x)n∈N, generated by

−te−

3t2 sec h( t

2)[4 + et(1 + 2et(−1 + et))

]2(−3 + 4 cosh t)(1 + (−2 + 4 cosh t) sinh t)

ext . (11)

It is easy to verify that (11) is the generating function of a sequence of Appellpolynomials. The first polynomials of the sequence are

p0(x) = −5, p1(x) = 29− 10x , p2(x) = −150 + 87x − 15x2,

p3(x) = 1279− 600x + 174x2 − 20x3, . . .

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 413: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

P. T., L–series and Hurwitz zeta functions associated with the universalformal group, Ann. Scuola Normale Superiore, IX, 1–12 (2010).There is a close correspondence among Formal Groups and AnalyticNumber Theory

I To any formal group law one can associate a family of generalizedBernoulli polynomials:

t

G (t)ext =

∞∑k=0

Bk (x)tk

k!, x ∈ R.

I a generalized Riemann zeta function (L-series) and a generalizedHurwitz zeta function,Let G (t) be a formal group exponential, such that e−at/G (t) is aC∞ function over R+, rapidly decreasing at infinity. The generalizedHurwitz zeta function associated with G is the function ζG (s, a),defined for Re(s) > 1 and a > 0 by

ζG (s, a) =1

Γ (s)

∫ ∞0

e−ax

G (x)x s−1dx . (12)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 414: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

P. T., L–series and Hurwitz zeta functions associated with the universalformal group, Ann. Scuola Normale Superiore, IX, 1–12 (2010).There is a close correspondence among Formal Groups and AnalyticNumber Theory

I To any formal group law one can associate a family of generalizedBernoulli polynomials:

t

G (t)ext =

∞∑k=0

Bk (x)tk

k!, x ∈ R.

I a generalized Riemann zeta function (L-series) and a generalizedHurwitz zeta function,Let G (t) be a formal group exponential, such that e−at/G (t) is aC∞ function over R+, rapidly decreasing at infinity. The generalizedHurwitz zeta function associated with G is the function ζG (s, a),defined for Re(s) > 1 and a > 0 by

ζG (s, a) =1

Γ (s)

∫ ∞0

e−ax

G (x)x s−1dx . (12)

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 415: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

I When a = 0, G (x) = 1− e−x , we obtain the standard Riemannzeta function:

ζ (s) =∞∑n=1

1

ns=

1

Γ (s)

∫ ∞0

1

ex − 1x s−1dx .

I When G (x) = 1− e−x , we obtain the classical Hurwitz zetafunction,

I Theorem 4.5. For any m ∈ N, the following property holds:

ζG (−m, a) = −BG ′

m+1 (a)

m + 1, (13)

where according to eq. (1), BG ′

m (x) is the m–th generalizedBernoulli polynomial associated with the formal group exponentialG ′(t) := −G (−t) (with α = 1).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 416: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

I When a = 0, G (x) = 1− e−x , we obtain the standard Riemannzeta function:

ζ (s) =∞∑n=1

1

ns=

1

Γ (s)

∫ ∞0

1

ex − 1x s−1dx .

I When G (x) = 1− e−x , we obtain the classical Hurwitz zetafunction,

I Theorem 4.5. For any m ∈ N, the following property holds:

ζG (−m, a) = −BG ′

m+1 (a)

m + 1, (13)

where according to eq. (1), BG ′

m (x) is the m–th generalizedBernoulli polynomial associated with the formal group exponentialG ′(t) := −G (−t) (with α = 1).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 417: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

I When a = 0, G (x) = 1− e−x , we obtain the standard Riemannzeta function:

ζ (s) =∞∑n=1

1

ns=

1

Γ (s)

∫ ∞0

1

ex − 1x s−1dx .

I When G (x) = 1− e−x , we obtain the classical Hurwitz zetafunction,

I Theorem 4.5. For any m ∈ N, the following property holds:

ζG (−m, a) = −BG ′

m+1 (a)

m + 1, (13)

where according to eq. (1), BG ′

m (x) is the m–th generalizedBernoulli polynomial associated with the formal group exponentialG ′(t) := −G (−t) (with α = 1).

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 418: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

II) The Tsallis entropy and the Riemann zeta function

The Tsallis entropy is the unique trace-form entropy possessing themultiplicative formal group as the composability law.

This group law is generated by the group exponential G (t) = e(1−q)t−11−q .

Now, if Re s > 1, and q < 1, we have the following result.Theorem 4.6.

1

Γ (s)

∫ ∞0

1e(1−q)t−1

1−q

ts−1dt =1

(1− q)s−1ζ (s) . (14)

In turn, the same group exponential generates the classical Bernoullinumbers and polynomials:

t

et − 1=∞∑k=0

Bktk

k!

text

et − 1=∞∑k=0

Bk(x)tk

k!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 419: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Classical Bernoulli

numbers

and polynomials

Riemann zeta function

Tsallis Entropy

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 420: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

III) The ζ-entropy

Problem. Can we construct entropies directly related with L-functions?Observe that ζ(x) ∈ A for x ∈ [0, 1).Definition 4.7. The functional

Sζ [p] :=kB

1− aζ ′(a)

W∑i=1

pi

(ζ(apσi )− pσi

σ+

1− ζ(a)

σ

)(15)

with a, σ ∈ (0, 1) will be called the zeta entropy.The entropy Sζ [p] satisfies the first three SK axioms and is weaklycomposable.

0.2 0.4 0.6 0.8 1.0

0.05

0.10

0.15

0.20

0.25

0.30

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 421: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

The Ramanujan entropy

The Ramanujan L-function is defined by the relation

L(s) =∞∑n=1

τ(n)

ns, Re s > 6,

where τ(n) is the Ramanujan tau function:∞∑n=1

τ(n)qn = q∏n≥1

(1− qn)24 = η(z)24, q = exp(2πiz), Im z > 0.

This series is absolutely and uniformly convergent, and can be analyticallycontinued to the whole complex plane.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 422: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Definition 4.8. The functional

Sτ [p] :=kB

1− aL′(a)

W∑i=1

pi

(L(apσi )− pσi

σ+

1− L(a)

σ

)(16)

will be called the Ramanujan’s τ -entropy. This functional is defined fora, σ ∈ (0, 1).The entropy Sζ [p] satisfies the first three SK axioms and is weaklycomposable.

0.2 0.4 0.6 0.8 1.0

0.05

0.10

0.15

0.20

0.25

0.30

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 423: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Resume

We have seen three different connections between Number Theory andNonadditive Entropies.

I NT I) The canonical probability distribution with the Hurwitz zetafunction

I NT II) The relation among generalized logarithms and Dirichletseries (and generalized Hurwitz functions). In particular, Tsallisentropy is directly related to the Riemann zeta function.

I NT III) The existence of nonadditive entropies directly constructedfrom L-functions.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 424: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Resume

We have seen three different connections between Number Theory andNonadditive Entropies.

I NT I) The canonical probability distribution with the Hurwitz zetafunction

I NT II) The relation among generalized logarithms and Dirichletseries (and generalized Hurwitz functions). In particular, Tsallisentropy is directly related to the Riemann zeta function.

I NT III) The existence of nonadditive entropies directly constructedfrom L-functions.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 425: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Resume

We have seen three different connections between Number Theory andNonadditive Entropies.

I NT I) The canonical probability distribution with the Hurwitz zetafunction

I NT II) The relation among generalized logarithms and Dirichletseries (and generalized Hurwitz functions). In particular, Tsallisentropy is directly related to the Riemann zeta function.

I NT III) The existence of nonadditive entropies directly constructedfrom L-functions.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 426: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Conclusions

I C I) The group-theoretical structure coming from thecomposability axiom determines crucially the form and theproperties of generalized entropies, in both cases of trace-formentropies (UGE) and non-trace form ones (Z-entropies).

I C II) In particular, the thermodynamical and information-theoreticalcontent of generalized entropies comes from the group structure.

I C III) The natural language to express the theory is that of formalgroup laws, as was elaborated by Bochner, Novikov, Serre, etc.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 427: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Conclusions

I C I) The group-theoretical structure coming from thecomposability axiom determines crucially the form and theproperties of generalized entropies, in both cases of trace-formentropies (UGE) and non-trace form ones (Z-entropies).

I C II) In particular, the thermodynamical and information-theoreticalcontent of generalized entropies comes from the group structure.

I C III) The natural language to express the theory is that of formalgroup laws, as was elaborated by Bochner, Novikov, Serre, etc.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 428: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Conclusions

I C I) The group-theoretical structure coming from thecomposability axiom determines crucially the form and theproperties of generalized entropies, in both cases of trace-formentropies (UGE) and non-trace form ones (Z-entropies).

I C II) In particular, the thermodynamical and information-theoreticalcontent of generalized entropies comes from the group structure.

I C III) The natural language to express the theory is that of formalgroup laws, as was elaborated by Bochner, Novikov, Serre, etc.

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 429: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Conclusions

I C IV) The same group structure is at the heart of manynumber-theoretical constructions: generalized Bernoullipolynomials, integer sequences, a large new class of L-series andL-functions, etc.

I C V) Nonextensive statistical mechanics can be related withL-functions!

I Main result: There are multiple connections among GroupTheory, Generalized Entropies and Analytic Number Theory,still to be explored!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 430: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Conclusions

I C IV) The same group structure is at the heart of manynumber-theoretical constructions: generalized Bernoullipolynomials, integer sequences, a large new class of L-series andL-functions, etc.

I C V) Nonextensive statistical mechanics can be related withL-functions!

I Main result: There are multiple connections among GroupTheory, Generalized Entropies and Analytic Number Theory,still to be explored!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 431: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Conclusions

I C IV) The same group structure is at the heart of manynumber-theoretical constructions: generalized Bernoullipolynomials, integer sequences, a large new class of L-series andL-functions, etc.

I C V) Nonextensive statistical mechanics can be related withL-functions!

I Main result: There are multiple connections among GroupTheory, Generalized Entropies and Analytic Number Theory,still to be explored!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 432: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Trace-form entropies

and

UGE

Universal Bernoulli

Polynomials

and Congruences

Formal groups

L-series

L-functions

Z-Entropies

Information Theory

Rota’s theory of

Delta Operators

Hyperfunctions

Algebraic Topology

Theory of Functional

Equations

Integer sequences

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY

Page 433: LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORYcomplex/Files/slidesptempesta.pdf · LECTURES ON GROUPS, ENTROPIES AND NUMBER THEORY Piergiulio Tempesta Universidad Complutense de

Number Theory and Statistical MechanicsPreliminaries: The Theory of L-functions

Bernoulli numbers and polynomialsFormal Groups and Number Theory

Universal congruences and integer sequencesPolynomial sequences with integer coefficientsFormal Groups and Analytic Number TheoryThe relation between the ζ function and Tsallis entropyThe zeta and Ramanujan entropies

Grazie!!

Piergiulio Tempesta, Universidad Complutense de Madrid and ICMAT GROUPS, ENTROPIES AND NUMBER THEORY