a counter-example to the hirsch conjecture...the mathematics of klee & grünbaum — seattle,...
TRANSCRIPT
![Page 1: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/1.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A counter-example to the Hirsch conjecture
Francisco Santos
Universidad de Cantabria, Spainhttp://personales.unican.es/santosf/Hirsch
The mathematics of Klee & Grünbaum — Seattle, July 30, 2010
1
![Page 2: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/2.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A counter-example to the Hirsch conjectureOr “Two theorems by Victor Klee and David Walkup”
Francisco Santos
Universidad de Cantabria, Spainhttp://personales.unican.es/santosf/Hirsch
The mathematics of Klee & Grünbaum — Seattle, July 30, 2010
2
![Page 3: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/3.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Two quotes by Victor Klee:
A good talk contains no proofs; a great talk contains notheorems.
Mathematical proofs should only be communicated inprivate and to consenting adults.
3
![Page 4: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/4.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
W A R N I N G
4
![Page 5: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/5.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
W A R N I N G
This talk contains materialthat may not be suited for all audiences.
4
![Page 6: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/6.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
W A R N I N GThis includes, but may not be limited to,mathematical theorems and proofs,
pictures of highly dimensional polytopes,and explicit coordinates for them.
4
![Page 7: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/7.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
W A R N I N GThis includes, but may not be limited to,mathematical theorems and proofs,
pictures of highly dimensional polytopes,and explicit coordinates for them.
4
![Page 8: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/8.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
W A R N I N GThis includes, but may not be limited to,mathematical theorems and proofs,
pictures of highly dimensional polytopes,and explicit coordinates for them.
4
![Page 9: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/9.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
W A R N I N GThis includes, but may not be limited to,mathematical theorems and proofs,
pictures of highly dimensional polytopes,and explicit coordinates for them.
4
![Page 10: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/10.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
W A R N I N GThis includes, but may not be limited to,mathematical theorems and proofs,
pictures of highly dimensional polytopes,and explicit coordinates for them.
4
![Page 11: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/11.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
W A R N I N GThis includes, but may not be limited to,mathematical theorems and proofs,
pictures of highly dimensional polytopes,and explicit coordinates for them.
4
![Page 12: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/12.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
W A R N I N GThis includes, but may not be limited to,mathematical theorems and proofs,
pictures of highly dimensional polytopes,and explicit coordinates for them.
4
![Page 13: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/13.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
W A R N I N GWe declare this room to be private for the following 45 minutes.
By staying in it you acknowledge to be an adult, and consent tobe exposed to such material.
4
![Page 14: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/14.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
W A R N I N GWe declare this room to be private for the following 45 minutes.
By staying in it you acknowledge to be an adult, and consent tobe exposed to such material.
4
![Page 15: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/15.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
W A R N I N GWe declare this room to be private for the following 45 minutes.
By staying in it you acknowledge to be an adult, and consent tobe exposed to such material.
4
![Page 16: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/16.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The graph of a polytope
Vertices and edges of a polytope P form a graph (finite,undirected)
The distance d(a,b) between vertices a and b is the length(number of edges) of the shortest path from a to b.
For example, d(a,b) = 2.
5
![Page 17: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/17.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The graph of a polytope
Vertices and edges of a polytope P form a graph (finite,undirected)
The distance d(a,b) between vertices a and b is the length(number of edges) of the shortest path from a to b.
For example, d(a,b) = 2.
5
![Page 18: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/18.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The graph of a polytope
Vertices and edges of a polytope P form a graph (finite,undirected)
The distance d(a,b) between vertices a and b is the length(number of edges) of the shortest path from a to b.
For example, d(a,b) = 2.
5
![Page 19: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/19.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The graph of a polytope
Vertices and edges of a polytope P form a graph (finite,undirected)
The diameter of G(P) (or of P) is the maximum distance amongits vertices:
δ(P) = max{d(a,b) : a,b ∈ vert(P)}.
5
![Page 20: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/20.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The Hirsch conjecture
Conjecture: Warren M. Hirsch (1957)For every polytope P with n facets and dimension d ,
δ(P) ≤ n − d .
Theorem (S. 2010+)There is a 43-dim. polytope with 86 facets and diameter 44.
CorollaryThere is an infinite family of non-Hirsch polytopes with diameter∼ (1 + ε)n, even in fixed dimension. (Best so far: ε = 1/43).
6
![Page 21: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/21.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The Hirsch conjecture
Conjecture: Warren M. Hirsch (1957)For every polytope P with n facets and dimension d ,
δ(P) ≤ n − d .
Theorem (S. 2010+)There is a 43-dim. polytope with 86 facets and diameter 44.
CorollaryThere is an infinite family of non-Hirsch polytopes with diameter∼ (1 + ε)n, even in fixed dimension. (Best so far: ε = 1/43).
6
![Page 22: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/22.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The Hirsch conjecture
Conjecture: Warren M. Hirsch (1957)For every polytope P with n facets and dimension d ,
δ(P) ≤ n − d .
Theorem (S. 2010+)There is a 43-dim. polytope with 86 facets and diameter 44.
CorollaryThere is an infinite family of non-Hirsch polytopes with diameter∼ (1 + ε)n, even in fixed dimension. (Best so far: ε = 1/43).
6
![Page 23: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/23.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Motivation: linear programming
A linear program is the problem of maximization / minimizationof a linear functional subject to linear inequality constraints.
The set of feasible solutions P = {x ∈ Rd : Mx ≤ b} is apolyhedron P with (at most) n facets.The optimal solution (if it exists) is always attained at avertex.The simplex method [Dantzig 1947] solves the linearprogram starting at any feasible vertex and moving alongthe graph of P, in a monotone fashion, until the optimum isattained.In particular, the Hirsch conjecture is related to thequestion of whether the simplex method is apolynomial-time algorithm.
7
![Page 24: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/24.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Motivation: linear programming
A linear program is the problem of maximization / minimizationof a linear functional subject to linear inequality constraints.
The set of feasible solutions P = {x ∈ Rd : Mx ≤ b} is apolyhedron P with (at most) n facets.The optimal solution (if it exists) is always attained at avertex.The simplex method [Dantzig 1947] solves the linearprogram starting at any feasible vertex and moving alongthe graph of P, in a monotone fashion, until the optimum isattained.In particular, the Hirsch conjecture is related to thequestion of whether the simplex method is apolynomial-time algorithm.
7
![Page 25: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/25.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Motivation: linear programming
A linear program is the problem of maximization / minimizationof a linear functional subject to linear inequality constraints.
The set of feasible solutions P = {x ∈ Rd : Mx ≤ b} is apolyhedron P with (at most) n facets.The optimal solution (if it exists) is always attained at avertex.The simplex method [Dantzig 1947] solves the linearprogram starting at any feasible vertex and moving alongthe graph of P, in a monotone fashion, until the optimum isattained.In particular, the Hirsch conjecture is related to thequestion of whether the simplex method is apolynomial-time algorithm.
7
![Page 26: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/26.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Motivation: linear programming
A linear program is the problem of maximization / minimizationof a linear functional subject to linear inequality constraints.
The set of feasible solutions P = {x ∈ Rd : Mx ≤ b} is apolyhedron P with (at most) n facets.The optimal solution (if it exists) is always attained at avertex.The simplex method [Dantzig 1947] solves the linearprogram starting at any feasible vertex and moving alongthe graph of P, in a monotone fashion, until the optimum isattained.In particular, the Hirsch conjecture is related to thequestion of whether the simplex method is apolynomial-time algorithm.
7
![Page 27: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/27.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Motivation: linear programming
A linear program is the problem of maximization / minimizationof a linear functional subject to linear inequality constraints.
The set of feasible solutions P = {x ∈ Rd : Mx ≤ b} is apolyhedron P with (at most) n facets.The optimal solution (if it exists) is always attained at avertex.The simplex method [Dantzig 1947] solves the linearprogram starting at any feasible vertex and moving alongthe graph of P, in a monotone fashion, until the optimum isattained.In particular, the Hirsch conjecture is related to thequestion of whether the simplex method is apolynomial-time algorithm.
7
![Page 28: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/28.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Motivation: linear programming
A linear program is the problem of maximization / minimizationof a linear functional subject to linear inequality constraints.
The set of feasible solutions P = {x ∈ Rd : Mx ≤ b} is apolyhedron P with (at most) n facets.The optimal solution (if it exists) is always attained at avertex.The simplex method [Dantzig 1947] solves the linearprogram starting at any feasible vertex and moving alongthe graph of P, in a monotone fashion, until the optimum isattained.In particular, the Hirsch conjecture is related to thequestion of whether the simplex method is apolynomial-time algorithm.
7
![Page 29: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/29.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Complexity of linear programming
There are more recent algorithms for linear programming whichare proved to be polynomial: (ellipsoid [1979], interior point[1984]). But the simplex method is still one of the most oftenused, for its simplicity and practical efficiency:
8
![Page 30: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/30.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Complexity of linear programming
There are more recent algorithms for linear programming whichare proved to be polynomial: (ellipsoid [1979], interior point[1984]). But the simplex method is still one of the most oftenused, for its simplicity and practical efficiency:
8
![Page 31: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/31.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Complexity of linear programming
There are more recent algorithms for linear programming whichare proved to be polynomial: (ellipsoid [1979], interior point[1984]). But the simplex method is still one of the most oftenused, for its simplicity and practical efficiency:
The number of steps to solve a problem with mequality constraints in n nonnegative variables isalmost always at most a small multiple of m, say 3m.
The simplex method has remained, if not the methodof choice, a method of choice, usually competitivewith, and on some classes of problems superior to, themore modern approaches.
(M. Todd, 2010)
8
![Page 32: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/32.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Complexity of linear programming
There are more recent algorithms for linear programming whichare proved to be polynomial: (ellipsoid [1979], interior point[1984]). But the simplex method is still one of the most oftenused, for its simplicity and practical efficiency:
The number of steps to solve a problem with mequality constraints in n nonnegative variables isalmost always at most a small multiple of m, say 3m.
The simplex method has remained, if not the methodof choice, a method of choice, usually competitivewith, and on some classes of problems superior to, themore modern approaches.
(M. Todd, 2010)
8
![Page 33: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/33.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Complexity of linear programming
Besides, the methods known are not strongly polynomial. Theyare polynomial in the “bit model” but not in the “real machinemodel” [Blum-Shub-Smale 1989]).
Finding strongly polynomial algorithms for linear programmingis one of the “mathematical problems for the 21st century"according to [Smale 2000]. A polynomial pivot rule would solvethis problem in the affirmative.
... in any case, ...Knowing the behavior of polytope diametersis one of the most fundamental openquestions in geometric combinatorics.
9
![Page 34: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/34.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Complexity of linear programming
Besides, the methods known are not strongly polynomial. Theyare polynomial in the “bit model” but not in the “real machinemodel” [Blum-Shub-Smale 1989]).
Finding strongly polynomial algorithms for linear programmingis one of the “mathematical problems for the 21st century"according to [Smale 2000]. A polynomial pivot rule would solvethis problem in the affirmative.
... in any case, ...Knowing the behavior of polytope diametersis one of the most fundamental openquestions in geometric combinatorics.
9
![Page 35: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/35.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Complexity of linear programming
Besides, the methods known are not strongly polynomial. Theyare polynomial in the “bit model” but not in the “real machinemodel” [Blum-Shub-Smale 1989]).
Finding strongly polynomial algorithms for linear programmingis one of the “mathematical problems for the 21st century"according to [Smale 2000]. A polynomial pivot rule would solvethis problem in the affirmative.
... in any case, ...Knowing the behavior of polytope diametersis one of the most fundamental openquestions in geometric combinatorics.
9
![Page 36: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/36.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Complexity of linear programming
Besides, the methods known are not strongly polynomial. Theyare polynomial in the “bit model” but not in the “real machinemodel” [Blum-Shub-Smale 1989]).
Finding strongly polynomial algorithms for linear programmingis one of the “mathematical problems for the 21st century"according to [Smale 2000]. A polynomial pivot rule would solvethis problem in the affirmative.
... in any case, ...Knowing the behavior of polytope diametersis one of the most fundamental openquestions in geometric combinatorics.
9
![Page 37: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/37.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Complexity of linear programming
Besides, the methods known are not strongly polynomial. Theyare polynomial in the “bit model” but not in the “real machinemodel” [Blum-Shub-Smale 1989]).
Finding strongly polynomial algorithms for linear programmingis one of the “mathematical problems for the 21st century"according to [Smale 2000]. A polynomial pivot rule would solvethis problem in the affirmative.
... in any case, ...Knowing the behavior of polytope diametersis one of the most fundamental openquestions in geometric combinatorics.
9
![Page 38: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/38.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Some known cases
Hirsch conjecture holds ford ≤ 3: [Klee 1966].n − d ≤ 6: [Klee-Walkup, 1967] [Bremner-Schewe, 2008]H(9,4) = H(10,4) = 5 [Klee-Walkup, 1967]H(11,4) = 6 [Schuchert, 1995],H(12,4) = 7 [Bremner et al. >2009].0-1 polytopes [Naddef 1989]Polynomial bound for network flow polytopes [Goldfarb1992, Orlin 1997]
10
![Page 39: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/39.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Some known cases
Hirsch conjecture holds ford ≤ 3: [Klee 1966].n − d ≤ 6: [Klee-Walkup, 1967] [Bremner-Schewe, 2008]H(9,4) = H(10,4) = 5 [Klee-Walkup, 1967]H(11,4) = 6 [Schuchert, 1995],H(12,4) = 7 [Bremner et al. >2009].0-1 polytopes [Naddef 1989]Polynomial bound for network flow polytopes [Goldfarb1992, Orlin 1997]
10
![Page 40: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/40.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Some known cases
Hirsch conjecture holds ford ≤ 3: [Klee 1966].n − d ≤ 6: [Klee-Walkup, 1967] [Bremner-Schewe, 2008]H(9,4) = H(10,4) = 5 [Klee-Walkup, 1967]H(11,4) = 6 [Schuchert, 1995],H(12,4) = 7 [Bremner et al. >2009].0-1 polytopes [Naddef 1989]Polynomial bound for network flow polytopes [Goldfarb1992, Orlin 1997]
10
![Page 41: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/41.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Some known cases
Hirsch conjecture holds ford ≤ 3: [Klee 1966].n − d ≤ 6: [Klee-Walkup, 1967] [Bremner-Schewe, 2008]H(9,4) = H(10,4) = 5 [Klee-Walkup, 1967]H(11,4) = 6 [Schuchert, 1995],H(12,4) = 7 [Bremner et al. >2009].0-1 polytopes [Naddef 1989]Polynomial bound for network flow polytopes [Goldfarb1992, Orlin 1997]
10
![Page 42: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/42.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Some known cases
Hirsch conjecture holds ford ≤ 3: [Klee 1966].n − d ≤ 6: [Klee-Walkup, 1967] [Bremner-Schewe, 2008]H(9,4) = H(10,4) = 5 [Klee-Walkup, 1967]H(11,4) = 6 [Schuchert, 1995],H(12,4) = 7 [Bremner et al. >2009].0-1 polytopes [Naddef 1989]Polynomial bound for network flow polytopes [Goldfarb1992, Orlin 1997]
10
![Page 43: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/43.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Some known cases
Hirsch conjecture holds ford ≤ 3: [Klee 1966].n − d ≤ 6: [Klee-Walkup, 1967] [Bremner-Schewe, 2008]H(9,4) = H(10,4) = 5 [Klee-Walkup, 1967]H(11,4) = 6 [Schuchert, 1995],H(12,4) = 7 [Bremner et al. >2009].0-1 polytopes [Naddef 1989]Polynomial bound for network flow polytopes [Goldfarb1992, Orlin 1997]
10
![Page 44: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/44.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Some known cases
Hirsch conjecture holds ford ≤ 3: [Klee 1966].n − d ≤ 6: [Klee-Walkup, 1967] [Bremner-Schewe, 2008]H(9,4) = H(10,4) = 5 [Klee-Walkup, 1967]H(11,4) = 6 [Schuchert, 1995],H(12,4) = 7 [Bremner et al. >2009].0-1 polytopes [Naddef 1989]Polynomial bound for network flow polytopes [Goldfarb1992, Orlin 1997]
10
![Page 45: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/45.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
General bounds
A “quasi-polynomial” boundTheorem (Kalai-Kleitman 1992): For every d-polytope with nfacets
δ(P) ≤ nlog2 d+2.
A linear bound in fixed dimensionTheorem (Barnette 1967, Larman 1970): For every d-polytopewith n facets:
δ(P) ≤ n2d−3.
11
![Page 46: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/46.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
General bounds
A “quasi-polynomial” boundTheorem (Kalai-Kleitman 1992): For every d-polytope with nfacets
δ(P) ≤ nlog2 d+2.
A linear bound in fixed dimensionTheorem (Barnette 1967, Larman 1970): For every d-polytopewith n facets:
δ(P) ≤ n2d−3.
11
![Page 47: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/47.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
General bounds
A “quasi-polynomial” boundTheorem (Kalai-Kleitman 1992): For every d-polytope with nfacets
δ(P) ≤ nlog2 d+2.
A linear bound in fixed dimensionTheorem (Barnette 1967, Larman 1970): For every d-polytopewith n facets:
δ(P) ≤ n2d−3.
11
![Page 48: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/48.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Polynomial bounds, under perturbation
Given a linear program with d variables and n restrictions, weconsider a random perturbation of the matrix, within aparameter ε (normal distribution).
Theorem [Spielman-Teng 2004] [Vershynin 2006]The expected diameter of the perturbed polyhedron ispolynomial in d and ε−1, and polylogarithmic in n.
12
![Page 49: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/49.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Polynomial bounds, under perturbation
Given a linear program with d variables and n restrictions, weconsider a random perturbation of the matrix, within aparameter ε (normal distribution).
Theorem [Spielman-Teng 2004] [Vershynin 2006]The expected diameter of the perturbed polyhedron ispolynomial in d and ε−1, and polylogarithmic in n.
12
![Page 50: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/50.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
13
![Page 51: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/51.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Theorem 1: The d-step TheoremKlee and Walkup, 1967
13
![Page 52: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/52.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Why is n − d a “reasonable” bound?
Hirsch conjecture has the following interpretations:
14
![Page 53: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/53.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Why is n − d a “reasonable” bound?
Hirsch conjecture has the following interpretations:
Given any two vertices a and b of a simple polytope P:
14
![Page 54: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/54.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Why is n − d a “reasonable” bound?
Hirsch conjecture has the following interpretations:
Given any two vertices a and b of a simple polytope P:
non-revisiting path conjectureIt is possible to go from a to b so that at each step we enter anew facet, one that we had not visited before.
non-revisiting path⇒ Hirsch.
14
![Page 55: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/55.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Why is n − d a “reasonable” bound?
Hirsch conjecture has the following interpretations:
Given any two vertices a and b of a simple polytope P:
non-revisiting path conjectureIt is possible to go from a to b so that at each step we enter anew facet, one that we had not visited before.
non-revisiting path⇒ Hirsch.
14
![Page 56: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/56.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Why is n − d a “reasonable” bound?
Hirsch conjecture has the following interpretations:
In particular: assume n = 2d and let a and b be two comple-mentary vertices (no common facet):
14
![Page 57: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/57.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Why is n − d a “reasonable” bound?
Hirsch conjecture has the following interpretations:
In particular: assume n = 2d and let a and b be two comple-mentary vertices (no common facet):
d-step conjectureIt is possible to go from a to b so that at each step we abandona facet containing a and we enter a facet containing b.
“d-step conjecture”⇒ Hirsch for n = 2d .
14
![Page 58: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/58.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Why is n − d a “reasonable” bound?
Hirsch conjecture has the following interpretations:
In particular: assume n = 2d and let a and b be two comple-mentary vertices (no common facet):
d-step conjectureIt is possible to go from a to b so that at each step we abandona facet containing a and we enter a facet containing b.
“d-step conjecture”⇒ Hirsch for n = 2d .
14
![Page 59: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/59.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The d-step Theorem
Theorem 1 (Klee-Walkup 1967)Hirsch⇔ d-step⇔ non-revisiting path.
Proof: Let H(n,d) = max{δ(P) : P is a d-polytope with nfacets}. The key step in the proof is to show that for any k :
· · · ≤ H(2k − 1, k − 1) ≤ H(2k , k) = H(2k + 1, k + 1) = · · ·
15
![Page 60: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/60.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The d-step Theorem
Theorem 1 (Klee-Walkup 1967)Hirsch⇔ d-step⇔ non-revisiting path.
Proof: Let H(n,d) = max{δ(P) : P is a d-polytope with nfacets}. The key step in the proof is to show that for any k :
· · · ≤ H(2k − 1, k − 1) ≤ H(2k , k) = H(2k + 1, k + 1) = · · ·
15
![Page 61: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/61.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The d-step Theorem
Theorem 1 (Klee-Walkup 1967)Hirsch⇔ d-step⇔ non-revisiting path.
Proof: Let H(n,d) = max{δ(P) : P is a d-polytope with nfacets}. The key step in the proof is to show that for any k :
· · · ≤ H(2k − 1, k − 1) ≤ H(2k , k) = H(2k + 1, k + 1) = · · ·
15
![Page 62: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/62.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The d-step Theorem
Theorem 1 (Klee-Walkup 1967)Hirsch⇔ d-step⇔ non-revisiting path.
Proof: Let H(n,d) = max{δ(P) : P is a d-polytope with nfacets}. The key step in the proof is to show that for any k :
· · · ≤ H(2k − 1, k − 1) ≤ H(2k , k) = H(2k + 1, k + 1) = · · ·
15
![Page 63: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/63.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The d-step Theorem
Theorem 1 (Klee-Walkup 1967)Hirsch⇔ d-step⇔ non-revisiting path.
Proof: Let H(n,d) = max{δ(P) : P is a d-polytope with nfacets}. The key step in the proof is to show that for any k :
· · · ≤ H(2k − 1, k − 1) ≤ H(2k , k) = H(2k + 1, k + 1) = · · ·
15
![Page 64: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/64.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The d-step Theorem
Theorem 1 (Klee-Walkup 1967)Hirsch⇔ d-step⇔ non-revisiting path.
Proof: Let H(n,d) = max{δ(P) : P is a d-polytope with nfacets}. The key step in the proof is to show that for any k :
· · · ≤ H(2k − 1, k − 1) ≤ H(2k , k) = H(2k + 1, k + 1) = · · ·
15
![Page 65: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/65.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The d-step Theorem
Theorem 1 (Klee-Walkup 1967)Hirsch⇔ d-step⇔ non-revisiting path.
Proof: Let H(n,d) = max{δ(P) : P is a d-polytope with nfacets}. The key step in the proof is to show that for any k :
· · · ≤ H(2k − 1, k − 1) ≤ H(2k , k) = H(2k + 1, k + 1) = · · ·
That is to say:1) H(n,d) ≤ H(n + 1,d + 1), for all n and d .2) H(n − 1,d − 1) ≥ H(n,d), when n < 2d .
15
![Page 66: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/66.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The d-step Theorem
Theorem 1 (Klee-Walkup 1967)Hirsch⇔ d-step⇔ non-revisiting path.
Proof: Let H(n,d) = max{δ(P) : P is a d-polytope with nfacets}. The key step in the proof is to show that for any k :
· · · ≤ H(2k − 1, k − 1) ≤ H(2k , k) = H(2k + 1, k + 1) = · · ·
2) H(n − 1,d − 1) ≥ H(n,d), when n < 2d :Since n < 2d , every pair of vertices a and b lie in acommon facet F , which is a polytope with one lessdimension and (at least) one less facet. Hence,dP(a,b) ≤ dF (a,b) ≤ H(n − 1,d − 1).
15
![Page 67: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/67.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The d-step Theorem
Theorem 1 (Klee-Walkup 1967)Hirsch⇔ d-step⇔ non-revisiting path.
Proof: Let H(n,d) = max{δ(P) : P is a d-polytope with nfacets}. The key step in the proof is to show that for any k :
· · · ≤ H(2k − 1, k − 1) ≤ H(2k , k) = H(2k + 1, k + 1) = · · ·
2) H(n − 1,d − 1) ≥ H(n,d), when n < 2d :Since n < 2d , every pair of vertices a and b lie in acommon facet F , which is a polytope with one lessdimension and (at least) one less facet. Hence,dP(a,b) ≤ dF (a,b) ≤ H(n − 1,d − 1).
15
![Page 68: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/68.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The d-step Theorem
Theorem 1 (Klee-Walkup 1967)Hirsch⇔ d-step⇔ non-revisiting path.
Proof: Let H(n,d) = max{δ(P) : P is a d-polytope with nfacets}. The key step in the proof is to show that for any k :
· · · ≤ H(2k − 1, k − 1) ≤ H(2k , k) = H(2k + 1, k + 1) = · · ·
2) H(n − 1,d − 1) ≥ H(n,d), when n < 2d :Since n < 2d , every pair of vertices a and b lie in acommon facet F , which is a polytope with one lessdimension and (at least) one less facet. Hence,dP(a,b) ≤ dF (a,b) ≤ H(n − 1,d − 1).
15
![Page 69: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/69.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The d-step Theorem
Theorem 1 (Klee-Walkup 1967)Hirsch⇔ d-step⇔ non-revisiting path.
Proof: Let H(n,d) = max{δ(P) : P is a d-polytope with nfacets}. The key step in the proof is to show that for any k :
· · · ≤ H(2k − 1, k − 1) ≤ H(2k , k) = H(2k + 1, k + 1) = · · ·
1) H(n,d) ≤ H(n + 1,d + 1), for all n and d :Choose an arbitrary facet F of P. Let P ′ be the wedge ofP over F . Then:
∀a,b ∈ vert(P), ∃a′,b′ ∈ vert(P ′), dP′(a′,b′) ≥ dP(a,b).
15
![Page 70: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/70.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The d-step Theorem
Theorem 1 (Klee-Walkup 1967)Hirsch⇔ d-step⇔ non-revisiting path.
Proof: Let H(n,d) = max{δ(P) : P is a d-polytope with nfacets}. The key step in the proof is to show that for any k :
· · · ≤ H(2k − 1, k − 1) ≤ H(2k , k) = H(2k + 1, k + 1) = · · ·
1) H(n,d) ≤ H(n + 1,d + 1), for all n and d :Choose an arbitrary facet F of P. Let P ′ be the wedge ofP over F . Then:
∀a,b ∈ vert(P), ∃a′,b′ ∈ vert(P ′), dP′(a′,b′) ≥ dP(a,b).
15
![Page 71: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/71.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The d-step Theorem
Theorem 1 (Klee-Walkup 1967)Hirsch⇔ d-step⇔ non-revisiting path.
Proof: Let H(n,d) = max{δ(P) : P is a d-polytope with nfacets}. The key step in the proof is to show that for any k :
· · · ≤ H(2k − 1, k − 1) ≤ H(2k , k) = H(2k + 1, k + 1) = · · ·
1) H(n,d) ≤ H(n + 1,d + 1), for all n and d :Choose an arbitrary facet F of P. Let P ′ be the wedge ofP over F . Then:
∀a,b ∈ vert(P), ∃a′,b′ ∈ vert(P ′), dP′(a′,b′) ≥ dP(a,b).
15
![Page 72: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/72.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The d-step Theorem
Theorem 1 (Klee-Walkup 1967)Hirsch⇔ d-step⇔ non-revisiting path.
Proof: Let H(n,d) = max{δ(P) : P is a d-polytope with nfacets}. The key step in the proof is to show that for any k :
· · · ≤ H(2k − 1, k − 1) ≤ H(2k , k) = H(2k + 1, k + 1) = · · ·
1) H(n,d) ≤ H(n + 1,d + 1), for all n and d :Choose an arbitrary facet F of P. Let P ′ be the wedge ofP over F . Then:
∀a,b ∈ vert(P), ∃a′,b′ ∈ vert(P ′), dP′(a′,b′) ≥ dP(a,b).
15
![Page 73: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/73.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Wedging, a.k.a. one-point-suspension
vF
w
u
16
![Page 74: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/74.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Wedging, a.k.a. one-point-suspension
v
d(a’, b’)=2
b’
F
w
d(a, b)=2
b
a
ua’
16
![Page 75: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/75.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Spindles
DefinitionA spindle is a polytope P with two distinguished vertices u andv such that every facet contains either u or v (but not both).
u u
vv
DefinitionThe length of aspindle is thegraph distancefrom u to v .
17
![Page 76: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/76.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Spindles
DefinitionA spindle is a polytope P with two distinguished vertices u andv such that every facet contains either u or v (but not both).
u u
vv
DefinitionThe length of aspindle is thegraph distancefrom u to v .
17
![Page 77: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/77.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Spindles
Theorem (Generalized d-step, spindle version)
Let P be a spindle of dimension d, with n > 2d facets andlength δ.Then there is another spindle P ′ of dimension d + 1, with n + 1facets and length δ + 1.
That is: we can increase the dimension, length and number offacets of a spindle, all by one, until n = 2d .
CorollaryIn particular, if a spindle P has length > d then there is anotherspindle P ′ (of dimension n − d, with 2n − 2d facets, and length≥ δ + n − 2d > n − d) that violates the Hirsch conjecture.
18
![Page 78: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/78.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Spindles
Theorem (Generalized d-step, spindle version)
Let P be a spindle of dimension d, with n > 2d facets andlength δ.Then there is another spindle P ′ of dimension d + 1, with n + 1facets and length δ + 1.
That is: we can increase the dimension, length and number offacets of a spindle, all by one, until n = 2d .
CorollaryIn particular, if a spindle P has length > d then there is anotherspindle P ′ (of dimension n − d, with 2n − 2d facets, and length≥ δ + n − 2d > n − d) that violates the Hirsch conjecture.
18
![Page 79: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/79.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Spindles
Theorem (Generalized d-step, spindle version)
Let P be a spindle of dimension d, with n > 2d facets andlength δ.Then there is another spindle P ′ of dimension d + 1, with n + 1facets and length δ + 1.
That is: we can increase the dimension, length and number offacets of a spindle, all by one, until n = 2d .
CorollaryIn particular, if a spindle P has length > d then there is anotherspindle P ′ (of dimension n − d, with 2n − 2d facets, and length≥ δ + n − 2d > n − d) that violates the Hirsch conjecture.
18
![Page 80: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/80.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Prismatoids
DefinitionA prismatoid is a polytope Q with two (parallel) facets Q+ andQ− containing all vertices.
Q+
Q−
Q
DefinitionThe width of aprismatoid is thedual-graphdistance from Q+
to Q−.
19
![Page 81: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/81.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Prismatoids
DefinitionA prismatoid is a polytope Q with two (parallel) facets Q+ andQ− containing all vertices.
Q+
Q−
Q
DefinitionThe width of aprismatoid is thedual-graphdistance from Q+
to Q−.
19
![Page 82: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/82.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Prismatoids
Theorem (Generalized d-step, prismatoid version)
Let Q be a prismatoid of dimension d, with n > 2d vertices andwidth δ.Then there is another prismatoid Q′ of dimension d + 1, withn + 1 vertices and width δ + 1.
That is: we can increase the dimension, width and number ofvertices of a prismatoid, all by one, until n = 2d .
20
![Page 83: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/83.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Prismatoids
Theorem (Generalized d-step, prismatoid version)
Let Q be a prismatoid of dimension d, with n > 2d vertices andwidth δ.Then there is another prismatoid Q′ of dimension d + 1, withn + 1 vertices and width δ + 1.
That is: we can increase the dimension, width and number ofvertices of a prismatoid, all by one, until n = 2d .
20
![Page 84: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/84.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The generalized d-step Theroem
Proof.
Q ⊂ R2
Q+
Q−Q̃−
Q̃ ⊂ R3
Q̃+
w
Q̃− := o. p. s.v(Q−)
Q+
w
o. p. s.v(Q) ⊂ R3
v
u
u
21
![Page 85: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/85.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Width of prismtoids
So, to disprove the Hirsch Conjecture we only need to find aprismatoid of dimension d and width larger than d . Its numberof vertices and facets is irrelevant!!!
QuestionDo they exist?
3-prismatoids have width at most 3 (exercise).4-prismatoids have width at most 4 [S., July 2010].5-prismatoids of width 6 exist [S., May 2010].
22
![Page 86: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/86.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Width of prismtoids
So, to disprove the Hirsch Conjecture we only need to find aprismatoid of dimension d and width larger than d . Its numberof vertices and facets is irrelevant!!!
QuestionDo they exist?
3-prismatoids have width at most 3 (exercise).4-prismatoids have width at most 4 [S., July 2010].5-prismatoids of width 6 exist [S., May 2010].
22
![Page 87: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/87.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Width of prismtoids
So, to disprove the Hirsch Conjecture we only need to find aprismatoid of dimension d and width larger than d . Its numberof vertices and facets is irrelevant!!!
QuestionDo they exist?
3-prismatoids have width at most 3 (exercise).4-prismatoids have width at most 4 [S., July 2010].5-prismatoids of width 6 exist [S., May 2010].
22
![Page 88: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/88.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Width of prismtoids
So, to disprove the Hirsch Conjecture we only need to find aprismatoid of dimension d and width larger than d . Its numberof vertices and facets is irrelevant!!!
QuestionDo they exist?
3-prismatoids have width at most 3 (exercise).4-prismatoids have width at most 4 [S., July 2010].5-prismatoids of width 6 exist [S., May 2010].
22
![Page 89: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/89.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Width of prismtoids
So, to disprove the Hirsch Conjecture we only need to find aprismatoid of dimension d and width larger than d . Its numberof vertices and facets is irrelevant!!!
QuestionDo they exist?
3-prismatoids have width at most 3 (exercise).4-prismatoids have width at most 4 [S., July 2010].5-prismatoids of width 6 exist [S., May 2010].
22
![Page 90: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/90.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Width of prismtoids
So, to disprove the Hirsch Conjecture we only need to find aprismatoid of dimension d and width larger than d . Its numberof vertices and facets is irrelevant!!!
QuestionDo they exist?
3-prismatoids have width at most 3 (exercise).4-prismatoids have width at most 4 [S., July 2010].5-prismatoids of width 6 exist [S., May 2010].
22
![Page 91: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/91.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
23
![Page 92: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/92.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Theorem 2: A non-Hirsch 4-polyhedronKlee and Walkup, 1967
23
![Page 93: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/93.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Combinatorics of prismatoids
Analyzing the combinatorics of a d-prismatoid Q can be donevia an intermediate slice . . .
Q+
Q−
Q ∩ HH
Q
24
![Page 94: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/94.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Combinatorics of prismatoids
. . . which equals the Minkowski sum Q+ + Q− of the two basesQ+ and Q−. The normal fan of Q+ + Q− equals the “superposi-tion” of those of Q+ and Q−.
+ 12
12 =
24
![Page 95: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/95.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Combinatorics of prismatoids
. . . which equals the Minkowski sum Q+ + Q− of the two basesQ+ and Q−. The normal fan of Q+ + Q− equals the “superposi-tion” of those of Q+ and Q−.
+ 12
12 =
24
![Page 96: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/96.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Combinatorics of prismatoids
. . . which equals the Minkowski sum Q+ + Q− of the two basesQ+ and Q−. The normal fan of Q+ + Q− equals the “superposi-tion” of those of Q+ and Q−.
=+ 12
12
24
![Page 97: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/97.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Combinatorics of prismatoids
So: the combinatorics of Q follows from the superposition ofthe normal fans of Q+ and Q−.
RemarkThe normal fan of a d − 1-polytope can be thought of as a(geodesic, polytopal) cell decomposition (“map”) of thed − 2-sphere.
Conclusion4-prismatoids⇔ pairs of maps in the 2-sphere.
5-prismatoids⇔ pairs of “maps” in the 3-sphere.
25
![Page 98: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/98.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Combinatorics of prismatoids
So: the combinatorics of Q follows from the superposition ofthe normal fans of Q+ and Q−.
RemarkThe normal fan of a d − 1-polytope can be thought of as a(geodesic, polytopal) cell decomposition (“map”) of thed − 2-sphere.
Conclusion4-prismatoids⇔ pairs of maps in the 2-sphere.
5-prismatoids⇔ pairs of “maps” in the 3-sphere.
25
![Page 99: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/99.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Combinatorics of prismatoids
So: the combinatorics of Q follows from the superposition ofthe normal fans of Q+ and Q−.
RemarkThe normal fan of a d − 1-polytope can be thought of as a(geodesic, polytopal) cell decomposition (“map”) of thed − 2-sphere.
Conclusion4-prismatoids⇔ pairs of maps in the 2-sphere.
5-prismatoids⇔ pairs of “maps” in the 3-sphere.
25
![Page 100: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/100.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Example: (part of) a 4-prismatoid
4-prismatoid of width > 4m
pair of (geodesic, polytopal) maps in S2 so that twosteps do not let you go from a blue vertex to a red vertex.
26
![Page 101: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/101.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Example: (part of) a 4-prismatoid
4-prismatoid of width > 4m
pair of (geodesic, polytopal) maps in S2 so that twosteps do not let you go from a blue vertex to a red vertex.
26
![Page 102: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/102.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The Klee-Walkup (unbounded) 4-spindle
Klee and Walkup, in 1967, disproved the Hirsch conjecture:
Theorem 2 (Klee-Walkup 1967)There is an unbounded 4-polyhedron with 8 facets anddiameter 5.
27
![Page 103: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/103.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The Klee-Walkup (unbounded) 4-spindle
Klee and Walkup, in 1967, disproved the Hirsch conjecture:
Theorem 2 (Klee-Walkup 1967)There is an unbounded 4-polyhedron with 8 facets anddiameter 5.
27
![Page 104: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/104.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The Klee-Walkup (unbounded) 4-spindle
Klee and Walkup, in 1967, disproved the Hirsch conjecture:
Theorem 2 (Klee-Walkup 1967)There is an unbounded 4-polyhedron with 8 facets anddiameter 5.
The Klee-Walkup polytope is an “unbounded 4-spindle”. Whatis the corresponding “superposition of two (geodesic, polytopal)maps” in a surface?
27
![Page 105: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/105.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The Klee-Walkup (unbounded) 4-spindle
d
fe
h
g
b
a
c
28
![Page 106: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/106.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The Klee-Walkup (unbounded) 4-spindle
c
fe
h
g
b
a
d
28
![Page 107: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/107.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The Klee-Walkup (unbounded) 4-spindle
28
![Page 108: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/108.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 4-dimensional prismatoid of width > 4?
Replicating the basic structure of the Klee-Walkup polytope wecan get a “non-Hirsch” pair of maps in the plane:
29
![Page 109: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/109.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 4-dimensional prismatoid of width > 4?
Replicating the basic structure of the Klee-Walkup polytope wecan get a “non-Hirsch” pair of maps in the plane:
29
![Page 110: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/110.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 4-dimensional prismatoid of width > 4?
Replicating the basic structure of the Klee-Walkup polytope wecan get a “non-Hirsch” pair of maps in the plane:
29
![Page 111: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/111.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 4-dimensional prismatoid of width > 4?
29
![Page 112: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/112.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 4-dimensional prismatoid of width > 4?
29
![Page 113: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/113.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 4-dimensional prismatoid of width > 4?
Surprisingly enough:
Theorem (S., July 2010)There is no “non-Hirsch” pair of maps in the 2-sphere.
Proof (rough idea of).Every pair of non-Hirsch maps on a surface necessarilycontains certain “zig-zag alternating cycles”, and no such cyclecan bound a 2-ball.
30
![Page 114: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/114.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 4-dimensional prismatoid of width > 4?
Surprisingly enough:
Theorem (S., July 2010)There is no “non-Hirsch” pair of maps in the 2-sphere.
Proof (rough idea of).Every pair of non-Hirsch maps on a surface necessarilycontains certain “zig-zag alternating cycles”, and no such cyclecan bound a 2-ball.
30
![Page 115: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/115.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 4-dimensional prismatoid of width > 4?
Surprisingly enough:
Theorem (S., July 2010)There is no “non-Hirsch” pair of maps in the 2-sphere.
Proof (rough idea of).Every pair of non-Hirsch maps on a surface necessarilycontains certain “zig-zag alternating cycles”, and no such cyclecan bound a 2-ball.
30
![Page 116: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/116.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 5-prismatoid of width > 5
But, in dimension 5 (that is, with maps in the 3-sphere) we haveroom enough to construct “non-Hirsch pairs of maps”:
Theorem
The prismatoid Q of the next two slides, of dimension 5 andwith 48 vertices, has width six.
31
![Page 117: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/117.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 5-prismatoid of width > 5
But, in dimension 5 (that is, with maps in the 3-sphere) we haveroom enough to construct “non-Hirsch pairs of maps”:
Theorem
The prismatoid Q of the next two slides, of dimension 5 andwith 48 vertices, has width six.
31
![Page 118: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/118.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 5-prismatoid of width > 5
But, in dimension 5 (that is, with maps in the 3-sphere) we haveroom enough to construct “non-Hirsch pairs of maps”:
Theorem
The prismatoid Q of the next two slides, of dimension 5 andwith 48 vertices, has width six.
CorollaryThere is a 43-dimensional polytope with 86 facets and diameter(at least) 44.
31
![Page 119: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/119.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 5-prismatoid of width > 5
But, in dimension 5 (that is, with maps in the 3-sphere) we haveroom enough to construct “non-Hirsch pairs of maps”:
Theorem
The prismatoid Q of the next two slides, of dimension 5 andwith 48 vertices, has width six.
Proof 1.It has been verified with polymake that the dual graph of Q(modulo symmetry) has the following structure:
H
C
D J
BA K L
I
E
F
G
31
![Page 120: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/120.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 5-prismatoid of width > 5
Q := conv
8>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>:
0BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB@
x1 x2 x3 x4 x5
1+ 18 0 0 0 12+ −18 0 0 0 13+ 0 18 0 0 14+ 0 −18 0 0 15+ 0 0 45 0 16+ 0 0 −45 0 17+ 0 0 0 45 18+ 0 0 0 −45 19+ 15 15 0 0 110+ −15 15 0 0 111+ 15 −15 0 0 112+ −15 −15 0 0 113+ 0 0 30 30 114+ 0 0 −30 30 115+ 0 0 30 −30 116+ 0 0 −30 −30 117+ 0 10 40 0 118+ 0 −10 40 0 119+ 0 10 −40 0 120+ 0 −10 −40 0 121+ 10 0 0 40 122+ −10 0 0 40 123+ 10 0 0 −40 124+ −10 0 0 −40 1
1CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCA
0BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB@
x1 x2 x3 x4 x5
1− 0 0 0 18 −12− 0 0 0 −18 −13− 0 0 18 0 −14− 0 0 −18 0 −15− 45 0 0 0 −16− −45 0 0 0 −17− 0 45 0 0 −18− 0 −45 0 0 −19− 0 0 15 15 −110− 0 0 15 −15 −111− 0 0 −15 15 −112− 0 0 −15 −15 −113− 30 30 0 0 −114− −30 30 0 0 −115− 30 −30 0 0 −116− −30 −30 0 0 −117− 40 0 10 0 −118− 40 0 −10 0 −119− −40 0 10 0 −120− −40 0 −10 0 −121− 0 40 0 10 −122− 0 40 0 −10 −123− 0 −40 0 10 −124− 0 −40 0 −10 −1
1CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCA
9>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>=>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>;32
![Page 121: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/121.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 5-prismatoid of width > 5
33
![Page 122: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/122.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 5-prismatoid of width > 5
33
![Page 123: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/123.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 5-prismatoid of width > 5
33
![Page 124: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/124.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 5-prismatoid of width > 5
33
![Page 125: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/125.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 5-prismatoid of width > 5
33
![Page 126: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/126.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 5-prismatoid of width > 5
33
![Page 127: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/127.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 5-prismatoid of width > 5
33
![Page 128: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/128.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 5-prismatoid of width > 5
Proof 2.Show that there are no blue vertex a and red vertex b such thata is a vertex of the blue cell containing b and b is a vertex of thered cell containing a.
34
![Page 129: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/129.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
A 5-prismatoid of width > 5
Proof 2.Show that there are no blue vertex a and red vertex b such thata is a vertex of the blue cell containing b and b is a vertex of thered cell containing a.
34
![Page 130: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/130.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Conclusion
Via glueing and products, the counterexample can beconverted into an infinite family that violates the Hirschconjecture by about 2%.This breaks a “psychological barrier”, but for applications itis absolutely irrelevant.
Finding a counterexample will be merely a small firststep in the line of investigation related to theconjecture.
(V. Klee and P. Kleinschmidt, 1987)
35
![Page 131: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/131.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Conclusion
Via glueing and products, the counterexample can beconverted into an infinite family that violates the Hirschconjecture by about 2%.This breaks a “psychological barrier”, but for applications itis absolutely irrelevant.
Finding a counterexample will be merely a small firststep in the line of investigation related to theconjecture.
(V. Klee and P. Kleinschmidt, 1987)
35
![Page 132: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/132.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
Conclusion
Via glueing and products, the counterexample can beconverted into an infinite family that violates the Hirschconjecture by about 2%.This breaks a “psychological barrier”, but for applications itis absolutely irrelevant.
Finding a counterexample will be merely a small firststep in the line of investigation related to theconjecture.
(V. Klee and P. Kleinschmidt, 1987)
35
![Page 133: A counter-example to the Hirsch conjecture...The mathematics of Klee & Grünbaum — Seattle, July 30, 2010 2 Introduction Theorem 1 Theorem 2Conclusion Two quotes by Victor Klee:](https://reader035.vdocuments.us/reader035/viewer/2022071422/611c48edff9d7c29fa439449/html5/thumbnails/133.jpg)
Introduction Theorem 1 Theorem 2 Conclusion
The end
T H A N K Y O U !
36