lower dimensional ads/cft correspondences and their...
TRANSCRIPT
Lower dimensional AdS/CFT correspondencesand their integrability
Alessandro Torrielli
University of Surrey, UK
Selected Topics in Theoretical High Energy PhysicsTbilisi, Georgia
September 23, 2015
Reviews −→ arXiv: 1012.3982 1012.4005 1104.2474with Beisert et al.
Recent −→ arXiv: 1211.1952 1303.5995 1306.2512AdS3 with Borsato, Ohlsson Sax, Sfondrini and Stefanski
−→ arXiv: 1406.2840 1505.06767with Pittelli, Prinsloo, Regelskis and Wolf
Recent −→ arXiv: 1407.0303 and to appearAdS2 with Hoare, Pittelli
Summary
1. Integrable string backgrounds
2. Three dimensional case: the AdS3 × S3 × T 4 story
3. Remarkable similarities, and some difference
4. Massless integrability
5. Two dimensional case: AdS2 × S2 × T 6 and long reps
STRING BACKGROUNDSMetsaev-Tseytlin 9805028
Classification by Zarembo 1003.0465 - see also Wulff 1505.03525
SEMI-SYMMETRIC SUPERSPACES
Supergroup cosets with Z4 symmetry for integrability (Lax pair)
In order to be consistent string theories:
• Beta function vanishes
• Sigma-model central charge equals 26 to balance bc ghosts
[one loop in α′ - planar (genus 0) analysis]
−→
Vanishing Killing form: psu(n|n), osp(2n + 2|2n) - and d(2, 1;α)
FIVE DIMENSIONS: AdS5/CFT4
AdS5 × S5 dual to N = 4 Super Yang Mills in 4D
Superalgebra from psu family: psu(2, 2|4)
32 supersymmetries
• Coset: PSU(2, 2|4)/[SO(4, 1)× SO(5)]
• Relation among parameters: gs = g2YM , λ = g2
YMN ∝ R4/α′2
• Central charge equals 26
• Ramond-Ramond 5-form ∼ N (rank of SYM U(N) gauge)
−→
• Spectral problem ”solved” (significant portion of people sitting all around you)
FOUR DIMENSIONS: AdS4/CFT3
AdS4 × CP3 dual to N = 6 Chern-Simons in 3D
Superalgebra from osp family: osp(6|4)
24 supersymmetries
• Coset: OSP(6|4)/[SO(3, 1)× U(3)]
• Relation among params: RCP3 = 2RAdS , λ = N/k ∝ R4/α′2
• Central charge equals 26
• R-R: 4-form ∼ N (U(N)× U(N) gauge), 2-form ∼ k (level)
−→
• Spectral problem “similar” (in a way) to five-dimensions - h(λ)
THREE DIMENSIONS: AdS3/CFT2
1. AdS3 × S3 × S3 × S1 dual to mystery CFT2 [Tong 1402.5135]
Superalgebra from osp family (deformation): d(2, 1;α)2
16 supersymmetries: large N = (4, 4)
• Coset: D(2, 1;α)2/[SU(1, 1)× SU(2)× SU(2)]
• Relation among parameters: R2AdS/R
2S+
= α = 1− R2AdS/R
2S−
• Central charge < 26, needs extra S1 CFT
• Can mix R-R & NS-NS Cagnazzo-Zarembo 1209.4049 Hoare-Tseytlin 1303.1037
−→ Babichenko-Stefanski-Zarembo 0912.1723
• Massless modes appear
THREE DIMENSIONS: AdS3/CFT2
2. AdS3 × S3 × T 4 dual to CFT2 [Ohlsson Sax, Sfondrini, Stefanski 1411.3676]
Superalgebra from psu family: psu(1, 1|2)2
16 supersymmetries: small N = (4, 4)
• Coset: PSU(1, 1|2)2/[SU(1, 1)× SU(2)]
• Relation among parameters: RAdS = RS , λ ∝ R4/α′2
• Central charge < 26, needs extra T 4 CFT
• Mixed flux
−→ Symmetric orbifold, Higher spins
• More massless modes
TWO DIMENSIONS: AdS2/CFT1
AdS2× S2×T 6 dual to mystery sup conf quantum mechanics
Superalgebra from psu family: psu(1, 1|2)
8 supersymmetries
• Coset: PSU(1, 1|2)/U(1)2
• Relation among parameters: RAdS = RS , λ ∝ R4/α′2
• Central charge < 26, needs extra T 6 CFT
• Massless modes Sorokin-Tseytlin-Wulff-Zarembo 1104.1793 Murugan-Sundin-Wulff 1209.6062
• Recently conjectured integrable S-matrix Hoare-Pittelli-AT 1407.0303
There are many more, with and without time in the coset
STANDARD PICTURE
String modes as particles −→ Toy model for QFT
2D INTEGRABLE SYSTEMS
Exact quantisation of solitons
AdS/CFT CORRESPONDENCE
Strings in 10D
Fields at same holographic point “Spins” on a chain
tr[A(x)B(x) d
dx iC(x)...
]−→ |AB C ′...〉
Example: su(2) sector - one loop
Z = φ1 + iφ2 W = φ3 + iφ4
O{α}(x) = tr[Z ...ZW ...WZ ...W ] −→ ↓ ... ↓ ↑...↑ ↓ ...↑
Corresponding Hamiltonian −→ su(2) Heisenberg spin-chain(’isotropic’)
WHAT THIS PRODUCES FOR US
Spectrum of an interacting CFT via 2D solvable model
Necessity of taking ’t Hooft limit (planar Feynman graphs)
COORDINATE BETHE ANSATZ [Bethe ’31] (Subseq. version: Algebraic BA)
Infinite spin-chain limit: 2-particle state
|ψ >=∑n1<n2
ψ(n1, n2) |Z ...Z V Z ...Z WZ ...〉
n1↑ n2
↑
ψ(n1, n2) = e ip1n1 + ip2n2 + S(p1, p2) e ip2n1 + ip1n2
S(p1, p2) S-matrix: magnon scattering
↓ ... ↓ ⇑ ↓ ... ↓ ... ↓ ... ↓ ⇑ ↓ ... ↓
Periodicity restored by Bethe Equations
AdS/CFT [Arutyunov-Frolov-Staudacher ’04; Beisert-Staudacher ’05]
RESIDUAL ALGEBRAS
Spin-chain sites: reps of the superconformal algebra
After fixing (BMN) vacuum the magnon (spin-wave) excitationscarry reps of smaller subalgebras
Reduced scattering problem
• Five dimensions: psu(2|2)2
• Four dimensions: psu(2|2)
• Three dim: psu(1|1)L × psu(1|1)R in S3 × S1Borsato, Ohlsson Sax, Sfondrini
[psu(1|1)L × psu(1|1)R ]2 in T 4and with Stefanski, AT
• Two dimensions: psu(1|1)2Abbott-Murugan-Sundin-Wulff 1308.1370
−→ all centrally extended [as first shown by Beisert in 5D]
• Quantum inverse scattering method
Massless magnons
New feature in three and two dimensions:integrability with massless particles
Separate story: to be discussed later
Apart from this, remarkable universality of structures
• Zhukovsky variables
• Centrally-extended superalgebras (also with vanishing Killing form)
• Braided coproduct
• Yangian-like symmetry firmly ubiquitous
• Secret (bonus) symmetry review de Leeuw et al 1204.2366
−→ exotic quantum group underlying all the cases
AdS3/CFT2: The T 4 casereview Sfondrini 1406.2971
Start from: AdS3 × S3 × S3 × S1︸ ︷︷ ︸ , 1R2
++ 1
R2−
= 1R2
AdS3
limiting caseα→1 T 4
Superconformal algebra: D(2, 1;α)× D(2, 1;α)︸ ︷︷ ︸×u(1)
Inonu-Wigner α→1 psu(1, 1|2)× psu(1, 1|2)× u(1)3 × u(1)
BMN spectrum: 8b + 8f masses: (1, α, 1− α, 0)× 2b + f → (1, 0)× 4b + f
BMN ground state: point-like string on light-like geodesic in AdS3 × S3
su(1|1)2: algebra relations and fund rep (1 copy) ad − bc = 1
{QL,GL} = HL QL|φ〉 = a |ψ〉 QR |φ〉 = 0{QR ,GR} = HR QL|ψ〉 = 0 QR |ψ〉 = c |φZ+〉{QL,QR} = C GL|φ〉 = 0 GR |φ〉 = b |ψZ−〉{GL,GR} = C† GL|ψ〉 = d |φ〉 GR |ψ〉 = 0
and barred
• Action on states: Dynamical Spin-Chain
(HL + HR) |p〉 = ε(p) |p〉 ε(p) =
√1 + 16 h2 sin2 p
2
C† |p〉 = c(p) |p Z−〉 C |p〉 = c(p) |p Z+〉
Z+(−): one site of the chain is added (removed)h algebr. freedom→ coupling
Interpreted as a non-trivial Hopf algebra:
E.g. C† ⊗ 1 |p1〉 ⊗ |p2〉 =
C† ⊗ 1∑
n1<<n2
e i p1 n1 + i p2 n2 | · · ·Z Z V Z · · ·Z︸ ︷︷ ︸n2−n1−1
W Z · · · 〉
(rescale n2) = c(p1) e ip2 |p1〉 ⊗ |p2〉
S ∆(C†) = S [C† ⊗ 1 + 1⊗ C†] = S [e ip2C†local ⊗ 1 + 1⊗ C†local ]
∆(C†local) = C†local ⊗ e ip + 1⊗ C†local[Gomez-Hernandez ’06; Plefka-Spill-AT ’06]
CROSSING as shown by Janik in 5D
Connects L with R, e.g.
Σ(QL(p)
)= C−1
[QR
]str(p) C
Σ −→ antipode and str −→ supertranspose
Crossing equations can be derived purely from Hopf algebra
Antipode crossing compatible with level 1 Yangian:
S-matrix is projection of universal R-matrix of Yangian of u(1|1)
Zhukovsky variables crossing: x± → 1x±
mass 1: x+ + 1x+ − x− − 1
x−= i
h
Crossing symmetry of R-matrix
(C−1 ⊗ 1)[RL,R
]str1( 1
x±1, x±2
)(C ⊗ 1) RL,L(x±1 , x
±2 ) = 1⊗ 1 etc.
RL,Lpq |φ〉 ⊗ |φ〉 = ρL,Lpq
x+q − x−p
x−q − x+p
e i(p−q)
4 |φ〉 ⊗ |φ〉
RL,Lpq |φ〉 ⊗ |ψ〉 = ρL,Lpq
x+q − x+
p
x−q − x+p
e−i(p+q)
4 |φ〉 ⊗ |ψ〉+ ρL,Lpq
x+q − x−q
x−q − x+p
ηpηq|ψ〉 ⊗ |φ〉
RL,Lpq |ψ〉 ⊗ |φ〉 = ρL,Lpq
x−q − x−p
x−q − x+p
e i(p+q)
4 |ψ〉 ⊗ |φ〉+ ρL,Lpq
x+p − x−p
x−q − x+p
ηqηp|φ〉 ⊗ |ψ〉
RL,Lpq |ψ〉 ⊗ |ψ〉 = ρL,Lpq e−i
(p−q)4 |ψ〉 ⊗ |ψ〉 etc.
Analogs of Janik’s equation (recalling the two copies)
ρL,Lpq ρR,Lpq =
(x+q − x+
p
x+q − x−p
)2
ρL,Rpq ρR,Rpq =
((x−q x+
p − 1)
(x+q x+
p − 1)
x+q
x−q
)2
etc .
Borsato, Ohlsson-Sax, Sfondrini, Stefanski, AT 1306.2512 solution
Perturbation theory: massive OK (massless not ok)Abbott, Aniceto, Beccaria, Bianchi, Forini, Hoare, Levkovich-Maslyuk, Macorini, Murugan, Roiban, Rughoonauth,
Tseytlin, Sundin, Wulff
Yangian: L-R encompassing version by Regelskis 1503.03799
AdS2: a different central extension of su(1|1)2Hoare-Pittelli-AT 1407.0303
su(1|1): algebra relations and fund rep (1 copy) Kac module
{Q,G} = H Q|φ〉 = a |ψ〉 Q|φ〉 = b |φ〉{Q,Q} = C G|φ〉 = c |ψ〉 G|φ〉 = d |ψ〉{G,G} = C†
Does not contradict Iohara-Koga theorem - psu(1|1) not simple
No constraint on H2 − CC † [For other AdS’s, = mass ∈ Z]
−→ long reps [short reps]
Shortening condition happens for H2 − CC † = 0 massless 1D indecomp block
Yangian and long representations Hoare-Pittelli-AT to appear
(cf. AdS5: de Leeuw, Arutyunov, AT 0912.0209)
EXACT S-MATRICES
{for reviews} [P. Dorey ’98; Arutyunov-Frolov ’09]
2D integrable massive S-matrices
• No particle production/annihilation
• Equality of initial and final sets of momenta
• Factorisation: SM−→M =∏
S 2−→2
(all info in 2-body processes)
Extrapolate from relativistic case
S 2−→2 = S(u1 − u2) ≡ S(u) [Ei = mi cosh ui , pi = mi sinh ui ]
Crossing symmetry S12(u) = S 21(iπ − u)
Yang-Baxter Equation (YBE) S12 S13 S23 = S23 S13 S12
Bootstrap S3B(u) = S32
(u + iθ1
)S31
(u + iθ2
)[Zamolodchikov-Zamolodchikov ’79]
• S-matrix real analytic: S(s∗) = S∗(s)Mandelstam s = 2m2(1 + cosh u) for equal masses
• S-matrix simple poles ↔ bound states
0 < s < 4m2 ↔ u = iϑ ↔ p = m sinh u = im sinϑ
DRESSING FACTOR
S(u) = Φ(u) S(u)
S(u) acts as 1 on highest weight state
• Dressing factor Φ(u) not fixed by symmetry, matrix S(u) yes
• Dressing factor Φ(u) constrained by crossing S12(u) = S−112
(u − iπ)
• Dressing factor Φ(u) essential for pole structure
S-MATRIX and HOPF ALGEBRA
{for review} [Delius ’95]
Algebraic treatment −→ relativistic & non (spin chains)
R : V1 ⊗ V2 −→ V1 ⊗ V2 R is called Universal R-matrix
Vi carries a representation of symmetry algebra A
Symmetry on ‘in’ states: coproduct
∆ : A −→ A⊗ A
[∆(a),∆(b)] = ∆([a, b]) (homomorphism)
(P∆)R = R ∆
P (graded) permutation P∆ ‘opposite’ coproduct ∆op (‘out’)
Lie (super)algebras can have ‘trivial’ coproduct
∆op(Q) = ∆(Q) = Q ⊗ 1 + 1⊗ Q ∀Q ∈ A
non trivial → quantum groups
Hopf algebra: coproduct + extra algebraic structurese.g. antipode (antiparticles) + list of axioms
The Yangian is an ∞-dim non-abelian Hopf algebra
{books } [Chari-Pressley ’94; Kassel ’95; Etingof-Schiffmann ’98]
(Σ⊗ 1)R = R−1 = (1⊗ Σ−1)R
R12R13R23 = R23R13R12
(∆⊗ 1)R = R13 R23 ; (1⊗ ∆)R = R13 R12
Universal R-matrix is abstract object whichgenerates all S-matrices by projecting into irreps
R −→fundam⊗ fundam Sfund1,fund2
−→b.state⊗ b.state Sbound1,bound2
−→inf .dim⊗ inf .dim Sinf .1,inf .2
Inclusive of (minimal) dressing factors
MASSLESS LIMIT[Zamolodchikov-Zamolodchikov ’92]
E = m cosh u p = m sinh u E 2 − p2 = m2
Lorentz boost shifts u: the two branches are connected
- 3 - 2 - 1 1 2 3p
0.5
1.0
1.5
2.0
2.5
3.0
E H p L
SEND m→ 0 E = m2
(eu + e−u
)p = m
2
(eu − e−u
)u = u0 + ν
m
2e |u0| = M = finite u0 → ±∞
Two branches of E =√p2
• left moving u0 → +∞
E = Meν+ p = Meν+ E = p ν+ ∈ (−∞,∞)
• right moving u0 → −∞
E = Me−ν− p = −Me−ν− E = −p ν− ∈ (−∞,∞)
- 3 - 2 - 1 1 2 3p
0.5
1.0
1.5
2.0
2.5
3.0
E H p L
• Left-left and right-right scattering does not occur (v = c)
• Many theorems of integrability do not hold
• Nevertheless, ∃ notion of Yang-Baxter: limiting S-matrices are solutions
Left-left and right-right S-matrix does not depend on M:
u1 − u2 = |u0|+ ν1,+ − (|u0|+ ν2,+) = ν1,+ − ν2,+
Left-right S-matrix does, and degenerates in the limit:
u1 − u2 = |u0|+ ν1,+ − (−|u0|+ ν2,−) = 2|u0|+ ν1,+ − ν2,−
• SLL and SRR do not depend on M
−→ removing M describes a CFT via L and R Yang-Baxter
Thank you very much