Summing up all genus free energy of ABJM matrix model
Seminar @ National Taiwan Univ.
Shinji Hirano (Nagoya Univ.)
Based on
H. Fuji, SH, S. Moriyama (1106.4631), O. Bergman, SH (0902.1743)
Introduction
Sum up all-loop all-genus free energy of strongly coupled 3d CFT!
3d CFT (ABJM theory) = U (N )k × U (N )−k Chern-Simons-Matter theory
• ’t Hooft coupling λ ≡ N/k 1
• Genus expansion gs ≡ 2πi/k = 2πiλ/N
ABJM theory (on S
3)
U (N )k × U (N )−k theory with 4 bifundamental matters:
SABJM = SCSk + SCS−k + SM where
SCSk = k Z
d3xTr
A ∧ dA + 2i
3 A3 − ¯λλ + 2Dσ
with DR of 4d N = 1 vector muliplet (Aµ, σ, λ, D)
SM = Z
d3x
4
X
I=1
Dµφ¯IDµφI − i ¯ψIγµDµψI + 3 4
φ¯IφI + · · · + Z
d2θW + h.c.
with bifundamental chiral multiplets (φI, ψI, FI)
ABJM Conjecture
Type IIB brane configuration of (UV completed) ABJ/M:
5 (1,k)
D3 N
D3 N+l
NS5
. &
. Gravity Gauge theory &
IR limit of M-theory lift IR limit of SYM + CSM
AdS4 × S7/Zk ABJM theory
AdS4/CFT3 ⊂ AdS/CFT ⊂ Gravity/Gauge duality:
M-theory on AdS4 × S7/Zk = U (N )k × U (N )−k Chern-Simons-Matter
When k 1, dimensional reduction 11d → 10d
Type IIA on AdS4 × CP3 = U (N )k × U (N )−k Chern-Simons-Matter
Result
FABJM (λ, N ) = log
2πC1Ai
"
√π 2
N λ
2
λ3/2ren
#2/3
+ O
e−2π
√
λ−241
up to worldsheet instantons O
e−2π
√
λ−241
λren = λ − 1
24 − λ2 3N2
Note:
(1) Stringy corrections : 1
√λ − expansion = α0 − expansion (2) QG loop corrections : 1
N2 − expansion = GN − expansion
PROSPECT:
Quantum Gravity test of AdS/CFT !
Technique: LOCALIZATION
Strong coupling λ 1 non-planar computation on gauge side is now possible!
Note: Integrability in AdS5 – exact computation from λ 1 to λ 1 BUT only at N = ∞
Gravity prediction of non-planar corrections (Bergman, S.H.)
M2-brane charge N shifted to
N → N − 1
24k + 1 24k
In terms of ’t Hooft coupling (IIA description)
λ → λ − 1
24 + λ2
24N2 (large N, k with λ = N/k)
Implying renormalization of AdS radius
RAdS = (32π2kN)1/6`p →
32π2k
N − 1
24k + 1 24k
1/6
`p
(1) M-theory (11d SUGRA action):
S11 = 1 2κ211
Z
d11x √
−G
R − 1
2|G4|2
− 1 6
Z
C3 ∧ G4 ∧ G4+ (2π)2 Z
C3 ∧ I8
+ SM 2
where I8 = 23·4!(2π)1 4
TrR4 − 14 TrR22
(2) M2 Maxwell equation:
d ∗ G4 = (2π)4(kN )δ8(x) − 1
2G4 ∧ G4 + (2π)2I8
(I) (localized) M2 brane source
(II) flux (discrete torsion – H3(S7/Zk, Z) = Zk) (III) higher curvature (Duff-Liu-Minasian)
(3) Membrane charge (M8 = C4/Zk and ∂M8 = S7/Zk):
QM 2 ≡ 1 (2π)4
Z
∂M8
∗ G4 = N − 1 2(2π)4
Z
M8
G4 ∧ G4 − χblk 24 where χblk bulk contribution to C4/Zk Euler characteristic
χblk(C4/Zk) = k − 1 k
Flow chart of free energy computation
3d N = 6 U (N )k × U (N )−k Chern-Simons-Matter conformal field theory
⇓
Localization
⇓
U (N ) × U (N ) ABJM matrix model
matrix potential = Gaussian + 2d Coulomb repulsion
⇓
Analytic continuation to Lens space matrix model (N, N ) → (N1, −N2)
⇓
Large N technique to find planar solution
⇓
Chain of dualities
Lens MM geom trans−→ Top A mirror−→ Top B
⇓
Holomorphic Anomaly Equation = recursion for higher genus free energies
⇓
Solve HAE neglecting O e−
√λ
⇓
All genus free energy up to worldsheet instantons
(1) Once genus zero free energy F0(λ) is given, HAE can be explicitly found (2) Together with genus one data, HAE can be solved (at least) recursively
We managed to sum it up !
Localization
∃ exact nilpotent quantum Gassmann-odd symmetry Q (such as SUSY &
BRST), partition function
Z = Z
Dϕ exp (−S[ϕ]) = Z
Dϕ exp (−S[ϕ] − tQV [ϕ])
(1) With QV ≥ 0 (choosing V cleverly) sending t → ∞
Path integral localizes on the locus {ϕ = ϕi | δS[ϕi] ≡ QV = 0}
(2) In fortunate situations, localization locus {ϕi=1,···,N} is finite dimensional Path integral reduces to finite dimensional integrals (Matrix Model) (3) Partition function one-loop exact (ϕ = ϕi + √1
tδϕ with t → ∞):
Z =
Z N Y
i=1
Dϕi v u u u t
det δ2∆S
δϕ2F [ϕiF] det δ2∆S
δϕ2B [ϕiB] exp (−S[ϕi])
ABJM application: (Kapustin-Willet-Yaakov, Hama-Hosomichi-Lee) (1) Nilpotent Grassmann-odd symmetry δ¯ (one of supercharges)
(2) Localization action ¯δV ≡ ¯δVgauge + ¯δVmatter δV¯ gauge = ¯δδ
2
X
A=1
Tr 1 2
λ¯AλA − 2DAσA
= SYM
δV¯ matter = ¯δδ ¯ψIψI − 2i ¯φI σ1 − σ2 φI − ¯φIφI
= Sm
where
LYM = Tr 1
4Fµν2 + 1
2(Dµσ)2 + 1
2(D + σ)2 + · · ·
(3) Localization locus
AAµ = φI = 0 , DA = −σA = const
(4) ABJM Matrix Model:
ZABJM = 1 (N !)2
Z N Y
i=1
dµi 2π
N
Y
a=1
dνa 2π
Q
i<j
2 sinhµ
i−µj 2
2 Q
a<b 2 sinh νa−ν2 b2 Q
i,a 2 cosh µi−ν2 a2 e−2gs1 (Piµ2i−P
aνa2)
where gs = 2πi/k
Luckily, these integrals can be done exactly!
Drukker-Mari˜ no-Putrov
Genus zero free energy of ABJM theory at strong coupling λ 1
F0(λ) = 4
√2π2
3 λ − 241 3/2
+ O
e−2π
√
λ−241
Precisely agrees with classical SUGRA result!
−F0(λ)/gs2 = SSUGRA[AdS4 × S7/Zk]
Higher genus – solving recursion
(1) H(olomorphic)A(nomaly)E(quation) (Bershadsky-Cecotti-Ooguri-Vafa):
∂I¯Fg = 1
2CI ¯¯J ¯Ke2KGJ ¯JGK ¯K DJDKFg−1 +
g−1
X
r=1
DJFrDKFg−r
!
where
GI ¯J = Im∂I∂JF0 CIJ K = ∂I∂J∂KF0
(2) ABJM case: moduli space coordinate XI = λ (one-dimensional)
(3) Turns out, neglecting worldsheet instanton, HAE yields
Fg0(x) = 1
4x4Fg−100 (x) + 12x − 1
12 x2Fg−10 (x) + x4 4
g−2
X
r=2
Fr0(x)Fg−r0 (x)
where x ≡ 1/√
2π2λ + · · ·
(4) (Modular) weight zero free energy
Fg := Fg[0]x3g−3 + O x3g−4 t≡−igsx3/2
−→ F (t) :=
∞
X
g=2
Fg[0]t2g−2
(5) It sums up to (1st step sum = weight zero sum)
F (t) = logh
2πC1e−3t22 t−13Ai(t−4/3)i
(6) 2nd step sum = gravity expectation (w/ ∗)
λ → λren = λ − 1
24 − λ2
3N2 ⇐⇒ x → y ≡ x
p1 + (gsx)2/6
⇓
Replace x in F (t) by y
⇓
FABJM (λ, N ) = log
2πC1Ai
"
√π 2
N λ
2
λ3/2ren
#2/3
+ O
e−2π
√
λ−241
This indeed solves HAE!
Conclusions and discussions
(1) Summed up all-loop all genus free energy of ABJM theory! (except for worldsheet instantons)
(2) Expect similar all genus free energy for ABJ theory (unequal rank U (N1)k × U (N2)−k theory)
w/ λren = λ − k(B2 − 1/4)/2 − 1/24− 1/(3k2)
(Aharony-Hashimoto-SH-Ouyang, DMP)
(3) Mismatch between field theory and gravity results at non-planar
λFTren = λ − 1
24 − λ2
3N2 vs. λGravityren = λ − 1
24 + λ2 24N2
Is it field theory or gravity?