Two loop scattering in the near flat sigma-model
Transcript of Two loop scattering in the near flat sigma-model
Two loop scattering in the near flat sigma-model
J. Minahan, Uppsala Universitet2007 Itzykson Meeting
Based on arXiv:0704.3891
T. Klose, T. McLoughlin, K. Zarembo, J.M.
Outline• Introduction
• The near flat limit of AdS₅xS⁵
• S-matrix in the near flat limit
• σ-model Lagrangian
• Computing mass-shifts to 2 loops
• Computing scattering amplitudes and comparison to BHL results
• Conclusions
Introduction• Exciting results over the last year
BES 0610251 BCDKS 0610248 BHL 0609044
• BHL finds solutions to Janik’s eq. 0603038
consistent with AFS 04060256 and HL 0603204
Consistent at 2 ws loops and higher?
• General AdS₅ x S⁵ world-sheet scattering initiated by KMRZ 0611169 using the light-cone Hamiltonian of FPZ 0603008 AFPZ 0609157
Same structure as the spin-chain 0511082
• One or more loops is computationally hard
• Look for a simplification (and yet is still nontrivial)
• Near flat limit proposed by MS 0612079
• Simplification of crossing eqs. 0603038, 0609044
-- even crossing has rational solution
• Explicitly show that the two loop near flatis consistent with BHL
The near-flat limit (Maldacena & Swanson)
g >> 1 p ∼ g−1/2
(p < 0)ε(p) ≈ −2gp− 14gp
+gp3
6
Rescale: p→ p/2g γ = 1/4gε(p) ≈ −p− 12p
+γ2p3
3
Let: p± =12(ε± p)
⇒ p− = −p p+ =1
4p−−
γ2p3−
12
Spin +2γ++
Single Magnon: ε(p) ≡ E − J =√
1 + 16g2 sin2 p
2g2 =
λ
16π2
Symmetry Algebra
SL(2,R) TransformedBeisert Algebra
Near Flat (after rescaling):
{QI−, QJ
−} = 2δIJp−
{QI−, QJ
+} = γ δIJp−(p− + 2p−l) + SU(2) factors
{QI+, QJ
+} = 2δIJ [p+ + γ2p−(p2−/3 + p−p−l + p2
−l)]
NormalSusy madefrom localcurrents
{QI−, QJ
−} = δIJκ−
{QI−, QJ
+} = δIJκ2 + SU(2) factors
{QI+, QJ
+} = δIJκ+ κ± = ε± κ1
κ1 + iκ2 = −i2geipl(eip − 1)
S-Matrix
x±(p) =1 +
√1 + P 2
Pe±iγp , P =
1γ
sin γp
SU(2|2)xSU(2|2) Structure (Beisert hep-th/0511089)
Zαα, Yaa,Ψαa,ΥaαImpurities:p
p′
p
p′
All terms are linear in γ =π√λ
Near Flat : x± = −1− 1p−± iγ p−
CrossingAntipode (particle antiparticle): x→ 1/x
Near flat: (p−, p+)→ (−p−,−p+)
Janik hep-th/0603038 σ12σ12 = h12, σ12 = eiθ(p,p′)/2
Beisert, Hernandez & Lopez hep-th/0609044
Near flat: heven12 =
√1− 4iγ p′2p2
p′2−p2 + γ2p′2p2
√1 + 4iγ p′2p2
p′2−p2 + γ2p′2p2
Near flat:
χodd(x, y) =
θodd(p, p′) =8γ2
π+O(γ4)
solution to half-flat crossing eq. consistent with even BHL result
eiθeven(p,p′) =1 + iγ p′pp′−p
p′+p
1− iγ p′pp′−pp′+p
e iθodd(p,p′)
1 + γ2 p2−p′2−
(p′−+p−p′−−p−
)2 S ⊗ S
Prefactor only has even powers of γ
e iθodd(p,p′)
1 + γ2 p2−p′2−
(p′−+p−p′−−p−
)2 S ⊗ S
Near flat action Maldacena & Swanson hep-th/0612079
After gauge fixing, imposing constraints and integrating out left-handed fermions :ψ+
• 8 right-handed susy’s for the free action are preserved by interactions
• 8 left-handed susy’s are modified by interactions
ψ ≡ ψ−
• Interactions break 2d Lorentz invariance (Breaking is mild)
• Interactions are quartic, only ∂− derivatives• Fermion propagator is ip−
p2 −m2
1-loop propagator
p >> p = 0
Because of the form of the interactions and fermion prop.only can appear in the numeratork−
∫d2k
(2π)2(k−)n
k2 −m2= 0 by Lorentz inv.if n > 0
Loop integrals with n = 0 all come from interactions with Y 2 − Z2
2-loop propagator
• Integrals are finite due to Lorentz inv. of free action
• Numerator factors appear because of fermion propagatorsor terms in interactions∂−
Useful identity:
✓(m = 1)
I000(p2) p2 = m2
n.b. There are still higher loop corrections
is evaluated at
for the 2-loop mass-shift
1-loop amplitudes
Permutation:
Forward Klose & Zarembo hep-th/0701240 :
Trace:
2-loop amplitudes
DoubleBubble
WineglassInverse
Wineglass
s
t
u
Bubbles
Double log
Absorbtive
2-loop amplitudes
DoubleBubble
WineglassInverse
Wineglass
s
t
u
Wineglasses
Double logs canceled!
First line is precisely what we want
Unwanted piece:
Jacobian FactorWave function ren.
of ext. legs
X X
✓
Conclusions• 2 loop 2→2 agreement with BHL in near flat limit
• Why do the individual 2-loop channels have the same relation to the tree-level amps?
• All loop proof ?
p+ = −γ2p3−/12
1/4p−
• Giant magnons : classical result is 2 loop correction
• 3 →3 scattering and factorization (Giangreco Marotta Puletti, Klose, Ohlsson Sax)
• Extend results to full AdS₅ x S⁵ ?
• Other 2-loop results by Roiban, Tirziu and Tseytlin