The BFKL Pomeron in AdS/CFT

Lorenzo Cornalba
University of Milano Bicocca & Centro Fermi
The IST String Fest Lisbon, June 2009
Introduction I
High energy phenomena probe relevant features of interactions
In at space string theory they are controlled by the leading Regge
trajectory of states
In perturbative YM theory (QCD and susy completions) dominated
by Pomeron exchange
High energy strings in holographic backgrounds will interpolate
between these two regimes
We will present a general formalism valid both at weak and at strong
coupling
At large t Hooft coupling, analyze graviReggeon exchange in AdS
and recover at space interactions
At weak coupling, use holography to correctly analyze known pQCD
results
High Energy Scattering in Flat Space I
Scattering in at space in the Regge limit s t
Single spin J exhange in the T-channel
Maximal spin dominates
J unbounded High s behavior controlled by leading Regge pole in
the complex J plane
/
tree
(s, t) (t) s
j (t)
Strings in at space
j (t) = 2 +

/
2
t
Violation of unitarity at ultra-high energies
High Energy Scattering in Flat Space II
Partial wave decomposition in the S-channel
/

J0
C
J
_
t
s
_
e
2i
J
(s)
(Im
J
(s) 0)
In the Regge limit
/ 2s
_
d
3
x e
iqx+2i (s,x)
_
q
2
= t
[x[ = 2J/

s
_
Eikonal resummation of tree interaction
/
tree
4is
_
d
3
x e
iqx
(s, x)
High Energy Scattering in Flat Space III
Eikonal limit when phase shift /
tree
(s, x) /s Gs/ [x[ 1. Multi
gravireggeon exchange of linear gravitons
In AdS
5
we expect similar physics with
Extra scale of the AdS radius
Transverse space H
3
instead of E
3
High Energy Interactions in AdS I
AdS
5
given by
X
2
=
2
( = 1)
X M
2
M
4
Boundary points given by rays
Q
2
= 0 Q Q
High Energy Interactions in AdS II
Boundary points behave as momenta for AdS interactions. External
wave functions
1
(2X Q)

_
de
i XQ

1
dual to CFT operator of dimension
Scalar amplitudes
A(Q
i
)
depend on invariants
Q
ij
=
_
Q
i
+Q
j
_
2
and are homogeneous
A( , Q
i
, ) =

i
A( , Q
i
, )
High Energy Interactions in AdS III
Consider CFT correlator
O
1
(Q
1
) O

1
(Q
3
) O
2
(Q
2
) O

2
(Q
4
)) =
1
Q

1
13
Q

2
24
/(cross ratios)
Cross ratios
Q
13
Q
24
Q
12
Q
34
Q
14
Q
23
Q
12
Q
34
Regge kinematics
Q
13
, Q
24
0
Basic example in ^ = 4 SYM
O
1
=Tr(Z
2
) O
2
= Tr(W
2
)
/ = 1 + N
2
/
planar
+
T Channel View I
Conventions
X =
_
X
+
, X

, x
_
M
2
M
4
x =
_
x
+
, x

, x
_
M
4
Scattering kinematics
Q
1
= (0, 1, 0) Q
3
= (q
2
, 1, q)
Q
2
= (1, q
2
, q) Q
4
= (1, 0, 0)
with
q, q Future light cone M
4
T Channel View II
Transverse conformal group
SO (3, 1) q, q vectors
SO (1, 1) q, q q,
1
q
Cross ratios

2
= q
2
q
2
cosh = q q
Regge kinematics
0 xed
T Channel View III
Tchannel exchange T
E,J
(, ) of a conformal block of
Energy E+2
Spin J
Euclidean OPE for 0
T
E,J

2+E
Scattering regime
T
E,J

1J

E
()

E
() propagator of energy 1 + E in H
3
General spin J contribution to correlator

1J
_
d
J
()
i
()
_

H
3
= 1 +
2
_
T Channel View IV
Maximal spin dominates. If spin J unbounded resum contributions.
Leading Regge pole at J = j () gives
/
planar

_
d
1j ()
()
i
()
In ^ = 4 SYM
j (, g)
(, g)
_
g
2
= g
2
YM
N
_
with j (, g) the spin of the twist two operator of dimension 2 + i
S Channel Eixonal Resummation I
Impact parameter representation
/ [q q[
4
_
M
dxd x e
iqxi q x
e
2i (x, x)
Cross ratios
S = [x[ [ x[
B impact parameter on H
3
No interaction
/ = 1 = 0
Lattice sum

S Channel Eixonal Resummation II
For large E, J replace

with
_
M
dxd x where
E =

S cosh (B/2) J =

S sinh (B/2)
Phase shift determined eikonally by planar amplitude

1
N
2
_
d S
j ()1
()
i
(B)
with
() = V
min
(, j ()) ()

V
min
(, j ())
High energy unitarity
Im(S, B) 0
Anomalous dimension 2/ of double trace O
1
O
2
of
dimension E and spin J for elastic AdS interactions
Momentum Space I
Correlator in momentum space
O
1
(p
1
) O

1
(p
3
) O
2
(p
2
) O

2
(p
4
))
with
t , p
2
i
xed s
AdS Poincar coordinates
x
+
, x

, x , r
..
H
3
External states
e
i

sx

e
i p
1
x
f
1
(r )
_
r
2
1/p
2
1
_
Amplitude
s
_
drd
2
x
r
3
f
1
(r ) f
3
(r ) e
i px

_
d r d
2
x
r
3
f
2
( r ) f
4
( r ) e
i p x
e
2i (S,B)
Momentum Space II
Phase shift depends on
S = sr r
cosh B =
1
2r r
_
r
2
+ r
2
+ (x x)
2
_
B
b
single
cross - ratio
Final answer
s
_
d
2
b e
i pb
e
2i (s,b)
with
e
2i (s,b)
= s
_
dr
r
3
f
1
(r ) f
3
(r )
_
d r
r
3
f
2
( r ) f
4
( r ) e
2i (S,B)
Large t Hooft Coupling I
Gravitational interaction in AdS
5
with
G

3
N
2
Along O
1
trajectory high energy states follow null geodesics in AdS
parameterized by ane parameter x
+
and labelled by x

and a point
x, r in H
3
with propagator
1

s
(x
+
j
x
+
i
)
(x

j
x

i
)
H
3
(x
j
, r
j
[x
i
, r
i
)
2 1
3
4
i
j
Large t Hooft Coupling II
For graviton exchange phase given by tree level interaction between
geodesics
(S, B) = i
_
dx
+
d x

3
S
2
(B)
with
2
(B) propagator in H
3
of scalar dimension 3
For g
2

j (, g) 2
(, g)

2
4 +
2
Anomalous dimension of double trace O
1
O
2
of large
dimension E and spin J
1
4N
2
(E J)
4
EJ
_
Exact in 1/N
2
_
Including String Eects I
Regge trajectory of the graviton j (, g) with
g
2
=

4

/2
= g
2
YM
N
Flat space limit
lim

j
_

t,
2
/
/
_
= 2 +

/
2
t
Energy momentum tensor j (2i , g) = 2
Decreasing intercept
j (, g) = 2
4 +
2
2g

Weak t Hooft Coupling I
Amplitude /
planar
(, ) computed to order g
4

g
2
2
2

1
(z, z) +
g
4
16
4
2 + 2z z z z
4z z

2
1
(z, z) z, z = e

+
g
4
16
4
z z
z z
_

2
(z, z)
2
(1 z, 1 z)
2
_
z
z 1
,
z
z 1
__
with

1
=
z z
z z
[2H
2
logz z H
1
] H
p
(z, z) = Li
p
(z) Li
p
( z)

2
= 6H
4
3logz z H
3
+
1
2
log
2
z z H
2
Weak t Hooft Coupling II
In scattering regime obtain spin 1 Regge pole
/
planar

g
4
8
2

2
sinh
2
()
( 0)
corresponding to
j (, g) 1 (g 0)
(, g) i
g
4
16
sinh
_

2
_
cosh
3
_

2
_
Small g and BFKL Pomeron Exchange I
High energy hadron scattering (with s [t[
QCD
) at weak g
dominated by hard perturbative Pomeron exchange
Quantum numbers of the vacuum
Twogluon color singlet state with ladder interactions (reggeized
gluons)
Spin 1 for g 0
j (, g) = 1 +
g
2
4
2
_
2(1)
_
1 +i
2
_

_
1 i
2
__
+
BFKL picture for ^ = 4 SYM for g 0
Small g and BFKL Pomeron Exchange II
Amplitude (y
i
M gluon positions in transverse 2d space)
/
_
H
3
dy
1
dy
3
dy
2
dy
4
y
4
13
y
4
24
V (q, y
1
, y
3
) F (y
i
)

V ( q, y
2
, y
4
)
Pomeron propagator F (y
i
) sum of transverse partial waves of spin n
and energy [n 1[ + 1. Only n = 0 term contributes for scalar
amplitudes
_
d

2
(1 +
2
)
2

Impact factor V (q, y

1
, y
3
) constrained by conformal symmetry to be
function of single cross ratio
u =
x
2
y
2
13
(2x y
1
) (2x y
3
)
Small g and BFKL Pomeron Exchange III
Computable in perturbation theory
V u
2
(1 (u))
Basis functions
_
d V ()
_
H
3
dy
5

We obtain
()
i
4
V ()
tanh

2

V ()
with
V () =

V () =
g
2
2
1
cosh

2
Saturation at Weak Coupling I
BFKL trajectory
j (, g) = 1 + o(g
2
)
(S, B) and () imaginary
Vanishing momentum transfer t = 0
s Q
2

Q
2
_
p
2
1
= p
2
3
_ _
p
2
2
= p
2
4
_
Focus on cross section
_
dr
r
3
f
1
(r ) f
3
(r ) r 1/Q
_
d r
r
3
f
2
( r ) f
4
( r ) r 1/

Q
_
d
2
b Re
_
1 e
2i (S,B)
_
(s, Q,

Q)
Saturation at Weak Coupling II
Integral over impact parameter (with S = s/Q

Q)
(s, Q,

Q)
1
Q

Q
_

[
lnQ/

Q
[
dB sinh B Re
_
1 e
2i (S,B)
_
For large B one has [[ 1. Phase shift 1 along saturation line

_
d e
lnS(j ()1)B(1+i )
so that
B
ln S
B
B
s
(S) lnS
= 0.06 g
2
+
0.14 (exp. value)
Saturation at Weak Coupling III
Deep Saturation
[lnQ/

Q[ B
s
(s/Q

Q)
Approximate black disk
1
Q

Q
_
B
s
(S)
[
lnQ/

Q
[
dB sinh B 1
B
ln S
B
Dominant Contribution
When B
s
lnS then
c
Q

Q
_
_
s
Q

Q
_

+
_
s
Q

Q
_

c
Q

_
Q

Q
_
with c, c,

from r , r integrals
Applications to DIS I
O
1
E&M current (photon) O
2
proton
Kinematics
s Q
2
/x

Q related to connement scale &mass of proton

(simulate connement with wavefunction in r )
Cross section for small x is
Q
2
F
2
_
x, Q
2
_
Geometric Scaling I
Near saturation
[lnQ/

Q[ B
s
(s/Q

Q)
cross section reads
_
d
2
b Im(S, B)
1
Q

QN
2
_
d ()
_
s
Q

Q
_
j ()1
_
Q

Q
_
i
B
ln S
B
At saddle point

1

Q
2

(1+i
s
)
1
2
=
_
Q
Q
s
_
2
with Q
s
=

Q
2
_
1
x
_ 2
1
Geometric Scaling II
In deep saturation

1
Q

Q
_
Q
x

Q
_

Q
2

1
2
Specic dependence on the scaling variable
Experimental evidence
0.001
0.01
0.1
1
10
0.0001 0.001 0.01 0.1 1 10 100
xed experimentally to be 0.138 0.021
Fit to DIS Data I
Expression for F
2
_
Q
2
, x
_
cQ

Q
_
_
Q
x

Q
_

+
_
Q
x

Q
_

cQ

_
Q

Q
_
Q/x
Q
10
5
10
4
10
3
10
2
10
1
10
-1
1 10 10
2
GeV
GeV
(i)
(ii)
(iii)
1
Weak coupling
Q > Q
min
0.7 1 GeV
2
Inside saturation
ln
Q
x

Q
> ln
Q

Q
(

Q 0.2 1 GeV)
3
Asymptotic linear regime
Q
x

Q
10

( 3)
Fit to DIS Data II
Minimize mean square deviation against experimental and simulated
data
0.126
c 0.13
c 0.14

1 (GeV)
Match experimental data in
rather large kinematical
range with 6% accuracy
0.5 < Q
2
< 10
x < 10
2
Q/x
Q
10
5
10
4
10
3
10
2
10
1
10
-1
1 10 10
2
GeV
GeV
Real Data I
0.5 1.0 2.0 4.0
Q [GeV]
0.5 1.0 2.0 4.0
0.2
0.4
0.6
0.8
0.2
0.4
0.6
0.8
0.2
0.4
0.6
0.8
F
2
/
Q
Y = 3.6 Y = 3.8
Y = 4.0 Y = 4.2
Y = 4.6
Y = 4.4
Y = 4.8 Y = 5.0
Simulated Data I
0.5 1.0 2.0 4.0
Q [GeV]
0.5 1.0 2.0 4.0
0.2
0.4
0.6
0.2
0.4
0.6
F
2
/
Q
Y = 3.6 Y = 3.8
Y = 4.0 Y = 4.2 4 . 4 = Y
Y = 4.8 Y = 5.0
6 . 4 = Y
0.2
0.4
0.6
Comments on Dipole Formalism I
Dipole phase shift
(s, r , r , b)
D
(s, r, r, b)
B
b b
single
cross - ratio
two
cross - ratios
Representation of
D
_
d () s
j ()1
T
i
(r, r, b)
where
T
i
(r, r, b)
_
[r[ [ r[ /b
2
_
1+i
for [r[ , [ r[ [b[
Comments on Dipole Formalism II
For r , r [b[
B lnb
2
/r r

_
d S
j ()1
(r r /b)
1+i

D
does not satisfy unitarity constraints (no asymptotic dipole
states)
Even if one assumes saturation at Im
D
1, a simple exponential
saddling for
D
is not possible for general r, r, b
For [r[ , [ r[ [b[ one obtains only the rst term in . A pure black
disk is then a poor approximation of experimental data
Impact Factors for Spin 1 Operators I
AdS graviton trajectory corresponds to n = 0 BFKL trajectory.
What about n 1 trajectories ?
Consider as external states spin 1 operators
O
A
1
O
A
2
Impact factor
V
mn
(q, y
1
, y
3
)
with
symmetric in m, n and y
1
, y
3
vanishing weight in q , y
1
, y
3
Full amplitude
1
[q[
2
1
[ q[
2
2

_
H
3
dy
1
dy
3
dy
2
dy
4
y
4
13
y
4
24
V
mn
(q, y
1
, y
3
) F (y
i
)

V
m n
( q, y
2
, y
4
)
Impact Factors for Spin 1 Operators II
Transverse SO (3, 1) conformal symmetry implies
V
mn
=
5

i =1
f
i
(u) F
mn
i
with
F
mn
1
=
mn
F
mn
2
=
q
m
q
n
q
2
F
mn
3
=
1
2
q
m
_
y
n
1
q y
1
+
y
n
3
q y
3
_

1
2
q
n
_
y
m
1
q y
1
+
y
m
3
q y
3
_
F
mn
4
=
q
2
4
y
m
1
y
n
1
(q y
1
)
2

q
2
4
y
m
3
y
n
3
(q y
3
)
2
F
mn
5
=
y
m
1
y
n
3
+ y
m
3
y
n
1
y
13
Impact Factors for Spin 1 Operators III
Conserved current with
1
= 3 and

q
m
_
1
q
6
V
mn
_
= 0
Projection on n = 0 and n = 2 trajectory
V
mn
= V
mn
0
+ V
mn
2
Construct n = 2 part using basis functions
V
mn
2

_
d T ()
_
H
3
dy
5

Spin 2 propagator from q to y

5
is unique and one can verify that

q
m
V
mn
2
= 0 q
m
V
mn
2
= 0 V
mn
2

mn
= 0
Impact Factors for Spin 1 Operators IV
Remaining four structures are n = 0 terms
V
mn
0
=
4

i =1
T
mn
i
S
i
(u)
with
T
mn
1
=
mn

q
m
q
n
q
2
T
mn
2
=
q
m
q
n
q
2
T
mn
3
= q
m

q
n
+ q
n

q
m
T
mn
4
= q
2

2
q
m
q
n
+
_
q
m

q
n
+ q
n

q
m
_

1
3
_

mn

q
m
q
n
q
2
_
q
2

q
Impact Factors for Spin 1 Operators V
Easy to determine S
1
, S
2
, S
3
S
1
= f
1
+
1
6
f
4
+
1 2u
6
f
5
S
2
= f
1
+ f
2
2f
3

1
2
f
4

1
2u
f
5
S
3
=
_
du
4u
2
(2uf
3
+ uf
4
+ f
5
)
Complex to disentangle S
4
and the n = 2 contribution
V
mn

= T
mn
4
S
4
+ V
mn
2
with
q
m
V
mn

= 0 V
mn

mn
= 0
Impact Factors for Spin 1 Operators VI
Use
m
V
mn
2
= 0 to determine
( 3) S
4
=
_
du
4u
2
_
3uf
4
+ 5f
5
+ u
2
(3u 2)f
/
4
+ u(u 2)f
/
5
_
with
= 4u
2
(1 u)
d
2
du
2
4u
2
d
du
To determine V
mn
2
note that V
mn

can be viewed as an innitesimal

variation of the metric on H
3
. The n = 0 term is then a combined
dieomorphism and Weyl transformation. Therefore, the
innitesimal Cotton tensor
C
abc
(q, y
1
, y
3
)
due to a metric uctuation V
mn

will only come from the n = 2 term

Impact Factors for Spin 1 Operators VII
Due to conformal invariance and the symmetries of the Cotton
tensor, we may construct a single Cotton function
C (u) =
_
q
2
_
2
(2y
1
q) y
13
y
a
1
y
b
3
y
c
1
C
abc
given explicitly by
C = (1 u)u
2
_
(3u 1)f
/
4
+ u(3u 2)f
//
4
+
u
2
2
(u 1)f
///
4

f
/
5
+ (1 2u)f
//
5

u
2
(u 1)f
///
5
_
From C one can immediately deduce T ()
Examples in ^ = 4 SYM I
Basic spin 1 operators with
1
= 3 in the free limit
Tr (

m
)
Tr
_

D
m

j
_
+ c Tr
_

ij

_
Tr
_

D
m

j
_
+ c
/
Tr
_

ij

_
Only the Rsymmetry current in the 15 of SO (6) is chiral and has
protected dimension. The other spin 1 operators (in the 15 and 1 of
SO (6)) acquire dimension in the g
2
limit
Very simple computation (compared to standard momentum space
techniques) allows to compute
Impact factor for scalar quark current Tr
_

D
m

_
3u
2
F
mn
4
+ 2u
3
F
mn
5
Examples in ^ = 4 SYM II
Impact factor for quark current Tr (

m
)
2u
3
F
mn
1
+ 4u
3
F
mn
5
The n = 0 contributions are dierent. The n = 2 contributions are
identical with Cotton function
C = 36u
3
(1 u) (1 2u)
Reinserting the SO (6) factors one has that the Rsymmetry current
has no overlap with the n = 2 trajectory
Basic Conjecture (tested already in more cases) : The SUGRA
chirally protected states in ^ = 4 SYM interact uniquely with the
n = 0 Pomeron / graviton trajectory