vladimir m. braun - ukr
TRANSCRIPT
KITP, QCD and String Theory Integrability in QCD and beyond - p. 1/35
Integrability in QCD and beyond
Vladimir M. BraunUniversity of Regensburg
thanks to
Sergey Derkachov, Gregory Korchemsky and Alexander Manashov
KITP, QCD and String Theory Integrability in QCD and beyond - p. 3/35
Outline
In this talk:
■ Renormalization in conformally-covariant form■ Integrability on the example of three-quark operators in QCD■ Violation of integrability: Mass gaps■ SUSY extensions: From N = 0 to N = 4
■ Further QCD example: Polarized deep-inelastic scattering
not included:
■ High energy scattering in QCD
■ Full N = 4 SUSY and gauge/string correspondence
KITP, QCD and String Theory Integrability in QCD and beyond - p. 4/35
Heisenberg spin chain
1926: Heisenberg
H
s=1=2
= �
L
X
n=1
�
^
S
n
�
^
S
n+1
�
14
�
1981-83: Generalizations for arbitrary spin
H
s
=
L
X
n=1
H(J
n;n+1
)
J
n;n+1
(J
n;n+1
+ 1) = (
^
S
n
+
^
S
n+1
)
2
H(x) = (2s+ 1)� (x+ 1)
Complete integrability
m
Number of degrees of freedom =number of conservation laws
� Conceptually interesting� Powerfull machinery:
Algebraic Bethe Ansatz (ABA)Method of separation of
Variables (SoV)
KITP, QCD and String Theory Integrability in QCD and beyond - p. 5/35
at the same time, in QCD . . .
1973-74: Anomalous dimensions of leading twist operators
N
= C
F
�
�3�
2
(N+1)(N+2)
+ 4 (N+2)� 4 (1)
�
1976-78: High-energy (Regge) asymptotics of scattering amplitides
E
BFKL
N;�
= 2 (1)�
�
n+1
2
+ i�
�
�
�
n+1
2
� i�
�
1991: Anomalous dimensions of quark-antiquark-gluon operators ,N
!1
5 10 15 20 25 3010
15
20
25
30
35
40 γN
N
Exact analytic solution
Hidden symmetry?
KITP, QCD and String Theory Integrability in QCD and beyond - p. 6/35
Example:
Twist-three operators
,
Leading-twistthree-particle parton
distributions
E.g. baryon distribution amplitudes B = N;�; : : :
h0jq(z
1
)q(z
2
)q(z
3
)jB(p; �)i = : : :
Z
1
0
dx
1
dx
2
dx
3
Æ(
P
x
i
� 1) e
�ip(x
1
z
1
+x
2
z
2
+x
3
z
3
)
'
B
(x
i
; �
2
)
q
"
q
"
q
"
) '
�=3=2
�
(x
i
; �
2
) ;
q
"
q
#
q
"
)
(
'
�=1=2
N
(x
i
; �
2
)
'
�=1=2
�
(x
i
; �
2
)
� quark fields “live” on a light-ray z2 = 0
KITP, QCD and String Theory Integrability in QCD and beyond - p. 7/35
Problem: Prolifiteration of degrees of freedom
Moments of distribution amplitudes , local operators:
'(x
i
)! '(k
i
) =
Z
Dx
i
x
k
1
1
x
k
2
2
x
k
3
3
'(x
i
; �
2
)
q(z
1
)q(z
2
)q(z
3
) ! (D
k
1
+
q) (D
k
2
+
q) (D
k
3
+
q)
Mixing matrix: N = k
1
+ k
2
+ k
3
0
2
4
6
8
10
12
14
16
0 5 10 15 20 25 30
E3/
2(N
)
N
γ n=0,1,...N=14
Example:qqq
� = 3=2
Rich spectrum of anomalous dimensions reflectscomplexity of genuine degrees of freedom
KITP, QCD and String Theory Integrability in QCD and beyond - p. 8/35
Conformal Symmetry on the Light-Cone: SL(2; R)
z ! z
0
=
az + b
z + d
; ad� b = 1 �(z)! �
0
(z) = �
�
az + b
z + d
�
� ( z + d)
�2j
�
j
q
= 1, jg
= 3=2 is conformal spin of the field
Generators obey the SL(2) algebra
L
�
�(z) = �
d
dz
�(z)
L
+
�(z) =
�
z
2
d
dz
+ 2j
�
z
�
�(z)
L
0
�(z) =
�
z
d
dz
+ j
�
�
�(z)
Casimir operators
L
2
= L
20
� L
0
+ L
+
L
�
L
2
�(z) = j(j � 1)�(z)
Summation of spins
L
2ik
=
P
�=0;1;2
(L
i;�
+L
k;�
)
2
L
2123
=
P
�=0;1;2
(L
1;�
+L
2;�
+L
3;�
)
2
� second-order differential operators on the space of (z
1
; z
2
; z
3
)
KITP, QCD and String Theory Integrability in QCD and beyond - p. 9/35
RG equations in SL(2)-covariant form
Light-ray operators
�
�
�
��
+ �(g)
�
�g
�
B = H �B ; B(z
1
; z
2
; z
3
) ' q(z
1
)q(z
2
)q(z
3
)
Two-particle structure:
H
qqq
= H
12
+H
23
+H
13
Renormalization = Displacement: Balitsky, V.B. 88’
H
12
B(z
1
; z
2
; z
3
) =
Z
D� !(�
1
�
2
) B(��
1
z
1
+ �
1
z
2
; ��
2
z
2
+ �
2
z
1
; z
3
)
where
R
D� =
R
d�
1
d�
2
d�
3
Æ(1��
1
��
2
��
3
), �� = 1� �
KITP, QCD and String Theory Integrability in QCD and beyond - p. 10/35
RG equations in SL(2)-covariant form — continued
Conformal symmetry ! [L
�;0
;H
12
℄ = 0
!(�
1
�
2
) = ��
2j
1
�2
1
��
2j
2
�2
2
'
�
�
1
�
2
��
1
��
2
�
�� = 1� �
Explicit calculation
1 2 3
) '(x) = Æ(x);
1 2 3
) '(x) = 1
KITP, QCD and String Theory Integrability in QCD and beyond - p. 11/35
RG equations in SL(2)-covariant form — Hamiltonian approach
B’FK’L 85’
write H as a function of Casimir operators L
2ik
= �
�
�z
i
�
�z
k
(z
i
�z
k
)
2
�
b
J
ik
(
b
J
ik
� 1)
1 2 3
=)
H
v12
= 2[ (J
12
)� (2)℄
H
�=3=2
qqq
= H
v
12
+H
v
23
+H
v
13
1 2 3
=)
H
(e)
1=2
=
1
L
212
=
1
J
12
(J
12
�1)
H
�=1=2
qqq
= H
�=3=2
qqq
�H
e
12
�H
e
23
Have to solve H
N;n
= E
N;n
N;n
— A Schr odinger equation with Hamiltonian H
KITP, QCD and String Theory Integrability in QCD and beyond - p. 12/35
Hamiltonian approach — continued
going over to local operators:
In helicity basis (BFKL 85’) obtain a
HERMITIAN Hamiltonianwith respect to the conformal scalar product
h
1
j
2
i =
Z
1
0
dx
i
Æ(
X
x
i
� 1)x
2j
1
�1
1
x
2j
2
�1
2
x
2j
3
�1
3
�1
(x
i
)
2
(x
i
)
that is, conformal symmetry allows for the transformation o f the mixing matrix:
0�
forward
non-hermitian
1A
| {z }
N + 1
)
0�
non-forward
hermitian
1A
| {z }
N(N + 1)=2
)
0�
hermitian
in onformal basis
1A
| {z }
N + 1
KITP, QCD and String Theory Integrability in QCD and beyond - p. 13/35
Hamiltonian approach — continued (2)
~ Eigenvalues (anomalous dimensions) are real numbers
~ “Conservation of probability” – possibility of meaningful
approximations to exact evolution
~ Systematic 1=N
expansion
E
N;n
= N
E
N;n
+N
�1
ÆE
N;n
+ : : :
N;n
=
(0)
N;n
+N
�2
Æ
N;n
+ : : :
with the usual quantum-mechanical expressions
ÆE
N;n
= jj
(0)
N;n
jj
�2
h
(0)
N;n
jH
(1)
j
(0)
N;n
i
etc.
KITP, QCD and String Theory Integrability in QCD and beyond - p. 14/35
Complete Integrability
� Conformal symmetry implies existence
of two conserved quantities:
[H; L
2
℄ = [H; L
0
℄ = 0
� For qqq and GGG states with maximum helicity and for qGq states at N
!1 there exists an
additional conserved charge
qqq
GGG
�=max
)
Q = [L
212
; L
223
℄
[H; Q℄ = 0
qGq
N
!1
)
Q = fL
2qG
; L
2Gq
g+
1
L
2qG
+
2
L
2Gq
[H; Q℄ = 0
BDM ’98
� Anomalous dimensions can be clas-
sified by values of the charge Q:
H = E
Q = q
=)
E = E(q)
� A new quantum number
KITP, QCD and String Theory Integrability in QCD and beyond - p. 15/35
WKB expansion of the Baxter equation ) 1/N expansion
G. Korchemsky ’95-’97
~ equation Q = q is much simpler as H = E
~ non-integrable corrections are suppressed at large N
~ can use some techniques of integrable models
E(N; `) = 6 ln � � 3 ln 3� 6 + 6
E
�
3
�
(2`+ 1)
�
1
�
2
�
5`
2
+ 5`� 7=6
�
�
1
72�
3
�
464`
3
+ 696`
2
� 802`� 517
�
+ : : :
� =
p
(N+3)(N+2)
An integer ` numerates the trajectories:(semiclassically quantized solitons: Korchemsky, Krichever, ’97)
� Integrability imposes a nontrivial analytic structure
KITP, QCD and String Theory Integrability in QCD and beyond - p. 16/35
WKB expansion of the Baxter equation ) 1/N expansion — continued
The ‘dispersion curve’ E(q)
-0.20 -0.10 0.00 0.10 0.2010.0
11.0
12.0
13.0
14.0
15.0
E(q
)
η3q/
Ν = 30
WKB
exactExample:qqq
� = 3=2
E(q) = 2 ln 2� 6 + 6
E
+
+ 2Re
3
Xk=1
(1 + i�
3
Æ
k
) +O(�
�6
)
Æ
k
are defined as roots of the cubic equation:2Æ
3
k
� Æ
k
� q=�
3
= 0
� =
p
(N+3)(N+2)
KITP, QCD and String Theory Integrability in QCD and beyond - p. 17/35
WKB expansion of the Baxter equation ) 1/N expansion — continued (2)
Double degeneracy:
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0 5 10 15 20 25 30
q/3
N
0
2
4
6
8
10
12
14
16
0 5 10 15 20 25 30
E3/
2(N
)
η
N
l = 0
l = 2
l = 7
Example:qqq
� = 3=2
E(N; q) = E(N;�q)
q(N; `) = �q(N;N � `)
BDKM ’98
KITP, QCD and String Theory Integrability in QCD and beyond - p. 18/35
The ground state solution ( q = 0) for the wave function of �
�=3=2
For evenN q = 0 is the solution with lowest energy
x
1
x
2
x
3
N;q=0
(x
i
) = x
1
(1�x
1
)C
3=2
N+1
(1�2x
1
)
+ x
2
(1�x
2
)C
3=2
N+1
(1�2x
2
)
+ x
3
(1�x
3
)C
3=2
N+1
(1�2x
3
)
The lowest anomalous dimension
E(N; q = 0) = 4(N + 3) + 4
E
� 6
Here x1
; x
2
; x
3
are fractions of the �3=2 momentum carried by the three quarks
KITP, QCD and String Theory Integrability in QCD and beyond - p. 19/35
Breakdown of integrability: Mass gaps and bound states — ”experiment”
Simplest case:
Difference between �
�=3=2 and N(�)
�=1=2
H(") = H
3=2
� "
�
1
L
212
+
1
L
223
�
H(" = 1) = H
1=2
Flow of energy levels
(anomalous dimensions):
KITP, QCD and String Theory Integrability in QCD and beyond - p. 20/35
Breakdown of integrability: Mass gaps and bound states — ”theory”
In lower part of the qqq spectrum
‘Perturbation’ hq
0
j
1
L
212
+
1
L
223
jqi �
1
lnN
Level splitting �
1
ln
2
N
=)
Of order � lnN lower levels
have to be rediagonalized
hq
0
j
1
L
212
+
1
L
223
jqi �
1
lnN
os(�
q
� �
q
0
)
Phases of the cyclic permutations: (123)
P
�! (312)
Pjqi = �
q
jqi
θq
Matrix ( lnN � lnN ):
1
lnN
0BBBBBB�
1 �1=2 �1=2 1 : : :
�1=2 1 �1=2 �1=2 : : :
�1=2 �1=2 1 �1=2 : : :
......
......
. . .
1CCCCCCA
diag
�!
1
lnN
0BBBBBB�
lnN 0 0 0 : : :
0 lnN 0 0 : : :
0 0 0 0 : : :
......
......
. . .
1CCCCCCA
! A mass gap !
KITP, QCD and String Theory Integrability in QCD and beyond - p. 21/35
Breakdown of integrability: Mass gaps and bound states — ”interpretation”
�
�=3=2 wave function Nucleon and �
1=2 wave functions
'
�
3=2
(x
i
)
�
=
1
X
N=0
'
�
0
N;n=0
�
�
s
(�)
�
s
(�
0
)
�
N;n=0
�
x
1
(1�x
1
)C
3=2
N+1
(1�2x
1
) + x
2
(1�x
2
)C
3=2
N+1
(1�2x
2
)
+x
3
(1�x
3
)C
3=2
N+1
(1�2x
3
)
�
'
�
1=2
(x
i
)
�
= x
1
x
2
x
3
1
X
N=0
'
�
0
N;n=0
�
�
s
(�)
�
s
(�
0
)
�
N;n=0
�
P
(1;3)
N
(1� 2x
3
)� P
(1;3)
N
(1� 2x
1
)
�
1 2 3
free motion
scalar diquarks?
1 2 3
1] -+ ( )3[
KITP, QCD and String Theory Integrability in QCD and beyond - p. 22/35
Supersymmetry: from N = 0 to N = 4
A. Belitsky, S. Derkachov G. Korchemsky and A. Manashov [hep-th/0403085]
� One expects that SUSY enhances integrability as one goes fro m N = 0! N = 4
? QCD-like integrability is extended and new features may appear
� Need a universal approach to treat simultaneously the opera tor mixing in various
N�extended SYM
The light-cone superspace formalism
■ Quantize SYM in the light-cone gaugeA+
(x) = 0
■ Decompose fermions and gauge fields into “good” (+
; A
?
) and “bad” (�
; A
�
) components
■ Integrate out nonpropagating (“bad”) fields and rewrite the action in terms of propagating (“good”) fields
S
N
= S
N
(
+
; A
?
)
■ Combine “good” fields into light-cone superfields �(x; �A)
Kogut, Soper’70
KITP, QCD and String Theory Integrability in QCD and beyond - p. 23/35
N = 4 light-cone SYM
Light-cone superfield formulation of N = 4 SYM Mandelstam, Brink et al. ’83
S
N=4
=
Z
d
4
x d
4
� d
4
�
�
8<
:
12
�
�
a
�
�
2
+
�
a
�
23
gf
ab
�
1
�
+
�
�
a
�
b
�
��
+
1
�
+
�
a
�
�
b
�
�
�
�
�
12
g
2
f
ab
f
ade
�
1
�
+
(�
b
�
+
�
)
1
�
+
(
�
�
d
�
+
�
�
e
) +
12
�
b
�
�
�
d
�
�
e
�
9=
;
Complex scalar N = 4 chiral superfield
� = �
�1
+
A+ �
A
�
�1
+
�
�
A
+
i
2!
�
A
�
B
�
�
AB
�
1
3!
"
ABCD
�
A
�
B
�
C
�
D
�
1
4!
"
ABCD
�
A
�
B
�
C
�
D
�
+
�
A
� Chiral superfield spans all helicities of the particles in th e supermultiplet
� The anti-chiral superfield is not independent
� All single-trace light-cone operators are built from a sing le chiral superfield
O(Z
1
; : : : ; Z
L
) = trf�(z
1
; �
1
) : : :�(z
L
; �
L
)g
KITP, QCD and String Theory Integrability in QCD and beyond - p. 24/35
. . . and back from N = 4 to N = 0
“Method of truncation”: SN=4
! S
N=2
! S
N=1
! S
N=0
Brink et al. ’83
) N = 2 light-cone chiral superfield
�(x; �) = �
�1
+
A(x) + �
A
�
�1
+
�
�
A
(x) +
i
2!
"
AB
�
A
�
B
�
�(x)
) N = 1 light-cone chiral superfield
�(x; �) = �
�1
+
A(x) + � �
�1
+
�
�
A
(x)
) N = 0 light-cone chiral (super)field
�(x) = �
�1
+
A(x)
� chiral superfield � and its conjugate �
� are independent
� Single-trace light-cone operators can be built from two sup erfields
Integrable sector O
hiral
(Z
1
; : : : ; Z
L
) = Trf�(Z
1
) : : :�(Z
L
)g
Non-integrable sector O
mixed
(Z
1
; : : : ; Z
L
) = Trf�(Z
1
) : : :
�
�(Z
L
)g
KITP, QCD and String Theory Integrability in QCD and beyond - p. 25/35
One-loop dilatation operator in chiral sector
A two particle kernel
H
k;k+1
=
k
k+1
= V
k;k+1
(1� �
k;k+1
)
acts as a displacement in the light-cone superspace
Z = (z; �
A
) (A = 1; : : : ;N )
V
12
O (Z
1
; Z
2
) =
R
1
0
d�
(1��)�
2
h
2�
2
O (Z
1
; Z
2
)�O (�Z
1
+(1��)Z
2
; Z
2
)�O (Z
1
; �Z
2
+(1��)Z
1
)
i
�(x) = �
�1
+
A(x)+: : :
� Projection operator, (�
12
)
2
= �
12
�
12
O (Z
1
; Z
2
) =
12
(1 + Z
12
� �
Z
) O (Z; Z
2
)
���
Z=Z
2
+
12
(1 + Z
21
� �
Z
) O (Z
1
; Z)
���
Z=Z
1
Compare this with QCD expression for B(z
1
; z
2
; z
3
) = q
"
(z
1
)q
"
(z
2
)q
"
(z
3
):
H
12
B(z
1
; z
2
; : : :) =
R
1
0
�d�
1��
h
2B(z
1
; z
2
)�B(�z
1
+ (1� �)z
2
; z
2
)�B(z
1
; �z
2
+ (1� �)z
1
)
i
[For quarks jq
= 1, for chiral superfield j�
= �1=2, hence different power of �]
KITP, QCD and String Theory Integrability in QCD and beyond - p. 26/35
Dilatation operator on the light-cone superspace Z = (z; �)
■ The one-loop dilatation operator has the same, universal form for N = 0, N = 1,
N = 2 and N = 4 SYM
◆ Dilatation operator displaces superfields in the superspaceZ = (z; �
A
)
◆ The number of superchargesN defines the dimension of the superspace
◆ Natural generalization of the QCD formulae Z = (z; �
A
= 0)
■ Two reductions of the N = 4 dilatation operator
◆ Expand in even light-cone coordinates, project on � = 0
! Maximal helicity gluon operators H = SL(2;R) magnet[Braun,Derkachov,Manashov; Belitsky; G.K ’98]
◆ Expand in odd �-coordinates, project on zi
= 0
! BMN-like operators: O
s al
= Trf�
A
1
B
1
(0) : : : �
A
L
B
L
(0)g H = SO(6) magnet[Minahan, Zarembo ’02]
■ In general: collinear subgroup of the superconformal symme try group:Heisenberg (super) spin magnet with the SL(2jN ) symmetry
KITP, QCD and String Theory Integrability in QCD and beyond - p. 27/35
N = 4 beyond the light-cone
The case of N = 4 SYM is still special:
■ All light cone operators belong to the SL(2jN ) sector since �
�(Z) is not dynamicaly
independent
■ Integrability extends to all operators (including those wi th “bad” components) and the collinear
subgroup SL(2jN ) extends to the full superconformal symmetry group PSU(2; 2;N ).
[Beisert, Staudacher, ’03]
Why?◆ Crucial: Superconformal symmetry relates operators of different twist
◆ Superconformal transformations applied to light-cone operators �iDN
+
�
i yield combinations of
operators of different twist but having the same anomalous dimension
◆ This is not the case in QCD where operators of different twists are believed to be dynamically
independent
■ Probably persists to all loops
KITP, QCD and String Theory Integrability in QCD and beyond - p. 28/35
Outlook
■ One-loop dilatation operator (evolution equations) in QCD is/are integrable in somesectors◆ Classical bremsstrahlung: A �
d!
!
d�
�
◆ Cusp anomalous dimension: for N !1
N
� �
usp
(�
s
) logN
◆ Integrability appears as a consequence of the existence of massless vector particles
■ Offers powerful machinery
■ Open problems in QCD context:◆ Analytic structure of the spectrum — Parton interpretation in higher twists
◆ FromN = 4 toN = 0: Rethinking of the role of non-quasipartonic operators; properties of
anomalous dimensions in all twists
◆ Beyond one loop: Formal conformal limit
A = A
onformal
+
�(g)
g
�A
Integrability? Particular models?
◆ Breaking of integrability vs. breaking of conformal symmetry: physics issues?
KITP, QCD and String Theory Integrability in QCD and beyond - p. 29/35
Addendum: Polarized deep-inelastic scattering
Cross section: � = pq=M = E
l
�E
l
0
d
2
�
ddE
l
0
(#"+"#) =
8�
2
E
2
l
0
Q
4
�
�
F
2
(x;Q
2
) os
2
�2
+
2�
M
F
1
(x;Q
2
) sin
2
�2
�
;
d
2
�
ddE
l
0
(#"�"#) =
8�
2
E
l
0
Q
4
x
�
g
1
(x;Q
2
)
�
1+
E
l
0
E
l
os �
�
�
2Mx
E
l
g
2
(x;Q
2
)
�
At tree level:
g
1
(x) =
Z
1
�1
d�
2�
e
i�x
hp; s
z
j�q(0)6n
5
q(�n)jp; s
z
i
g
T
(x) =
1
2M
Z
1
�1
d�
2�
e
i�x
hp; s
?
j�q(0)
?
5
q(�n)jp; s
?
i
— quark distributions in longitudinally and transversely polarized nucleon, respectively
KITP, QCD and String Theory Integrability in QCD and beyond - p. 30/35
Polarized deep-inelastic scattering — continued
Beyond the tree level
hN(p; s)j�q(z
1
)G(z
2
)q(z
3
)jN(p; �)i =
= : : :
Z
1
�1
dx
1
dx
2
dx
3
Æ(x
1
+x
2
+x
3
) e
�ip(x
1
z
1
+x
2
z
2
+x
3
z
3
)
D
q
(x
i
; �
2
)
�q G q ) g
p�n
2
(x
B
; �
2
)
�q G q
GGG
)
) g
p+n
2
(x
B
; �
2
)
quark-gluon correlations in the nucleon
KITP, QCD and String Theory Integrability in QCD and beyond - p. 31/35
Renormalization of quark-gluon operators, flavor non-sing let
H
qGq
= N
H
(0)
�
2
N
H
(1)
;
H
(0)
= V
(0)
qg
(J
12
) + U
(0)
qg
(J
23
) ;
H
(1)
= V
(1)
qg
(J
12
) + U
(1)
qg
(J
23
) + U
(1)
(J
13
) :
where
V
(0)
qg
(J) = (J + 3=2) + (J � 3=2)� 2 (1)� 3=4 ;
U
(0)
qg
(J) = (J + 1=2) + (J � 1=2)� 2 (1)� 3=4 ;
V
(1)
qg
(J) =
(�1)
J�5=2
(J � 3=2)(J � 1=2)(J + 1=2)
;
U
(1)
qg
(J) = �
(�1)
J�5=2
2(J � 1=2)
;
V
(1)
(J) = (J)� (1)� 3=4 ;
U
(1)
(J) =
12
[ (J � 1) + (J + 1)℄� (1)� 3=4 :
KITP, QCD and String Theory Integrability in QCD and beyond - p. 32/35
Open inhomogeneous spin chain
5 10 15 20 25 3010
15
20
25
30
35
40 γN
N
Exact analytic solution
� The lowest level is special
and is separated from the rest of
the spectrum by a “mass gap”
� This special level was found by ABH 91’ and it determines the e volution of g
2
(x;Q
2
) to the
leading logarithmic accuracy
� It corresponds to the particular combination of quark-anti quark-gluon operators that can be
reduced to the quark-antiquark twist-three operator using equations of motion
Detailed study:
Belitsky; Derkachov, Korchemsky, Manashov ’00
KITP, QCD and String Theory Integrability in QCD and beyond - p. 33/35
Flavor singlet qGq and GGG operators
BDM ’01
10
15
20
25
30
35
40
45
0 2 4 6 8 10 12 14 16 18 20
EN
,α
N
qGq levels: crosses GGG levels: open circles
The highest qGq level is separated
from the rest by a finite gap and is
almost degenerate with the lowest
GGG state
� Interacting open and closed spin chains
KITP, QCD and String Theory Integrability in QCD and beyond - p. 34/35
Flavor singlet qGq and GGG operators: WKB expansion
Energies: j = N + 3
E
low
q
= N
�
2 [j℄ +
1
j
�
12
+ 2
E
�
�
2
N
�
ln j +
E
+
34
�
�
2
6
+ : : :
�
E
high
q
= N
�
4 ln j + 4
E
�
19
6
�
19
3j
2
�
+
23
n
f
+O(1=j
3
)
E
low
G
= N
�
4 ln j + 4
E
+
13
�
�
2
3
+
1
j
�
2�
2
�18
3
(ln j +
E
)�
�
2
3
��
+O(ln
2
j=j
2
)
To the leading logarithmic accuracy, the structure functio n g2
(x;Q
2
) is expressed through the
quark distribution
g
2
(x;Q
2
) = g
WW
2
(x;Q
2
) +
12
X
q
e
2q
Z
1
x
dy
y
�q
+
T
(y;Q
2
)
KITP, QCD and String Theory Integrability in QCD and beyond - p. 35/35
Approximate evolution equation for the structure function g
2
(x;Q
2
)
Introduce flavor-singlet quark and gluon transverse spin di stributions
Q
2
d
dQ
2
�q
+; S
T
(x;Q
2
) =
�
s
4�
Z
1
x
dy
y
h
P
T
(x=y)�q
+; S
T
(y;Q
2
) + P
T
qg
(x=y)�g
T
(y;Q
2
)
i
Q
2
d
dQ
2
�g
T
(x;Q
2
) =
�
s
4�
Z
1
x
dy
y
P
T
gg
(x=y)�g
T
(y;Q
2
)
with the splitting functions
P
T
(x) =
�
4C
F
1� x
�
+
+ Æ(1� x)
�
C
F
+
1
N
�
2�
�
2
3
��
� 2C
F
;
P
T
gg
(x) =
�
4N
1� x
�
+
+ Æ(1� x)
�
N
�
�
2
3
�
13
�
�
23
n
f
�
+N
�
�
2
3
� 2
�
+N
ln
1� x
x
�
2�
2
3
� 6
�
;
P
T
qg
(x) = �4n
f
�
x� 2(1� x)
2
ln(1� x)
�
: