CERN-TH/98-188

KUL-TF-98/21
hep-th/9806087

The Holographic Weyl Anomaly

arXiv:hep-th/9806087 v2 17 Jun 1998

M. Henningson

Theory Division, CERN

CH-1211 Geneva 23, Switzerland
[email protected]

K. Skenderis

Instituut voor Theoretische Fysica, KU Leuven

Celestijnenlaan 200D, B-3001 Leuven, Belgium
[email protected]

Abstract
We calculate the Weyl anomaly for conformal field theories that can be described
via the adS/CFT correspondence. This entails regularizing the gravitational part of the
corresponding supergravity action in a manner consistent with general covariance. Up
to a constant, the anomaly only depends on the dimension d of the manifold on which
the conformal field theory is defined. We present concrete expressions for the anomaly
in the physically relevant cases d = 2, 4 and 6. In d = 2 we find for the central charge
c = 3l/2GN , in agreement with considerations based on the asymptotic symmetry algebra
of adS3 . In d = 4 the anomaly agrees precisely with that of the corresponding N = 4
superconformal SU(N) gauge theory. The result in d = 6 provides new information for
the (0, 2) theory, since its Weyl anomaly has not been computed previously. The anomaly
in this case grows as N 3 , where N is the number of coincident M5 branes, and it vanishes
for a Ricci-flat background.

CERN-TH/98-188
KUL-TF-98/21
June 1998
1 Introduction
At low energies, the worldvolume theory on N coincident p-branes in M-theory or string
theory decouples from the bulk theory and can be studied on its own. In some cases, the
worldvolume theory constitutes a conformal field theory (CFT). This is true, for example,
for D3-branes in type IIB string theory and for five-branes in M-theory, which give rise to
the d = 4 N = 4 superconformal SU(N) gauge theory and a d = 6 (0, 2) superconformal
field theory, respectively. It has recently been conjectured by Maldacena [1], following
earlier work on black holes [2]–[6], that these conformal field theories are dual to M-
theory or string theory in the background describing the near-horizon brane configuration.
This equivalence may also be inferred by observing that the brane configuration can be
mapped to its near-horizon limit [7] by means of certain duality transformations [8].
However, this argument is as yet incomplete, since these duality transformations are
not fully understood, as they involve the time coordinate. The correspondence between
string theory in a specific background and conformal field theories is a realization of the
holographic principle advocated by ’t Hooft [9] and Susskind [10] in that it describes
a (d + 1)-dimensional theory containing gravity in terms of degrees of freedom on a
d-dimensional hypersurface.
The conjectured correspondence was clarified by Gubser, Klebanov and Polyakov [11]
and by Witten [12] as follows: the supergravity background is a product of a compact
manifold and a (d + 1)-dimensional manifold Xd+1 with a boundary (“horizon”) Md .
The conformal field theory is defined on Md . There is a one-to-one relationship between
operators O of the conformal field theory and the fields φ of the supergravity theory.
In particular, gauge fields in the bulk couple to global currents in the boundary. The
presence of a boundary means that the supergravity action functional S[φ] must be
supplemented by a boundary condition for φ parametrized by a field φ(0) on Md . The
partition function is then a functional of the boundary conditions
Z
(0)
Zstring [φ ] = Dφ exp (−S[φ]) , (1)
φ(0)

where the subscript φ(0) on the integral sign indicates that the functional integral is over
field configurations φ that satisfy the boundary condition given by φ(0) . The conjecture
states that the string (or M-theory) partition function, as a functional of φ(0) , equals the
generating functional of correlation functions in the conformal field theory:
 Z 
ZCF T [φ(0) ] = exp dd xOφ(0) . (2)
Md

The fields φ(0) act as sources for the operators of the conformal field theory. Notice that
the bulk theory only sees, through the boundary values of its fields, the abstract conformal
field theory and not the elementary fields that may realize it. The partition function

1
(2) may also be viewed as describing the coupling of conformal matter to conformal
supergravity [13]. The sources φ(0) constitute conformal supermultiplets.
The relationship just described is conjectured to hold for any number N of coincident
branes. However, in most cases one can reliably compute the string partition function
only for large N. The reason is that the backgrounds involve RR forms whose coupling
to perturbative strings is through D-branes. Therefore a complete string calculation is
rather difficult to perform. However, if the number of branes is large, the characteristic
length scale of the supergravity background is large compared to the string scale (or the
Planck scale in the case of M-theory), and one can trust the supergravity approximation.
In addition the string coupling may be chosen small. Under these circumstances, the
string partition function reduces to the exponential of the supergravity action functional
evaluated for a field configuration φcl (φ(0) ) that solves the classical equations of motion
and satisfies the boundary conditions given by φ(0)
 
tree
Zstring [φ(0) ] = exp −S[φcl (φ(0) )] . (3)

An operator of particular importance in any conformal field theory is the energy-

momentum tensor Tij , i, j = 1, . . . , d. The corresponding bulk gauge field is the metric
Ĝµν , µ, ν = 0, . . . , d on Xd+1 . In the supergravity backgrounds under consideration,
the metric Ĝµν does not induce a unique metric g(0) on the boundary Md , because it
has a second-order pole there. However it does determine a conformal equivalence class
or conformal structure [g(0) ] of metrics on Md . To get a representative g(0) , we pick
a function ρ on Xd+1 with a simple zero on Md and restrict ρ2 Ĝµν to Md . Different
choices of the function ρ yield different metrics on Md in the same conformal equivalence
class. The field that specifies the boundary condition of the metric is thus a conformal
structure [g(0) ]. This means that, at least naively, the trace of the energy-momentum
tensor decouples.
In this paper we wish to determine the dependence of the boundary theory partition
function (or zero-point function) on a given representative g(0) of the conformal structure.
In other words, we shall study whether the trace of the energy-momentum decouples.
Since we examine correlation functions that only involve the energy-momentum tensor,
the only relevant part of the bulk action is the gravitational one. Therefore we set all other
fields to zero. At tree level, we then need to solve the classical supergravity equations of
motion on Xd+1 subject to the conditions that the metric Ĝµν on Xd+1 induces a given
conformal structure [g(0) ] on Md and all other fields vanish there. In the theories under
consideration, this means that Ĝµν fulfils Einstein’s equations
1
R̂µν − Ĝµν R̂ = ΛĜµν , (4)
2
with some cosmological constant Λ and that all other fields vanish identically on Xd+1 .
According to a theorem due to Graham and Lee [14], up to diffeomorphism, there is

2
a unique such metric Ĝµν . (Actually the theorem has been proved for the case when
Xd+1 is topologically a ball Bd+1 so that Md is a sphere S d and the conformal structure
[g(0) ] on Md is sufficiently close to the standard (conformally flat) one.) The conformal
field theory effective action (strictly speaking, the generating functional of the connected
graphs) WCF T [g(0) ] = − log ZCF T [g(0) ] is then given by evaluating the action functional

S[Ĝµν ] = Sbulk + Sboundary

"Z #
1
q   Z q
d+1
= (d+1)
d x det Ĝ R̂ + 2Λ + dd x det g̃ (2Dµ nµ + α) (5)
16πGN Xd+1 Md

for this metric. Here g̃ is the metric induced on Md from Ĝ, and nµ is a unit normal vector
to Md . The bulk term is of course the usual Einstein-Hilbert action with a cosmological
constant. The inclusion of the first boundary term is necessary on a manifold with
boundary in order to get an action that depends only on first derivatives of the metric
[15]. The possibility of including the second boundary term with some coefficient α was
first discussed in [13].
The above description might seem to indicate that the conformal field theory effective
action WCF T [g(0) ] only depends on the conformal equivalence class of the metric on Md .
This is of course as it should be in a truly conformally invariant theory. However, the
action functional (5) does not make sense for the metric Ĝµν determined by (4) and the
boundary conditions. Indeed, the bulk term of the action diverges because of the infinite
volume of Xd+1 . The boundary terms are also ill-defined, since the induced metric g̃ij on
Md diverges because of the double pole of Ĝµν . The action should therefore be regularized
in a way that preserves general covariance, so that the divergences can be cancelled by
the addition of local counterterms. As we will see shortly, this regularization entails
picking a particular, but arbitrary, representative g(0) of the conformal structure [g(0) ] on
Md . In this way, one obtains a finite effective action, which, however, will depend on the
choice of this representative metric. Conformal invariance is thus explicitly broken by a
so-called conformal or Weyl anomaly. The anomaly, which is usually perceived as a UV
effect, thus arises from an IR-divergence in the bulk theory. This is an example of a more
general IR-UV connection that applies to holographic theories [16].
In this paper, we will calculate the Weyl anomaly for conformal field theories that can
be derived from a supergravity theory, as described above. In the next section, we will
describe the regularization procedure and the computation of the anomaly in general. In
the last section, we evaluate the anomaly in the physically relevant cases d = 2, 4, 6. For
d = 2 and d = 4 we compare with the known anomaly for the adS3 boundary conformal
field theory and the d = 4 N = 4 superconformal SU(N) gauge theory respectively,
and find perfect agreement. For d = 6 there is no corresponding calculation of the Weyl
anomaly, so our result provides new information about the (0, 2) superconformal field
theory.

3
2 The regularization procedure
A regularization scheme that preserves general covariance was described in [12]. As
discussed above, up to diffeomorphisms, there is a unique Einstein metric Ĝ on Xd+1
that induces a given a conformal structure [g(0) ] on the boundary Md . We now pick a
metric g(0) on Md in the given conformal equivalence class. According to a theorem due
to Fefferman and Graham [17], there is a distinguished coordinate system (ρ, xi ) on Xd+1
in which Ĝ takes the form
l2 −2
Ĝµν dxµ dxν = ρ dρdρ + ρ−1 gij dxi dxj , (6)
4
where the tensor g has the limit g(0) as one approaches the boundary represented by
ρ = 0. The length scale l is related to the cosmological constant Λ as Λ = − d(d−1)
2l2
.
Einstein’s equations for Ĝ amount to
 
ρ 2g ′′ − 2g ′g −1 g ′ + Tr(g −1g ′ )g ′ + l2 Ric(g) − (d − 2)g ′ − Tr(g −1 g ′ )g = 0
 
(g −1)jk ∇i gjk
′
− ∇k gij′ = 0
1
Tr(g −1g ′′ ) − Tr(g −1g ′ g −1g ′ ) = 0, (7)
2
where differentiation with respect to ρ is denoted with a prime, ∇i is the covariant
derivative constructed from the metric g and Ric(g) is the Ricci tensor1 of g.
In the case when d is odd, these equations can be solved order by order in ρ so that

g = g(0) + ρg(2) + ρ2 g(4) + . . . , (8)

where the tensor g(k) is given by some covariant expression in the boundary metric g(0) ,
its Riemann tensor and the corresponding covariant derivative. Throughout this paper,
a subscript in parentheses on a quantity indicates the number of derivatives with respect
to xi . In the case when d is even, this procedure breaks down at order d/2 in ρ, where a
logarithmic term appears:

g = g(0) + ρg(2) + . . . + ρd/2 g(d) + ρd/2 log ρ h(d) + O(ρd/2+1 ). (9)

−1
The tensors g(k) for k = 0, 2, . . . , d−2 are again covariant. The same is true for Tr(g(0) g(d) )
−1
but not for the complete tensor g(d) . Finally, Tr(g(0) h(d) ) vanishes identically.
The regularization procedure now amounts to restricting the bulk integral to the
domain ρ > ǫ for some cutoff ǫ > 0 and evaluating the boundary integrals at ρ = ǫ. The
(d+1)
regulated action evaluated for the metric Ĝ is thus (16πGN )−1 dd x L, where
R

dZ −d/2−1
q
L = dρρ det g
l ǫ
1
Our conventions are as follows Rijk l = ∂i Γjk l + Γip l Γjk p − i ↔ j and Rij = Rikj k .

4
 !
−d/2 2d q 4 q q 
+ρ − det g + ρ∂ρ det g + α det g 
. (10)
l l


ρ=ǫ

In the first term, which arises from the bulk part of the action, we have used the fact
4
that Ĝ is an Einstein metric so that R̂ + 2Λ = − d−1 Λ = 2d
l2
.
√
For d odd, it follows from (8) that det g is a power series in ρ with covariant
coefficients. For d even, this is true up to and including the ρd/2 terms. (The higher-
order non-covariant corrections will play no role in the sequel). The Lagrangian (10) can
therefore be written as
q  
L= det g(0) ǫ−d/2 a(0) + ǫ−d/2+1 a(2) + . . . + ǫ−1/2 a(d−1) + Lf in (11)

for d odd, and as

q  
L= det g(0) ǫ−d/2 a(0) + ǫ−d/2+1 a(2) + . . . + ǫ−1 a(d−2) − log ǫ a(d) + Lf in , (12)

for d even, where Lf in is finite in the ǫ → 0 limit. All the a(k) coefficients are covariant, so
the divergent terms can be cancelled by subtracting covariant counterterms, as promised.
The logarithmic divergence that appears for d even comes only from the bulk integral.
After subtraction of the divergent counterterms, we are left with a renormalized ef-
(d+1)
fective action (16πGN )−1 dd x Lf in with a finite limit as ǫ goes to zero. Its variation
R

under a conformal transformation δg(0) = 2δσg(0) for an infinitesimal parameter function

δσ is of the form Z q
δLf in = − dd x det g(0) δσA, (13)
Md

and we would like to calculate the anomaly A. For d odd, A in fact vanishes, whereas
for d even
1
A= (d+1)
(−2a(d) ). (14)
16πGN
To see this, we note that for a constant parameter δσ, the regulated Lagrangian (11) or
(12) is invariant under the combined transformation δg(0) = 2δσg(0) and δǫ = 2δσǫ. The
terms proportional to negative powers of ǫ are separately invariant, so the variation of
the finite part plus the variation of the logarithmically divergent term (for d even) must
q
vanish. Since log ǫ transforms with a shift and det g(0) a(d) itself is invariant, we get
(14).
On general grounds [18, 19], the coefficient a(d) that appears in the anomaly (14) must
be of the form  
i
a(d) = dld−1 E(d) + I(d) + Di J(d−1) , (15)
where E(d) is proportional to the d-dimensional Euler density and I(d) is a conformal
invariant. These terms are referred to as the type A and the type B anomaly, respectively,
i
in [19]. The dimension of the space of conformal invariants grows with d. The Di J(d−1)

5
term, where Di is the covariant derivative constructed from the boundary metric g(0) ,
is trivial in the sense that it can be cancelled by the variation of a finite covariant
counterterm added to the action. To see this, notice that a covariant counterterm will
be, in particular, scale invariant. Making the parameter of the scale transformation
local amounts to computing the Noether current for scale transformations. Thus, the
result of the variation is δσDi J i . However, local scale transformations are just Weyl
transformations. Thus terms of the form Di J i can be obtained by variation of covariant
counterterms.
The coefficients of the various independent contributions (properly normalized) in
(15) are closely related to renormalization group equations, and they reflect the matter
content of the superconformal theory. Using Ward identities one can relate them to
Schwinger terms in the OPEs of the energy-momentum tensor [20]–[23]. For a recent
application, see [5]. Our results for d = 6 can be similarly used to determine Schwinger
terms in the OPEs of the (0, 2) theory.

3 Evaluation of the anomaly

In this section, we will perform the above procedure in the physically relevant cases
d = 2, 4, 6 and give concrete formulas for the quantities E, I and J i appearing in (15).
As we have mentioned in the previous paragraph, the logarithmic divergence comes only
from the bulk integral. It is completely straightforward to obtain a(d) . One only needs to
√
expand det g up to appropriate order in ρ. In the formulae below, we further simplify the
−1
result by eliminating Tr(g(0) g(n) ), for n > 2, by using the third equation in (7). We raise
−1
and lower indices with the boundary metric g(0) and its inverse g(0) . The Riemann tensor
i
and the covariant derivative constructed from g(0) are denoted R jkl and Di , respectively.

3.1 d = 2 and the asymptotic symmetry algebra of adS3

Calculating
−1
a(2) = l Tr(g(0) g(2) ) (16)
and decomposing it according to (15), we get
1
E(2) = R
4
I(2) = 0
i
J(1) = 0. (17)

(There is in fact no non-trivial conformal invariant I in this dimension.) Writing the

anomaly in the form
c
A=− R (18)
24π
6
(in our conventions a free boson contributes to the anomaly −1/24πR) we thus get
3l
c= (3)
. (19)
2GN

This agrees with the value of the conformal anomaly c as computed in [24] by considering
the asymptotic symmetry algebra of adS3 .

3.2 d = 4 and N = 4 super Yang-Mills theory

In this case one finds

1 
a(4) = l3 −1
[Tr(g(0) g(2) )]2 − Tr[(g(0)
−1
g(2) )2 ]
2
1 ij 1 2

3
= l − R Rij + R . (20)
8 24
Notice that this expression vanishes for a Ricci-flat background. A check on our calcu-
lation is whether (20) can be rewritten in the form (15). Indeed, this is possible, and we
obtain
1  ijkl 
E(4) = R Rijkl − 4Rij Rij + R2
64 
1 1

I(4) = − Rijkl Rijkl − 2Rij Rij + R2
64 3
i
J(3) = 0. (21)

(Up to a constant, I(4) is in fact the unique conformal invariant with four derivatives
in this dimension, namely the Weyl tensor contracted with itself.) We now use the fact
(5) (10)
that gN = gN /V ol(S 5 ), where V ol(S 5 ) = l5 π 3 is the volume of the compactification
(10)
five-sphere of radius l, and GN = 8π 6 gstr
2
is the ten-dimensional Newton’s constant (the
α ’s cancel out in the Maldacena limit). Furthermore, l is related to the number N of
′

D3-branes as l = (4πgstr N)1/4 [1]. Putting everything together, we get

N2  
A=− E(4) + I(4) . (22)
π2
This should be compared with the conformal anomaly of the d = 4 N = 4 superconformal
SU(N) gauge theory. The conformal anomaly of a theory with ns scalar fields, nf Dirac
fermions and nv vector fields is [25]
1 1
− 2
(ns + 11nf + 62nv )E(4) − (ns + 6nf + 12nv )I(4) . (23)
90π 30π 2
The anomaly of the N = 4 SU(N) super Yang-Mills multiplet is equal to N 2 − 1 (as all
fields are in the adjoint) times (23) for ns = 6, nf = 2 and nv = 1. Thus, in the large N
limit we obtain exact agreement with (22). This is perhaps surprising since the result (23)
is derived using free fields whereas our result is about the full interacting N = 4 SU(N)

7
superconformal field theory. This indicates that there must be a non-renormalization
theorem that protects these coefficients.
Various orbifolding procedures [26, 27] change the volume of the compactification
space and also give rise to other gauge groups. It is easy to check that the anomalies still
work out correctly.

3.3 d = 6 and tensionless strings

Following [18], we introduce


(K1 , . . . , K11 ) = R3 , RRij Rij , RRijkl Rijkl , Ri j Rj k Rk i , Rij Rkl Riklj ,
Rij Riklm Rj klm , RijklRijmn Rkl , R Rimnl Rj mn k ,
 mn ijkl
R R, Rij Rij , Rijkl Rijkl . (24)

The six-dimensional Euler density is then proportional to

E0 = K1 − 12K2 + 3K3 + 16K4 − 24K5 − 24K6 + 4K7 + 8K8 (25)

and
19 57 3 7 9 3
I1 = K1 − K 2 + K 3 + K 4 − K5 − K 6 + K8
800 160 40 16 8 4
9 27 3 5 3
I2 = K1 − K2 + K3 + K4 − K5 − 3K6 + K7
200 40 10 4 2
1
I3 = K1 − 8K2 − 2K3 + 10K4 − 10K5 − K9 + 5K10 − 5K11 (26)
2
form a basis for conformal invariants with six derivatives.
We have
1 3

a(6) = l5 −1
[Trg(0) g(2) ]3 − Tr[g(0)
−1 −1
g(2) ]Tr[(g(0) g(2) )2 ]
8 8
1

−1
+ Tr[(g(0) g(2) )3 ] − Tr[g(0)
−1 −1
g(2) g(0) g(4) ] . (27)
2
Evaluating this expression we obtain
l5 1 3

a(6) = − RRij Rij + R3 + Rij Rkl Rikjl
64 2 50
1 1 1

+ Rij Di Dj R − Rij Rij + R R . (28)
5 2 20
Observe that the above expression vanishes in a Ricci-flat background. The next task
is to put this expression in the form (15). This is a nice check on our calculation. The
result is
1
E(6) = E0
6912
8
 1 10 1 1
 
I(6) = − I1 − I2 + I3
1152 3 6 10
i 1
J(5) = − [−R D Rmjkl + 2(Rjk D i Rjk − Rjk D j Rki )]
ijkl m
1152
1 ij 17
+ R Dj R + RD i R (29)
720 11520
(7) (11)
We now use the fact that GN = GN /V ol(S 4 ), where V ol(S 4 ) = Rsph 4
(8π 2 /3) and
Rsph = lP lanck (πN)1/3 is the radius of the compactification sphere. In addition, the eleven-
(11)
dimensional Newton’s constant is equal to GN = 16π 7 lP9 lanck , and the characteristic
length l is l = 2lP lanck (πN)1/3 . Putting every together we get for the anomaly

4N 3  i

A=− E(6) + I(6) + D i J (5) . (30)
π3
The anomaly for a (0, 2) tensor multiplet has not yet been calculated. However, we see
that the anomaly grows as N 3 , in agreement with considerations based on the entropy
of the brane system [28, 5]. This growth is presumably related to the appearance of
tensionless strings when multiple fivebranes coincide.

Acknowledgements
We would like to thank each other’s institute for hospitality and financial support during
part of this work.

References
[1] J.M. Maldacena, The Large N Limit of Superconformal Field Theories and Super-
gravity, hep-th/9711200.

[2] S.S. Gubser, I.R. Klebanov and A.W. Peet, Entropy and Temperature of Black 3-
branes, Phys. Rev. D54 (1996) 3915, hep-th/9602135.

[3] I.R. Klebanov, World Volume Approach to Absorption by Non-dilatonic Branes,

Nucl. Phys. B496 (1997) 231, hep-th/9702076.

[4] S.S. Gubser, I.R. Klebanov and A.A. Tseytlin, String Theory and Classical Absorp-
tion by Threebranes, Nucl. Phys. B499 (1997) 217, hep-th/9703040.

[5] S.S. Gubser and I.R. Klebanov, Absorption by Branes and Schwinger Terms in the
World Volume Theory, Phys. Lett. B413 (1997) 41, hep-th/9708005.

9
 [6] K. Sfetsos and K. Skenderis, Microscopic derivation of the Bekenstein-Hawking en-
tropy formula for non-extremal black holes, Nucl. Phys. B517 (1998) 179, hep-
th/9711138.

[7] H.J. Boonstra, B. Peeters and K. Skenderis, Duality and asymptotic geometries,
Phys. Lett. B411 (1997) 59, hep-th/9706192; Branes and anti-de Sitter spacetimes,
hep-th/9801076.

[8] S. Hyun, U-duality between Three and Higher Dimensional Black Holes, hep-
th/9704005.

[9] G. ’t Hooft, Dimensional reduction in quantum gravity, in Salamfest 1993, p.284,

gr-qc/9310026.

[10] L. Susskind, The World as a Hologram, J. Math. Phys. 36 (1995) 6377, hep-
th/9409089.

[11] S.S. Gubser, I.R. Klebanov and A.M. Polyakov, Gauge Theory Correlators from
Non-Critical String Theory, hep-th/9802109.

[12] E. Witten, Anti-de Sitter space and Holography, hep-th/9802150.

[13] H. Liu and A.A. Tseytlin, D = 4 Super Yang-Mills, D = 5 gauged supergravity, and
D = 4 conformal supergravity, hep-th/9804083.

[14] C.R. Graham and J.M. Lee, Einstein Metrics with Prescribed Conformal Infinity on
the Ball, Adv. Math. 87 (1991) 186.

[15] G.W. Gibbons and S.W. Hawking, Action integrals and partition functions in quan-
tum gravity, Phys. Rev. D15 (1977) 2752.

[16] L. Susskind and E. Witten, The Holographic Bound in Anti-de Sitter Space, hep-
th/9805114.

[17] C. Fefferman and C.R. Graham, “Conformal Invariants”, in Elie Cartan et les
Mathématiques d’aujourd’hui (Astérisque, 1985) 95.

[18] L. Bonora, P. Pasti and M. Bregola, Weyl Cocycles, Class. Quant. Grav. 3 (1986)
635.

[19] S. Deser and A. Schwimmer, Geometric Classification of Conformal Anomalies in

Arbitrary Dimensions, Phys. Lett. B309 (1993) 279, hep-th/9302047.

[20] H. Osborn and A. Petkos, Implications of Conformal Invariance in Field Theories

for General Dimensions, Ann. Phys. 231 (1994) 311, hep-th/9307010.

10
[21] D. Anselmi, M. Grisaru and A. Johansen, A Critical Behaviour of Anomalous Cur-
rents, Electric-Magnetic Universality and CF T4 , Nucl. Phys. B491 (1997) 221, hep-
th/9601023.

[22] J. Erdmenger and H. Osborn, Conserved Currents and the Energy Momentum Ten-
sor in Conformally Invariant Theories for General Dimensions, Nucl. Phys. B483
(1997) 431, hep-th/9605009.

[23] D. Anselmi, D.Z. Freedman, M.T. Grisaru, and A.A. Johansen, Nonperturbative For-
mulas for Central Functions of Supersymmetric Gauge Theories, hep-th/9708042.

[24] J.D. Brown and M. Henneaux, Central charges in the canonical realization of asymp-
totic symmetries: An example from three-dimensional gravity, Commun. Math.
Phys. 104 (1986) 207.

[25] M.J. Duff, Twenty years of the Weyl anomaly, Class. Quant. Grav. 11 (1994) 1387,
hep-th/9308075 and references therein.

[26] S. Kachru and E. Silverstein, 4d Conformal Field Theories and Strings on Orbifolds,
Phys. Rev. Lett. 80 (1998) 4855, hep-th/9802183.

[27] E. Witten, Baryons and Branes In Anti-de Sitter Space, hep-th/9805112.

[28] I.R. Klebanov and A.A. Tseytlin, Entropy of Near-Extremal Black p-branes, Nucl.
Phys. B475 (1996) 164, hep-th/9604089.

11