Professional Documents
Culture Documents
North-Holland, Amsterdam
S T R I N G L O O P C O R R E C T I O N S T O BETA F U N C T I O N S
We study the problem of finding the beta functions, and the associated spacetime effective
action, for interacting open and closed strings propagating in background fields. String loop
divergences play a crucial role in this problem. Cancelling them against sigma model divergences
gives a consistent set of loop-corrected beta functions, which can be derived from a simple
generalization of the string-tree-leveleffective action. This suggests the existence of new string
theories which are conformallyinvariant only after all world sheets have been summed.
1. Introduction
tive answer would shed useful light on the deeper symmetry structure of string field
theory, and would probably suggest new ways to look for string-compatible vacua.
Lovelace [8] and Fischler and Susskind [9] have suggested a way of generalizing
the renormalization group beta functions of sigma models to include string loop
effects. The idea is that, if the loop expansion is defined by the Polyakov path
integral sum over world sheets of different topology, nonstandard sigma model
counterterms are required to remove the divergences in the modular parameter
integrations over inequivalent world sheets of the same topology. Letting the
renormalization group flow be defined by the sum of the standard and nonstandard
counterterms defines loop-corrected beta functions whose vanishing can be taken as
the conformal invariance conditions of a generalized sigma model. Fischler and
Susskind showed that, by using this strategy in a very simple contest, it was possible
to see the cosmological constant of closed bosonic string theory appearing as a
one-loop correction to the beta function for the background metric field.
In this paper we extend this very promising idea to more elaborate contexts where
it is possible to apply nontrivial internal consistency checks. We regard this as quite
important, since the regulation and renormalization procedure used to define the
new counterterms is rather ad hoc and not guaranteed to maintain the spacetime
gauge symmetries of string theory. Specifically, we determine the first corrections to
closed string beta functions due to open string loops, with the effects of open and
closed string background fields included. With this knowledge of background field
dependence, we can make a nontrivial check of the mutual consistency of the
loop-corrected beta function equations. We do find consistency and find further-
more that the full equations may be derived from a loop-corrected spacetime action
of very reasonable form. It is probable that this action generates the appropriate
loop-corrected S-matrix elements of string theory. In short, the entire set of
relationships between sigma models, beta functions and the string theory S-matrix
seems to survive the passage from tree to loop amplitudes. The precise nature of the
conformal invariance of the underlying sigma model, apart from the observation
that it involves cancellation of conformal invariance violations between different
world sheets, remains only partially understood.
The paper is organized as follows: In sects. 2 and 3 we determine the tree-level
beta functions for open and closed strings in general backgrounds. This actually
allows us to infer what the open string loop corrections to the closed string beta
functions must be. In sect. 4 we review, and slightly improve, the Fischler and
Susskind computation of the first closed string loop corrections to beta functions. In
sect. 5 we find the counterterms needed to renormalize open string loop divergences
in general background fields and derive the associated loop corrections to closed
string beta functions. In sect. 6 we discuss the meaning and internal consistency of
our results and draw conclusions. In an appendix, we present an alternative
derivation of our key results by more conventional operator methods (as opposed to
the sigma model methods used in the body of the paper).
C G. Callan et al. / String loop corrections to B-functions 527
1
Sc = - - f d2z [!/y'gabg~v ( X ) CgaX~Obx v
4 ~rot'
where X" is the spacetime coordinate of the string, 3'ab is the world sheet metric,
and R (2) is its curvature. Equations of motion for the backgrounds are obtained by
imposing conformal invariance or, equivalently, demanding that the renormalization
group beta function for each field vanish. Among other things, this makes the
physics independent of conformal transformations on the world sheet metric. At
string tree level and to leading order in a', the beta functions are [4]
f i b = trvX H , 1
where Hx~ . = 3Vixb,, ]. Remarkably, the equations obtained by setting these beta
functions to zero are equivalent to the equations of motion for the spacetime action
where D is the spacetime dimension. The relation between the variational equations
for this action and the above beta functions is
3
sclosed
eel - ¢~-e*fl*,
3
3b~,,
8 / ,closed =
+ ½gp,~/"eft grg-e*flgp • (2.4)
derivatives of the background fields [6,10, 7]. This whole procedure generates the
conditions on closed string background fields arising from tree-level, or classical
string physics.
Our main concern in this paper will be the way in which the background field
equations are modified by non-classical, or string-loop, effects. As mentioned in the
introduction, we have found it very useful to approach this problem indirectly, by
studying tree-level open strings in general backgrounds. This gives us very useful
information about string loop physics because of the well-known fact that the basic
interaction between open and closed strings is really a loop phenomenon.
The problem of an open string coupled to open string backgrounds has been
considered by several authors [11-14]. The appropriate sigma model is
1 0
So= f d 2z ll~yabOaX ~*ObXJ*-~dsAt,( X)-~s X ~ , (2.5)
where ds is the length element around the boundary induced by "lab- In this case,
the beta function was found to be
This result is valid to all orders in a' and to lowest order in derivatives of F. Setting
/3A= 0 gives an equation of motion equivalent to that which follows from the
Born-Infeld action
Sef pen=
f
f dDx ¢det(1 + 27ra'F) (2 7)
.~,eff =
~A ~ vopen
-¢det(1 + 2~ra'F)[1
_
(2~ra r F ) 2 ],v
-1 u
fl]- (2.8)
This is compatible with the observation that the effective action and the beta
functions are not expected to be identical, but rather to satisfy the relation
8seff/~A~ = v A
for a nonsingular matrix X,, [6,15]. For our purposes, it will be necessary to know
the beta functions for the open string in a general background of both closed and
open strings. The relevant results and representative calculations are presented in
the next section.
C G. Callan et al. / String loop corrections to fl-functions 529
The coupling of an open string to all massless backgrounds, arising from both
closed and open strings, is described by a sigma model of the form
S = m 1 f d 2 z [ v ~ 3 , . b g . . ( X ) O~X.ObX ~
4 ~ra'
--ieabb.~(
X) O~X g ObX ~ - ½a'Cy-R(2)~ ( X ) ]
1 [ o ] (3.1)
+ 2ira' ds iA.(X)--XU-os ½a'keo(X) '
where k is the extrinsic curvature of the boundary and A t has been rescaled to
include a factor of 1/2~ra'. Both the world sheet and spacetime metrics are taken to
be euclidean. The coupling of the dilaton to the boundary curvature is needed
because e e is the coupling constant of the string theory, and therefore must multiply
the entire Euler density. If we work in a conformal gauge 3'ab = eZ°~ab and let r
denote the coordinate along the boundary, then v~-R (2) = 2 [] p and k ds = - Onp dr,
where O, is the exterior normal derivative. Then the world sheet action may be
rewritten as
1
S [ X ] = ~ a , f d2z[g.~(X) OaX"O.X ~
i
+ 2rra---7~d'rA~,(X) O.~X~'. (3.2)
1
812, (1--s[xl + -47fdazOoo[~v~G¢(x) Gx"+ V . O ( X ) D ~ " +
v
O(~2)]
1
+ ~ f d2z [g..(.Y) N ~ " N a ( ~ + (3~b--e ~b)
i
+ 47r-a'~dr [(b(X') + F( ~-'))..~" D~"
1
where ~ is the generalized curvature for the connection F + IH, namely
and we define
D o e = (g". aa + r ~ a a ~ ) ~ ~,
Comparing with [13], we notice that on the boundary, F.. is replaced by F~. + b~..
Along the lines of [13], one can use the terms g.. O~" 0 ~ ~ and (b + F)..~" 8.~" in
(3.4) to define the propagator. At open string tree level, we may take the world sheet
to be the upper half-plane, with z = r + io. Then we obtain the propagator
G . . ( z , z') = -
1[
~a G.lnlz - z'l 2
In the limit where g, b, and F are constant, this propagator is exact. Using (3.5) to
compute the tadpole graph of fig. la gives a contribution to flf analogous to the
previous result (2.6), namely
~7"(b+ F ) ~ [ g - ( b + F)2] -1
from the graph fig. lb. If we use the exact propagator to compute this graph and use
the equations of motion (3.3), we find an addition
~x [ b+F
½(b+ F)u,H °[g_-~+F)2]xp
6 2 1 1 1
g f d z 4= Oap v.epOoX" = - -VV.p/OoY"OoE"-4=
1
+ ~8(o(z)) V,, 0,2".
+½(F+B),,H~XO[ F+b ]
g_~++b)2jxo+½V~ep(r+b),~. (3.6)
- e~'/Z(det(g + b + F ) [ g - (b + F)Z],~lfl~
with flA as defined in (3.6). Therefore, the vanishing of this variation is equivalent
to/3 A = 0 .
Although the open-string and closed-string beta functions are logically indepen-
dent of one another (after all, they arise from different sigma models evaluated on
different world sheets), it is quite natural to guess that the proper way to describe
interacting open and closed strings is simply to add the associated actions, (3.7) and
(2.3), together (with an as-yet-undetermined relative coefficient x). The proposed
full effective action is therefore
=f d~x [ 7 r g e * ( R - ~2H 2 - (Vq~) 2 - 2V2q~)+ t¢ e¢/Z~det(g + b + F ) ] .
(3.8)
Since the original terms in the effective action are of second order in derivatives of
the fields, while the added terms are dimensionless, we note that K will be of order
a'-1. Note that (3.8) is invariant under the combined transformation
This implies that A, can be totally absorbed in b,~ by a Higgs mechanism of a type
familiar in supergravity theories [17]. This effective action exactly reproduces the
open string beta function equations, but not those of the closed string. There are
additional terms in the variational equations for g,,, bu~, and q5 which amount to
gauge-field source terms for the closed-string fields. Specifically, we find
8
g
S eft
t°t~l --- -v/ge*fl ~' + ½x e 't'/2 ~/det(g + b + F )
8
--+2g#v~-~ "eft (b+F)2]~ '
(3.10)
where fig, fib, and fl* are given by (2.2). The new terms are certainly reasonable
(after all, gauge fields must act as a source for gravity), but are not visible in the
standard treatment of closed string beta functions: As noted in refs. [13] and [14],
C.G. Callan et al. / String loop corrections to S-functions 533
the presence of a boundary does not change the beta functions of the closed string
massless fields. When the relevant graphs are computed, it is found that the only
possible changes arise from the boundary terms in the propagator (3.5). However,
these terms do not contribute to the divergences which give the closed string beta
functions. In particular, the boundary contribution to G~,(z, z') is finite as z ~ z' in
the interior of the world sheet, so the logarithmic tadpole divergences responsible
for the one-loop contributions to/3g and/3 ° are unchanged. It is straightforward to
check that the other contributions to (2.2) are unchanged as well. The only
difference is that/3 ~ is now the coefficient of the entire Euler density, rather than
R ~2), in the trace of the sigma model stress-energy tensor. Therefore, if the new
terms in the closed string beta functions are correct, they must arise from as-yet-
unaccounted-for physics.
An essential clue is that the new terms contain a factor of e ~/2 relative to the
original closed-string beta functions and must therefore be regarded as string-loop
corrections. In the next section, we will discuss a general proposal for computing
such corrections to beta functions and, in the rest of the paper, we will use it to
show that all the equations of motion derived from (3.8) are interpretable as
loop-corrected beta functions.
operators must be added to Sc to yield finite results. The general form of that action
is
8Sc - l ° g A
2¢r
fd2z[f~V~Sg..(X) OoX"ObX"-½a'v/-~R'2'Seo(X)] (4.1)
where A is a cutoff parameter and 8g, 8~ are functions of the background fields.
They define the renormalization group beta functions, whose vanishing is the
condition for sigma model conformal invariance.
In performing the sum over two-dimensional world sheets to get the string loop
expansion, the conformal invariance of the sigma model makes it only necessary to
sum over conformally inequivalent surfaces. For a given genus world sheet, this
leaves a finite-dimensional parameter space, known as Teichmiiller space, to in-
tegrate over. The crucial points for our purposes are that the Teichmiiller parameter
integrations in general do not converge; that the divergences come from boundaries
of the parameter space where topological fixtures (such as handles) shrink to zero
size; that the divergence coming from shrinking a single fixture away is proportional
to the insertion of a simple local operator on the lower-genus world sheet obtained
by removing the shrunken fixture altogether [18]. For the sigma model correspond-
ing to flat empty spacetime, the divergence associated with shrinking away a single
handle is equivalent to the insertion of the operator
log A
2~r C~I~,~OoX/zO~Xv, (4.2)
8S~ °°p
= log2--~fd2zC~,~OaX
A .
OaX (4.3)
evaluated on the sphere, it gives a divergent contribution which precisely cancels the
divergence of the torus. Since it is already a one-loop effect, it would, to this order,
be neglected in calculating the torus contribution. Thus 8Scl°°p renormalizes the
string-loop divergences in very much the same way that 8S~ renormalizes the
ordinary field theory divergences. We must make two adjustments to the above
counterterm: In a general coordinatization of flat spacetime, the Minkowski metric
*/,~ must be replaced by the general metric g,~ and, in the presence of a constant
dilaton field, we must include a factor e ~ to account for the well-known topologi-
cally-determined dependence of the path integral on the dilaton zero mode (the
torus behaves as constant while the sphere behaves as e~'; thus the counterterm must
behave as e -~ in order that, when evaluated on the sphere, it reproduce the
divergence of the torus!). The result of these corrections is
8S2oop : f d2z
log____AA aox.aox (4.4)
2~r
v g 1
o = v = v.( - - ) 2) (4.6)
fields, and with equal and opposite U(1) gauge charges assigned to the inner and
outer boundaries, can be shown, using the techniques of [13] to be
where
We choose equal and opposite boundary charges in order that the states propagat-
ing around the loop be electrically neutral. If we consider only orientable world
sheets (i.e. if we do not include such surfaces as the projective plane or the Mrbius
strip in our world sheet summation), it is always possible to choose the boundary
charges such that all internal lines in loop diagrams are neutral. The other possible
choice, of equal charges on both boundaries, has been considered in [11]. It gives
different results, corresponds to different physics and will not be considered here. If
the background fields are not constant, the propagator is the same, but nontrivial
vertices, involving derivatives of background fields, appear in the expansion of the
sigma model action. We will ignore the effects of such terms, and our results should
be thought of as the first term in a systematic expansion of powers of spacetime
derivatives of background fields.
The expansion of the propagator in the limit as a approaches zero is
g+ b+ log(1 - zU)
,uv
-a
[(z
2 gjzv 7 + --
z
-}-
g+b ~
( 1)1
ZZ'+ --
z,T'
+O(a4). (5.3)
Substituting the first two terms in (5.3) into (5.1) gives the propagator on the unit
disk, with the boundary condition O,G,,= -¢x'g,~ at ]z] = 1. A straightforward
calculation shows that the a 2 terms are equivalent to the insertion at z = 0 on the
unit disc of the operator (see fig. 2)
2a2(g-b-F) ~
a' g + b + F ~vOzX O~X. (5.4)
This means that a scattering amplitude on the annulus of inner radius a may be
538 C.G. Callan et al. / String loop corrections to fl-functions
Fig. 2. A small hole in the annulus is mimicked by an operator insertion on the disk.
where Z disk is the vacuum amplitude on the disk without a gauge field. For the
annulus, with the choice of equal and opposite boundary charges the result is [13]
= [lda
Jo -a-T[1 + ( D - 2 ) a 2 + O ( a 4 ) ] (5.7)
is the zero-field partition function for the annulus, including the ghost determinant.
Note that (5.5) is claimed to be proportional to the spacetime effective action for the
background gauge field! The background field dependence of any partition function
is in fact very divergent and this simple finite result is obtained by a careful use of
zeta-function regularization. It is in accord with the general arguments of refs. [11]
and [8] that the Polyakov path integral for a background field sigma model can be
defined to be the spacetime effective action.
The expansion of the integrand of (5.7) in powers of a 2 should, according to the
arguments given above, correspond to an expansion in insertions on the disk of
operators of increasing dimension. With a little algebra we can show that (5.7) is
reproduced by
z~mnulus = [1 d a
Jo a 3 ( ( 1 - 2a2)¢det(1 + F ) )disk
[2
8S~°°P= f d2z ~-log ACdet(1 + F) ~ ~,OzX~'O~X~
+ ( A 2 - 21ogA)¢det(1 + F ) ]. (5.9)
540 C.G. Callan et al. / String loop corrections to fl-functions
The first term is a counterterm for the standard dimension-two operators in the
sigma-model action and will lead, via the Fischler-Susskind type of argument, to
loop contributions to the beta functions of the massless closed string fields. The
second term is a counterterm for dimension-zero operators, which are used in the
sigma-model action to describe the interaction with background tachyon fields [21].
This term should therefore generate loop corrections to the tachyon beta function!
Since the tachyon is not present in realistic string theories, we will not concern
ourselves further with this type of term.
Our final task is to include the effect of closed string backgrounds on these
counterterms. If we work to lowest order in derivatives of background fields, the
vacuum amplitudes (5.5) and (5.6) become
2-71°g Afd2z~det(l+g_l(b+F))e_~,/2(g-b-F
g+b+F ) 8zX~O~X~" (5"11t
Following the approach of ref. [9], we treat it as a new counterterm for the sigma
model couplings to the metric and antisymmetric tensor fields and read off from it
(by separating the symmetric and antisymmetric parts) the following open string
loop corrections to the beta functions fig and fib:
Comparing with (3.10), we see that these loop corrections to the closed string beta
functions are precisely those implied by the conjectured action (3.8) if we take the
undetermined coefficient ~ to have the value - 8 7 r / a ' . Note that since we have not
kept track of the world sheet curvature, or world sheet ghost field, dependence of
the loop counterterms, we cannot directly identify the loop correction to f * . We
C G. Callan et al. / String loop corrections to fl-functions 541
can, however, infer what it must be by a consistency argument of the type which led
to (4.7) and will necessarily get the same equation as found by varying (3.8).
Before turning to a discussion of these results, we would like to present a brief
description (details are in the appendix) of another way to derive the crucial
background field dependence of (5.9). The divergences of open string loops can be
interpreted as due to zero mass states of the closed string disappearing into the
vacuum. In the appendix, we give a general technique for calculating the transition
amplitude, T, for this process: One first builds a Fock-space representation of the
operator V(A*, F~), made of the closed string creation operators A*, that sews a
boundary, with boundary conditions appropriate to the presence of background
gauge field strength F,., onto a closed string world sheet. We find that
times a BRST ghost factor (A.26). For F = 0, this reduces to a result of Ademollo
et al. [22]. The transition amplitudes of interest to us are the expectation value of
this operator between the massless closed string states and the vacuum. We see that
for an annulus in an external electromagnetic field both the graviton and the
antisymmetric tensor field have vacuum transition amplitudes that are respectively
the symmetric and antisymmetric piece of
This reproduces the graviton and antisymmetric tensor part of (5.9). In the appen-
dix, we will show how, by including ghost operators in the above considerations, it
is possible to reproduce the dilaton piece as well. It is reassuring that the potentially
ill-defined background field dependence of the overall normalization of this quan-
tity can be derived in two completely independent ways. Since overall vacuum
stability of the closed string requires that loop amplitudes cancel against the tree
vacuum amplitudes, which are precisely the original sigma model beta functions [8],
the full stability condition will be precisely the loop-corrected beta functions derived
above.
27t
Aflg=-a7 ( g~ - ~ g ~ F 2 + 2F2 + . . . ) ,
4~
- .- (6.1)
7 G + ..
Appendix
unit mass. Let u,(q), n = 0,1, 2,... be its normalized position space wave functions
{ ~ /1/4
u.(q)=[~) 2-"/2(n?)-Z/2H.(~l/2q)e ~q2/2, (A.1)
where H,(x) are Hermite polynomials. We are interested in the functional integral
over a finite euclidean time 0 ~< • ~< t,
(In this expression, the zero point energy has been subtracted.) Inserting (A.1) and
doing the sum gives a gaussian in qi, qf, whose coefficient matrix is the inverse of
the Neumann function restricted to the boundaries [11, 31].
Eq. (A.3) exhibits the factorization of the two boundaries, but we would like to
express it in terms of the creation and annihilation operators A* and A, rather than
Hermite polynomials. Therefore we look for an operator V(At, S) satisfying
£ 1
dqu.tq)"' e -s[ql - ~ (OIA"V(A*,S)[O) (1.4)
for any given function S[q]. Using the Hermite polynomial generating function
£ ~nV.
n=0
("'~"exp{-Jo~qZ+(2wl'/2qz -"I
' u,(q)z"=--) ,z (A.~
turns the right-hand side of (1.4) into
(A.7)
C.G. Callan et al. / String loop corrections to fl-functions 545
Fig. 3. The annulus is mapped into a rectangle. The inner boundary becomes the top.
displaying the factorization of the two boundaries in a convenient form. Notice that
the position variable inside the boundary action S[q] is scaled by (2w) -1/2 in the
integral (A.7) for V(A*, S).
We now extend these formulae to the string by multiplying the separate oscillator
pieces. The conformal transformation (shown in fig. 3)
r + io = - l o g z
0~<o~<2~r, O~r~t= - l o g a.
= +
~[~,(r)e i"° + ~7~(r)e~'"], (A.9)
m=l
1 -". 0~ ~ 2 0~>/2 ]
Sint 47rolf "10 o
(A.IO)
m=l [ d r d'r
Writing
~ - ~-(~ + iX~m) (A.11)
Now we consider the boundary action. The two relevant boundaries are r = 0,
r = t. If we describe them clockwise, they will have opposite charges. Let us consider
the ~--- t boundary alone, with a positive charge so o runs from 0 to 2rr. Then by
(3.4) with ¢ ~ o, the boundary action for a constant field-strength is
-i r2~r v 0
S,=--/0
4~ra'
doF~ (o,r ) ~o ~'( °' ,r). (A.12)
i,, v = l m = l
Eq. (A.1) and therefore also (A.7) require that the coordinate Q be real. We
therefore first associate creation operators Bm
~* and C~* with the coordinates ~ and
X~m in (A.11), and then combine them by
- (ArIA-t) (1.15t
in vector notation. The complete operator (the product of (A.7) over all modes) in
this notation is
+(Atl~)+(~IXt)-(AtlAt)}. (A.16)
Since only creation operators are present, everything commutes. We do the gaussian
C G. Callan et al. / String loop corrections to fl-functions 547
= ~ ( 0 ) l o g x + 51 D~(
, 0)
=-½1ogx+.... (A.17)
There is one factor [det(1 + F)] -1 for each m, so we finally get
( D ~=1
v(a*'F~")=[det(l + F)]'/2exp~,~=l£ ~, A~,(1-F]
rn~ 1+Flay A~* / (A.18)
-mj"
This displays how the left and right moving modes A~,, are coupled together by
boundary reflections. For F~ = 0 it reduces to a formula of [22]. This fixes the
numerical constant dropped in (A.17), since they derived their formula from open
string unitarity.
To see the physical significance of this operator, consider an annulus e - ' ~< Iz [ ~<1,
with N open string tachyons of momenta Kj entering the outer boundary at z = e'°:
as in fig. 3. The inner boundary contains only a charge interacting with the external
field F~,. Then (A.18) is the contribution of the inner boundary. The corresponding
operator V(A, So) for the outer boundary is obtained by taking [32]
N
So = -i E K~(oj, O) (A.19)
j=l
e -'°t(L°+L°) , (A.20)
with
O oo
Lo+L0= ~ ~ [mlA~m*A~. (A.21)
/*=1 m = - o 0
The Teichmiiller parameter (inner radius) is e-t. Including the integration measure
from the ghosts [19, 31] gives the amplitude
8 Kj fo dte2tfl (1-e-2"t)2<OIV(A'S°)e-'°tfL°+E°)V(At'F)IO>"
n=l
(A.22)
After expanding the ghost partition function, we can do the t integral to get a series
548 C.G. Callan et aL / String loop corrections to fl-functions
Fig. 4. Factorization of fig. 3. Open string particles go via a closed string propagator into the vacuum.
Here I g'(A*)) is the oscillator wave function of a closed string state of mass M~,
[33].
Eq. (A.23) is shown graphically in fig. 4. The solid lines are open string tachyons,
the wavy line is a closed string. The blob is the amplitude for the closed string to
disappear into the vacuum. If the closed string state has zero mass this can really
occur, giving a divergence in (A.23). Taking [33]
eq. (A.18) gives the graviton-to-vacuum transition induced by the applied field-
strength F..,
C±((I, T) = ~ +e m(~'+io)
c ,?,
b ++(o', T) =
_ _
~ b
.Z
e
+m(~'+io)
_
, (A.27)
are the Fourier expansions of the BRST ghost and antighost fields of the closed
string. They satisfy the algebra
where (q'[ is its oscillator wave function. The relevant wave functions, including
ghost content, have been calculated by Siegel and Zwiebach [33]. Let us call h,, the
zero mass state whose wave function is pure matter and q the zero mass state whose
wave function is pure ghost:
('l'h."l = ½(OIA~AL~,
(ff'n I = (01 (b~c~ - c~b~). (A.30)
Now the sigma model beta functions are conventionally defined by their lagrangian
counterterms [4], without reference to the physical graviton and dilaton which are
mixtures of h~, and 7) [33]. The composite fields corresponding to the wave
functions (A.30) can be deduced from the reduction formulae [8] at z = e -(t÷jo) = 0.
References
[1] C. Lovelace, Phys. Lett. 135B (1984) 75
[2] P. Candelas, G.T. Horowitz, A. Strominger and E. Witten, Nucl. Phys. B258 (1985) 46
[3] E.S. Fradkin and A.A. Tseytlin, Phys. Lett. 158B (1985) 316; 160B (1985) 69
[4] C.G. CaUan, D. Friedan, E.J. Martinec and M.J. Perry, Nucl. Phys. B262 (1985) 593
[5] A. Sen, Phys. Rev. Lett. 55 (1985) 1846
[6] C.G. Callan, I. Klebanov and M.J. Perry, Nucl. Phys. B272 (1986) 111
[7] R. Nepomechie, Phys. Rev. 32D (1985) 3201
[8] C. Lovelace, Nucl. Phys. B273 (1986) 413
[9] W. Fischler and L. Susskind, Phys. Lett. 173B (1986) 262
[10] M. Grisaru, A. van de Ven and D. Zanon, Phys. Lett. 173B (1986) 423;
M. Freeman, C. Pope, M. Sohnius and K. Stelle, Phys Lett. 178B (1986) 199
[11] E.S. Fradkin and A.A. Tseytlin, Phys. Lett. 163B (1985) 123
[12] A.A. Tseytlin, Nucl. Phys. B276 (1986) 391
[13] A. Abouelsaoud, C.G. Callan, C.R. Nappi and S.A. Yost, Nucl. Phys. B280 [FS18] (1987) 599
[14] H. Dorn and H.-J. Otto, Z. Phys. C32 (1986) 599
[15] A.A. Tseytlin, Phys. Lett. 176B (1986) 92;
D.J. Gross, in Unified string theories, eds. M. Green and D.J. Gross (World Scientific, 1986), p. 391
[16] E. Braaten, T.L. Curtright and C.K. Zachos, Nucl. Phys. B260 (1985) 630
[17] E. Cremmer and J. Scherk, Nucl. Phys. B72 (1974) 117;
O. Foda, Utrecht preprint (1986)
[18] V. Alessandrini and D. Amati, Nuovo Cim. 4A (1971) 793
[19] J. Polchinski, Comm. Math. Phys. 104 (1986) 37
[20] B. Ratra, M. Peskin and L. Susskind, private communication
[21] C.G. Callan and Z. Gan, Nucl. Phys. B272 (1986) 647
[22] M. Ademollo, R. D'Auria, F. Gliozzi, E. Napolitano, S. Sciuto and P. di Vecchia, Nucl. Phys. B94
(1975) 221
[23] O. Alvarez, Nucl. Phys. B216 (1983) 125
[24] S. Das and B. Sathiapalan, Phys. Rev. Lett. 56 (1986) 2664
[25] M. Dine, N. Seiberg, X.G. Wen and E. Witten, Nucl. Phys. B278 (1986) 769
[26] C. Callan, Z. Gan and M. Perry, Princeton preprint
[27] G. Chapline and N. Manton, Phys. Lett. 120B (1983) 105
[28] P. Mansfield, Nucl. Phys. B283 (1987) 551;
E. Martinec, Nucl. Phys. B281 (1987) 157
[29] D. Friedan and S. Shenker, Nucl. Phys. B281 (1986) 509
[30] L. Clavelli and J. Shapiro, Nucl. Phys. B57 (1973) 490
[31] A. Cohen, G. Moore, P. Nelson and J. Polckinski, Nucl. Phys. B267 (1986) 143;
F. Jimenez, J. Ramirez, M. Ramon-Medrano and G. Sierra, Madrid preprint (1986)
[32] J.L. Gervais and B. Sakita, Phys. Rev. D4 (1971) 2291
[33] W. Siegel and B. Zwiebach, Nucl. Phys. B263 (1986) 105
[34] M. Kato and K. Ogawa, Nucl. Phys. B212 (1983) 443