December, 1985

1. TWO DIMENSIONAL CONFO RMAL FIELD THEO RY

Introduction to Two Dimensional Conform;a--r~=::::-:-...-:J In these two lectures I will describe some aspects of conformal and
super·
and Superconformal Field Theory conformal two dimensional field theories and the string a~d superst ring
theories
buill from them. The lectures are meant to be read in conjunction with
the notes
of Daniel Friedan1 11. A large number of other contributions to this
volume also
discuss related issues.
Stephen H. Shenker
En_rico Perrf'!i ~
In this first lecture I will discuss some of the rich structu re present
Jame1 PNnci Imtitute1 in two-
Gild D~~rlrnenl of PA.,ic• dimensional conformally invariant field theories. My approach follows
that of
Unioe,itll oj- Chicago, Chicago, Rlinoi• 60691 Belavin, Polyakov, and Zamolodchikov[BPZ[I 21.

The central object in the study of such theories is the stress energy
tensor
T.~. Euclidean invariance (we work in Euclidean
ABSTR ACT siguature throughout) implies
conservation, and scale invariance implies tracelessness.
Some of the basic properties of conformal and superconfonnal
field theories in two dimensions are discuased in connection with the a.T.~ =o (1.1)
string and superstring theories buill from the01. In the first lecture
the stress-energy tensor, the Viraaoro algebra, highest weight slates, These equations imply the Cauchy-Riemann equations for the combin
ations
primary fields, operato r product coefficients, bootstr ap ideas, and
unitary and degenerate representations of the Viraaoro algebra are T(z) = T 11 - T22 + iT12 T(z) = T11 - T22 - iT12 ( 1.2)
discussed. In the second lecture the basic structu re of superconformal
Correlation functions of T(T) are meromorphic (anti-meromorphic)
two dimensional field theory ia sketched and then the Ramond Neveu- functions.
Schwarz formulation of the superstring is described. Some of the Analyticity, bose symmetry, the spin-2 nature of TaD and the require
ment
issues involved in constructing the fermion vertex in this formali that Toft generate translations and scale transformations force the operato
sm r prod~
are discussed. ucl of T with itself to have the form

T(z)T( w) ~ ~ ~1__ +. 2T(w) + ~T ___ __!__ +.. (1.3)

2 (z- w)' (z- w)
2 iJw (z- w)
To appear in the Proceedings of the Workshop on Unified String Theorie
s, The number cis a parame ter of the given theory related to the trace
Institu te for Theoretical Physics, Santa Barbara, July 29- August 16, anomalyi31.
1985. T obeys an analogous equation with the same value of c.
The most useful operato r interpretation for a scale invariant theory is
radial
quantizationl•l. Dilation plays the role of time translation, radial orderin
This work was supporte d in pari by DOE gr~& QE~FG02-84ER-45U.4,
g that
NSF gran& PHY~
8451285 and ibe Alfred. P. Sloao foUoda&ioO.
2
of time ordering. In such a quantization the •time zero surface• can be chosen space will decompose into irreducible representations of Ibis algebra which will
as the unit circle and operatoi'B defined by integrating fields on this surface. In be very large. In fact there are certain theories whose entire Fock space consists
particular we can define operatoi'B that generate conformal transformations. In of a finite number of such representations. The properties of all states in an irrep
two dimensions there are many such transformations since any analytic mapping are related lo each other purely by lhe Virasoro algebra.
is conformal. A basis for the infinitesimal conformal transformations is What do these representations look like? Choose lhe hermitian operators
Lo and Lo diagonal;
fn complex. (1.4)

Lolh > = hlh > , Lolli > = Iiiii > h,li real. ( 1.8)
The operator generating such a transformation is <.L. + l.L. where

L. =
2 ~; f dz z"+'T(z) , L. = ~
2n
j dz zn+~f'(z). (1.5)
(We will usually suppress the L algebra since ils properties are analogous to
the L algebra. The slate space is buill up of tensor products of irreps of the
This quantization gives the following hermiticity properties: two commuting algebras.) Equation (I. 7) shows I hal for n > 0 L_. raises £ 0
by n units, L+n lowers L0 by n units. We can then construct •highest weight"
(1.6) representations in the standard fashion: choose an eigenstate of Lo, lh > thai is
annihilated by all the L+•· Then the sel of stales formed by applying products of
Note that n = 0, < real in ( 1.4) is a dilation zo Lo +Lo is the radial "hamiltonian".
the {L-n} is a representation space for the algebra. These states are conveniently
The commutation relations of these operatoi'B can be determined131 from the
organized into levels by £ 0 eigenvalue. L- 1 lh > is al level I; L: 1 lh >, L_ 2 lh >
operator product expansion (1.3)
are at level 2; and so on. This set of states is called a Verma module; h is the
weight of the irrep, lh > is a "highest weight" slate. In dual theory, it is called
(1.7)
a physical slate (if h = I).
The L's obey the same algebra as the L's, and the L's and L's commute. The Such states exial in a quantum field theory. The hamiltonian, £ 0 + Lo is
algebra (1.7) is called the Virasoro algebral•l. The fii'BI term on lhe RHS is lowered by L+n· The lowest •energy• state, the vacuum 10 >, must be anni~
just what we would expect from the composition of transformations (1.4). The hilaled by all the L+• - il is a highest weight slate. Scale invariance implies
second •central iierm" with coefficient c is the consequence of the projective thai £ 0 annihilates 10 > as well; £ 0 10 >= 0. This implies translation invariance,
representation of the transformations in the Hilbert space of lhe quantum field £_,10 >= 0, since
theory.
__..---._
We see that scale invariance and a local conserved slress·energy tensor j!flPiy ( 1.9)

the existence of an infinite dimensional algebra lhal acts on I he slate !~Of• pf

the theory. This infinite dimensional algebra is lhe source of much of die magic We should stop for a minute and contrast the situation in two dimensions
' with that in higher dimensions. There the conformal algebra is finite dimensional,
in two dimensional conformally invariant field theories. In particular -~he state

3 4
the analogous objecl in two dimensions is the closed sub algebra Lo, L±., L0 , 1.±1 is created by a field 4>t>l (z) at z = 0 where
called SL 2 (C) that we have shown annihilates the vacuum. These generate the
infinitesimal fractionallinearlrana formaliona . The irrepa of S L.( C) are much (Ll4)
smaller than those of Viraaoro. There are in general an infinite number of S L2
and
irreps in eacli Yiraaoroimp; ihe< Viraaoro algebra ties together their behavior.

There ia more in a quantum field theory than states; there are local field
L.(z) =
2 ~i / dw (w- z)H 1 T(w). ( 1.15)

operators aa well. In a<< ~-"d•conformally invariant quantum field theorythere What are these fields? It is clear that
are special fields called primary fielda121 characterised by their operator products iN
L-t(z)4>(z) = {)z • (1.16)
with T(z)
114>(w) 1 84>
T(z)4>(w, w) ~ (z-w )• + -( - ) 8- tu
z-w
(1.10) Other descendant fields are composites of the stresa"energy tensor with 4>. The
The absence of higher order poles distinguishes primary from other acaling fields. properties of all these fields and not just the derivatives (1.16) are organized by
The coefficienla are fixed by lhe requirement thai T generate dilations and trano- the Virasoro algebra.
lations. We need to know more about fields than their action on the vacuum. We
Equation (1.10) gives the following commutation relatione need ~o know their matrix elements between arbitrary states. This information
will be contained in three point correlation functions. Let {0.} be arbitrary
(1.11) fields, primary or descendant.
Observing that L. w111 obey a similar formula with h replaced by 1i and special-
izing lo n = 0 serves to identify h + ii aa the scaling dimension of 4> and h - ii ( 1.17)

aa its euclidean spin (L0 - L. generales rotations). The commutation relations contains the information we seek. But we can evaluate (1.17) using the operator
(1.11) are alao the conditions lhal 4> must obey to be a vertex operator in dual
product expansion
theory (along with a constraint on h).

From (1.11) and the highest weight properties of the vacuum we find, for Op(z)O,(O) ~ L; Cp, 6 0 6 z-•.-•,+•• (1.18)
n>O, '
L. 4>(0)10 >= 0; Lo 4>(0)10 >= h4>(0)IO >
where the Cp,, are the operator producl coefficients and the z dependence is
(1.12)
supresaed. So the three point function (1.17) is
So 4>(0)10 >= Ill > ia alao a highest weight alate with weight h. Primary fields
are those thai create higlieal weight elates from the vacuum. L; Cp,,z.-•.-•,+>• (O,(z,)O,(O)) . (1.19)
What kind of fields create descendants? It is easy to see from (1.5) thai '
The matrix element information is related to the operator product coefficients
L..... L•,4>(0)IO > (L!3) and completely determined two point functions. II is useful to notice that the

5 6
two point function of descendants of different primaries vanishes (in fact SL(2) and similarly for z 1 ---. z4 and 22 --+ Z3 . The equality of these expressions is
primary is enough to ensure this.) a necessary consistency requirement It is just the constraint of duality in the

The key simplification is that the operator product coefficients of deacon· corresponding string four point amplitude. The sums over descendants can be

dania are determined by those of the primaries. Imagine a. three point function done, in principle mechanically, turning this requirement into algebraic equations

of two primaries ;., ;, and one descendant of primary ;, on primary field operator product coefficients. The in general difficult technical
problem of evaluating the descendant contribution is just the calculation of the
(1.20) conformal blocks of BPZI21. These equations provide strong constraints on con·
formal quantum field theories. Polyakovi<l originally proposed using them in a
By using the commutation relations (1.11) and the highest weight properties of
•conformal bootstrap• to solve for confonnally invariant systems in d dimen·
the vacuum we can relate (1.20) to derivatives of the three point function of the
sions. The dramatic simplification in two dimensions121 is that there are many
primaries. So operator product coefficients of descendants can be determined
fewer fields to consider - just the primaries. The difficulties are first the techni·
from those of the primaries mechanica.Uy.
cal one of evaluating the conformal blocks and second the possibility of a large
The complete information necessary lo specify a. conformally invariant two number of coupled primaries.
dimensional quantum field theory ia the Vira.aoro highest weight representation
BPZ noticed thai there are certain special c, h values where things simplify
content {h., li.} and the operator product coefficients c.., between the primary
enormously. Gervais and Neveul 71 also found these values. One way to see Ibis
fields creating them. In string theory these operator product coefficients are just
structure is to study the constraint imposed by requiring lhal lhe slate space be
the three point couplings between particles created by the "vertex operators• or
a Hilbert space i.e., having a positive inner productl 81. The basic constraint is
primary fields.
that each vector in each Verma module mus\ have non·negative norm squared.
There are constraints on this information. For instance, consider the four
Equation (1.9) already tells us that L_ 1 lh > is nol an acceptable state unless
point function h ~ 0. At higher levels in a Verma module we need to check the matrix of inner
(1.21) products of the basis vectors, the contravariant form introduced in reference 9,
Take z 1 --. z 2 and z~ -+ z4 and use the operator product expansion to evaluate. and verify that there is no negative eigenvalue. An important tool is the Kacl 101
We find (schematically) formula for the determinant of Ibis matrix of inner products.

As an example consider the second level. We have a basis of two states,

(;.~.~,;.) = E c... c,,~(o.o~) ( 1.22)
·~
L:,lh >and L_,lh >and hence a two by two matrix of inner products. There is
Where a, fJ range over aU fields, primaries and descendants. Now evaluate an- no negative eigenvalue for c ~ 1,h ~ 0 (a result true al a.U levels), but for c <I

other way, taking z 1 --+ z 3 and z2 --+ z4 . We find there is a region of h where one occurs. The boundary curve of this region is
given by
(~.;,;,~.) = E c.,.c,.~co.o~)
., (1.23) h = 116 (5- c± J(c- 1)(c- 25)) (1.24)

7 8
 and on it there is a sero eigenvalue, conesponding to an eigenvector
(p(m+ I) -qm)' -I

Jnull >= (L~, + aL_,)Jh >

3
a=--
4h+ 2
(1.25)
" 4m(m +I)
(1.30)
p=l. .. m-1 q = l. .. p
lhal satisfies This discrete series contains a number of interesting two dimensional statistical
< vJnull >= 0 (1.26) mechanical critical points. The unitarily result in combination with lhe differen-
for all vectors Jv > •. II ia nol hard lo see lhallhe firsllime a null vector appears tial equation technique and effective methods lo solve the equalions1 131 reduces
. il eorreapoitda,lo a highest weight alate within lhe Verma module. lhe study of conformally invariant lwo dimensional quantum field theories with
c < I lo a finite algorithm. Goddard , Kent and Olivel 141 have shown lhal all
BPZ pointed oul a remarkable properly of such null vectors. Let
elements of the list (1.29,1.30) are in fact unitary by construct ing manifestly

Jh >= ;(o)Jo > unitary realizations for all of them.

(1.27)
The finol nontrivial member of lhe series is al c = 1/2. The allowed h values
be a highest weight alate and have a null vector al second level. Then lhe n point
are 0,1/16,1 /2. A Maiorana fermion has c = 1/2 and provides a realization
fund ion
of lhe 0,1/2 representations. The full representation content of lhal theory is
< oJ;(z1 ) ••• ;(z.)(L~, + aL_,);(o)JO > (1.28)
("' = ,, + i¢,)
vanishes because the vector is null. Bul (1.28) can also be evaluated by commut- field (h,li)
ing lhe Virasoro operators lo lhe right using (1.11). This gives a linear differential ¢ (!. 0)
equation lhallhe n point function must satisfy! This properly of degenerate rep-
~ (O,j) (1.31)
resentations (i.e., those containing null vectors) of infinite dimensional algebras
\'lib (j,j)
is very importan t. For example, Knizhnik and Zamolodchikovl 111 have used Ibis I (0,0)
idea in the Kac-Moody context to calculate correlation functions of sigma models
The nonvanishing operator product coefficients are: (• = ¢¢)
with w..... zumino term at their fixed pointl121.
The fllllunita rity analysis181 shows that all unitary representations for c < I ¢¢ I
are degenerate, giving a physical motivation for studying the degenerate case. In ~~ I
fact the unitary representations occur only at a diserete set of points given by:
.
¢~ -• I
( 1.32)

6
• = 1 -. -m~(m--'-+-:lcc) m= 2,3,_4, ... (1.29)
The magnitudes of the coefficients are all one. This then is the complete speci-
fication of Ibis very simple theory. Next lecture we will find a place for the thus
•ror· vec:tofS within the Verma li:aodule this ia immediate. For vedors outside it folloWs
if the far absent h = 1/16 representation.
state spac:e ia composed entirely of highe.t weigh& irreps. Uoitarity is auftlc:ieo&
for this. I
thank C. Thoro for raiainr this point aod D. Friedao for &he uoitarity observatioo.

9
10
 2. SUPERCONFORMAL FIELD THEORY AND SUPERSTRINGS We now discu88 boundary conditions. First we introduce a new set of coordinates
via the conformal mapping w = < + iu = log z. The z-plane maps onto a
cylinder with ~ime• T and periodic •space• u. Dilation in z is translation in
In lhia lecture I will describe IIOIIle aspect of auperconformal field theories
in two dimensions and I he fermionic airing buill from them. Again, lhia !eel ure T 10 radial quanti1ation on the plane is just conventional quantization on the
should be read in conjunction with Daniel Friedan's notesi•J. Much oflhe material cylinder. There are two natural boundary conditioll8 for the ferm.ionic fields on

diacuased here can be found in references 15--19. the cylinder, period or anti-periodic. We refer lo the periodic case as Ramond
boundary conditions, anti-periodic as Neveu-Schwars. Translated back lo lhe
We work in two dimenaional auperspacef201 and use complex coordinates
z-plane (remembering f(dz)'(dz)' ia conformally invariant) antiperiodic fermi
z,l,:t,O. We will often leave the (:t,O) dependence implicit. The ouperconformal
(half-integer h) fields are single valued; periodic fermi fields are double valued,
properliea of a general theory are determined by properties of the auper alresa-
.p --+ -.P along a cui beginning al the origin. From (2.2) we see lhal the mode
energy tensor
numbers n for fermionic fields are integers in Ramond and half integers in Neveu·
T(z,l) = Tr(z) + ITs(•) (2.1)
Schwarz. The algebra (2.4) with n integral ia called the Ramond algebra121 1,
where Ts is the ordinary boaonic olreao tensor and Tr ia ilo fermionic super- with n half integral the Neveu-Schwarz algebra1221. The fullsuperconformal field
partner. Tr is a conformal primary field with h = 3/2, 1i = 0. The momenta of theory contains both Ramond and Neveu-Schwars sectors.
T(z, 8) form the auperaymmelric extensions of the Viraooro algebra with opera·
Notions of highest weight stales and descendants apply in this situation is
tors Gn,Ln well. The vacuum is a highest weight slate for the reasons discussed in the last
Tr(z) = ! L: .,-•-•t•a. lecture and is in lhe NS sector, as we shall see. Other highest weights in the
2 •
(2.2)
Ts(z) = L z-•-• L• . NS sector are created by superconfonnal primary fields (often called conformal
n
superfields) acting on the vacuum. A highest weight state in the R sector must
The commutation relations follow from the operator product
. be created by an operator that changes the boundary conditions of the fermionic
'C/4 I D,T 9, fields as it ads on the NS vacuum. Such an operator can be visualized as the
T(z,I,)T(z2 ,82 ) - _.- + -3 -T(z.,l
9 1,
2) + - - + - 82 T (2.3)
~~, 2 Zu 2 zu Zn endpoint of a cui in the fermionic fields and is called a spin fieldi"l. Super6elds
lake NS lo NS and R lo R; apin fields take NS lo R and R lo NS.
where Cis the trace anomaly, normalised so that C = lc, z,, =z, - z,- 8,8, and
912 = 91 - 92 • They are We now begin a discuBBion of the Ramond Neveu-Schwarz fennionic string121 ·221.
II ia constructed out of D free super6eldsi201X• coupled lo 2-d supergravityi23 ·"1.
(L.,L.J = (m- n)Lm+n + ~(m3 - m)6•,-• We gauge 6x lo superconformal gauge. In the critical dimension D = 10 this
m leaves the free superconformal field theory of the X• and I he Faddeev-Popov
(L., G.J = (2" - n)G•+• (2.4)
ghostsl20 •26 •201. The component expansion of X• is given by
{G., G.}= 2L•+• + i(m 2
- ~)6•.-•·
X•(z,B,z,B) = X• + 91/J" + 6ij,• + 9BF" (2.5)

II 12
 where X• is a free massless scalar, r/1" and ¢• are lefl and right handed Majorana- This is a superconformal primary if p·IC = 0 and baa h = 1/2 if p2 = 0. This vertex
Weyl fermions and F is auxiliary. Free field equations of motion lei us work wilh operator creates a massless parlicle. Loren Ia properties show lhal il corresponds
chiral auperfields to a massless vector. Note that the 8 integral of the tachyon vertex is odd under
X•(z,8) = X"(z) + 8~(z) (2.6) ( -1)' where F is sheet fermion number, while the integral of the massless vector
wilh lwo point function vertex is ( -1 jF even. We can eliminate the tachyon in Ibis theory by projecting
on (-I)' even operators. This projection is called lhe GSO projection1281.

(2.7) We now lurn lo the Ramond aeclor. Here il is more convenient lo look al
stales al firal, rather than lhe operators thai create them. Fermionic fields have
In Ibis example
1 . integer moding so lhe expansion of r/J"(z) is
T{z,8) = - DXD2X (2.8)
2
where D is lhe super-covariant derivative
r/J"(z) = L: r/J:z-•-l (2.13)
n<Z

The operators r/1: obey commutation relations

8 8
D= 88 + 8 8z · (2.9)
{ r/1::., r/J:) = -6,.+ng"•
(2.14)
Vertex operators V lhal create physical stales in lhe Neveu-Schwan aeclor (Lo, r/1::.) = -mr/1::.
are superconformal primary fields. The amplitude is formed by integrating vertex First focus on the zero modes 1/JC that commute wUh the •hamiltonian" L •
0
operator correlation functions wilh lhe invariant measure dzd8' . In order lhal The first pari of equation (2.14) restricted lo the zero modes forma a Clifford
algebra. Because the zero modes commute with L0 ib eigenspaces must form
j dzd8 V{z,8) (2.10) representations of lhe Clifford algebra. In lhe lowest energy sector, in particula r,
tP:are just realized as gamma matrices, and the lowest energy states form a
be superconformal invariant V musl have conformal weight h = 1/2. A first apace-time apinor. n is in this way that spac~time
spinor indices enter the
example of a vertex operator is
R - N S siring. (-I)' anlicommulea with r/J: so ( -1 )' acls like 'J"+I on ground
ipX-
t • (2.11) slates. The masses of these stales are determined by their L eigenvalue. On the
0

This exponential has conformal weight h = p2 /2 so p2 = 1. This means m 2 = -I ; z-plane in the R sector translation invariance is clearly absent - there is a cut
this vertex operator creates a tachyon. A aecond example is emanating from the origin. As we saw last lecture this means that L cannot
0
be aero on lhe vacuum. L0 can be computed from the r/J propagat or on the cui
(2.12) plane using, e.g.,
I
< O(Lo(O >= < OI21L+h L_,IO >~
•rn Che closed Type U s'rin' there will be u in'e!l'al ove.r J, Das weli. In the heterotic stnngl271
there is no i aod only u iote!l'alion on :. (2.15)
/ dz / dw z' < T(z)T{w) >

13 14
 (See reference 1). The resull ia
This islhe supersymm elry relation Q 2 = H where Q = G 0 and H = L 0 -cf16. A
D supersymm elric vacuum G 0 j0 >= 0 has
< OILoiO >= 16. (2.16) Lo eigenvalue c/16. So, assuming sheet
aupersymm elry is no\ broken (\rue because we are jus\ dealing wilh free field
This h value of 1/16 (per Ramond fermion) gives a realization of I he 1/16 unilary
theories)
represenla lion al c = 1/2 we zaw lasl lime.
(2.21)
We need Ia consider lhe ghoal syalem12'•2b.2t( as well. The auperconfo rmal
From (2.18) CTOT = 0 if D = 10 80 we have massless parlicles lhere. (-1)F
gauge fixing gives rise Ia ghoal auperfielda
acls like 1 11 on ground stales 80 the net space-lime chirality ia just Willen's
B(z, S) = p(z) + S6(.o) (-I)' indexl291 for Ramond sheet aupersymmelry. The nonvanishing of \his
(2.17) index provides a criterion in more general superconfo rmal theories correspond ing
C(.o,S) = c(.o) + S'J(z)
lo string compaclifi calions for I he presence of massless fermions.
where 6, c are the anlicommu ling ghoals of lhe ordinary siring wilh h values of
2 and -I. p and 1 are their superparln ers wilh h values of 3/2 and -1/2. The We saw in the Neveu-Sch wan sector thai the GSO projeclion was useful lo
lrace anomaly values of these syslems are as follows: eliminate the tachyon. In the Ramond sector on the massless slates il is just a
chirality projection 1 11 = +1. The resulting ma88less states are a massless vector
cx=D
and a massless Majorana- Weylspino r- aN= I d = 10 super·mull ipletl28 1. This
c0 = D/2 projecled theory is the superstring .
(2.18)
Ct,t = -26
We note that the GSO projection is necessary to preserve modular invariance
.,,, = +11 on the lorus if we want both Ramond and Neveu-Schwarz seclors in the theory.
When D = 10 , <TOT = 0 and the lheory ia consislenl. The ground slale energies Summing over different u boundary conditions combined with u +-+ .,. interchang e
in the Ramond sector are as follows: (/J, 1 are sheel spinors so their boundary (a modular transforma tion) implies summing over 1' boundary conditions. This
condilions shift with sector) sum just implement s the GSO projeclion.
x• p2/2
.,. D/16
(2.19)
To compute scattering amplitudes of fennions we need to know the vertex
operators thai create the space-lime spinor stales. The matter pari of Ibis op-
b,c -I erator was already conslrucle d in the earlier epoch of siring lheory1 301. This
p,.., 6/16 vertex operator must act on the vacuum which is in the Neveu-Schwarz sectort
So when D = 10 lhe physical stale condition ( Lo)TOT = 0 implies P 2 = 0, i.e., and creale a slate in the Ramond sector. This tells us thai the fermion vertex
massless space-lime spinors. Thia resull has a simple interpretat ion in terms of operator must change the boundary conditions of the "". It opens a cut changing
sheet supersymm elry. The algebra (2.4) conlains lhe relalion e.g., cylinder antiperiodi c to periodic boundary conditions. This is parallel to the

~=Lo--
c
(2.20)
16 fHt>re tbis Y61:':Uum is a null state that does not correspond to a physical part ide.

15
16
 conslruclion oflhe Ising disorder variable out offermions1311. The h = 1/16 value
Because we have the full z, t dependence here this gives h = 1/8 = 2/16, as
for one fermion is related to the Ising (dis)order exponent. This is the simplest
advertised.
example of a "spin" fieldi 1 ~1.
For D = 10, roughly speaking, we can do the same thing1 171. Group the
We now review an intuitively appealing way to compute correlations of such
10 fennione into 5 groups of 2 and then bosonize. We can then make a vertex
operators . Conaiderl321 the caae D = 2. A cui acroaa which t/> 1 __, -'/1 1 and
operator representalionl••l of the spin fields asl
Y, 2 __, -t/>2 can be represented as an 80(2) gauge field concentrated along the
cut with field "Strength only at the ends of the cui. Changing the position of the 8. = i•·• (2.24)
cut is just a gauge tranaformaliont. The field strength is adjusted to give a "
where~=(;, ... ;,), ii = (±~ ... ± ~). The 32 poaaibilities fill out aD= 10
phaae for parallel transport around the endpoint. The spin field is like a point
Majorana spinor. The conformal weight h of 8. is liil 2 /2 = 10/16. Correlation
magnetic vortex on the world sheet. The GSO projection eliminates operators
funclions can also be computed using the 80(9,1) Kac-Moody algebra generated
thai see the •string" of Ibis vortex. The projected theory is local.
by currents y,•y,• and the nuU vector differential equation in the Viraaoro-Kac-
We now can boaonise1331 the two Majorana equal one Dirac fermion and Moody semi-direct productl171.
rewrite the gauge field coupling lo the fermion current io
This operator has h = 10/16, not h = I necessary for the fermion vertex
operator to have the correct conformal properties: that the integral f dzVr should
be conformally invariant. The resolution of this problem involves the ghosts, as
ia __, <a~a~;
suggested by Goddard and Olive. This is reasonable since the spinor ghosts
I d'{Aaia-.. I d'{;<a~{J~A. (2.22)
change boundary condition at the cut as weD as the .p•. There should be a spin
=I d'{;B field for the ghosts included in the vertex as well. The Ramond ghost ground
stale of energy 6/16 should be created by a conformal field with that weight.
where B is the field strength, B = ~6({- z)- ~6({- w). We see that we only
If we denote that spin field by E we can construct a candidate h = I massless
need to compute the ratio of the partition function of a free scalar field with
fermion vertex as
po.inl sources to that of one without, i.e., a correlation function of exponentials
of free 8Calars. These witt just be power laws. In fact the answer for the two p · 'JU = 0, p
2
=0. (2.25)
point fullclion is
This turns out to be about half right. The full working out of these ideas is
1
(2.23) described in references 1,18,19.

The structure of superconformal field theories described above is quite gen-

*The gause 8eld cu aetuaUy be coupled to the anomalous left or risht banded fermion curreob eral. It tells us how to construct spin fields and hence fermion vertex operators
alone. Movinc the nt "loially involves ca.uce lransforma tioas in regions of zero 8eld strength
-:here the a.oomaly vaobh_es.

I We must really include a cocycle here.

17
18
in general superconformal theories, for instance those deacribing compadified REFERENCES
superstrings. In fad the first indication of the general structure came from a
different direction: a study of the unitary repreaentatioDB of the Ramond and
1. D. Friedan, lecture notes in this volume.
Neveu-Schwara algebras. Wel151 found that there were diacrete aeries of unitary
repreaentatioDB for each algebra with the eame values of c but different valuee of 2. A.A. Belavin, A.M. Polyakov, and A. B. Zamolodchikov, Nuc. Phys. 1!211
h. This strongly suggested the existence of theories containing both typea of rep- (1984) 333.
reaentations in the way described above. This has been verified expliciUyi15,MJ
3. D. Friedan, in 1982 Lea Houches summer school Recent Advance• in Field
for the first two element• of the discrete aeries. The first one correspond• to
Theorp and Stati•tical Mechanin, J. B. Zuber and R. Stora, eds., North-
the Ising model with vacancies-a eystem that has already been realieed in the
Holland( 1984).
laboratory!

4. cf. S. Fubini, A. J. Hanson, and R. Jackiw, Phys. Rev. ll1 (1973) 1732.

Adtnowledgemeuta
5. M. Virasoro, Phys. Rev. Ill (1970) 2933; I. M. Gelfand and D. B. Fuchs,
I would like to thank J. Cohn, E. Martinec, and Z. Qiu for enjoyable col- Funds. Anal. Prilozhen 2 (1968) 92; in the context of string theory,
laboration and valuable diacuzsions. I would also like to thank P. Di Vecchia, the Schwinger term in the conformal algebra wae discovered by J. We is,
L. Dixon, J. Harvey, C. Thorn, C. Vafa and A. Zamolodchikov for enlightening unpublished (cf. S. Fubini and G. Veneziano, Ann. Phys. IN (1971) 12).
remarks. I am especially grateful to D. Friedan for counUees conversations about
6. A. M. Polyakov, Zh. Eksp. Thor. Fi•. 57 (1969) 271; !Sov. Phys. JETP
all aspects of this subject.
a!! (1970) 1511.

7. J-L. Gervais and A. Neveu, Nuc. Phys. lll22 (1982) 50; B2!!2 (1982) 125;
B224 (1983) 320; B238 (1984) 125; B238 (1984) 301;

8. D. Friedan, z. Qiu, and S.H. Shenker, Phys. Rev. LeU. ll (1984) 1575;
in Vertez Operator~ in Mathematic• and Phy•in, J. Lepowsky et.al. (eds),
Springer-Verlag (1984).

9. R. Brower, Phys. Rev. ~ (1972) 1655; P. Goddard and C. Thorn, Phys.

Leu. 40B (1972) 235.

10. V. G. Kac in Proc. lot. Cong. Math., Helsinki, 1978 and in Group The-
or-.tical Methodo in Phy&ic•, edited by W. Beiglbock, A. Bohm, Lecture

19 20
 Noles in Physics Vol. 94 (Springer-Verlag, New York, 1979), p. 441; B.
25. E. Martinec, Pbys. Rev. DZ8 (1983) 2604.
L. Feigin and D. B. Fuch., llJD.tho Anal. Prila.ben. 16 (1982) 47 (JUncl.
Anal. Appl. 1§ (1982) 114). 26. M. Kalo and K. Ogawa, Nuc. Pbys. Dll2 (1983) 443; S. Hwang, Pbys.
Rev. DZ8 (1983) 2614;K. Fujikawa, Pbya. Rev. lln (1982) 2584.
11. V. Kniahnik and A. B. Zamolodchikov, Nuc. Pbya.f!l li (1984) 83.
27. D. Groaa, J. Harvey, E. Marlinec, and R. Rohm, Pbys. Rev. Lett. M
12. E. Willen, Comm. Malb. Pbya. 02 (1984) 451. (1985) 502, Nuc. Pbys. ~ (1985) 253, and Princeton preprinl (July
1985).
13. B. L. Feigin and D. B. Fuchs, Moscow preprinl (1983); V. L. Dolaenko and
V. Faleev, Nuc. Pbys. 11240 (1984) 312. 28. F. Gliozzi, D. Olive, and J. Schork, Nuc. Pbya. B122 (1977) 253.

14. P. Goddard, A. Kenl and D. Olive, DAMTP preprinl (1985). 29. E. Willen, Nuc. Pbya. ll2l!l (1982) 253.

15. D. Friedan, Z. Qiu and S.H. Shenker, Pbya. Lell. lllB (1985) 37. 30. C. Thorn, Pbys. Rev.IM (1971) 1112; J. Schwarz, Pbya. Lell. 37B (1971)
315; E. Corrigan and D. Olive, N. Cim. llA (1972) 749; E. Corrigan and
16. M. Bershadsky, V. Knizhnik, and M. Teilelman, Pbys. LeU. llli! (1985) P. Goddard, N. Cim.)!A (1973) 339; L. Brink, D. Olive, C. Rebbi, and J.
31. Schork, Pbya. Lell. iii! (1973) 379; D. Olive and J. Scberk, NucJ. Pbys.
' 17. J. Cohn, D. Friedan, Z. Qiu and S.H. Shenker, in preparation.
11M (1973) 334; J. Schwan and C. C. Wu, Pbys. Lell. flB (1973) 453;
~

~
E. Corrigan, P. Goddard, D. Olive, and R. Smith, Nucl. Pbys: lm7 (1973)
~
18. D. Friedan, E. Martinec, and S.H. Shenker, Pbys. LeU. 160B (1985) 55. 477.

19. D. Friedan, E. Martinec, and S.H. Shenker EFI preprinl 85-89, November 31. L. P. Kadanolf and H. Ceva, Pbys. Rev. .83 (1971) 3918.
1985. 32. A. Luther and I. Peechel, Pbys. Rev.BU (1975) 3908; J. B. Zuber and C.
20. L. Brink and J. 0. Winnberg, Nuc. Pbys. B103 (1976) 445. llzykson, Phya. Rev. Uti (1977) 2875.

33. ct. S. Coleman, Phys. Rev. llll (1975) 2088.

21. P. Ramond, Pbys. Rev. D3 (1971) 2415.
34. M. Halpern, Phys. Rev. .lll2 (1975) 184; T. Banks, D. Horn and H. Neu-
22. A. Neveu and J. Schwarz, Nuc. Pbys. B3l (1971) 86.
berger, Nucl. Phys. JllM (1976); I. Frenkel and V. Kac, lnvenl. Math. 62
23. L. Brink, P. DiVecchia, and P. Howe, Phys. Lell. MD (1976) 471; S. Deeer (1980) 23; G. Segal, Comm. Math. Phys. ~ (1981) 301; P. Goddard and
and B. Zumino, Pbys. Leu. MD (1976) 369. D. Olive, in Vertez Operator> in Mathematics and Phy•ic•, J. Lepowsky
el.al. (edo), Springer-Verlag (1984).
24. A.M. Polyakov, Pbys. Lell.l23 B (1981) 207, 211.
35. D. Friedan and S.H. Shenker, in preparation.

21
22