A Review of Topological Strings

Imran Parvez Khan

Department of Physics,
CIIT, Islamabad.
Email:imraan.pervez@gmail.com.
Submitted on September 30, 2011
Abstract
We have reviewed a series of papers ([6] - [12]) by Edward Witten
1
,
which laid the foundations of topological strings and topological eld
theory. We have concisely presented the hallmarks of this theory, es-
pecially focussing on the topological aspects. We extracted invariants,
and discussed the cohomological eld theory. Then we studied the re-
lationship of Calabi-Yau spaces with the cohomological eld theories.
After studying supersymmetry and R-symmetry, we partially proved
the topological eld theory to be a cohomolgical eld theory. We twist
the N = 2 supersymmetric sigma model to obtain the A/B-models of
the topological eld theory. Finally coupling of these models to gravity
is sketched.
KEYWORDS: supersymmetry, spinor, holomorphic, correlator, chern class.
1
School of Natural Sciences, Institute for Advanced Study, Princeton NJ 08540.
i
Contents
1 Emergence 1
1.1 Chern-Simons Theory . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Cohomological Field Theories . . . . . . . . . . . . . . . . . . 2
1.2.1 Semi-classical limit . . . . . . . . . . . . . . . . . . . . 3
1.3 Calabi-Yau Manifolds . . . . . . . . . . . . . . . . . . . . . . . 3
1.3.1 K ahler Metric . . . . . . . . . . . . . . . . . . . . . . . 3
1.3.2 Ricci-at K ahler manifold . . . . . . . . . . . . . . . . 4
1.3.3 Moduli Spaces . . . . . . . . . . . . . . . . . . . . . . . 4
2 Topological Strings 6
2.1 Supersymmetry . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.1 N=2 supersymmetric sigma model . . . . . . . . . . . 6
2.1.2 N=2 SUSY and the K ahler Metric . . . . . . . . . . . 8
2.2 R-symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Twisted N=2 theories . . . . . . . . . . . . . . . . . . . . . . 10
2.4 The A-model . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.5 The B-model . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3 Coupling to Gravity 13
4 Conclusion 13
1 EMERGENCE 1
1 Emergence
The stregnth of topology lies in its quality of being metric-independent. Be-
cause this helps to nd invariants of any theory.
1.1 Chern-Simons Theory
The Chern-Simons action, with a gauge group G on a three-manifold M is
given by
S =
k
4
_
M
Tr(A dA +
2
3
A A A),
where k is the coupling constant, and A is a G-gauge connection on the trivial
bundle over M. Notice that the Chern-Simons action does not involve the
metric, which implies that the Chern-Simons theory is topological, and we
can look for invariants. One invariant is found to be the partition function,[2]
Z(M) =
_
[DA]e
iS
. (1)
One more invariant can be constructed from the Chern-Simons theory as
follows[1]. Given a Lie algebra element A, we have a Lie group element g
such that
g = e
A
Now consider a path (t) inside M. Given the instantaneous tangent vectors
on ,
d
dt
t
For the entire closed curve ,
g = lim
t0
_
exp
_
A(
d
dt
(0)t
_
exp
_
A(
d
dt
(t)t
_
exp
_
A(
d
dt
(2t)t
_

_
From the Baker-Campbell-Housdor theorem, e
U
e
V
= e
U+V
if and only if [U, V ] =
0 But here we may face anti-commutativity, therefore we express,
g = P exp
_

A
where P is the path ordering. The element g is dened as the holonomy of
A. The trace of holonomy of A is an invariant, called the Wilson loop W

(A):
W

(A) = Tr P exp
_

A (2)
1 EMERGENCE 2
Equations (1) and (2) are very important invariants, which boosted the study
of this subject in its early days.
1.2 Cohomological Field Theories
If the action S has no boundary
2
, i.e.,

S = 0, then notice that this property

could serve as an invariant of the theory. Which by Noethers thoerem implies
the existence of an invariance operator, say Q, such that for an operator O

O =
_
i[Q, O] = i(QO OQ) if at least one operator is bosonic,
i{Q, O} = i(QO +OQ) if both operators are fermionic.
But this theory is not cohomological yet, to make it so, we impose the fol-
lowing:
Nilpotency of the invariance operator Q: Q
2
= 0,
Closure of all O
i
s under the action of Q
3
: [Q,O
i
]

=0
If the invariant states |j satisfy Q|j=0, then in particular for an
unbroken symmetry, the vacuum satises
Q|0 = 0.
These facts enable us to deduce that any correlator containing an oper-
ator of the form O = [Q, L], i.e. a Qexact operator, is automatically
zero
0|, O
1
. . . [Q, L]
p
m. . . O
n
|0 = 0.
This, infact implies that the physical operators of this theory are given
by the cohomology ring of Q, viz.,
O O + [Q, L]

.
2
We like to study topological strings over Riemann surfaces without boundary.
3
[,] represents both commutator and anti-commutator
1 EMERGENCE 3
The nal requirement for a cohomological eld theory is that the energy-
momentum tensor must be Q-exact:
T

S
h

= [Q, G

, (3)
for some operator G

. Last requirement shows the metric independence of

the theory. So, in order to ensure (3), we use such Langrangean which itself
is Q-exact:
L = [Q, V ]

hence S = [Q,
_
V ]

,
for some operator V .
1.2.1 Semi-classical limit
If we explicitly include the Plancks constant in our description, then the
quantum measure in (1) reads
exp
i
h
_
Q,
_
M
V
_

.
One can show that
d
d h
O
i
1
O
in
= 0.
That is, the correlators are independent of h, and therefore we can calculate
them exactly in the classical limit.
1.3 Calabi-Yau Manifolds
1.3.1 K ahler Metric
A (local) K ahler metric on a certain patch is dened to be
g
i

j
=
K(z, z)
z
i
z
j
(4)
where K is known as the K ahler potential. The K ahler metric belongs to the
category of Hermitean metrics, such that
g
ij
= 0 = g
i

j
.
If the patch is an m-dimensional ball, then notice that the Kahler metric
is nothing but second derivative of the K ahler potential, and hence we can
1 EMERGENCE 4
invoke the Poincare lemma, showing the equivalence of (4) to the integrability
condition
4
:
g
i

j
z
k
=
g
k

j
z
i
. (5)
It is not dicult to show that in notation of n forms, (5) becomes
d = 0, (6)
where
2i

K
implies
i

j
=
ij
= ig
i

j
, while all other components of vanish. It follows
that a global K ahler manifold is a complex manifold with a Hermitean metric
satisfying (6).
1.3.2 Ricci-at K ahler manifold
From dierential geometry, we know that given a metric we can evaluate the
Ricci(contracted curvature) tensor, R

on it. For the K ahler metric, we nd

R
ij
and R
i

j
to vanish. In particular,
R
i

j
= 0. (7)
Now a Calabi-Yau manifold is dened to be a Ricci-at Kahler manifold, i.e.,
one for which (7) holds.
1.3.3 Moduli Spaces
Theorem 1 Ricci-atness implies that the rst Chern class, c
1
(M) of the
manifold M vanishes.
In [3] and [4], the converse of Theorem 1 was established for K ahler metrics,
stated as under:
Theorem 2 Given a complex manifold M with c
1
(M) equal to zero, then
there exists at most one Ricci-at K ahler metric in each K ahler class.
From the above theorems, we can say that given a certain topological space,
a complex structure and a K ahler class on it, we can obtain a unique Calabi-
Yau metric. Hence the Calabi-Yau moduli space is dened to be the space
of all possible Kahler classes and complex structures on M.
4
Similarly for the anti-holomorphic derivatives.
1 EMERGENCE 5
Figure 1: A 6-dim Calabi-Yau manifold. A 2-dim Calabi-Yau manifold is a
torus.
2 TOPOLOGICAL STRINGS 6
2 Topological Strings
2.1 Supersymmetry
Supersymmetry a.k.a SUSY, generators transform as fermions under the
Lorentz group. An N = n supersymmetric theory has n such spinor of super-
charges. In two-dimensions, the Lorentz group is SO(2) = U(1), so we use
the N = (p, q) when there are p irreducible spinor supercharges with positive
U(1)-charge, and q irreduible spinor supercharges with negative U(1)-charge.
2.1.1 N=2 supersymmetric sigma model
We begin with a Riemann surface M, the manifold on which our theory is
dened. Then consider the maps such that
: MX, (8)
where X is the target space
5
. We dene a set of elds
i
on M by the
compositon

i
= x
i
,
where x
i
are the local coordinates for X. The action, in this case is so dened
as to minimize the area of the Riemann surface, M. For the case of N = 2
supersymmetry,
S =
_
M
dzd z
_
1
2
g
ij

j
+
i
2
g
ij

+
i
2
g
ij

i
+

j
+
+
1
4
R
ijkl

i
+

j
+

_
.
(9)
Here g
ij
is the metric of X, R
ijkl
is the Riemann tensor, and is the covariant
derivative (cf. equation 11). The s transform as spinors w.r.t manfold
M rotations, and as vectors w.r.t Lorentz transformations on X. The +
and - subscripts denote the positive U(1)-charge and negative U(1)-charge
respectively. Equation (9) gives the action for our nonlinear sigma model.
We now introduce the notion of canonical bundle, B. Given an n-dimensional
manifold, the (n, 0)-forms
6
=
i
1
,...in
dz
1
. . . dz
n
(10)
5
physically speaking, X is space-time.
6
(n, m)-forms are F = F
i1...in,j1...jm
(z, z)dz
i1
. . . dz
in
d z
j1
. . . d z
jm
.
2 TOPOLOGICAL STRINGS 7
are called canonical bundle
7
. Such structure is also called a complex line
bundle. On a Riemann surface, the (complex) line bundle is the space of
(1,0) forms, which is dual to the bundle of holomorphic vectors (vectors of
the form v = v
z

z
) T
1,0
M Hence, we make the identication
B = T
1,0
M= T
0,1
M,
where the last expression employs the idea of anti-canonical bundle, intro-
duced in footnote [7] . Formally,

K as the bundle of (0, n)-forms is an anti-
canonical bundle. Further we make the following identications,
K
1,0
= T
1,0
= T
0,1
,
K
0,1
= T
1,0
= T
0,1
.
A (1, 0)-form transforms as follows, under a U(1) gauge transformation:
z z

= ze
i

z

z
=
z
e
i
If

e
i/2
, then (8) implies that the positive U(1)-charges transform as
square roots of (1, 0)-forms, and negative U(1)-charges transform as square
roots of (0, 1)-forms. Thus we may dene line bundles K
1/2
and K
1/2
, which
can be thought of as square roots of the canonical and anti-canonical bun-
dles, respectively. In terms of global data, this means that the transition
functions dening the spinors are elements of U(1) that square to the tran-
sition functions dening K and K. K
1/2
and K
1/2
, are the line bundles in
which these spinors live. [5] In terms of the index of the local coordinates x
i
of X, spinors
+
,

can be dened as:

i
+
(M, K
1/2

(TX)),

(M, K
1/2

(TX)),
where (M, E) means space of sections of the bundle E, and

is the pull-
back map of the tangent bundle TX onto the worldsheet M. The covariant
derivative on the spinor is dened as

i
+
=
i
+
+
i

i
jk

k
+
. (11)
7
It wouldnt be inapt to introduce anti-canonical bundle here, which is simply any
(0, n)-form.
2 TOPOLOGICAL STRINGS 8
Similarly,

=
i

+
i

i
jk

, (12)
where we have used the Christoel symbol on X. Now the theory exhibits
N = 2 supersymmetry under the following transformations:

i
= i

i
+
+ i
+

i
+
=

i
i
+

i
km

m
+
,

=
+

i
i

k
+

i
km

.
Here the transformation parameter
+
is a holomorphic section of K
1
/2 and

is an anti-holomorphic section of K
1
/2. holomorphism/anti-holomorphism
is a necessary condition for the
+
/

to be pulled through the derivatives

upon varying the Lagrangean.
2.1.2 N=2 SUSY and the K ahler Metric
Now consider if the target space X in (8) is a complex manifold. Analogically,
we know that a real 2n-dimensional manifold can be described by a family of
charts {U

, z
i

, z

}, where i runs from 1 through n, such that the transition

functions do not mix the holomorphic and anti-holomorphic coordinates, i.e.,
given an intersection U

,
z
i

= z
i

(z
1

, . . . , z
n

), and z

= z

(z

, . . . , z
n

).
This implies that we may write the nonlinear sigma model in (9) be decom-
posing the elds as

i
= {
i
,
[
i,

i
},
i

= {
i

}, g
ij
= {g
i

j
, g

ij
}.
Now this decomposition, would be consistent with the complex structure of
X, if and only if the metric on X is such that the parallel transport
8
of
vectors preserves the decomposition
TX = T
1,0
X T
0,1
X.
The metric that satisfy this condition is the K ahler metric given by (4),
because for
g
i

j
=
K(z, z)
z
i
z
j
,
8
The Christoel symbol terms in Covariant derivative expressions (11) and (12).
2 TOPOLOGICAL STRINGS 9
only
i
jk
and

k
are non-zero, while all other components vanish. This
enables us to rewrite our model (Equation. (9)) as
S =
_
M
dzd z
_
1
2
g
i

j
+
1
2
g

ij

j
+ ig
ij

D
z

+ ig
ij

i
+
D
z

j
+
+
1
4
R
i

jk

l

i
+

j
+

_
.
The corresponding supersymmetry rules are

i
= i

i
+
+ i
+

i
= i

i
+
+ i
+

i
+
=

i
i
+

i
km

m
+
,

i
+
=

b
ari i
+

k m

m
+
,

=
+

i
i
+

k
+

i
km

=
+

b
ari i
+

k m

m
+
,
As mentioned in the introduction, this model is known as N = 2 super-
symmetry, because there two spinors of supercharges, or equivalently two
holomorphic and two anti-holomorphic SUSY parameters. The spinors and
parameters are sections of the following bundles:

i
+
(M, K
1/2

(T
1,0
X)),

i
+
(M, K
1/2

(T
0,1
X)), (13)

(M, K
1/2

(T
1,0
X)),

(M, K
1/2

(T
0,1
X)),

+
,
+
(M, K
1/2
),

(M, K
1/2
).
2.2 R-symmetry
Besides supersymmetry, the sigma model remains conditionally invariant
under two more symmetries. They are known as the vector R-symmetry
and the axial R-symmetry, their generators being denoted by F
V
and F
A
,
respectively. These two symmetries act on fermions only, and are dened as
follows:
e
iF
V
{
i

} {e
i

, e
i

},
e
iF
A
{
i

} {e
i

, e
i

}.
2 TOPOLOGICAL STRINGS 10
The total supersymmetery variation is expressed as follows:
= i

Q
+
+ i
+
Q

+ i

Q
+
+ i
+
Q

,
where Q

and Q

are supersymmetry generators. These generators obey

the following Lie-algebra:
{Q

, Q

} = P H, (14)
where P and H are the Euclidean versions of the generators of space and
time translations. Let M, be the generator of Euclidean Lorentz boosts, i.e.,
SO(2) rotations. Then it satises the following Lie-algebra:
[M, Q

] = Q

and [M, Q

] = Q

. (15)
So in the same analogy, we have a Lie-algebra corresponding to the R-
symmetry generators:
[F
V
, Q

] = Q

, (16)
[F
V
, Q

] = Q

,
[F
A
, Q

] = Q

,
[F
A
, Q

] = Q

.
We conclude this section with the following important result:
Result 1 The vector R-symmetry is conserved for any K ahler target spaces.
But the axial R-symmetry is conserved if and only if the target space is Calabi-
Yau.
2.3 Twisted N=2 theories
Dont forget that our N = 2 supersymmetric eld theory is not yet a com-
pletely cohomological eld theory. From (13), we note that,
{Q
+
+Q

, Q
+
Q

} = 2H,
{Q
+
+Q

, Q
+
+Q

} = 2P,
{Q
+
+Q

, Q
+
Q

} = 2H,
{Q
+
+Q

, Q
+
+Q

} = 2P.
2 TOPOLOGICAL STRINGS 11
Dening Q
A
= Q
+
+Q

and Q
B
= Q
+
+Q

, we can nd two operators which

square to zero, such that the Hamiltonian and the momentum are Q
A/B
-
exact. Now we try to make our theory a cohomological eld theory. Notice
from (14) and (16) that the operators M, F
A
and F
V
have exactly same types
of commutators with the other operators, and have vanishing commutators
among themselves. Therefore, we may dene a new operator,
M
A
= M F
V
or M
B
= M F
A
relating the Lorentz transformation properties of the new elds with the old
ones.
Now consider the following commutators:
[M
A
, Q
+
] = 2Q
+
, [M
B
, Q
+
] = 2Q
+
,
[M
A
, Q

] = 0, [M
B
, Q

] = 2Q

,
[M
A
, Q
+
] = 0, [M
B
, Q

] = 0,
[M
A
, Q

] = 2Q

, [M
B
, Q

] = 0.
Notice that for M
A
, the operator Q
A
has become a scalar, similarly for M
B
,
the operator Q
B
serves as a scalar. And therefore the corresponding symme-
try operations can now be dened on any arbitrary curved worldsheet. This
construction is called twisting, and we may conclude that these twisted
theories can be viewed as cohomological eld theories.
2.4 The A-model
In the last section, the transformation M
A
= MF
V
corresponds to what is
known as the A-model
9
in the literature. After making this transformation,
spinors become

i
+

i
z
,

i
+

i
,

i.
z
The A-model action now reads
S
A
= 2t
_
M
(g
i

j
+g
i

j
+ig
ij

i
z

j
+ig
i

j
z

i
+
1
2
R
i

jk

i
z

j
z

l
),
9
So named, because of its relation to Type IIA strings.
2 TOPOLOGICAL STRINGS 12
where we have introduced a coupling constant t. With respect to the old
supersymmetry transformations (13),

and
+
are now scalars, and
+
and

are sections of the canonical and anti-canonical line buncle, respectively.

Now we can have a globally supersymmetric system, by doing away with
the
+
and

, and keeping the scalar SUSY parameters only. With the

parameters denoted by and , the transformations are:

i
= i,

i
= i

i,

i
z
=
z

i
i

k m

m,
z

i
z
=
z

i
i
k

i
km

m,
z

i
=

i
= 0.
2.5 The B-model
As in the previous case, the transformation M
B
= MF
A
corresponds to the
B-model
10
in the literature. After making this transformation, the spinors
become:

i
+

i
z
, g
i

j
(

j
+

)
i
,

i
z
,

i
+

i
,
The action for the B-model then reads,
S = t
_
M
(gi

j
z

i
+ig
i

i
(
z

i
z
+
z

i
z
)+i
i
(
z

i
z

i
z
)+
1
2
R
l
i

jk

i
z

k
z

l
).
The action is invariant uder the following supersymmetry transformations
11
:

i
= 0,

i
= i

i
,

i
z
= ,
i
z
= d
i
,

i
=
i
= 0.
10
So named because of its link with the Type IIB strings.
11
We have set = for simplicitys sake.
3 COUPLING TO GRAVITY 13
3 Coupling to Gravity
Hitherto, we have assumed a xed worldsheet geometry, viz., a xed Riemann
surface. What if we allow to integrate over dierent worldsheet geometries
? Infact this integration over metrics, is what is referred tp as quantum
gravity/geometry. The idea is to couple the sigma models to worldsheet
geometry.[13]
4 Conclusion
We have reviewed the story of topological eld theory, beginning from Chern-
Simons theory upto the A and B models. The question that which of the
many Calabi-Yaus is the string theory Calabi-Yau is still open. We developed
the A and B models of topological strings which correspond to the type IIA
and type IIB strings respectively. In the end, we discussed how to couple the
sigma models to gravity, an issue which is the principle motivation behind
all string theories.
REFERENCES 14
References
[1] Vonk, Marcel. A mini-course on topological strings, (2005),
[arXiv:hep-th/0504147v1].
[2] Mari no, Marcos. Chern-Simons Theory and Topological Strings,
(2005), arXiv:hep-th/0406005v4
[3] Calabi, E. On K ahler manifolds with vanishing canonical class, in
Algebraic geometry and topology: a symposium in honor of S. Lef-
schetz, eds. R. H. Fox, D. C. Spencer and A. W. Tucker, Princeton
University Press, (1957).
[4] Yau, S.-T. On the Ricci curvature of a compact K ahler manifold and
the complex Monge-Amp`ere equation, I, Comm. Pure Appl. Math.
31, 339-411 (1978).
[5] Collinucci, Andres and Wyder, Thomas. Introduction to topological
string theory, (2007).
[6] Witten, Edward. Topological Gravity, Phys. Lett. B 206, 601 (1988).
[7] Witten, Edward. Topological Sigma Models, Commun. Math. Phys.
118, 411 (1988).
[8] Witten, Edward. Topological Quantum Field Theory, Commun.
Math. Phys. 117, 353 (1988).
[9] Witten, Edward. On The Structure Of The Topological Phase Of Two-
Dimensional Gravity, Nucl.Phys.B 340, 281 (1990).
[10] Witten, Edward. Introduction To Cohomological Field Theories, Int.
J. Mod. Phys. A 6, 2775 (1991).
[11] Witten, Edward. Mirror manifolds and topological eld theory, in Es-
says on mirror manifolds, ed. S.-T. Yau, International Press 1992,
120-158 [arXiv:hep-th/9112056].
[12] Witten, Edward. Chern-Simons gauge theory as a string theory, Prog.
Math. 133, 637 (1995) [arXiv:hep-th/9207094].
[13] Hori, Vafa, Klemm, et al., Mirror Symmetry, Clay Mathematics In-
stitute (2003)