Академический Документы
Профессиональный Документы
Культура Документы
Author:
Carlos Jonathan Ramrez Valdez
Thesis Advisor:
Dr. Hector Hugo Garca Compean
Physics Department
ACKNOWLEDGEMENTS
iii
iv
ABSTRACT
vi
RESUMEN
vii
viii
CONTENTS
1 Introduction
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3 Quantization
3.1 Fundamentals . . . . . . . . . . . . . . . . . . .
3.1.1 Morse Function . . . . . . . . . . . . . .
3.1.2 Lagrangian and Coisotropic submanifolds
3.2 Quantization Problem . . . . . . . . . . . . . .
3.3 Brane Quantization . . . . . . . . . . . . . . . .
3.3.1 A new perspective on quantization . . .
4 Path Integral Quantization of G/H
4.1 Basic Construction . . . . . . . . . .
4.1.1 Analytic Continuation . . . .
4.1.2 Simplest Cycle of Integration
4.1.3 Another Cycle of Integration .
4.2 Sigma Model . . . . . . . . . . . . .
4.3 Boundary conditions of the A-model
ix
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
5
5
6
7
10
.
.
.
.
.
.
15
15
15
17
17
19
19
.
.
.
.
.
.
21
21
22
22
23
25
28
CONTENTS
4.4
29
Conclusions
35
Bibliography
36
CHAPTER 1
INTRODUCTION
The quantization of a classical system is a very important topic in physics, if we take
any classical system, it does not exist a unique quantum version of it. Some problems
to quantize a physical system will be explained in the section 3.2. Many authors
are making efforts for a better understanding of the meaning of quantization, one of
these efforts is called brane quantization [1].
This approach to quantization relies on the equivalence between two structures,
a symplectic manifold, (M, ), that describes completely a classical system, and
an A-model of a complexification of M . The so reached quantization of (M, ) is
precisely the space of strings with boundary conditions on two A-branes, Lagrangian
the first and coisotropic the other. The Lagrangian A-brane is supported on M , while
the symplectic structure defines the coisotropic A-brane. With this in mind we
want to inquire in the case when the phase space M is a coset G/H. And with this
extend the scope of the theory in [2].
An interesting way to introduce this approach is to start from the Feynmans
formalism of the quantum mechanics [2]. That is the reason to review it in chapter 2.
We also clarify this formalism by quantizing the Lie group SU (2). After doing this,
we delve in some needed definitions and describe the problem of quantizing a physical
system in chapter 3.
Subsequently, in chapter 4 we developed the main theory, i.e. we applied the
method given in [2] to cosets G/H. The procedure is briefly described as follows, we
1
Chapter 1. Introduction
take a complexification of the coset phase space. In doing this, we have to carefully
choose the integration cycles. Since, if we do not pick them up correctly then there
convergence problems could exist. We can choose different cycles of integration,
but we are particularly interested in one that has a direct interpretation as a two
dimensional quantum field theory in which the target is the complexification of our
phase space. The output theory would be a topologically twisted A-model with a
fermionic symmetry and a coisotropic brane as the cycle of integration.
[
In our work we are interested in the case when the target is a complex coset G/H
where G is a Lie group. With this in mind, we first establish the structure of the
path integral in curved spaces [3][4]. The elements that appear in the exponential
of the path integral for these spaces are a term pa dq a and an effective Hamiltonian, where pa and q a are the usual canonical coordinates while the last term arise
from the ordering problem that we will explain in section 3.2. Fortunately, the term
of our interest, pa dq a , is the same independently of the choice of the order of operators.
Once we clarify the starting point, we can follow the theory developed in [2]. We
do a complexification of our phase-space (and as consequence the complexification
of our integration space) duplicating our degrees of freedom. The main difference
between this work and [2] is the quotient structure of the initial phase space. As
[ of the quotient G/H has to be a
in [5] we suppose that the complexification G/H
Kahler manifold, this implies that we have a compatible metric with a complex
structure J, and a Kahler form A.
As mentioned, the complexification of the phase space, implies a complexification
of our integration space, but this is not necessarily our final integration cycle because
the exponential of the path integral could be divergent, then, we have to build
our integration cycles properly. As a first approach we construct a simple cycle of
integration. This is achieved by constructing a new submanifold (G/H)0 with the
same structure of G/H and a with different integration cycle. This new construction
could arise from a different classical system, but it reaches the same complexified space.
On the other hand a different integration cycle can be constructed through the
Morse flow equation. Through this we can narrow the divergence term on the path integral, and also add a new evolution parameter. Then, the associated action becomes
into a non-linear sigma model. Next step is to analyse the corresponding boundary
conditions for the coset structure to relate the holomorphic and anti-holomorphic parts
of the bosonic and fermionic fields. With this relations we reach a Kazama Suzuki
2
model with type A boundary conditions [6]. In our case this conditions correspond to
a coisotropic D-brane wrapping around our integration cycle.
The final theory reached in this work is a topological twisted A-model with
fermionic symmetry [5] constructed from a coset theory G/H.
Finally, we apply the theory to an example, we take the Lie group SL(2, R) and
the subgroup U (1) to form the coset the coset G/H = SL(2, R)/U (1). Next, we
obtain the cycles of integration via the Morse flow, and we build the Lagrangian and
the coisotropic A-branes.
Chapter 1. Introduction
CHAPTER 2
PATH INTEGRAL ON GROUP
MANIFOLDS
In this chapter we review the Feynman path integral formalism of Quantum Mechanics
in curved spaces. Also, we quantize a group manifold as an example, being this
SU (2).
2.1
Path Integral
The year 1925 can be seen as the beginning of modern quantum mechanics marked
by the two almost simultaneously published papers of Heisenberg and Schrodinger.
In the former it is proposed the formalism of matrix mechanics, while the latter
proposes the formalism of wave mechanics. Schrodinger was the first in showing that
the two formulations are physically equivalent. Both of these approaches were combined heuristically by Dirac into a more general formulation of quantum mechanics.
The mathematically rigorous development of this general formulation of quantum
mechanics was carried out by von Neumann.
The general formulation of quantum mechanics developed by Dirac and von
Neumann lay on the Hamiltonian formalism of classical mechanics. The Lagrangian
formalism did not have place in this general formulation of quantum mechanics,
except in the derivation of Schrodingers wave equation from the Hamilton-Jacobi
equation.
5
The first hint of the possible importance of the Lagrangian in quantum mechanics
was given by Dirac, he remarked
that the quantum amplitude hqt |qt0 i corresponds to
R
i t
Ldt
the classical quantity e ~ t0 . This remark led Feynman in 1941 to a new formulation
of quantum mechanics.
Feynman and Hibbs gave a heuristic introduction to this new approach to Quantum Mechanics, whereas Schulman provided a more rigorous introduction to the
Feynmans path integral on configuration space and considered a number of applications of the method in different fields of physics.
This new approach can not lead us to new discoveries that the ordinary quantum
mechanics could not reach. However, the Feynmans formalism brings new ways to
resolve some issues in easier manner and performs a better understanding of the
connection of quantum mechanics and classical mechanics [7].
2.1.1
Dp(t)Dq(t)e ~
pa dq a H(pa ,q a )dt
(2.1)
where pa and q a are the usual canonical coordinates of the phase space and H(pa , q a )
the corresponding Hamiltonian, but this is for a flat space, and we want the Feynmans
path integral over a phase space of a Riemannian manifold M , because we want to
extend the scope of the the theory in [2].
First, let g be a metric on a manifold M . If we try to construct the Feynmans
path integral from the formulation of operators of the quantum mechanics, we arrive
to the ordering problem. This is the arbitrariness in choosing the order of operators
in the Hamiltonian. Then, the path integral depends on this choice of order. Taking
into account this fact, the Feynmans path integral in curved spaces can be writing as
Z
Dp(t)Dq(t)e ~
pa dq a Hef f (p,q)dt
(2.2)
00
(2.3)
where q 0 and q 00 are consecutive elements of the partition that is used to calculate the
path integral, and the potential corrections V given by
~2 ab d c
g ac bd R ,
8m
~2
a b g ab 2g 1/4 LB g 1/4 ,
=
8m
~2
=
R,
6m
VW eyl =
VSR
VAS
(2.4)
where R is the scalar of curvature, abc are the Christoffel symbols and LB 1 is the
Laplace-Beltrami operator.
2.2
In this section we will quantize the Lie Group SU(2) with the Feynmans path integral
of Quantum Mechanics.
SU (2) is the set of special unitary matrices of order 22. This set is isomorphic to
S , so we will consider a parametrization of the 3-sphere in R4 . This parametrization
is given by the euler angles, as follows
3
LB f |g|1/2 i |g|1/2 g ij j f , where f is a scalar function, gij the metric and g his determinant
u1
u2
u3
u4
+
= cos
cos
,
2
2
+
sin
,
= cos
2
2
= sin
cos
,
2
2
= sin
sin
,
2
2
(2.5)
[0, ]
(2.6)
[0, 2]
(2.7)
[0, 2]
(2.8)
with this parametrization the Lagrangian of a free particle over SU (2) is given by
1
1
L = I u 2 = I u1 2 + u2 2 + u3 2 + u4 2 ,
2
2
where I is the inertial moment.
(2.9)
On the other hand, the isomorphism between SU (2) and S 3 is given by the
following relation
u1 + iu2 iu3 u4
(u1 , u2 , u3 , u4 )
,
(2.10)
iu3 + u4 u1 iu2
in terms of the euler angles an element of SU (2) has the following form
!
+
cos 2 ei 2
i sin 2 ei 2
,
+
i sin 2 ei 2 cos 2 ei 2
(2.11)
+
T r (U ) = cos cos
.
2
2
2
8
(2.12)
(2.13)
cos 2 ei 2
i sin 2 ei 2
+
i sin 2 ei 2
cos 2 ei 2
!
,
(2.14)
when considering the product in the relation (2.12), we obtain the following equations:
cos = cos a cos b sin a sin b cos (a b ) ,
e
[sin a cos b cos a sin b cos (a b ) i sin b sin (a b )] ,
sin
(2.16)
1
a
b i
a
b i
cos cos e 2 (a b +a b ) + sin sin e 2 (a b a +b ) . (2.17)
=
2
2
2
2
cos 2
ei =
ei
+
2
(2.15)
ia
Furthermore, we can bring U to a diagonal form by doing a similarity transformation, we will obtain a diagonal matrix that transforms Ua into Ub by a rotation
around of his principle axis. Considering them as initial and final states, respectively,
the transformation have the meaning of the evolution of the system by a rotation
around this principle axis. The diagonal matrix has the form
i
e2
0
.
(2.18)
0 ei 2
An easy way to see this, is that the characteristic polynomial is of degree two, so
it has two real or complex roots. The eigenvalues have to be the same for any , so, if
we take = 2 and = 2 the matrix U gives rise to a rotation matrix of dimension
2 2 and we know that there does not exist any real eigenvalues that diagonalize
such a matrix. Then, the only option is to have complex eigenvalues. Hence, the
roots have to be complex conjugate of each other. In addition, the determinant of
the matrix has to be one, this implies that the only option for the matrix is to have
the form of equation (2.18).
By the properties of the trace, we know that the traces of the (2.11) and (2.18)
have the same value. Thus we obtain
+
= cos cos
,
(2.19)
2
2
2
this equation enables us to relate the inner product with only one angle, that
corresponds to the parameter of the diagonal matrix.
cos
2.2.1
The path integral is constructed through slices of time. That implies that the most
important part is to find his form at very small times and build the complete path
integral by the method that we wish. Then for very small time, we propose as usual
the semiclassical approximation [8]
Kj = eiSj ,
(2.20)
where Sj is the action of the system and Kj is the kernel (propagator) between uj1
and uj at short times. For the free particle case we can calculate Sj by integrating
the equation of motion,
I
(
uj uj1 )2 ,
2
and the fact that |
u| = 1 we rewrite the action as
Sj =
(2.21)
I
(1 cos j ) .
(2.22)
As we know, we have to include the ordering correction in the action [3][8], so the
Kernel at short times has to be,
I
Kj = exp i
(1 cos j ) +
,
(2.23)
8I
Sj =
we just found the kernel for SU (2) at short times. Now, we could try to calculate
the path integral with the kernel (2.23) but it does not have a practical form, and so
we need to do some changes before proceeding it.
Using the identity [3][8]
e
with = , u =
u cos
k=
1 X
1
1
2
exp ik + u
k
,
2u k=
2u
4
(2.24)
iI
we can change the cosine term in the exponential as follows,
I
Kj = ei( + 8I ) ei cos j
i
i i X ikj cos 2I
2 1
e 8I
e
e (kj 4 ) ,
2I
k =
I
10
l
i i X
i
1
i 2 1
2
8I
Kj e
lj
exp
(lj + 1) +
exp
eimj ,
2I
2I
4
2I
4
l=0
m=l
(2.25)
the difference of the exponential with the indexes lj can be simplify with the following
property [8][3]
"
#
2 2
Y
x
(a b) x
eax ebx = (a b) xe(a+b) 2
1+
,
(2.26)
4s2 2
s=1
then
X
l=0
1
i
1
2
2
lj
exp
(lj + 1) +
4
2I
4
"
#
2
i
1
4I
(2lj + 1)
(2lj + 1) exp
lj +
sin
2I
2
(2lj + 1)
4I
"
#
2
i
1
(2lj + 1) exp
lj +
.
2I
2
i
exp
2I
i
2I
i
2I
imj
mj =lj
lj
X
lj
X
mj =lj nj =lj
(2.27)
so, the kernel reaches the form
Kj
i
2I
Pmljj nj
2 X
lj
lj
X
i
(2lj + 1) e 2I
N2
m
where AN = 2~
, and for our case we have I = m|
u|2 = m, ~ = 1 and
dxj = sin j dj dj dj . Replacing our kernel in eq. (2.28), we get
K (ua , ub , T ) = lim
N
I
2i
N2+1 Z Y
N
dj ei
PN +1
l=1
Sl
(2.29)
j=1
N
I
2i
N2+1 Z Y
N
dj
j=1
N
+1
Y
Kl ,
(2.30)
l=1
Z
dj sin j
Pnl j mj j
dj
2
1
l l m m n n ,
(cos j ) Pmj+1 nlj+1 j+1 (cos j ) =
2lj + 1 j j+1 j j+1 j j+1
(2.31)
with the purpose of simplifying the calculations, we define the following new variables
X
lj
lj
X
X
X
lj =0 mlj nlj
i
(2.32)
(2.33)
N
I
2i
N2+1 Z Y
N
dj
j=1
N
+1 X
Y
l=1
j
j Kj1
Kjj ,
(2.34)
the kernel is now in a very manageable form. In order togain a better understanding
of his structure let us analyse two successive elements of the product in the integrand.
Z XX
j
j+1
j j+1 Kj1
Kjj Kjj+1 Kj+1
d
j
j+1
Z
XX
j
j+1
j j+1
j j+1 Kj1 Kj+1
Kj K j
d
=
j
j+1
(2.35)
XX
1
j
j+1
j j+1 Kj1 Kj+1
lj lj+1 lm lm+1 ln ln+1
=
j
j+1
2lj + 1
X
1
j
j
Kj+1
,
Kj1
=
j2
j
2lj + 1
then, we do over the above procedure until we obtain (doing the integration N times),
X
1
1N +1
1
2lj + 1
N
K01 KN1 +1 ,
(2.36)
finally, we obtain the kernel for two arbitrary points in the group, in other words we
quantized the SU (2) group. The final form of the kernel is then
K (ua , ub , T )
=A
iT
l(l+1)
2I
(2l + 1) e
l
l
X
X
l
l
ei(ma +na ) ei(mb +nb ) Pmn
(cos a ) Pnm
(cos b ) .
m=l n=l
l=0
(2.37)
We have reached the quantized version of SU (2), from here we can directly obtain
the quantization of SU (2)/U (1). Equation (2.37) can be reduce to the quantization
of SU (2)/U (1) through the identification of a = b , thus obtaining the following
result
K (ua , ub , T ) = A
X
l=0
iT
l(l+1)
2I
(2l + 1) e
l
X
n=l
13
14
CHAPTER 3
QUANTIZATION
In this chapter, we will introduce some fundamentals, like the notion of Lagrangian
manifolds and coisotropic manifolds. The aim is to establish the essential tools to
build the A-model.
3.1
Fundamentals
3.1.1
Morse Function
(3.1)
Chapter 3. Quantization
The flow equation eq. (3.1) implies that the Morse function h is always decreasing
along any non-constant flow.
In the neighborhood of a critical point p, with the Hessian matrix of h diagonalized, we can find a system of normal coordinates wi centring at p, such that in these
coordinates the Morse function h can be expand as h = h0 + ei wi + O(w3 ) and the
metric as g ij = ij + O(w2 ), where h0 = h(0) and the ei s are the eigenvalues of the
Hessian matrix at p.
Then, the flow equation to a first order in w will become
dwi
= ei wi
ds
(3.2)
wi = ri eei s
(3.3)
with solutions
for these solutions, we can ask ourselves at what limit s reaches the critical point
p = 0, i.e. wi = 0 for any i. The answer depends on the sign of the eigenvalues ei . For
a non-trivial solution the limits at which s reaches a critical point are s and
s . Each solution leaves some free parameters. Without lost of generality we
now focus in the solutions s , then we have ip indeterminate values because
we have to fix ri = 0 for each ei > 0.
Thus the solutions at p = 0 form a family of dimension ip , now we define the
subspace Cp of Z as the set of all values of h at s = 0 with flows that begin in
s .
Since h is a decreasing function, its maximum value is h(p) and this value also
belongs to Cp because it is the value of the trivial flow. Hence, the maximum value
of Cp is h(p).
Now consider the case when Z is a complex manifold, with complex dimension n
with a function h = Re S where S is holomorphic. Then,
local form of h near
Pn the
i 2
to a non-degenerated critical point p is h = h0 + Re i=1 (z ) , with local complex
coordinates z i . Since Re z i = xi y i the stable and unstable directions for h are paired,
then the subspace Cp of each p is middle dimensional. In this situation the Cp generates a basis that classifies cycles on which h is bounded above and goes to at .
16
3.1.2
3.2
Quantization Problem
In this section we will define the motivation of this work. We will establish some of
the problems of quantization [1].
Usually, the passage from classical to quantum mechanics is through the correspondence principle. This is a recipe that establishes the bridge between the two
worlds. Consider for example the Hamiltonian formulation of classical mechanics
in which one of its big achievements is the introduction of canonical transformations, (x, p) (X, P ), that preserve the Poisson brackets {x, p} = {X, P }. Here the
recipe consist in replacing the last with commutators. Let a and b be two classical
observables, Oa and Ob their associated operators, then the recipe is summarized as
follows
1
[Oa , Ob ],
(3.4)
i~
however, this simple description has some problems even for a basic phase space
i.e. the plain space R2n . For example, for a classical system, two different sets of
coordinates related by a canonical transformation are equivalent, but the associated
{a, b}
17
Chapter 3. Quantization
quantum systems for each set of coordinates can be different, and these could not be
connected through an unitary transformation U , i.e. a change of basis between the
different associated Hilbert spaces.
Another difficulty is the so called ordering problem, let f (x, p) be an observable
on the phase space and his associated operator Of . The function f could have a
combination of products of x and p that generates an ambiguity in the choosing of
his operator Of . This problem is avoided only if the function f is linear or a linear
combination of powers of only one variable. In this case we have a simple relation
eq. (3.4) given by
[Of , Og ] = i~O{f,g} .
(3.5)
The two mentioned problems are related because a well defined observable function
does not remain invariant under canonical transformations, so these different choices
lead to different quantizations.
Now, consider a phase space M of dimension 2n. Still, if we can identify locally
M with R2n , we can not quantize the manifold since the quantization of R2n is not
unique. Even, if we have a prequantum line bundle, there does not exist a recipe to
instruct us how to do the quantization of M , because we do not know how to glue
the local quantizations of the charts of M , so we need this information as a prerequisite.
Then, we can not quantize M without more information than the symplectic
structure and a prequantum line bundle, regrettably the additional structure required
it is unknown. Hence, there does not exist a general theory of quantization.
In practice, the quantization is an ambiguous notion. An example in which we
know what quantization means is the R2n space with an affine structure (that is
equivalent to establish the meaning of our linear observable functions). This structure
does not need extra information to be quantized. The usual procedure is to split
the linear functions in coordinates and momenta and then regard these to act as
multiplication and differentiation respectively. The resulting Hilbert space admits
an straightforward action of the symplectic group Sp(2n, R), or instead, of his double cover, without any dependence on the splitting between coordinates and momenta.
Another important example is the quantization of a cotangent bundle M = T U
for a manifold U with a given symplectic structure. This structure can be quantize in
a direct way through half-densities on U . A similarly procedure exist for quantization
18
3.3
3.3.1
Brane Quantization
A new perspective on quantization
Let us start with a symplectic manifold, (M, ), endowed with a prequantum line L as
in the geometric quantization, this line bundle is complex with an unitary connection
of curvature .
be a complexification of M with an antiholomorphic involution , such that
Let M
19
Chapter 3. Quantization
with curvature Im deIn particular, the choice of the line bundle L M
as support. The
termines a canonical coisotropic brane in the A-model with all M
quantization of (M, ) is achieved by standing that the associated Hilbert space H
is the vector space of (Bcc, Bl) strings, where Bcc refers to a canonical coisotropic
A-brane and Bl to a Lagrangian A-brane. This definition gives a vector space depending on the choice of Bl. Aldi and Zaslow shown in examples [9] that the explicit
construction of this vector space is similar to quantization. Finally, the resulting
,L, and Bl.
Hilbert space H, depends only on the choices of M
Only one thing remains unsolved, we need a natural way to quantize a large class
of functions on M , i.e. to define the operators on H. The functions on M that can
that have a suitable behavior
be naturally quantized are holomorphic functions on M
at infinity. In some cases these holomorphic functions could be only the functions of
polynomial growth.
20
CHAPTER 4
PATH INTEGRAL QUANTIZATION OF
G/H
4.1
Basic Construction
l
Dp(t)Dq(t) ei (pl dq Hef f dt) u1 (t1 )u2 (t2 ) . . . un (tn ).
(4.1)
where U is the space of cycles of integration, pl (t) and q l (t) are cyclic variables of
period 2 over the phase space of G/H and ui (t) are observables. For our purpose
21
4.1.1
Analytic Continuation
[ with an
Here we complexify our quotient G/H to get a complex manifold G/H
extended closed holomorphic (2, 0) symplectic form W = A + if and require the
. That we can
existence of an antiholomorphic involution such that (W ) = W
interpret as the minimal requirement to develop the theory [5][2][10][11]. Schematically
[
G/H G/H,
f W,
the definition of implies the constraint Re W |G/H = 0.
[ as g .
Hereinafter we will denote the real components of our coordinates on G/H
We know that W is closed, then, it can be written as W = d, where = dg .
For convenience we write down = c + ib. This implies that dc = A and db = f .
Thus the path integral is rewritten as
Z
H
of U as U.
4.1.2
Next, we will construct some cycles of integration, with the purpose of to have
[ and
middle dimension we take a middle dimensional sub-manifold (G/H)0 G/H
as the free loop space over (G/H)0 , i.e. parametrizes maps from S 1 to (G/H)0 .
We have to
H handle carefully the real part of the exponent of the path integral
(4.2), h = Re dg , because it is unbounded by above. With the purpose to avoid
22
4.1.3
In order to obtain integration cycles with h bounded by above, we use the Morse
flow
eq. (3.1). In our case the Morse function will be the real part of the exponent
H
h = Re dg = c dg .
(4.4)
the Morse flow equation will be
g (s, t)
h
= .
s
g
(4.5)
now, we have to find the space of critical points of the Morse function that are given
by
I
h = A g dg = 0.
(4.6)
since A is non degenerated we obtain dg = 0, and so the set of critical points is
the set of constant curves, i.e. there exist a one-to-one relation between the constants
[ Then the set of critical points can be identified with
curves and the points on G/H.
[ Let us name this copy as G/H
[ . Later on, we chose a middle dimensional
G/H.
[
space V G/H and we consider all the solutions in the half-line that starts in V .
23
|g| =
g g dt.
(4.7)
(4.8)
If is K
ahler, then I represents a complex structure.
24
4.2
Sigma Model
We have a complexified path integral eq. (4.2), but we want to define the integration
only over the disk D. For this we insert a Lagrangian multiplier T that restricts
the integration over the coordinates related through the Morse flow equation (4.8).
[
T is a one-form on D with values in the pullback of the cotangent bundle of G/H,
obeying T = ?I T .
First we rewrite the condition on eq. (4.8) as follows
U = dg + ?I dg ,
(4.9)
where ?ds = dt and ?dt = ds, then we insert the restriction in the path integral on
eq. (4.2) as a Dirac delta,
Z
H
where
1
(U ) = |det(U/g)|
DT ei
R
D
T U
(4.11)
and OV (0) is a closed operator that impose the constraint that the point z = 0 is
mapped to s = 0 or s = (the critical set given by the Lagrangian submanifold).
Now, we want to introduce in the path integral fermions and with different
fermionic quantum number +1, 1, respectively, with kinetic energy that is the
linearization of the equation U = 0. This term added has as a consequence the
cancellation of the bosonic determinant up to a sign, in our case this sign can be
taken as +1 since CV is connected.
Therefore are zero-forms that take values in the pullback of the complexified
[ of G/H,
[ and are (anti)holomorphic one-forms with
tangent bundle TC (G/H)
[ Also obey
values in the pullback of the complexified cotangent bundle TC (G/H).
25
DgDDDT e
D (T ?U
dg
n
Y
uk (tk ),
(4.13)
In the case of a symmetric Kahler coset group we know from [10][11][12] that the
fermionic symmetry transformations of the twisted topological A-model are
g
+
+
= i(g+ + g),
= = 0,
= g 1 z g,
= + (zg)g 1 ,
(4.14)
where is the fermionic susy parameter and g is the embedding map of to G,
[ , into the splitting T
are the projectors of complexified tangent space TC of G/H
due to the almost complex structure J. = and = , where are
Weyl fermions with values on TC . The set of equations eq. (4.14) can be rewritten as
g = ,
= T ,
= T = 0,
(4.15)
with the
where we used the exponential map g = eg X (X are the generators of G)
[ and also the following
purpose of introduce the coordinates g of the target G/H,
relations
= i(+
+
),
= + + ,
T = ( z + + z) g .
(4.16)
The action is invariant by exact terms, then, let us add the following exact term
Z
Z
Z
?T =
T ?T +
R
2 D
2 D
4 D
(4.17)
4
D 2
26
(4.18)
DgDD e D ( 2 U ?U i D + 4 R ) e
dg
n
Y
(4.19)
Then, the bosonic kinetic energy is
2
1
s g s g + t g t g + A s g t g ,
U ?U =
dsdt
2
D
D
(4.20)
let us rewrite the RHS of eq. (4.20) with the following identity
s g s g + t g t g = ij Tr (g 1 i g)(g 1 j g) ,
(4.21)
1
2
dsdt A s g t g =
A=
dc =
D
c = h.
(4.22)
(4.23)
now, by adding the term that comes from pa dq a on eq. (4.2) to the term proportional
to h, and fixing = 2 we obtain
I
I
I
h+
(c + ib) = h + h + i
b=i
b.
(4.24)
D
Thus, without consider the terms with fermions, the exponential on eq. (4.19) is
found to be
Z
I
1
ij
1
1
Tr(g i g)(g j g) + i
b.
(4.25)
2 D
D
this is the W ZW action in an open non-linear twisted sigma model of type A.
27
Tr(g i g)(g j g) A + i
b
2 D
D
D
(4.26)
where = 1 2 .
[ with a
Now we introduce the associated coisotropic D-brane L defined as G/H
connection b and a curvature f . Therefore the action, eq. (4.26), is a WZW action
describing open strings atached to a D-brane, or more precisely, to an A-brane. In
the next section we will argue that this is precisely a coisotropic A-brane BG[6].
If we choose 1
2
A = ,
b = .
here we can ask ourselves, whether it is possible to reach an N = 2 superconformal
[ With the purpose to clarify this point we introduce
field theory from a coset G/H.
the following theorem [10].
Theorem 1. When rank G = rank H, a necessary and sufficient condition for a
model on a coset G/H to have N = 2 superconformal symmetry is that G/H is
kahlerian.
4.3
D-branes can be studied by using the techniques of perturbative string theory. They
can also be described in terms of boundary conditions of open strings, or as boundary
states in the closed string sector. The concept of a boundary state describes how
closed strings are emitted or absorbed on the D-brane world volume.
For the study of bosonic D-branes on cosets G/H we require certain boundary
conditions [5] for the bosonic and fermionic fields given by
z ),
I(z) = RI(
z ),
(z) = R(
28
(4.27)
where T X|Y = N Y T Y and F = db, i.e. the curvature of a gauge field b. Also R
satisfies the following conditions
R R = ,
(4.28)
R R A = A ,
(4.29)
R R R f = f ,
(4.30)
(4.31)
where k refers to the parallel components and to the normal components of the
metric along the cycle. In contrast, eq. (4.29) implies that
A |k = 0 or A | = 0.
(4.32)
The vanishing of A and his non-degeneracy implies that the dimension of the
[ So we have a Lagrangian submanifold of G/H.
[
cycle has to be 12 G/H.
[ we do not have perpendicular components for
In the case when Y = G/H
[ that correspond to a
neither for A. Then we have an A-brane supported on G/H,
coisotropic A-brane with a connection b.
4.4
As an example, we consider the coset SL(2, R)/U (1) as the symplectic structure that
we want to quantize. The complexification of SL(2, R)/U (1) is SL(2, C)/U (1)C , and
29
(4.34)
A0 = A A,
(4.35)
where
A = 2i
dz dz
A = 2i
,
(1 + zz)2
dz d
z
;
(1 + z z)2
(4.36)
),
sin ei
x1 + ix2
=
,
1 x3
1 cos
(4.37)
x1 + i
x2
sin ei
z =
=
.
(4.38)
1 x3
1 cos
Eq. 4.37 represents a conformal transformation between the 2-sphere and the flat
plane with compactification points.
A0 from eq. (4.35) vanishes when one restricts to the diagonal submanifold
2
Sdiag
= {S 2 S2 |z = z(and z = z)}.
30
(4.39)
then
f0 =
1
1
d,
sin d d + sin d
4
4
(4.40)
and
2
f 0 |Sdiag
=f =
1
sin d d.
2
(4.41)
z
z
z
z
dz + 4i
d
z 2i
d
z
4i
dz,
1 + z z
1 + z z
1 + zz
1 + zz
(4.42)
2
it is immediate to check that on the Sdiag
c0 vanishes, that is
2
c0 |Sdiag
= 0,
(4.43)
c0 .
(4.44)
(4.45)
b0 = (1 cos )d + (1 cos )d
4
4
2
wich on the diagonal Sdiag
with = and = we get
1
2
b0 |Sdiag
= (1 cos )d.
2
(4.46)
(4.47)
1
0
0 sin2
0 =
0
0
0
0
(4.48)
0
0
0
0
,
1
0
0 sin2
(4.49)
(4.50)
d
h
= ;
ds
h
d
= ;
ds
d
h
= sin2
ds
h
d
= sin2
ds
Now, we can construct the A-branes for our example. First we recognize the
submanifolds, we have a Lagrangian submanifold, this being S 2 , and the coisotropic
submanifold has to be S 2 S 2 . Thus, by eq. (4.27) the corresponding boundary
conditions are [6]
Lagrangian A-brane,
2
2
R = idN S 2 (g f |Sdiag
)1 (g + f |Sdiag
),
(4.51)
Coisotropic A-brane,
R = ( f |S 2 S 2 )1 ( + f |S 2 S 2 ),
(4.52)
33
34
CONCLUSIONS
The prospects are, add a nonzero effective Hamiltonian to the theory and analyse
more examples like the Quaternion-Kahler symmetric spaces.
36
BIBLIOGRAPHY
[1] Sergei Gukov and Edward Witten. Branes and quantization. Advances in
Theoretical and Mathematical Physics, 13(5):14451518, 2009.
[2] Edward Witten. A new look of the path integral of quantum mechanics.
arXiv:1009.6032v1, September 2010.
[3] A. Demichev M. Chaichian. Path Integral in Physics, Vol. I. Springer, 1998.
[4] F. Steiner C. Grosche. Handbook of Feynman Path Integrals. Institute of Physics
Publishing, 2001.
[5] Sonia Stanciu. D-branes in kazama suzuki models. Nuclear Physics B, 526(13):295310, August 1998.
[6] Anton Kapustin and Dmitri Orlov. Remarks on a-branes, mirror symmetry, and
the fukaya category. Journal of Geometry and Physics, 48(1):8499, October
2003.
[7] Wolfgang Tome. Path Integral on Group Manifolds. World Scientific, 1998.
[8] D. Peak and A. Inomata. Summation over feynman histories in polar coordinates.
Journal of Mathematical Physics, 10(8):14221428, August 1969.
[9] M. Aldi and E. Zaslow. Coisotropic branes, noncommutativity, and the mirror
correspondence. Journal of High Energy Physics, 2005(06), June 2005.
37
BIBLIOGRAPHY
[10] Y. Kazama and H. Suzuki. Characterization of n=2 superconformal models
generated by the coset space method. Physics Letters B, 216(1-2):112116,
January 1989.
[11] Y. Kazama and H. Suzuki. New n=2 superconformal field theories and superstring
compactification. Nuclear Physics B, 321:232268, 1989.
[12] Edward Witten. The n matrix models and gauged wzw models. Nuclear Physics
B, 371:191245, 1992.
38