Lecture 1

Quantum Field Theories: An

introduction


 
The string theory is a special case of a quantum field theory (QFT). Any QFT deals


with smooth maps of Riemannian manifolds, the dimension of is



the dimension of the theory. We also have an action function defined on the set
Map

of smooth maps. A QFT studies integrals
$ % '& )(+* &-,
 !#" (1.1)

Here (+* &-, stands for some measure on the space of paths, . is a parameter (usually
Map

%
Map 
/ 021 is an insertion function. The
657<9=8/: should
very small, Planck constant) and
;
 tobetheinterpreted
number
 4
 3 as the probability amplitude of the contribution

>0?A@ 
of the map integral. The integral
$ED &
4    BC (1.2)

is called the partition function of the theory. In a relativistic QFT, the space  has a
Map

Lorentzian metric of signature GF

#HI
KJKJJ4
/H . The first coordinate is reserved for time,

> @ 
the rest are for space. In this case, the integral (1.1) is replaced with
65798/: % M& G(;* &N, J
Map 4  7L 3 (1.3)

Let us start with a O -dimensional theory. In this case  is a point, so &
P is
a point QSR  and 
<TU1 is a scalar function. The Minkowski partition function

>V@  WB98/: D Q J
of the theory is an integral

 7L 3 (1.4)

Following the Harvard lectures of C. Vafa in 1999, let us consider the following
example:

1
2 LECTURE 1. QUANTUM FIELD THEORIES: AN INTRODUCTION

X
=
 [ e
 [
Example 1.1. Recall the integral expression for the -function:

XY 'Z @ \^])_ 4`Kba D ] @dc \f]Gg/_ 4`Kbaih D ] J (1.5)

This integral is convergent for Re MZ kj O but can be meromorphically extended to the

whole plane with poles at ZlRSmn \ . We have

XY 'Z HdoB @ ZBXY MZ p

XY oB @ o
q
Xr co @ts u J
By substituting ]
@ s v ]Gw in (1.5), we obtain the Gauss integral:
 [ D @ XY @zy u J
[  4x/aih ] v ` 8g` g v
(1.6)

 v v
formula (1.6) make sense, as a Riemann integral, for any complex with Re 0{ O .
Although in the substitution above is a positive real number, one can show that
v
v
When Re Ej O this is easy to see using the Hankel representation of XY 'Z as a
v
contour integral in the complex plane. When is a pure imaginary, it is more delicate
v @|u , we can use
and we refer to [Kratzer-Franz], 1.6.1.2.
Taking
D~} @ D ]
~€‚aih
to define a probability measure on 1 . It is called the Gaussian measure. Let us compute

>  9ƒ @  [ WB† D} @  [ W WB† D Q J

the integral

[ 7L…„ [  € h)‡ L6„

Here ƒ
@ o7ˆ . We have  
> 9ƒ @  [ )F u Q g Ž [ i’“ƒQb”  ˆ‚•—– D Q J
[Š‰p‹Œ ‘ \

[
Obviously,

Q g™˜ `  š€ W h D Q @ O J

[

[
Also

W D @ H o ™
ˆ u @ ž
o B
 ž
Ÿ K
 
  c ›ŠF oB @
Q Q XY 9› c ˜ ‡œ c‚u ˜
[ g˜  € h h

 3
c
u c › #›– – @V  c › ™ˆ c‘u ˜

˜ g˜
  c › @ c c › #–– @ o –¡ c c;› ¢  ¡ c ›£c F c ¢ K ¡ cc7¢
where

˜› › c
is equal to the number of ways to arrange › objects in pairs. This gives us

> Mƒ @ orH Ž [ GF o7  ƒ g    M¤ •¥ ˆ c •¥ #– c‘u ”  J

‘ (1.7)

c `
arrange ¤ • objects in pairs is the same as to make a labelled 3-valent
graph X with • vertices by connecting 1-valent vertices of the following disconnected
Observe that to

graph:

b1 b2 b 2n

c2 c
c1 2n

a1 a2 a 2n

Fig. 1

X %
This graph comes with labeling of each vertex and an ordering of the three edges
9 9X X ¨
@ c §¦ MX % 9X @
c •—
¦§¦§ M MX X @ Ÿ~•
emanating from the vertex. Let be such a graph,
Ÿ%
be the number of its vertices
and
•
be the number of its edges. We have
for some . Let
, so that

© 9X @ ) Fž% ’ªƒ « # !– ¬ c‚u o ? !¬ J

MX
> Mƒ @ orH Ž © MX p

Then

¬
where the sum is taken over the set of labeled trivalent graphs. Let be the number e­ 9®
© ¯ M® @ A­ M® © 9X
of labelled trivalent graphs which define the same unlabelled graph when we forget
­A 9®
about the labelling. We can write , where
® is the number of

> 9ƒ @ o°H Žp± © ¯ 9® #

labelling of the same unlabelled 3-valent graph . Thus

where the sum is taken with respect to the set of all unlabelled 3-valent graphs. It is

c p– Ÿ-–² 

easy to see that

­A 9® @ ³ ¥• Aut M® g
4 LECTURE 1. QUANTUM FIELD THEORIES: AN INTRODUCTION

±
© ¯ M® @ Gc Fž¥• ’“ #ƒ – c‚ g u  c  •¥³ #– Ÿ-–² g  @ c‚u ) F? ¤‘± ’“ ƒ ³ «   J
so that

” Autc 9® Aut M®

c‘u , then multiply •

GF¤~’ , toofeach
number edge a factor o7ˆ
Given an unlabelled 3-valent graph with vertices, we assign to each vertex a factor
all the factors and divide by the

contribution of this graph to the coefficient at ƒ . For example, the graph

g
symmetries of the graph. This gives the Feynman rules to compute the

contributes )F¤‘’   †
g g €` ` ` g @ F´µ €” † and the graph
contributes )F¤‘’ g   † µ
@ F·¶ † J The total coefficient at ƒ is F † . This
coincides with the coefficient g €` ` at ƒ g in`G¸> € 9ƒ given by the formula (1.7).g `G¸`)€¹
Recall that the Principle of Stationary Phase says that the main contributions to the
integral
~º WB D
7L6»B¼ &½ iQ Q
when ¾ goes to infinity comes from integrating over the union of small comapct neigh-
borhoods of critical points of ¿À iQ . More precisely we have the following lemma:
Lemma 1.1. Assume &½ 9Q has a compact support and ¿À iQ has no critical points
%
% •
ÁÂ6à [ ¾   [ WB &Å 9Q D Q @ O J
on . Then, for any natural number ,

»7ĕ  [  L6»B¼ @ O . Integrating by parts,

Proof. We use induction on . The assertion is obvious for •

’  [ WB7Æ &Å 9Q w D Q @ ’ Æ &½ iQ WBÈÈ [ [ H  [ WB &Å 9Q D Q @

we get

¾ [  L6»B¼ ¿À 9Q w#Ç ¾ ¿À 9Q w/Ç  L»7¼ È [  L6»B¼

  
[
6W D J
[ BL6»7¼ &½ iQ Q
Multiplying both sides by ¾
 , we get
‡Å`
ÁÂ6à [ ¾   [ WB &Å 9Q D Q @ ’ Á6Âà [ ¾   [ W7Æ &½ iQ w D Q J
½
‡ ` [  6
L B
» ¼ [  
L 7
» ¼ ž iQ w#Ç
»7Ä  »BÄ 
Applying the induction to the function Æ 56WWBi É w we get the assertion.
¼ Ç
 5

¿À iQ Q
KJJKJ4
QÊ
¿À 9Q ¿ 9Q ` Ë Q
Thus if has finitely many critical points , we write our function

Í
Ì
as a sum of functions
i
Q L
with support on a compact neighborhood of
½
& i
Q L L
•Îj O
and a function which has no critical points on the support of and obtain, for
Ð
 [ 6W98/: D @ Ž <Ï°Ð WB98/: D HeÑ  pJ
any ,

&½ 9Q Q 7L6¼ &½ iQ Q M.

[ 7L…¼
 L @ D HVo , where D
Now let us consider a QFT in dimension 1. Usually we write (
o @ O HV@ o is the
quantum mechanics. In this case, we take to be equal to , Ò
is the space-dimension, and is the time-dimension. A QFT in dimension
 1 * O , or  ` 1—ˆ m
K

o
parametrized by ] . A map Ó
< Ô
 Õ
  is path in  (infinite, or finite, or a loop). The
action is defined by
@ 
ž ] ×Ö ] #
šØ ] ) D ]

)

where
ÙfÚ1 is a smooth function defined on the tangent space of  (a La-
grangian). Ö The expression D ;
‘ Û
Ù =
 ¨
 Ö Ù  ] p
šØ ] ) D ] is a density on  equal to the composition
For example, take 
of the differential
@ 1  so that and .

Ù Ö  @ 1 ÝÜ 1  with coordinates 9Þ
Þ Ø . For
v 
any 9Þ
Þ Ø and a map ß
š*
à , á1 , 9ß° ] p
ßYØ ] ) is obtained by replacing Þ with
Ö Ø Ø
Þ withpointßY ] of. the functional ž iß°Ö ) satisfies the Euler-Lagrange equation
ß° ] Aandcritical
â p
Ø ) @ ] D â #
Ø ) #J
âãÖ iß° ] ßY ] D ] â ãÖ Ø 9ß° ] ßr ] (1.8)

For example, let us take theL Lagrangian L

Ž  ã Ø F % ã
KJJKJ
㠏
 Lg ` (1.9)

L`
› D g D QÅ] ] @ F0ä % 9QÅ ] ™ #J
Then we get from (1.8)

g
Thus a critical path satisfies the Newton Law; it gives the major contribution to the

 ]
@]Gw Q R
G  @ Q åæ ]
Q¥ç ]Gw
Q w
+
PQ  Q w R
partition function.
Fix and . Let be the space of smooth maps
such that ] ]Gw è w
. The integral

Ëe ]
Q¥ç ] w
Q w @   W‚é Ʌ WBÉ!  L 3 !ê a iM8: (;* ] , (1.10)

a a that a particle in the position Q at the

moment of time ] moves to the position Q w at the time ]Gw .
can be interpreted as the “probability amplitude”
6 LECTURE 1. QUANTUM FIELD THEORIES: AN INTRODUCTION

%
Let us compute it for the action defined by the Lagrangian (1.9) with . We  @ 1
The space

åæ ] QÅç ] w Q w

shall assume that the potential function is equal to zero.
is of course infinite-dimensional and the integration over

* ]
]Gw , @ ­ åæ ]
QÅç ]Gw
Q w @ ƒ @ ]Gw F ] ™ˆ ­ ­
such a space has to be defined. Let us first restrict ourselves to some special finite-
dimensional subspaces of . Fix a positive integer and subdivide the
time interval into
] ]
]
KJJKJ
]G ë
]Gë
equal parts of length by inserting inter-
G
] w Q @
mediate points
@
Q
Q g
KJKJJ4
Q ë
Q ë` Q g w * 1
, ‡Å`
. Let us choose some points
ì
Û* ]
]Gw , q1 ` 
‡Å`
in and consider the path such that
its restriction to each interval ] L ] L ‡½`
is the linear function

] @ Q H Q ] L ‡Å` FEF Q] L ] F ] #J
L L L ‡½` L  L
It is clear that the set of such paths is bijective with 1 ë
a function Ì
åæ ]
Q¥ç ]Gw
Q w lí1 over this space to get a number ` and soî weë . can integrate

define (1.10) as the limit of integrals î ë when ­ goes to infinity. However, this limit
Now we can

may not exist. One of the reasons could be that î ë contains a factor ï ë for some
constant ï with ð ïÍð jño . Then we can get the limit by redefining î ë , replacing it
with ï ë î ë . This really means that we redefine the standard measure on 1 ë
   
D D
 the measure Q on 1 by ï 4` Q . This is exactly what we are1  going 4to`

do. Also, when we restrict the functional to the finite-dimensional space
É À
N ëØ D ]` byof
replacing

piecewise linear paths, we shall allow ourselves to replace the integral ò

its Riemann sum. The result of this approximation is by definition the a right-hand a Ö side
in (1.10). We should immediately warn the reader that the described method of giving
a value to the path integral is not the only possible.

 [ JKJKJ  [ * ’ª› ëŽ
We have

Ë= ] Q¥ç ] w Q w ë [ @ Á  6
 à c ƒ  iQ FÎQ g ,óï  ë D Q g JKJKJ D Q ë J
Ä  [  [ ‰p‹Œ L ` L L ‡Å`
 D
Here Q
KJKJJ4
Q ë are vectors in 1 and Q is the standard measure in 1 . The number

(1.11)

ï shouldg be chosen to guarantee convergence L in (1.11). Using (1.6) we have

[ [
6W W  W W †  D Q @  g Æ W † Ç h W W †  D Q @
[ x œ  h h#4x h  h g [   x h kô œGõ h ô köh œ  h g
 [ 
@ W W †   Wg D Q @ y c u v W W †  J
  öh œ  h [   x h   öh œ  h

[ * v
Next

F c 9Q FÎQ ” g F v iQ ” FÎQN÷ g , D Q ” @
[£‰p‹Œ `

 7

@  [ * F Ÿ c v Æ Q F Q H Ÿ Qb÷ g F vŸ iQ FEQb÷ g , D Q @ y c‚Ÿ uv † W Wù™ J

[ø‰K‹Œ ” ` Ç ` ”  ö œ h

[ * v
Thus

F iQ FÎQ g g F v 9Q g FEQ ” g F v iQ ” FÎQ ÷ g , D Q g @

[ú‰K‹<Œ `

y c u v y c‚Ÿ uv * F vŸ iQ FEQb÷ g , @ y Ÿ u v g * F vŸ iQ FEQb÷ g , J
‰K‹<Œ ` g ‰p‹Œ `
 [ * v ëŽ
Continuing in this way, we find
u v
F  iQ FÎQ g , D Q g JJKJ D Q ë @ y ­ v ë ë 4` * F ­ 9Q FÎQ ë g ,

[Š‰K‹Œ L L ‡Å` 4
 ` ‰K‹<Œ ` ‡Å`
 v @ ˆc J L `
› ’“ƒ If we choose the Ðconstant ï equal to ï @¨û ˜ Ð hœ
then we will
g € L6„ü
where
be able to rewrite (1.11) in the form

Ë= ]
QÅç ] w
Q w @ýû c‚u › ’“­+ƒ hœ þ  ô Éiÿ ô " h @úû c‚u ’# ›] F ] hœ  þ  ô É É9ÿÿ ô " h J (1.12)
  

ü h w ü h "
We shall use Ëe ]
Q¥ç ]Gw
Q w to define a certain Hermitian operator in the Hilbert
g  1—
. D}Recall that for any manifold  with some Lebesgue measure
D} the
space g
space
Ö
Ö to zeroconsists
tions equal
of square integrable complex valued functions modulo func-
on the complement of a measure zero set. The hermitian inner

¿ @  ¯¿ D~} J
product is defined by



 

Example 1.2. An example of an operator in g 

D~} is a Hilbert-Schmidt operator:

Ö
Ù ¿À iQ @  Ëe 9Q
- ¿À - D}

where Ë= iQ
- R g Ü } Ü }
 
is the kernel of Ù . In this formuladwe integrate
Ö Ë= iQ
-


keeping Q fixed. By Fubini’s theorem, for almost all Q , the function

}is -integrable. This implies that Ù M¿ is well-defined. Using the Cauchy-Schwarz 

  
inequality, one can easily checks that

ð6ð Ù ¿—ðð g @  ð Ù ¿—ð g D~} ð6𠿗ðð g   ð Ëe 9Q

 ð g D~}D}

 
8 LECTURE 1. QUANTUM FIELD THEORIES: AN INTRODUCTION

Ù
 
i.e., is bounded, and

6ð ð Ù ðð g @  \ ð6ð 6ð ٠𠿗¿—ð6ð ðð g

 
- D}D~} J
  ð Ë= iQ ð g


¼ Œ g 

   
We have


)Ù ¿ @  û  ¿À - Ëe iQ
- D}


iQ D} @   Ë= iQ
- ¿À - 9Q D}D} J
 

ü
This shows that the Hilbert-Schmidt operator is self-adjoint if and only if

Ë= iQ
- @ Ë= 
Q

outside a subset of measure zero in 

Ü .
In quantum mechanics one often deals with unbounded operators which are defined
only on a dense subspace of a complete separable Hilbert space  . So let us extend
the notion of a linear operator by admitting linear maps  where is a dense (  (
Ù
linear subspace of  (note the analogy with rational maps in algebraic geometry). For
( Ù
Ù
such operators we can define the adjoint operator as follows. Let denote the
domain of definition of . The adjoint operator  will be defined on the set Ù
( Ù @ R
\  W k š ð Ù ð6 iðQ QÀ# ð
ð ð J
  
   !

Œ

  
 

Take R ( Ù 

. Since ( Ù
is dense in  the linear functional

Q  Ù i Q #

R Ù i Q #
@ Q

extends to a unique bounded linear functional on  . Thus there exists a unique vector
Ù
( Ù Ù p @ Ù Ù ( ÙÛ @ ( Ù K
"
  $"%
# such that . We take " for the value of  at . Note that
&
is not necessary dense in  . We say that is self-adjoint if
Ù &

( Ù Ù ( ÛÙ

and . We shall always assume that cannot be extended to a linear operator on
a larger set than . Notice that cannot be bounded on since otherwise we
Ùú
@ 
can extend it to the whole  by continuity. On the other hand, a self-adjoint operator
Ù
( ÙÛ
  is always bounded. For this reason self-adjoint linear operators with

@ g 1
D Q and define the operator
('
 are called unbounded linear operators.
Example 1.3. Let us consider the space 
Ö
Ù ¿ @ ’G¿ w @ ’ DD ¿ Q J
Obviously it is defined on the space of differentiable functions with square integrable

1
D Q 1k
D Ùt
~(Š
derivative. This space contains the subspace of smooth functions with compact support

¿;R
which is known to be dense in
self-adjoint. Let (
. Since 
Ù Ö g ¿wR Ö g Q
. Let us show that the operator
,
 is

 D @ 
¿\ a w 9Q ¿À iQ Q ð ¿À ] ð g F|ð ¿À 9O ð g F \ a À¿ 9Q ¿ w 9Q D Q
 9

] GF /
H ] * Á6ÂÃ [ ¿À ] ð ¿À iQ ð g
aÄ
is defined for all . Letting go to ) , we see that exists. Since

¿
R (

is integrable over , this implies that this limit is equal to zero. Now, for any

  

Ù ¿
 @ Á Â6à [  \ a ’G¿ w 9Q 9 Q D Q @ Á6Âà [ û ’G¿À ] 9Q ÈÈ ‡ [[ F  \ a ’G¿À iQ w 9Q D Q @
, we have

È

   

aÄ  a Ä  ü
@ ÁÂ6à [ \ a ¿À iQ ’ iQ D Q @ M¿
™Ù  pJ
w +

-, aÄ
This shows that ( t( Ù K and Ù is equal to Ù on ( . The proof that (
 @ ( Ù  

be two copies of the space g 
D} . Let Ù  É be the Hilbert-Schmidt
is more subtle and we omit it.

operator `  g defined by a kernel Ë= Ö ]

Q¥ç ]Gw Q w which has a a
]Gw as real parameters:
Let . .

` g Ù '& iQ @  Ë=
QÅç
Q &½ iQ D~} J
. .
]
É  ] ]w w w 
aa
Suppose our kernel has the following properties:

 É
(M)

Ëe ]
QÅç ] w
Q w²w @ a  Ë= ]
Q¥ç ] w
Q w Ëe ] w
Q w ç ] w
Q w D~}  D ] w
] ] w ç
a


(N)

ð Ë= ]
Q¥ç ] w
Q w ð g D}  @ o ç

Ë= ]
Q¥ç ] w
Q w @ Ë= ] g
Q¥ç ] wg
Q w if ] wg F ] g @ ] w F ] ç
(T)

(C) for any &

R
`  `
D~} , the function ` `

/
gÖ
] ] @  9Q Ëe ]
Q¥ç ] w
Q w &½ 9Q w D}  D~}
Â6Ã
10 / 32

is continuous for ]Gw j ] and

6
Á É ] @ 0
& pJ
0  / 

When Ë is defined by the path integral, a Ä property

ai‡ e
Å* (M)
] ]Gw , is2taken
 from

¥* ]
]Gw ,   from Q to Q w and a path
¥* ]Gw
]Gw ,Q 2to Q w isfromequalQ w toto
as one of the axioms of
QFT. It expresses the property that any path

Q w . Property (N)` says that the total probability amplitude ofg a particle to move from Q
a sum of paths

to somewhere is equal to 1. Notice that property (N) implies that the operator Ù  É is
unitary. In fact,
 Ù KÙ D} @ aa
 a a É & a a É 
/
10 LECTURE 1. QUANTUM FIELD THEORIES: AN INTRODUCTION
 û 

D} û 

D} D~} @
  Ë= ] QÅç ] w Q w &Å 9Q  ü  Ëe ] Q¥ç ] w Q w iQ  ü 
4/

 û 



D} 
D } @  D~} J
 &½ 9Q  Ëe ] QÅç ] w Q w Ëe ] Q¥ç ] w Q w  ü iQ   &Å 9Q 9Q 
/ /

Theorem 1.1. Let ] p

] R 1 \ be a family of unitary operators in a Hilbert space
Now we use the following Stone-von Neumann’s Theorem:
76
5


Á6ÂÃ all\
ÌÍR ] @ , the
function
. Assume that

]  Í
Ì ] @
Í ]  is continious for ] j O and
(i) for 8
9
:
 ;
9
;8 5
49

(ii) for aall Ä ] ‡

]Gw R 1 \
] H ]Gw @ ] ]Gw pJ
;8
6 =<
5 5 5

( @ SR
ÁÂ6Ã \ Í ] FÓÒ exists

Then
> 5 !

is dense in and the operator defineda Ä by‡

8 
] 8

@ ’ ÁÂ6à \ ] FÓÒ


aÄ ‡ ]
.?8 8

] @ 7L a
] { O J
is self-adjoint. It satisfies
@
5

Ù  @  
] { ]\
Applying this to our situation, we obtain that
@

 L aªba is called the Hamiltonian operator asso-

A

a a
A

ciated to Ëe ]
Q¥ç ]Gw
Q w .
for some linear operator . The . operator .

Ëe ]
QÅç ] w
Q w @ýû c‚u ’# ›] F ] hœ Æ ’'›Óc i Q ] w F FÎ] Q g J
We would like to apply the above to our function

w ü ‰p‹Œ w Ç
Unfortunately we cannot take the function Ëe ]
Q¥ç ]Gw
Q w to be the kernel of a Hilbert-
Schmidt operator. Indeed, it does not belong to the space g 1 g
Q Q w . In particular
D D

Ö (T) is obviously true and
(C) is true if one restricts to functions & from a certain dense subspace of g 1— .
property (N) is not satisfied. One can show that (M) is OK,

Ö
/

The way about this is as follows (see [Rauch]).

1
defined on the Schwartz space Û 1— 1—  of smooth functions with all derivatives
First let us recall the notion of the Fourier transform in . It is a linear operator

tend to zero faster than any power of ð Q½ð as Q

B
Ö g C,

 [ . It is given by the formula

M&½ iQ ) ž
@ &À @ s o c‘u  L W &½ 9Q D Q J



D -E H

[
GF

Here are some of the properties of this operator:

 11

D
 1° 0  1° is an unitary operator;
` M&½ 9Q ) @ M&½ )FžQ ) ;
(i) B B
D D

(iii) '&Å 9Q w
@ ’ &½ ™ ;
(ii)
D E

(iv) &À w
@ GFž’ªQš&½ 9Q ) ç

F ;F
E #D

k @ s c‚u M& , where

;F

(v) '&
D / KD D /

[
JI

& 9Q @ &½ iQ F - - D J / 4/

[
LI


Let us show that our function Ëe ]
Q¥ç ]Gw
Q w is the propagator for the Schrödinger equa-
tion
â â
’ â ] H › c â Q g ½ ]
Q k
@ ’ H › c WBW @ O
½ MO
Q @ ¿À 9Q R Ö g 1— #J
g @ o . Supose a ¿À 9Q R Û 1° . Let us find the solution in
8
8 M8 8

We take for simplicity ›

 1— using the Fourier transform. Using property (iii), we get @ F ` ’ g (we
NB
E E

use the Fourier transform only in the variable Q ). Integrating this equation a with g initial
B 8
F 8

condition À 9O

@ ¿À , we get ½ ]
@ 8 ¿½ #J
La h g
OE H E

Taking the inverse Fourier transform, we  get

E E
8 F GF 8 F GF

@ @ o  [
½ ]
Q 4`   L a h 8 g ¿½ ™ s c‘u  kœ L a h™‡ L W ¿À D J
#D H E H H E

[ h
8 (1.13) ;F ;F F

Clearly, À 9O
Q
@ M
¿ @ ¿À 9Q pJ Of course,  we have still to show the existence
of a solution. We skip the check that formula (1.13) gives a solution in  1° . This
PD D
8

 ` B

 ] k
Û 1—  Û 1° #
¿À iQ r ½ ]
Q #J
defines us a linear operator (the propagator)

We would like to show that it is an integral operator and find its kernel. Let Ë= ]
Q
@
B B 8

4` `   L a h 8 g . Then
D H

 g [ €
 - D @  [ û  [ o
Q

Ëe ] QÍF ¿À s ‘
c u 8 g W  D ¿À  D @ H R H

[ [   La h  L  ü

[
F

[  û [ o  
- D W 8 g D @ 8 g ¿À ) @ À ]
Q #J
s ‘
c u À
¿ R H H H D H E

[ [ L  L   La h `   L a h

F ;F 8

  ü
8 not belong
to Û 1° . A way about it is to consider this function as a distribution   L a h g does
H
Unfortunately, this computation is wrong since the function
B and extend the
Forier transform to distributions.
12 LECTURE 1. QUANTUM FIELD THEORIES: AN INTRODUCTION

ï [ —1 \
¿ 1
Recall that a distribution is a continuous linear functional on the space of
smooth functions with compact support (test functions). Any function which can be
integrated over any finite closed interval (but not necessary over the whole ) can be

[ D

considered as a distribution. Its value at a test function is equal to

¿À M& @ À¿ 9Q &½ 9Q Q
[
where the bar denotes the complex conjugation.  Such a distribution is called a regular

shall denote the value of a distribution ¿ on a test function & by

distribution or a tempered distribution. The rest are called singular distributions. We

@ [ D J
¿À M& &½ iQ —¯¿ 9Q Q
[

If ¿ is a regular distribution defined by a function ¿À iQ from 1— , then
g
¿À M& ¿ &  @
p J 
Ö
An example of a singular distribution is the delta-function  iQ°F whose value at a test
v
v
function & is equal to &½ . It is also denoted by . A linear operator Ù
~(Š
S

g 1—
with ï
[ —
1
\ = (
, (
NT Ù p
&
extends to the spacex of distributions by the formulaÖ
S

Ù ¿À '& @ ¿À Ù & #J 

If ¿;R gÖ 1— , viewed as a regular Ù ¿À M& @ Ù &

¿ @ &
)Ù ¿
distribution, we have
    

For example let Ù

so the two definitions agree.
@ @ W be defined on the space of functions with@ square integrable
Ù F Ù and for any distribution ¿ , ¿ w M& ¿À )F0& . If ¿ is a
VU
U
&

tempered distribution defined by an integrable differential function ¿ such that ¿ w also

derivative. We have

defines a tempered distribution, then the formula of integration by parts shows that this

transform is an example of an operator defined on  1° with

definition agrees with the usual definition of derivative.
D @ the, weFourier
Since
D
transform of a distribution ¿ by
B

4` can define the Fourier



D
'¿ M& @ ¿À 4` M& ) pJ
D

tion of a regular distribution ¿ and an element & of Û 1° by the formula

All the properties (i)-(v) extend to distributions. In property (v) we define the convolu-

¿ M& @ ¿À k& pJ
B

/ /

v with Re v k{ O
v @ O ,
LI I

Lemma 1.2. For any R '

 x W h @ s o c v   W h 8 ÷ x J
XW

D
 13

Proof. Assume first that Re v Ij O

. Then the function W
 x h belongs to  1° and, B
using the Gaussian integral, we obtain
 [  [ Ð
D
o
 x W h @ s c‚u  x W h   L W D Q @ s c‚u  4x W ‡  h   h 8 ÷ x D Q @H o Y H

[ [ hö
 
os ss uv 8 ÷ @ s o 8 ÷ J
c‚u   h x c v   h x H H

Therefore, for any &;R ï

[ 1— \ ,
@ o  [
D
 4x W h M& s c v   h 8 ÷ x &Å D J H [Z

[
GF F

Consider the both sides as functions of . When Re Îj O each side is a holo-

v@ v
morphic function, and, for Re { O

v v O , are continuous functions. The unique

'

v v @ O . This proves the lemma.

for Re ž{ O

continuation principle for holomorphic functions implies that the two sides are equal
'

Ë= ]
Q @ ` s o c‘u   L a h 8 g @ s o c‘u   L a    h 8 g @ s c‚o u ’  L ô h J
Now we can use the lemma to set
D H #D H


] h
[
Property (v) of Fourier transform gives us

ž ] ¿ Ëe ]
Q k¿ Ëe ]
QÍF - ¿À - D @
@ @

[
I


[ o ÿ @  [

- - D J
¿À - D Ë= Q ¿À
[ s c‘u ’ ]  L  ô h " h [ ]
 \


Thus we see that the integral operator with the kernel Ë= ]
Q
-
 @ Ë= ]
QIF - p
] j O

is well-defined as an operator ž ] on the space  1° . Now observe that

ž ] ¿ @ À ]
Q @ `   L a h 8 g ¿À ) @
] ] ]
8
] ] D H E
GF
]

] H
8 g ¿Å @ ¿À @ ¿ J ] ] ] ] ]

  La h
E E
GF GF

This shows that ž ] is a unitary operator. In particular, ž ] is bounded on  1° (of

norm 1) and hence continuous. It is known that  1À is dense in 1— . Thus we can
B

Ög B
14 LECTURE 1. QUANTUM FIELD THEORIES: AN INTRODUCTION

extend ž ]
by continuity to an unitary operator on the whole space
Ö g
1— . It satisfies
the property

'ž ] ¿ @   L a h 8 g M¿ pJ
D H D

ž ] ¿ @  L a ¿
] j O
Using property (iii) we get
@

@ F W J This justifies the claim that Ëe ]

QÅç ]Gw Q w is the kernel of the operator
ž ]Gw F ] @   L h ha É ša  on Ö g 1° .
U
where . U
@

Finally let us try to justify the following formula from physics books:

Ë= ]
QÅç ] w Q w @ Q½ð   L  a É ba  ð Q w (1.14)
 @


First of all for any &

from a Hilbert space , physicists employ the bra-ket notation
/

&Àð k
@ &
#J


 /^ 
/^

If Ù is a linear operator in , then

&Àð Ù ð
@ &
™Ù pJ


 /_  `/_

Let & be a normalized eigenfunction of an operator Ù with an eigenvalue . Physicists

denote it by ð (although it is defined only up to a factor of absolute value one). To
ba c

c

Åð6ð Ù ðð } @ ½ð Ù ð } #J
simplify the notation they set
   
c c

Consider an operator (the position operator)

 1— Û g 1— #
¿  Qš¿ J
d

Ö 1— B

It is a self-adjoint operator Û 1— 
g . Its eigenfunctions do not belong to the
space g 1— but rather to the space of Ö distributions.
B

WeÖ have

'& @ & @ 9Qš& @ v &½ v @ v '& pJ

d d

x x x x v
S S S S

Q;R  1 É x we have, according to physicist’s notation, ð Q @ W . Now we have to

compute
d
Thus can be considered as an eigendistribution of with eigenvalue . Thus for any
S

. Recall that we can view it as an integral operator with kernel Ëe ] QÅç ]Gw
Q w


W
S

  L a ša on the Schwartz space  1° which obviously contains ï [ 1— \ . We have

S

 [

àp D @

àK #

defined B

Ë= QÅç Q  iQ F Q Ë= Q¥ç
[ ] ]w w w w ] ]w
S


 15


ð   L  a É ba  ð @ 7ð Ë= ]
Q¥ç ] w
àK
@ 9Ëe ]
QÅç ] w
v ) @ Ëe ]
v ç ] w
àp pJ
 


v x @ Q
/à @ Q w we get formula (1.14). Wex have to understand it as
S Sfe Sfe S

Taking

W   L  a É ba  W É @ Ëe ]
Q¥ç ] w
Q w #J
S
@
S

For any function
 

% and a point ] R  one can consider a function on the
set Map 
 


 W

% * ] ,ª '& @ % M&½ ] ™ #J
defined by

% %
Let
KJKJJ4
 be functions on  and ]
JKJKJ¥
]  R  , one can consider the integral
` % *
JKJKJ¥
%  *  
@  ` % * JKJJ %  *  657 ( J
, ], 64  ` ] ` ,ª '& ] ,ª '&  L 3 &
` ]`
 

The right-hand side is called the path integral with insertion functions
KJKJJ4
 . The
% %
Map

•
left-hand-side is called the correlation -function. In the example above`

W * ] ,
WBÉ * ] w , @ Ëe ]
Q¥ç ] w
Q w #J

S S


Exercises

> 9ƒ @  [ W W ù D Q J
1.1 Find the Feynman rules to compute

[   )h ‡ L…„
Compute the coefficient at ƒ .

g
1.2 Show that the distribution Í ]
Q
-
@ Ëe MO
] ç™Q
 (defined to be zero for ] O )
g

 

F co WW @  ]  iQæF - #

is a generalized solution of the equation

a
8 b8 S S

1.3 Show that, for any { O , the function W is a generalized eigenfunction of the
(you have to give the meaning of the right-hand-side).

—
1 L
a

operator ’ W in
Ö g and any generalized eigenfunction
c
U
U coincides with one of these

1.4 Find the Fourier transform and the derivative of the Dirac function  9Q§F .
functions.
v S
16 LECTURE 1. QUANTUM FIELD THEORIES: AN INTRODUCTION
Lecture 2

Partition function as the trace of

an operator

Recall that the trace Tr Ù Ù

of an operator in a finite dimensional Hilbert space  is
Ù
 
K 
J K
J 4
J

equal to the sum of the diagonal entries of a matrix of with respect to any basis. If

`Ù @ Ž  Ù
pJ
we choose an orthonormal basis , then

Tr


 L L

(2.1)

If Ù is a normal operator (e.g. Hermitian or

% L ` unitary), then one can choose an orthonor-
mal basis of consisting of eigenvectors of Ù . In this case

Tr Ù
@ Ž D

 b
hc c (2.2)
a  Sp 

where Sp Ù is the spectrum of Ù (the set of eigenvalues) and is equal to the

D ic

Ù @
dimension of the eigensubspace corresponding to the eigenvalue . Notice that c

j k l  J
U a

‰  b
c
a  Sp 

@ Á ) pJ
This gives

Tr Ù
m j


k 

‰
(2.3)

dimensional Hilbert spaces. We shall briefly discuss them. First assume that Ù is a
There are several approaches to generalize the notion of the trace to operators in infinite-

One chooses a basis

JKJKJ4

KJKJJ and sets
bounded operator. First we try to generalize the definition of a trace by using (2.1).

`  Ž [ Ù
p

Tr 
Ù @ 

 L L


L 17`
18 LECTURE 2. PARTITION FUNCTION AS THE TRACE OF AN OPERATOR

Ù
if the series convergent. If the convergence is absolute, then this definition does not
@

depend on the choice of a basis. In this case is called a trace-class operator. For
 D}
Ö g ç
example, one can show that Tr Gn(o Tr hopn if both n and o are trace-class. An

Ëe 9Q

example of a trace-class operator is a Hilbert-Schmidt operator in the space .



D} J
If is its kernel, then

Tr Û
Ù @
 eË 9Q Q
Ù
When is a self-adjoint Hilbert-Schmidt operator, the two definitions coincide. This

D} be@ a finite

follows from the Hilbert-Schmidt Theorem.

³  set o~
JKJKJ
™• equipped with the measure
D} @

1 with inner product
 !

Ö g
Example 2.1. Let Gn


n . Then


&
@  & ¯  D~} @ Ž  v¯ à


/^ q/

@ v K

K
J 
J 4
J
v #

@ à 

K
J K
J 4
J /

à L ` L L
-
where & @  /  . It is clear that Ëe 9Q can be identified

with a matrix Ë ` and `

s

L Ù &½ 9’ @ Ž  v

hr

 L r s s
s

so that Ù is a linear operator defined by the matrix Ë . Then its trace is equal to
`

Ù @ Ž  J
Tr
 s
r s4s

This agrees with definition (2.1) when we take the standard orthonormal basis of 1 .
` 
tors in g 

D~} like a differential operator. One tries to generalize definition (2.3).

As we have already mentioned, in physics one deals with unbounded linear opera-

Notice Ö that

Á ÙÛ @ Ž  Á @ F D  D  _ È  \ J
 L L Z` L È _
tm`j k m
u c

‰
hc

`
Now for any Ù such that L has a basis of eigenvectors of Ù one can define the
zeta-function of Ù as follows. Let O JKJJ  JKJJ be the sequence of


positive eigenvalues of . One sets Ù `[


c
g 
c

c


MZ @ ~Ž  ›  

v

_ c

where ›  is the multiplicity of  , i.e. the dimension of the eigensubspace of eigen-

`
@ g 
D~} , where  is a compact manifold
vectors with eigenvalue  . When
c

of dimension and
D Ù c 
Ö
is a positive elliptic differential operator of order w , one can show
 19

v
MZ MZ j D ˆ
O
that  is an analytic function for Re w and it can be analytically extended

Ù @   É  \  J
to an open subset containing . In this case we define
j k x y

‰ %
This obviously agrees with (2.3) when is finite-dimensional. Also it is easy to see

Ù @  \  ÙÛ #J
that for any positive number c

j k x y j k

‰ ‰ @ -Âà % .
hc c (2.4)

This of course agrees with the finite-dimensional case because 9O

v j

Example 2.2. Consider the operator Ù

@ F W which acts on the space g ' , where 

@ °
1 ˆ ‚
c u D hh
U

measure ‚ m with the usual measure Q descended

‘
c u Ö that` in this
U

`D the length of  is equal to , i.e.  is the circle

w to the factor. Note

Q corresponds to the choice ` Ù of metric  onWB8 the

)• ` circle determined w of radius . The measure w

R m
by its radius. The

consists of numbers •Åˆ g

)• R m g € \ with
normalized eigenvectors of are
`  L ›  @Vc . Thus Q . The positive part of the spectrum
z

[
w C| {z

v
'Z @dc ‘Ž  •Åˆ  g_ @tc g_ c Z p

 w w
v

`
where MZ is the Riemann zeta function. It is known to be an analytic function for
c
§
j 7
o ˆ . This agrees with the above since  is one-dimensional and Ù is an
v

Re MZ
`
MO @ F co
w MO @ F co Á c‘u J
elliptic operator of second order. We have
v v m

w 9O @ F c Á c‘u #

Thus
v tm
 w

w GF D D Q g @ c‚u g J
and

g
det (2.5) w

Example 2.3. Let us consider the path integral when 

@ 1°ˆ c‘u m and  @ 1°ˆ c‘u 7m . 3}
w

o  g DJ
We use the action

 @
ž c \ € w ] g ]
~

A map ;
Pá extends to a map of the universal coverings E
~1|á1 . It satisfies

] @ ] H c‘u rH c‚u • for
/some integer • (equal to the degree of the map of


 be the set of maps corresponding to the same • .

3}

oriented manifolds). Let Map

  w

It is clear that each R Map 

 /


  can be uniquely written in the form
•
] @ ] HÓ \ ] @  ] ÅHÓ \ ] #
}
w
20 LECTURE 2. PARTITION FUNCTION AS THE TRACE OF AN OPERATOR

\ ] satisfies \ ] H c‘u 3} @ \ ] , hence belongs to g k . The value of 

where
Ö
o  g
on such is equal to

4 @
ž c \ €
~

\w ] g H c \w ] “ w ] ¥HÝ w ] g D ] J
 g \ ª D @ •  g \ D @ • \ c‚u \ ™ @ J
We have

\ € w ] w ] ] \ € w ] ] ì F 9O O
~ ~
w w 3}
} }

u 
ž  @ • g g H co \ g € \w ] g D ] J
Thus
~
w
}

> @  !ê‚ ( @ Ž h ªh  º !ê‚ ( ÀJ

The Minkowski partition function is


  7L 3  €L   7L 3

ƒ

Map ‚ Map
{€

 º ê7 (Í @ 
c‘u
D

Observe that we must have

64  7L 3  Ëe MO Q¥ç Q Q

3}

Map

Ëe ]
Q¥ç ] w Q w @ c‚u ’#o F  L W É  W h 8 g  a É ba 

where

]w ]
„

to be consistent with the previous computation of the path integral.

>d@ Ž os  D Q @ s Ž
This gives

J
  € L h h c‘u ’ 
 

’  €L h h
ƒ w ƒ
} }
‚ ‚
{€ {€

Ž W  @ s o Ž  8W J
Now we apply the Poisson summation formula

  š€ h Q   š€ h
Taking Q
@ ˆ ’ , we get {€ {€

> g
@ s Q Ž  W @ Ž  8™W @ Ž
}

 8 J (2.6)
w
~

  € h   € h   € L h h

}
;w
z

@
{€ {€ {€

@ g 
~:…

eigenfunctions in g  are the functions    L €

h h . Itswenormalized
~K@
Let us compute the trace of the operator
W78  . € ByL (2.1), … 

Ö `  L
/

g
have Q

@ Ž  ð h  @€ Ž 8 J z

Tr g
~K@  / ~ … {z ~

L €   € L h   š€ L h ™h …  / 

{€ {€ z
 21

>d@ Tr g p

Comparing this with (2.6), we see that

~K@

@ F . U
L €
where .
` h
Remark 2.1. Ifg wea h repeat the computations for the Euclidean partition function (replac-
U

ing  with ’G ) we get

>?
@ s Ž  8 @ Ž  8 J ~ ~

  € h ™h   š€ h ™h
} w
w }
{€ z {€ z

> ?
@ s > ?
ˆ pJ
This shows that
} w } }
w } w

If we modify the partition function by inserting the factor o‚ˆ

s , we get
> ?
@ > ?
ˆ pJ
w

} } }
w w

This is the first glimpse of the T-duality.

9¾ @ Ž  € L  h »
¾ R @ Q H ’ R
æj O J
Let †

 { !
?5 W

be the modular form associated to the quadratic form iQ

@ Q ( equal to the value at
{€

g
d

GF o‚ˆ ¾ @ )Fž’ª¾ œ i¾ pJ
zero of the Riemann theta function in one variable). It satisfies the functional equation
† †

h
> ?
@ 9’ ˆ g #J
(the proof uses the Poisson summation formula). Observe that †

} }
w w

This is our first encounter with the theory of modular forms.

>d@  û F co  \ g € Q w ] g D ] (
º
There is another way to compute the partition function for the action

Q½ ] p

4  ‰p‹Œ ü
where 
Map
}

 g  g D @
is the circle of radius . Notice that

\€
~
D @ È
Q w ] g ] QÅ ] Q w ] È \g € F \ € Q½ ] Q w²w ] ] F~
~

QÅ ] p
Q½ ] w!w   J

‡

h3œ
>?Ý@  º W~ ª W~ 

Thus
ˆ @

    a a
G‰
Map
22 LECTURE 2. PARTITION FUNCTION AS THE TRACE OF AN OPERATOR

@ F . The integral U

g` a h h  ~W  ª šW~  ( #

where . U

  a a QÅ ]
ˆ
 G‰

@ @ Gaussian integral since QÅ ] #

™Ù QÅ ] is a
% Ö ' ` . If  and Ù is a positive-definiteÙ self-adjoint
% 1


can be thought as a generalization of the
quadratic form in g
 º  D JKJJ D  @  º W
operator, we could use the orthogonal change of variables to diagonalize and write

Q Q ˆ‹Š Š
W D Q JKJKJ D Q  @
   œ hœ   h `
 ‰ a a  

Ð Ð

` 
ŒŽŒŽŒ

  [ W D @  y u @ u  8 @ u  8 @ o Ù J
 [  h QL 
l a
  sg l

g Ù ‰ u žhœ  „ j k
j k

L` L L` L ‰
c c

L `
@ ß
™Ù ß is positive definite. L To get rid of u let us change the measure on 1
Here we assumed that all eigenvalues are positive, or equivalently, that the quadratic c

D D
d  

replacing Q with
` gº € Q so that
form
Q

 s u  D Q JJKJ D Q  @ Ù J ˆ‹Š Š j

  ` ‰  hœ
 ‰ k


v ø
Ù

can write any element &eR  &  , where '&  is an orthonormal basis
Now, for any normal positive definite operator in a Hilbert space , we   
as a sum
v
of eigenvectors of Ù . The coordinate  is an analog of the Q coordinate from above.
 u

  ( @ o 8 JL
This motivates the following definition

ï «  det w u
4` g
ˆ‘ ‘ d
’ ‰ “9
(2.7)

Here the measure (+* , is defined up to some multiplicative constant ï . In fact we will
9

be defining the correlation functions by the formula

% ] #
JKJJ
%  ]  @£ò % ] K %  !ê‚  ] (+ *  L 3 ê7 (;* ,

` ` ò  L3 ,
 

` `
@ F in g 1Yˆ c‚u . However, not all of its eigenvalues are positive.
so the choice of the constant will not matter. We would like to appy this to the op-
3}

h h formÖ the nullspace of this operator. If we decompose

U
erator . U

v a
as a sum    of normalized eigenvectors, then coefficients  will be analogs
Constant functions
v each vector

of the coordinates in 1 . So, we can write

u
 /

@
v \ O our space as the product of the space of
o at \ @ o‚ˆ s c‚u is equals c‚u to s c‘u s .c‚u Thus@t. The
constant functions and functions with coefficient of the constant function

D ˆ c‚u s J
/ 3} 3}

functions is equal to ò \ g
€ >d@ c‘u Q c‘u s @ s So, using (2.5), we obtain
the integral over the space of constant
3} }
w

4` ˆ J
{z
3} } }
w w
 23

This agrees with the computations in example 2.3 if we switch from the Minkowski
partition function to the Euclidean one.

  ] Ö @ ‘
ÙÚ
] 
H 1 ]
Here is another application of the Gaussian integral for quadratic functionals. Con-
sider the action functional defined by some Lagrangian . We know
Y
]
”
that its stationary points are classical solutions. Write S , where
”
is a classical solution. Then

  @ ž ¥H co g  ÈÈÈ ê  ê  g H terms of higher order in ½J

” S

g
GS S
S q•


This gives a semi-classical approximation:

i’Gž 4 ˆ . )(+* , Ž i’Gž ˆ . p* c‚uo ’ g  ÈÈÈ ê  ê , œ J

” j k S

‰p‹Œ ‰p‹Œ ‰ g h
—–
S •
classical solutions

Example 2.4. We take 

@ 1°ˆ c‘u and  @ 1 . The Lagrangian is
This approximation is exact when the action is quadratic in .

ã Ö
ã Ø @ co ã Ø g F g ã g
Ö ™˜

>  @  +( * ] , û c’  Ø ] g F g g ] D ] J
and the Minkowski partition function is

p‰ ‹Œ 
š˜

ü
  is given by
It is called the path integral of the harmonic oscillator. The kernel of the operator
A @

  L aªša  ê   W (;* û ’  Ø

\

@
Ë= ] ç ] Q ê a  œ  ] , c  ] g F g g ] D ] J

` ‰p‹Œ
š˜

a
A 3R
ü
Choose a critical path cl for the action defined by our Lagrangian and decompose
the action in the Taylor expansion at cl .

ž ] ) @ ž ¥ H co g  ÈÈÈ ê  ê ] F #J S

g cl cl (2.8)
S cl

D â â ã â âã @ H @ J
The classical path is a solution of the Lagrangian equation:

D ] ÖØ F Ö ] w g ] O ˜

Its solution satisfying the initial condition ] \

@ š
) ] @ Q is

  `
] @ û Â 00 ] ] ` F F ] ] \ H Q û Â 00 ] ] F F ] ] \ \ J
 tm  tm
˜ ˜
 m  tm

` ü ` ü
cl
˜ ˜
24 LECTURE 2. PARTITION FUNCTION AS THE TRACE OF AN OPERATOR

û g H Q g 0 ] F ] \ F c Q
The value of the action functional on the classical solution is

@ o  D @
ž c a œ Æ ] w g F g ] g ] J
›[œ 

c  0 ] F` ] \

˜ ˜
cl cl ™˜ cl
Ç
 m
ü
a of the action functional `
A ˜

g  ] ™ @  a œ ] D D g H g ª ] D ] J
The second variation is
S

g S

a
A
]g ˜

Ëe ] \
ç ]
Q @ i’Gž ) Ü
Thus we can rewrite (2.8) in the form


 ê   W (+* û ’  ` ‰p‹Œ ) D g H ) D J

cl

ê a  œ  ] , ‰K‹<Œ F c a œ ] F ] D ] g g ] F ] ] ü
3R
cl ˜ cl

a a @
A A

Now let us make the variable change replacing ] F ] with ] . The limits in the
path integral change to ] \ @
] \O . @The paths
@ ] F ] \ satisfying @
cl

with the period Ù

` ] we have] O . Using the generalization of

we integrate over are periodic

 ê   ` \ (+* û ’  ` D g H “ D @
the Gaussian integral to functional integrals

ê a  œ  \ , ‰p‹Œ F c a œ ] D ] g g ] ] ü ˜

 a º (+* a
A A

, )
F c o  À
’ ( N @ c‘uo ’ (æ
j

  ‰p‹Œ  hœ
3‡ k

Map
h ‰
( @ F DD g F g J
where

( ]g \ @ @
™˜

 • u ] ˆ7Ù , where • Rtm \ and Ù @ ] F ] \ . The ] corresponding

] O areeigenvalues
`
The eigenfunctions of satisfying the condition the functions
@ • u ˆ‚Ù g F g . We know that`
 tm

equal to 
&| are

[ [ [
c ™˜

j k
c‘uo ’ ( @ ‘ c‚uo ’ ) u •Åˆ7Ù g F g @ ~ c‘uo ’ • u ˆ7Ù g ‘ o F • g Ùu g #J
l l l ˜

‰ g g
š˜

`
Of course here we use a “physicists’s argument” since we don’t have the right to write ` `
the product as the product of two infinite products , one of which is divergent (see
the next remark for an attempt to justify @ the argument). Now we use that the first
product corresponds to the action with ˜ O . So to be consistent with our previous
û [ c‘uo • u ˆ7Ù g  œ @ Ëe ] \
O-ç ]
O @ s c‚ou Ù J
computation we must have

‘ ’ ’
l

ü h `
`
 25

&½ ] *] \
] , &½ ] \ @ ½& ] @ O F ` Wh U

g€L h
Note that if we compute the product using the zeta function of the operator U

[ ` `
on the sapce of functions on satisfying we get

~
l
c‚uo ’ • u 7ˆ Ù g @ c‘u ’ Ù ï J
`
ï @ ï s c‚u
The two computations disagree. The way out of this contradiction is the choice of the
normalizing constant which we used to define the Gaussian integral. It shows that
we have to choose . Now we use the Euler infinite product expansion for the

 ٠@ [ o ٠pJ
sine function:

Ù ‘ F • gg u g g
 tm
;˜ l ˜
˜

`
Ëe ] \
ç ]
Q @ i’Gž ™ c‚uo ’ (æ ` 8 g @
From this we deduce that

j k

` ‰p‹Œ ‰ cl

i’Gž ) c‚u ’  ٠` 8 g @ Æ u œ s o  F L -8 g 9’Gž ) pJ

˜ ˜
Ÿž

‰p‹Œ Ç h   g L ‰p‹Œ
cl  tm cl
ž

Let us rewrite ž in the following form

˜

ž ] ™ @ ’ c û iQ g H g oo HF   g L F o QF   L @
cl

ž ž£
i˜  

 gL   gL ü
cl ¢¡
ž ž

’ c Æ iQ g H g o H c Ž [ g  F Q o H Ž [ g  @
‘   L   L ‘   L Ç
i˜ ¤ ž ž ž
¡

`[ `
Ž
’ c û F| iQ g H g ÅH c Q g H c g F Q  J
h˜ ¥
‘ \ L  L ü g
¡
ž ž£

Using this we obtain

Ë= ] \
ç ]
Q @ Æ u œ s o  F L -8 g Ü

˜
¦ž

` Ç h  gL ž

Ž [ c H c
§
F c F| iQ g g ‘ \ Q g g F Q   L   g L  @
˜û H 
H ¥ ž ž
¨

‰K‹<Œ ü ¡

Ž [ š  i  iQ
- @ N8 ˆ u šW 98 HdJKJJ4J
  L g   h)‡ h g
„

‘   L ‡ hœ
ž G  ž ž R
n ˜

`
26 LECTURE 2. PARTITION FUNCTION AS THE TRACE OF AN OPERATOR

Now recall that the kernel of a Hilbert-Schmidt unitary operator can be written in

Í iQ
- @ Ž  ¯  iQ   p

the form

/ 4/
g c
{€

where  is the normalized eigenfunction with the eigenvalue  . In our case, the
¥   . Thus the eigenvalues of the
/
c

Î
• H
@

Hamiltonian  are  L u 0 ÷ ` s . We shall  see L in‡Ûthehœ lecture that the eigenvectors of

  ©ž
eigenvalues of must be equal to

 @
iQ ˆ `\ 8  g \ -° Q@   W h ˆ -u 8 g
where
. ˜
8 are@ the\ Hermite
. are
• @ /

O , we get 9Q
©˜
š 
 W 9
.
i
Q
- #
J ˜ polynomials. ž
.

  hG‡ h g
„ R
/ 4/ ž

When This checks the ˜ n
first term.

Exercises
® ±
O @  g  JJKJ
MZ Z O @ O
 

MZ `  _
2.1 Let c c be a non-decreasing sequence of positive real numbers.
v  6 provided that this sum converges for Re `ª
[ @ É \  J `
Define u c and has
a meromorphicx continuation to the whole complex plane with no pole at . Set

L  that  for any ­ {to , [  @  KK [  .

«
c

L (i)` Prove 
ë L ë ‡Å` L


(ii) Give a meaning to the equality

L ` – L @ s ‘
c ` u .
g c ic c c c

2.2 Let Ù
@ 1  ˆ X be a • -torus. Here X @ m HVJKJJ m  and
JKJJ
 are linear 

independent vectors in 1 .
 ` @  â Ð`
+˜ ^˜ ˜ ˜

D}
D} is induced by the standard volume formÖ on 1 F  . L  ` Wg in Ö g Ù0
where
(i) Compute the trace of the Laplace operator šu

> ? Ù for the maps from  @ 1°ˆ c‘u

to Ù with the action defined by ž 
@ ò  ðð w ] ðð g D ] , where is any lift of
3}
(ii) Compute the Euclidian partition function
|
1 U
 1  g ` ¬ 

@ 
1
 QARÓm for all QARÝX . Use the Poisson summation
to a smooth map .
(iii) Let X R
 ­ Ÿ !

Ž ¿À iQ @ Ž ¿½  p

formula
E

W ‘¬ 4` ‘¬ 
n
R  %®

 >
where ¿+R  1 , ¿ is its Fourier transform and
@ *
KJJKJ
 , , to relate
? > ?
Ù and Ù K .
E j k

‰ `
:B n ˜ ˜

the partition functions

9Q
- and g 9Q
 from Example 2.4 to find the eigenfunc-



tions 9Q and 9Q . `
2.3 Compute the terms n n

g
/ /

`
Lecture 3

Quantum mechanics

The quantum mechanics is a O Hdo dimension QFT.

Let us recall the main postulates of quantum mechanics. A quantum state of a
system is a line in a separable Hilbert space  . It can be represented (not uniquely) by
a vector / of norm 1. To each observable quantity (like position, momentum or energy)
one associates a self-adjoint operator n in  (an observable). A measurement of an

GF
-,
observable n depends on the given state / and is not given precisely but instead there

to ¯—° ic / ç @ 6ð ð   k1 6ð ð g
° hc
4/   k
is a probability that the value belongs to a subset
, where ° ic is the spectral 
c . This probability is equal
function of n , an operator-

  
valued measure on . In the case when n is a compact operator,  has an orthonormal
   @ ²±
n  

basis of eigenvectors of n with eigenvalue c . Then ° ic u a  a  ,

  ð 0
  ð g
²±
where  is the orthogonal projector 
operator to the subspace W . Thus, for any
simple eigenvalue c of n , / 
can be interpreted as the probablility that the
observable n takes value c in the state / .
ð
0
Åð }
often writes c  for /

& ð
In physics literature one a norm 1 eigenvector a of n with

½ðð6ð }
 /   /   
eigenvalue c and rewrites ba in the form c . Also one writes c instead of
o
 
c .
0
 


The probability amplitude (a complex number of absolute value ) is defined to

&

 /  / 
be c . The function c c is called  the wave function of the state / with
respect to n . The inner product of two states
/^ is interpreted as the probability
&
amplitude that the state changes to the state / . Its absolute value is the probablity
of this event. Note that by Cauchy-Schwarz inequality, this number is always less or
equal to 1 and it is equal to 1 if and only if the two states are equal (as lines in the
Hilbert space).

@ 
@ =[ Q D}  —

The expectation value of n in the state / is defined to be

[ @  
 ³  /
/_ ³
n n (3.1) °

D~}  is the measure on 1 defined by D}   9¦ M¦ 

.
@ g 1r
D Q and @ is the position operator corresponding
³ ³  4/
/_
where ° ° °

Ö coordinate Q . It is defined by M¿ @ Qš¿ . @ This is an un-

d
Example 3.1. Let  n
d

bounded self-adjoint linear operator. We know from Lecture 1 that ð

to the measurement of the

(although c Sfa

27
28 LECTURE 3. QUANTUM MECHANICS

iQ #
@  i Q D Q @ pJ
they do not belong to the space  but rather to the space of distributions). We have
 /  / /
S´a S´a ic

. So, in the realization of as Ö g 1

D Q , a state iQ is interpreted as the wave
d /
The probability amplitude of the value of the observable in a state is equal to c
/ /

function of the state with respect to the observable . Any R

ic 
/ d /

@
can be written as µ

D J / /
hc S a c

&
Of course this has to be understood as the equality of distributions. For any test function

[
û  [ D ½& iQ D Q @ [ û [ D D @
we have

/
&½ iQ Q
/

[
[ [ [
hc S´a c ic Sfa c

  ü   ü
[ D @ M& #J
&½ / /

[
hc hc c

 d

@ º
The expectation value of is equal to
 d 
³
½ð ð g D J
c
/
hc c

Ö g 1
D Q ™ à
x
Consider the delta-function
d
S as a state (although it does not belong

to .
Then the probability of to take a value in the state S is equal to S Sfe  . The inner
x W  8 x
product is of course not defined but we can give it the following meaning. We know

g ` €‘a   x h g a ]
that S is equal to the limit of tempered distributions Q
x
when tends to

k
@ Á6Âà \ s c‘o u W  8 g @ Á6Â6à \ s c‚o u   8 g J
zero. Thus we can set


]    x h a a Ä ]   x h a
 e

x
S S e S e

à @ a Ä
Q is a variable, we get

W @ Á6Â6à \ s c‚o u W  8 g @ @  9Q§F v pJ
When


x @ 1°a Ä ˆ c‚u ]  4x h a x


S S S S

g  @ ’ W .
@ 
¿  `   L WB8 with eigenvalue • . We have
3} U
Example 3.2. Let
Ö  and equal to the momentum operator n
~
U

g€
Its eigenvectors are the functions Q ~

o 
iQ p
¿  @ s c‚u \ g € ¯ 9Q   L  WB8 D Q @ v¯ 

~
 / ~
 /
3}
 29

@m •   ¯   •+ v v¯ 
on Sp •
is the -th Fourier coefficient of / . So, the wave function of / is the function
. The probability that takes value at the state / is equal to ð  ðg .
  @ Ž • ðv  ðg J
The expectation value is equal to



 ³

{€

Ð
The dynamics of a quantum system is defined by a choice of a self-adjoint opera-
tor . , called the Hamiltonian operator. In Schrödinger’s picture the operators do not

@ $ 0J
change with time, but the states evolve according to the law
@

a  a
/ /

Here . is a fixed constant, the Planck constant. Equivalently, ] is a solution of the /

’G. D D ] @  ] #J
Schrödinger equation:

/ f/

a
.

In Heisenberg’s picture, the states do notÐ change with Ð time but the observables evolve
I ] @   $ a    $ a J
according to the law
@ @
n n

D
We have the Hamiltonian equation:

D ] ] @ * ] #
S, :

n (3.2) n .

where
*
l, : @ . ’ Ð F #J
n o Gn
<
o o
<
n

@ $ and hence the corresponding state

a to the line spanned by ) doesa not  change a with time, i.e. iQ describes a
/ / a /
If is an eigenvector of , then
.
/ /
(equal
stationary state. Usually one measures observables at the stationary states of . .

There are two ways to define a quantum mechanical system. One (due to Feynman)
@ 
D~}
uses the path integral approach. Here we take  as in Lecture 1 and
define the Hamiltonian by means of the path integral. The choice here is the action Ög
functional. It is defined in such a way that its stationary paths describe the motions
of a classical mechanical system. Another approach is via quantization of a classical
@
ÅÙÛ  1
mechanical system. Recall that the latter is defined by a Lagrangian
D * Öv
/à ,
 #J

which, in its turn, defines an action functional on the space Map

ž  @ p
šØ ) D J
Ö ] ] ]
e

x
A critical point of this functional defines a motion of the mechanical system. The
equations for a critical point are called the Euler-Lagrange equations. If one chooses
30 LECTURE 3. QUANTUM MECHANICS

ã ã ã
ã
Åã Ø @ ã K
JKJJ4
㠏 ç @ ã Ø
K JJK` J4
K
¥JKã JØ  J4
in Ùin  (so and
local coordinates ã corresponding
that Ø
the
@ , the local coordinates
S s

` ` â D â @
@ L o
KJKJJ4
)•—J L
equations look as L¶
¶¸·i¹

Ð âšãÖ F D ] â ãÖ Ø O ’ (3.3)

ã ] p
ã Ø @   . For example, L atL a path ]  ] p
Ø ] ™ in ÙÛ given by ã @
L
Here the left-hand side is evaluated


Ù 
U

L L W a a·U if we assume that the restriction of to each tangent

Ö
metric on  . A critical path becomes a geodesic. Ö
space is a positive-definite quadratic form, we can use to define a Riemannian


is a function on the cotangent bundle Ù K .

Another way to define the classical mechanics is via a Hamiltonian function which

Recall that any non-degenerate quadratic form

% on a vector space

% defines a d

a linear map 
% . If we% view
d

bilinear form, and4hence ` on% theas dual


quadratic form space a quadratic form as a symmetric

4`
% isUequal to the maximum`(if j O ) or the minimum

 d

see that, for any ÝR

, then is the inverse map. Let us
1 defined by
 d d

O ) of the function Ì
d
º , Gº

Ì  @  F  #J
(if ²»

9 9 d 9

% @ % @ 1 
and  @  , then  @
» º

 @  for some
 9 f9 d 9
If we choose the coordinates so that º º

äg` Ì +F  @ symmetric O , hence @ matrix4` and

9 %9
n . Thus to find an extremum we must have n
q9 9

Ã Ì ¥  @  +F c  @ co  @ pJ o
²» º n we get n º

«‹
“¼ d

` 4` 4` 4` %

9
‘ ²» º n º =º n º 3º n º hº


for any function ¿ on such

d

that its second differential is non-degenerate. This is 4called ` the Legendre transform of
Using this one can generalize the construction of

¿ . By definition, for any AR % , 

Leg '¿
@  FÓ¿À < p

9 9

@ D ¿ , where D ¿
% ^1 is
hº º

the differential of ¿ at the point R

where is the implicit function of defined by
% . In order that this function be defined
º º
‘ ‘
9

D g ¿À  ) @ O .weIn
general, the implicit function is a multivalued function, so the Legendre
j k 9 p'

‰
have to satisfy the conditions of the Implicit Function Theorem:

is defined only locally in a neighborhood of an extremum point of the function  F

9
Gº transform

¿À We .
9
º
9

the local coordinates in the cotangent bundle Ù p by Ö

shall apply the Legendre transform to the Lagrangian function . We denote


ã
@ ã
KJKJJ4
㠏 ç
KJJKJ
 #

KJJKJ4`
 are Ùtaken ` to be the dual of the coordinate
¯ G¯ ¯

functions
ã Ø K

K
J 
J 
J 

ã Ø
 in the tangent 
` bundle
where the fibre coordinates ½¯ ¯

KJKJJ4
` in Ù W . The Legendre
and can be identified with a basis
Ҧ

œ
¶ 

Ž  Ö
transform of is equal to

A ã
@  ã Ø F Ö ã
ã Ø p

¶¸· ¶¸·

L` LL
. ¯ ¯
 31

ãØ K
JJKJ4
 defined by the equation
L @ ââ ãÖ Ø ã
Å`ã Ø #
’ @ o~
JKJJ4
)•
where are the implicit functions of ½¯ ¯

Ù ÚU1 L is calledL the Hamiltonian associated to the Lagrangian

¯

Ö

The function . .
As we have explained before, in order it is defined the Lagrangian must satisfy some
conditions.

]
Using the Hamiltonian one can rewrite the Euler-Lagrange equation for a critical

Ø
@ D D @ F â A⚠ã ã

ã Ø
@ D D ã @ â Aâ ã
J
path of the action defined by the Lagrangian in the form:

] L UÙ K L
 L #
] L
¯ . ¯ . ¯

L
¯

Î
satisfiesL the above equations

¯

Ò ] ] which
ã

<


&
Ù K Õ1 . The projection
after we compose it with the coordinate functions
Here a solution is a path  

 (i.e. the composition with the projection map Ù KÚá )


¯



is the path ] describing the equation of the motion. The difference between the
of the path to the base


is a second order ordinary differential equation on ÙÛ and the second one is a first
Euler-Lagrange equations and Hamilton’s equations is the following. The first equation

order ODE on Ù p which has a nice interpretation in terms of vector fields.

&

of its tangent bundle Ù J Let ï [ Ù denotes the set of vector fields.

Recall that a (smooth) vector field on a smooth manifold is a (smooth) section
[
¾

obvious structure of a vector space. For each smooth function &dR|ï

F ¾

[ Ù by the formula
¾
It has an

differentiate & along R ï

one can ;¾

( M& iQ @ Ž â â Q &

F ¾

L
F

where QdR
K

K
J 
J 4
J

9Q Q  are local coordinates in a neighborhood L L of Q , and are
the coordinates of` - iQ R Ù W W K


J K
J 4
J
W of Ù W .L We
¿¾ F
F with respect to ¾ the basis ¶ ¶  ¾

œ
¶ ¶

( M& @ D &½ p

also have
H ;F

where we consider smooth 1-forms as linear functions on vector fields. This defines a

(ý
ï [ Ù — End Mï [ ™ #J
linear map

It is easy to check that ( M&

@ ( M& ×H§( k & , so that the image of ( lies in the
¾ ;¾

subspace of derivations of the algebra ï

[ . Given a smooth map e
Å* v
à ,  ,
H / H /

G¾
H /

¾
and a vector field we say that satisfies the differential equation defined by (or is
F F

D
@ D  Å â â @ - ™
an integral curve of ) if F

D] v
for all eRÎ
/àp #J
] » F hº º

tation in terms of the canonical symplectic structure on the manifold

@ Ù  J
The vector field on the right-hand-side of Hamilton’s equations has a nice interpre-
¾

32 LECTURE 3. QUANTUM MECHANICS

PR g which is a non-degenerate bilinear form on each Ù W . If we view W as

Recall that a symplectic form on a smooth manifold is a smooth closed 2-form ¾
˜ ÁÀ
Ù G¾
W  Ù W @ Ù W , then its inverse defines a linear isomorphism ;¾ ˜


-Ù  Ù W . Varying Q we get an isomorphism of vector bundles ÙÛ 

 

Ù p i Q and byW the pull-back

a linear map ¾ ¾ ¾
 
ž ¾ ¾


@ ï [ Ù ° ï [ Ù #J
of sections an isomorphism of the space of sections

`  1 , its differential D Ì is a 1-form on , i.e., a

 

Given a smooth function Ì

ž À G¾ ¾ ¾

section of the cotangent bundle Ù . Thus, applying

¾ ¾

D Ì of the tangent bundle, i.e., a vector field. It is called the Hamiltonian

 Â
we can define a section ;¾

@
ž

Ù 
 with
Â

Ì
vector
ã ã
ž

coordinates
field defined by


the
and Ì is the Hamiltonian function A
function .
¯
We apply this to the

situation
when
. We use the symplectic . ¯
¾
&

@ Ž Dã D J
form given in local coordinates by

L L
˜ ¯

b
Ù
R W 9 Ä L ŸÃ

W š
 @ Ž D ã < D  F Ž D ã  D < pJ
For any G¾

9
Ä 9 Ä Ä 9

L L L L
˜ ¯ ¯

L L
In particular,
â
â @ â
â @
W âã âã W â â O for all ’
N
˜
s
˜
¯ s
iÅ

L â â L â â
¯

W âã Ð
â F W â Ð
âã @ J
@ S s

D @ L
D ã @ F L
hence L
˜ ^˜
¯ s ¯ s

This shows that  VÂ

Lâ â L â â â â
ž ¯ ¶ ž ƶ

D @ û Ž âã D ã H Ž â D @ Ž GF âã â H Ž â âã J

¶¸· ¶[Ç

  . . . .

L L L Lto theL ü Hamiltonian

ž . ž ¯

L L vectorL field L  L defined L

¯ ¯ ¯

So we see that the ODE L corresponding È G.

 M¿ @ ¿
J
by is the vector from the right-hand-side of Hamilton’s equations. We have
.
É !
È G. .

Let
be a symplectic manifold. For any two functions ¿
R|ï
;¾ ˜
[ one

G¾

¿
@ 0 D ¿ p
D - ) pJ
defines the Poisson bracket
  ! Â
 

By definition of we have 0 ¿ p

D @ D ¿À , so that
˜ ž ž

 Â

¿
@ D ¿À D - ) @ D  M¿ @ F D ¿  pJ
ž ˜ ž F ;F
  ! Â  Â  Â 
ž ž ž
 33

The Poisson bracket defines a structure of Lie algebra on ï [ satisfying the addi-
G¾

¿
@ ¿
×H ¿
šJ
tional property :
 Ê !   ! Ê  ËÊ ! 

ã showã (Darboux’s Theorem) that it is always possible to choose local coor-

dinates
KJKJJ


KJJKJ4
 such that
One can

` ` @ Ž Dã D J
¯ ¯

L L L
˜ ¯

¦Ã

â â â â
¿
@ Ž âšã ¿ â F â ¿ âã pJ
In these coordinates
 
  !

L L L L L
¯ ¯

@ ã
ã @ O
ã
@ J
For example,
 ! > !  !
¯ s s ¯ s S s

L L L L
¯ (3.4)

D¿ @ ¿
J
The corresponding differential equation (= dynamical system, flow) is

D] 
.
!

A solution of this equation is a path ;
N*

à , á such that
v
D ¿À ] ) @
) pJ
D] ¿ ] 
.
!

This is called the Hamiltonian dynamical system on ã (with respect to the Hamiltonian
function ). If we take ¿ to be coordinate functions
on
@ Ù  , we obtain ¾

Hamilton’s equations for the critical path ;
N*

à , ´ L in
.
v L  . ¯ ¾


The flow of the vector field ¿ 

a by the formula ¿ is a one-parameter group of operators

on æ  defined
  !
.

a M¿ @ ¿ iQ @ ¿À ] )
5 Ì

a a
5

where ] is the integral curve of the Hamiltonian vector field ¿  ¿

with the
@
 !

initial condition MO Q , The equation for the Hamiltonian dynamical system defined
.

D¿ @ ¿
J
by is
.

D ]a  !

a
(3.5) .

Here we use the Poisson bracket defined by the symplectic form of  .

¿A. . RÝThisï [ assignment

a self-adjoint R .properties.
A quantization of a mechanical system is defined by assigning to any observable

¼@ Á6Â6Ã
;¾ operator This operator may contain a parameter
n 
must satisfy some natural For example:

 : \ *
,:
¼ Ä ¼
n&Í Î[Ï (3.6)
n n Î
34 LECTURE 3. QUANTUM MECHANICS

Under the quantization the Hamiltonian function of the mechanical system becomes a
self-adjoint operator . , called the Hamiltonian operator of the quantized system. We

  @  @ : Á6Â6Ã \ *
, : @ : ÁÂ6Ã \ *
, : J
have
@ @ @

¼ ¼ ¼ as the quantized ¼ action of the Hamil-

nÑÐ nÍ Ï n n n .

Thus the linear map T*

, : is interpreted Ä Ä
tonian vector field  . The analog of the dynamical system (3.5) is the Hamiltonian
n n .
È h.
equation in quantum mechanics (3.2).
ã system is given on the configuration space Ù 1

with coordinate functions


For example, when a mechanical

L@ L Ð   @ Ð @
we need to assign some operators to the coordinate
¯

’ o
KJJKJ4
)•—J
functions:
d

By analogy with (3.4), weL should haveL

n n
· Ç

* 
  , : @ *
, : @ O
*
  , : @ d d d

 
s s s S s

L @  Hilbert L space % andL operatorsL

R E % satis-
(3.7)

% 1
d

Ö g
 andã define L L
So we have to find an appropriate 
fying (3.7). We take
â
& &
 
&  ’“. âã & J
d

L self-adjoint L L operators. L
Recall that these are unbounded
d  
L ofL quantization of a classical mechanical system given by a
The operator (resp. ) is called the position (resp. momentum) operator .
Let us give an example

Ö ã
ã Ø @ c—o i› ã Ø g FΛ g ã g ) p

harmonic oscillator. It is given by the Lagrangian

µ˜

where › is the mass and is the frequency. The corresponding Hamiltonian function
˜

A ã
@ ã Ø F co i› ã Ø g FE› g ã g @ co › g H › g ã g #

is
¯
. ¯ ¯ :˜ :˜

where we used that

@ @ › ã Ø to express ã Ø via . The function e
ã can be
¯ ¶
‡
¯ . ½¯
viewed as the total energy ¶=
of· Ò the system.

DgQ @ F Q J
The corresponding Newton equation is

D ]g g ^˜

shall assume for simplicity that ›

@ o . The Hamiltonian operator can be written in
So the motion does not depend on the mass but the total energy does. In the sequel we
.

@ co   g H c g g @ v<v F rc . @ v /v H . c

the form
˜ d Ó ˜ Ó Ô˜
. (3.8)
 35

v @ s o c H ’   #
v @ s o c FΒ  
where
d Ó d
;˜ ©˜ (3.9)

are the annihilation and the creation operators. We shall see shortly the reason for
*
  , @ Fž’G.
these
d
names. They are obviously adjoint to each other. Using the commutator relation

* v
v , @ * v
v , @ O
q* v
v , @ .

, we obtain

*
v , @ F0. v
q*
v , @ . v J
Ó Ó Ó
Ԙ (3.10a)
Ó Ó
. Ԙ . Ԙ (3.10b)

This shows that the operators o~

v v form a Lie algebra  , called the extended .

Ó

gÖ 1—
Heisenberg algebra.
So we are interested in the representation of the Lie algebra  in .
v
v
Suppose we have an eigenvector / of . with eigenvalue c and norm 1. Since is
adjoint to , we have
Ó

½ðð ð6ð g @ 0

c
/  /
.
/_ @ 
v v ¥H 0
. c @ ðð v 6ð ð g H . c 6ð ð ðð g J
 / Ó /^  / Ԙ /_ / Ԙ /

This implies that all eigenvalues c are real and satisfy the inequality

{ .c J (3.11) c
Ԙ

The equality holds if and only if

v @ O . Clearly any vector annihilated by v is an /

eigenvector of with minimal possible absolute value of its eigenvalue. A vector of

.

Denote a vacuum vector by ð O . Because of the relation *

,
norm one with such a property is called a vacuum vector.
v @ F0. v , we have


v @ v FÓ. v @ ÍFÝ. v 0J
. Ԙ

/ / / /

v is a new eigenvector with@ eigenvalue FP. . Since eigenvalues

. . Ԙ ic Ԙ

This shows that


/
v O for some •P{ O . Thus
v v @ O c Ԙ

v
are bounded from below, we get that
and /
is a vacuum vector. Thus we see ‡Å` that the existence of one eigenvalue of is
/ /

v 
Now if we start applying to the vacuum vector ð O , we get, as above, eigen-
equivalent to the existence of a vacuum vector.

vectors with eigenvalue : Ý

H •
Ó 

. . So we are getting a countable set of eigenvectors

g
Ԙ

 @ v  ðO / Ó 

with eigenvalues 
@ g  . J It is easy to see, using induction on • that that
/ @ Ԙ

6ð  ð6ð g •—– '. J After renormalization
ðeigenvectors
c
g ‡Å`
Ԙ
we obtain a countable set of orthonormal

ð • @ M. o  •—– v  ð O #
• @ O
o~
c
KJJKJJ
=
„
Ԙ
(3.12)
Ó 
36 LECTURE 3. QUANTUM MECHANICS

1— ð•
One can show that the closure of the subspace of
an irreducible representation of the Lie algebra  .
spanned by the vectors = is
Ö g
The existence of a vacuum vector is proved by a direct computation. We solve the

s c v @ FE’   @ ã EH . D D ã @ O
differential equation
/
/ d / /
©˜ ˜

and get

ð O @ . ùœ   h 8/: J
 ˜ ž
(3.13) ·

In fact, we can find all the eigenvectors

ð • @ Æ . u ` 8 ÷  s 㠈 s . p

= ˜

Ç
. ˜

 9Q @ s c o  •—– Æ QÍF D D Q    ô h
where

Ç h
.

•
functions  iQ
ÿ is complete, i.e., forms an orthonormal basis in the Hilbert space
is a Hermite polynomial of degree . It is known also that the orthonormal system of

g 1— . ð O Thus ô h • an irreducible representation of with unique vacuum

we constructed
h
.

Övector p J
•§H ` . .The vectors ð are all orthonormal eigenvectors of with eigenvalues

 =
.

g function (3.13) gives the probability amplitude that a particle occupies the
Ԙ

position Q on the real line in the vacuum state of the system.

The

According to Example 3.1, the value of the function ð • at is equal to the proba-
bility that the observable takes value at the state ð • .
=
c
d =
c

of are 
@ •ÍH . and their multiplicities are equal to 1. So
Finally let us compute the partition function of the Hamiltonian . The eigenvalues .

g`
. c Ԙ

@ Ž [   M: @ ã

Tr
 L a ‘ \  L a ‡ hœ o F hœ ã
@
՞
(3.14)

ã@ : J
 La
՞
where

Exercises

ð • ) Ù 1—
3.1 Consider the quantum mechanical system defined by the harmonic oscillator. Find
 
ã Å

ã Ø @ i` › g F on , where ã @ O
ã
the wave function of the moment operator at a state = .
3.2 ã Consider the Lagrangian Ö Ø ã
for R MO
v and
ã @
5
o otherwise. g Quantize this mechanical system, solve the
5 5

Schrödinger equation and find the stationary states of the Hamiltonian operator.
 37

À
¿ i
Q @ W.
3.4 Let
@ F id g be the expectation value of the operator F
3.3 Compute the Legendre transform of the function

id (the dispersion of an observable at the state ). Prove the Heisenberg’s

³    ³  ³
Ö hn Gn n Gn

g
  ³ /
n n

Î{ c°o ð *
×, ð J
Uncertainty Principle

³ ³  
³
Ö Gn Ö Go n o
38 LECTURE 3. QUANTUM MECHANICS
Lecture 4

The Dirichlet action

Â6à zjzo
 @ Ù Ü­ Ù P1  ­
j
Now we shall move to QFT of dimension larger than 1, i.e. , for example,

where ,
0
Ù

or
`
and is a manifold of positive dimension which we
&
<| @ o~
KJJKJ
)•
&½ ] Q ] R
QÓ RÝ­ ­ O ­
shall assume for simplicity to be orientable. A map is given by a function

M & ` ] p
KJKJJ
&  ] ) ž
ÙÔ´ 
× !

&½ ] Q  
. Note that when is -dimensional, say , we
can view as a vector function and get the quan-
1 o~
K JJKJ1
)• Ü o
KJKJJ
)•
tum mechanics on ! . For example, if our QFT is a harmonic oscillator, passing from
 •
­
!
corresponds to considering harmonic oscillators. Replacing

by positive-dimensional means that we consider the whole manifold of


ÙTá1
harmonic oscillators!


Recall that any QFT is defined by an action functional on the space of paths. In

M&
D & k
~ÙeÙ ¨ Ù ÌÍÖ M&  D} ÙÛ
a one-dimensional theory we defined by a Lagrangian . The pull-back
D} Ö 
of under the map

® Â6à dj ok

is a function
on
ÌÍ
, so for any density

on (i.e. a section of j top  ) we can multiply to get a density on
ÌÍ M&
on ÙÛ
which we can integrate. If
and a density is a function on 7ØqÙ Ú ®  ÙÛk
this is not true anymore since
. So the definition of the Lagrangian
is a function

has to be changed. We are not going into a rigorous mathematical discussion of this

¦
definition referring to Deligne-Freed’s lectures at the IAS.

î Ê M¦ ¦ ¦ Q
KJJKJ4
Q 
Recall that the jet bundle of order r of a fiber bundle over a manifold ¾ is a


JKJJ4
 Ê
vector bundle whose local sections are local sections of together with their

H’ JJKJH ’ Ê
o ’ JJKJ ’ ` Ê • î @ 9¦

 L œ L ` O
partial derivatives up to order r . Let be a local frame of and
local coordinates on ¾ . A local frame of z is a set where

     
. Let Û
Û
½
& i
Q ¦L œ L
´Û be Û the
ŒŽŒŽŒ ¬Ü corresponding

` `
r

Ê 

& î 9¦
coordinate functions. Any local section Û
of ŒŽŒŽcan
Œ Ü
be uniquely extended
 Û

â
to a section of such that

&À iQ ™ @ â Q L œ ‡ JKJK‡ J L â Q& iQ #J

 Û

Lœ L
Û ÝŒŽŒŽŒ Ü

L L
Let be the space of sections of î Ê 9¦ (fields,œ and their partial derivatives). Roughly
ŒŽŒŽŒ Ü

speaking a Lagrangian of order is a smooth map from î Ê 9¦ to the space of densities

D Ü

on  . We will be usually dealing with Lagrangians of the first order. Then we can write
r

39
40 LECTURE 4. THE DIRICHLET ACTION

@ 9 Q

ð D  QÀð J
a Lagrangian as Þ

Ö L L



Û Û

So, the action will be defined by a formula

 Ð
 M& @  9Q
&
W & #J â
Þ

Û Û

â â
One can generalize the Euler-Lagrange equations to the higher-dimensional case: Þ Þ

â F â â @ O
} @ o~
KJJKJ4
Âà  J j

L L
(4.1)
Û

place when we evaluate the left-hand side on & .

Û
Þ
Here we assume that the equality takes
We also assume here that is of the first order.

First, a little of linear algebra. Let

Let us consider an example, which will

% © bebe very much relevant to the string theory.
two vector spaces equipped with non-

form on the space of linear maps Lin

by
degenerate bilinear forms and , respectively. We can define a symmetric bilinear
% © Ê 

¿
& @ Tr M¿ & #
   <

where ¿ I

©  % is the adjoint map with respect to the bilinear forms and (i.e.
@ b
¿  ) for% any R % , R © ). Let© us explain this definition.
'¿À  #
a basis
 Ê 

Choose
 9 Ä

KJKJJ4
 in and a basis
JKJJ4
in . Let be the matrix of
Ê 9  Ä 9 Ä

in the first basis` and be the matrix of in` the second

Ê
F
˜ basis. Let be the matrix of
F

ß ß .

¿ with@ respect to the bases, and is the same for & . Then the matrix of ¿ is equal to g
o
n


` a so ¿
& @


@ v_à

n . n g

 s


 ` a L a _
 Ê s 

@ v p
@ à p
@ #
@ L anda we employ the physics
Tr (4.2) G. n gCo

@  , where
s s

by ¿À 
s

` ¿
KJisJKJthe
L  map
Ê  s

L
where n o . g

L
is an element of the dual basis
summation notation. Assume L that
. Then@ v ˜
defined
@ L
 . Similarly, take & L
9
F
9
ß s F

defined by &½ < @ É < w ` ¿

& É É ˜ L %
© ;F F S S s

© @ @


% , we getL ¿
& É É andL6L
9 9  Ê  s4s
. Then we F get ß s
. If we identify Lin

L
É L @ É J
áà à à
with F ß s F ß s

L L É L6L É % ©
 à à  Ê  s4s
F ß s F ß s

is equal to the Kronecker

+à

the basis
From this we deduce that the matrix of the bilinear form on with respect to

L4` defines an inner product on % . So, our inner product on % % 4` and© could
à
;F ß s product of the matrices . The . g
matrix .
©   à

Now we are ready to globalize. Let be a metric on  and be a metric on  .


be taken as the definition of the tensor product of the inner product on and on .
Ê 

 '& @ ð D &Àð g D} @ Tr D & D & D} J

Define the Lagrangian Þ

32  < 32
(4.3)
 41

Here
D} 32
is the volume form defined by the metric Ê
and the adjoint
D& 
of
D& is
defined with respect to the metrics Ê and  .

 D D~}
The corresponding action

@
  M&  ð ½& ð g  (4.4)

If is given in local coordinates Q

KJJKJ4
Q by the matrix and is given
is called the Dirichlet action.
K


J K
J 4
J

U

` then (4.4) can be rewritten in the

Ê Ê 
»ãâ

`
  
in local coordinates by the matrix

 š â & â & D JJKJ D J

form Û{ä

@
  M&  ð ð œ â Q â Q Q j
Q k Ê Ê »ãâ ä 

‰ h
Û U

`
(4.5)
—» â

D~} @ ð š ð D Q  JKJJ D Q . Recall that

Û{ä
à Ã

%
the volume form on a vector space is equal to ‰ JJKJ hœ , where
JKJKJ¥
 is an
2 j k Ê U

`
This follows from (4.2) and the fact that

orthonormal basis.  `  `  Ã Ã
à Ã

ští w preserving the met-

Observe the following properties of the Dirichlet action:

(A1) (isometry invariance) For any diffeomorphism º

  M& @  bÉ M& ç

rics,
<
º

(A2) (locality) if  is glued together from  and  along their boundaries, then
g
  M& @   `  '&Àð   g æ

`
@ W œ is a h new metric
(A3) (conformal scaling) if w on  , and-Â Ã  w is@t
 does not change if and only if  thec J new
” å
Ê Ê
j

Proof. Let &

% @ ÙÛ= © be a linear map of inner product spaces© and
% w  %
action, then the action functional

be an isometry of inner product spaces. Then, for any w R

% w
R , 9 Ä
º

&½ w ) p
@ r w p
&  « @ w
4` M&  ™ « É J
 9
Ä($ç  9  Ä   9  Ä


@
Gº º º

This shows that M& & , hence

4
 `
<  < 

'& ) @ Tr M& & @ Tr M& & #J

º º

Tr ) M&
`
<  < < <  < <  <
º º hº º

&

that º ™ð6ð D &Àðð g D~} bÉ @ ðð D M& ð6ð g D}  J Property
Applying this to the case

when are the maps of the tangent spaces, this implies
<
(A1) now follows from the stan-
º
º

dard properties of integration of differential forms. Property (A2) is obvious from the

@ @ 1  with metric defined

definition of the action. Property (A3) follows easily from formula (4.5).

Example 4.1. Let

by a matrix  s 9Q  1
with a local coordinate and
Ù W] @ 1   . Define a metric on  by
L
in the canonical basis in
42 LECTURE 4. THE DIRICHLET ACTION

ð ð o ð6ð g @ v . Let ¿ 
1  1  be defined by the vector function M¿ ` ] #

JKJJ
¿  ] ) .
Ê

Thenx
v v v v
ðð D ¿ ð6ð g @ ¿ w Í M¿À v ) ¿ w @ ð6ð ¿ ðwð o ð6ð ð6ð g Jg

£ ~
o
x@ 4 gx @ Ø D Ê

So, if we take ʐè

g
g ` , we obtain the action ž g` ò  ] g ] which
GS

Â6à tj o , I do not know any geometric meaning of the Dirichlet action. How-
we used in the previous lectures. Û{ä

ever, let us see that for a fixed metric on  one can always choose a metric on
j
If
 such that the action acquires a very nice meaning. In fact, the metric is chosen 
Ê
Ê

compute its variation in the direction at

@ \
to minimize the action. Let us consider the Dirichlet action as a function of and
S
Ê Ê Ê
Ê

‘ ÈÈÈ @ 'ž \ H ƒ š FÝž \ ™ ™ˆ ƒ

Ȑ
S Ê Ê Ê
Ê $S
S

@ O . Note that, for any invertible matrix and any square matrix of the
Žé é A

where ƒ
g n o

ð H ƒ ð @ ð Ið H ƒ Tr 4` ð Ið J
same size, we have

n Ëo n Gn o n

ð H ƒ ð œ @ ð Ið œ orH cƒ Tr 4` ™ #J
Thus

h h
n Ëo n Gn o

H ƒ ` @ 4` FӃ 4` 4` J

Also

hn $o n Ën opn

@ ââ & ââ &
Let

Q Q »{â Û ä 

@ . The matrix is the matrix of the metric &  . It can be viewed as the
» â Û{ä

metric on the image of  under the map & (called the world-sheet). Then
 
and »{â


ž  @  ð ð œ Tr 4` 4 D~}  Ê Ê

h
(4.6)

and

ƒ ‘ ÈÈÈÈ  M& @ S
S
Ê
é é A

 * \ H \ \ H b 4 D~}  \ \ 4 D~} @
 ð ƒ ð hœ Tr ƒ 4`  F  ð ð hœ Tr 4` 
Ê Ê Ê Ê Ê Ê
ËS ËS
 43

@  ð \ ð co Tr \ š Tr \  F Tr \ \  ) D~}  J
 hœ ` ` ` `
Ê Ê Ê Ê Ê Ê—Ê
S S

Set 
@
F \ \ . Then

 ` 4` `
Ê Ê Ê—Ê

‘ ÈÈÈ M& @  ð \ ð û co Tr \ \ \ ) Tr \  F Tr \ \ 4 D~}  @

S S

Ȑ  hœ 4` 4` 4` 4` 4` ü

S Ê Ê Ê Ê¦Ê Ê Ê Ê¦Ê
Ê S S
S
é é A

 \ û o \ \  D} J
 ð ð hœ Tr  ` F c Tr ` ü 
Ê Ê Ê Ê
S

Since this must be zero for all possible , this implies that a critical metric \ satisfies Ê Ê

F co \ Tr \`  @ O J
S

Ê Ê
(4.7)

ð ð ` 8 @ co Tr \4`  ð ð ` 8

This implies

$ Ê Ê $

where (
@ Â6à  . Plugging this in formula(4.6) we get
j


  M&Åç \ @tc  ð ð ` 8 ð ð h ÿ D}  J Ê $ Ê Ôê
ê

We get a wonderful fact: if (

@£c , and the metric onh  is chosen to be critical, the
sheet &Å k in the metric induced from the metric of  .
action has a simple geometric meaning. It is equal to the twice the area of the world

In physics the latter action is called the Nambu-Goto action and the Dirichlet action
is called the Brink-DiVecchia-Howe-Desse-Zumino action, or the Polyakov action for
short.
 ðð
v
Remark 4.1. In the case when the metric on is Lorentzian, we have to replace Ê
ð…F ð v
 
JKJKJ¥
 
with Ê . Also physicists use the metric to “lower the indices”. If  s is the matrix of
a metric  in a basis
` v
L L
the vector u s u  s s
L L 
L L
then for any vector
is denoted by u

L L L  M& @  ð ð â â & â â & D} J
. In this notation formula (4.5) can be rewritten as

 hœ L Q Q 
s »
Ê Ê â
s

L
Ù @ Ù M& @ F co
Remark 4.2. The tensor
D Q D Q J s s Ê s Ê s

L L L L ä[Û
ä[Û

if is a critical metric for the action  M&Åç š . Observe that

is called the energy-momentum tensor. By (4.7) this tensor is equal to zero if and only
Ê Ê

Ù @ ð ð F co s
D Q D Q @ ð ð o F (c D Q D Q J
s s s

L L  hœ L L L L  hœ L L L
Ê s Ê Ê s Ê Ê s Ê s Ê Ê s s

L äqÛ
äqÛ
¦Ã
44 LECTURE 4. THE DIRICHLET ACTION

Ù @ is trace-less@ if ( @tc .

In particular,
Assume  1
 1 and the metrics ¥
are the standard Euclidean metrics.
 Ê 

Ù @ F co Tr 4 ™ D Q D Q J
Then

s s

L L
Let us write down the Euler-Lagrange equations for the Dirichlet action in the case
when the metrics Ê and  are flat (i.e. Ê »ãâ and  are constant functions). We get the
Û{ä

â
equations

â Q g &â Q @ O
} @ o
KJKJJ4
™(SJ
Ê »{â  ä (4.8)

@ 1 
/ @ 1 , and this is just the Laplace equation with
» â

In the case when 

Û{ä

respect to the metric . When Ê @ , its solutions are harmonic functions. In

Ê
»ãâ S »{â
Ê
»{â

general, the Euler-Lagrange equation for the Lagrangian (4.3) can be written in an

( D & @ O

invariant form:

@ Ù  & pÙ with respect

Tr

where ( is the covariant derivative of a section & of ¦

D & à  2

to the natural Riemannian connection defined on the bundle ¦ .

@ Consider@ the special case when @ 
@ 1 g , with coordinates iQ
Q @ ]
Q , and
Ê

Ê ~
o
diag F
7
o  1 ` g becomes
@ * â â & g FP ââ & g , D ] D Q J
»{â . Take . Then the Lagrangian density
Þ

] Q
â & â & @ J
The Euler-Lagrange equation is

â g ] F â gQ O
g g @ @ o)
Notice the analogy with the Lagrangian for a harmonic oscillator (with › ˜

@ D D Q g FÎQ g D ] J
Þ

]g
We can view &Å ]
Q as the displacement of the particle located at position Q at time ] .
The Euler-Lagrange equation for the scalar field &½ ]
Q can be thought as the motion
equation for infinitely many harmonic oscillators arranged at each point of the straight
line.
Ê 
If is the flat Lorentzian metric in 1 defined by the diagonal matrix diag * F o~
o~
KJJKJ
Ko ,

@& â â & @ F â â g & F Ž  ââ g & @ O J

the Euler-Lagrange equation for a scalar field with Dirichlet action is

]g  g Q g
ë
ä
ä
ä ä
 45

ë
The operator is called the D’Alembertian operator or relativistic Laplacian.

@ co â & â &§FΛ g & g p

A little more general, if we take the Lagrangian

Þ

ä
ä

& H › g& @ OJ
the Euler-Lagrange equation is the Klein-Gordon equation
ë


In many quantum field theories is a fibre bundle over and is   & r
& £ 
a section. When
&
is a g -bundle with some structure group g a map is called a
classical field, otherwise is called a non-linear ì -field. For example, when
o is 

š  
 Ü  á& 
the trivial vector bundle of rank , a classical field is called a scalar field. Of course
any map can be considered a section of a fibre bundle, the trivial bundle
.

 DQ 
Example 4.2. An example of a classical field is a gauge field or a connection on a

@ ¥•
principal g -bundle over . It is defined by a 1-form u n
L L Ù
on with values in the
adjoint affine bundle Ad hg . In other words, it is a section of the bundle :à Ad Gg .

bÙÛ
9¦ p
¦
For example when g GL , a gauge field is a map of vector bundles n

M¿4Z @ - M¿ Z H ¿ MZ p

End where is a smooth vector bundle of rank w over . It satisfies

n ;F F bn GF

where ¿ is a local smooth function and Z is a local section of ¦ . It is clear that the

is a section of an affine bundle. Each connection defines the Lie -valued 2-form,
difference of two connections is a morphism of vector bundles and thus a connection
n Gg

Ì @ D H co *
, @ Ž Ì D Q D Q J
the curvature form,

s s

L L
° n n n

Here Ì with values in Lie . We define the Lagrangian

¦Ã
s

density L on
is a local function on ¾ hg
2

@ Ì ‚Ì J
í by setting Þ

hn ° I °

@
D}

bundle ¦ equipped with a metric . It is determined by the property

Here is the star-operator on the space of differential forms with values in a vector
I

  î
º I{º º

Riemannian metric on Ù and the metric on ¦ ) and

D} is the volume form defined by
 Ë
where is a natural bilinear form on the space of such forms (determined by the Ã

the metric on . 
The Euler-Lagrange equation for the gauge fields is the Yang-Mills equation:

Ž â â Ì HV*
Ì , @ O
@ o~
JKJJ4
D J
U
s

 QL
s s ïÅ

L
n (4.9)

L` L
46 LECTURE 4. THE DIRICHLET ACTION

There are two approaches to quantization in higher-dimensional QFT. First uses

@
@ Ù Ü  w  Â6à ٠@ o 
/ ( 
functional integrals which generalize the path integrals.
Ù
D
Consider a space of fields j Map on a -dimensional manifold . We
R
& &;R   @ ] Ü  w  ¿ R ] Ð
’ @ o~
c J
assume that , where ( a “time factor”). For each we denote
D > ! D

aa a  
by the restriction of a field to . Let  be the space of fields


U  1
D D

L
on obtained by restrictions of fields from . Fix two fields 
D


Consider an action and set

M¿
¿ g @ ¼ h  L 3 657 (+* &-,
Ð
}

`
(4.10)

where we integrate over the space of fields¼ œ & on  such that &
@ ¿ . We use some
measure (+* &N, on  . a L @
Observe the obvious analogy with our previous definition where we take  w
D

point and  is the set of maps = .

 ! D

of functions on  , the integral oper-

D

ator with kernel (4.10) defines a linear map a

Now if we consider some Hilbert space  

Ù

Ù M¿ @  M¿
  - G(+* ,
É
}


   
 

  

aœah aœ ah aœah
  ið ð

\  \ such that
Map

Ù @  
It also defines a self-adjoint Hamiltonian operator .  
@

a œ a h map   L a h ba œ Ü  ,

@ ð  aœ  ah J
which can be used to define a Hermitian   W
  @

g ` g   L a h ba @ œ ` h
 4ñ
ð ð ð ð 

}
 has special meaning for ¿ ¿ . The integral
¼ œ ¼ h Tr   @  ` g > (+* ¿b,
The kernel

 É   ¼  ¼
@

  L a h ša œ
is the trace of the operator  
Map
@

   L a #h
ba @ œ o~.
JKItJJ4is
)called the partition function of the theory.
•—
be a local quantum  field at a point   on
generally, let ’
 (aMore
  L
ò

L on  which depends only on &Å and derivativesL of

of a local field is the functional &  '&½   ) , where À
<Úá1
D

& at ). An example
local field is a functional

ôó ó

>    ç k @  ¼ 657   M& G(;* &N,

is a function on . We set

h
l

¼œ ¼h L L  L 3
L L L
iò ò(4.11)

¼œ
>  
  JKJJ     @ ¼ œ  >¼ h  L kL ç k J
This leads to the correlation function

 hò

` `
ò ò (4.12)

¼œ ¼h
 47

 b É @
 w

D
We can use (4.11) to define a Hermitian form on the space  of functions on

 
Map
 
ç] ]g g
}
/

/ @ g M¿ g %õKö
M¿ >    ç k G(+* ¿ , (+* ¿ g , J
/ 
£õbö
/

` ` ` ` ¼œ ¼h `  
hò iò
 

h in . Thus œ it defines a linear operator

g
/ /

`
This is still linear in and half-linear ò ÷

  ç ]
] g g
@ g ð   L  a ša     L  a ba  J
such that
}  / @ @
/
/ /  ñ

` œ it go to infinity,
` @ ] F ] by hletting h œ i.e.` define
hò ò ÷

We can get rid of the parameter ]

ÁÂ6Ã [   ç
g @ Á6`ÂÃ [ ð   pJ  / @ @

g `   a Ä g   L a 7L a `
} /
/ / 
hò ò ÷

a Ä
In this way we get a local operator Í   in the Hilbert space . It is called the vertex
operator associated to a functional .
ò ÷ 
ò

we assume that 
@ Ù Ü  w . For any field &
<= we denote by
Another approach to quantization generalizes the one we used for the harmonic
â \ the partialAgainderivative
oscillator.
in the time variable. By analogy with classical mechanics we
introduce the conjugate momentum field Þ

u ]
Q @ â \ M& pJ S

@ ) 5 g Fz 5W g we obtain uú@ 5 .
Þ S

For example, when '&

g` a : 
¶ ¶ ¶ We can also
introduce the Hamiltonian functional
o a ¶ ¶ ¶

@ u Ø
Þ

A '& c  É & F Q D J .

Then the Euler-Lagrange equation is equivalent to the Hamiltonian equations for fields

½& Ø ]
Q @ u ]
Q
S
Sã. uØ ]
Q @ F

‘&½ ] Q S
Sã.

u
J &
where the dot means the derivative with respect to the time variable. Here we consider
and as independent variables in the functional . and use the partial derivatives of
.
& u
&
To quantize the fields and we have to reinterpret them as Hermitian operators
ã u
in some Hilbert space  which satisfy the commutator relations (remembering that
is an analog of and is an analog of ).
ãØ
* ]
Q p
]
- , @ .’  iQ§F - #

0 ø

S

(4.13)

* ]
Q #
]
- , @ * ]
Q #
]
- , @ O J
Ž0 Õ0
ø $ø 
(4.14)
48 LECTURE 4. THE DIRICHLET ACTION


}
Here we have to consider 0 ø as operator valued distributions, i.e. a continuous linear

functionals on the space of test functions on equipped with some measure with
values in the space of operators in a Hilbert space  . Any function on with values 
D}
in the space of operators in  which is integrable with respect to some operator-valued
3ù


measure defines a distribution

&   Ù iQ &½ iQ D} J 3ù

The commutator of two operator valued distributions is a bilinear form on the space of
test functions:

'&
Y * Ù ' & #
)Ù g , J
/ /

`
* M& #
, @  &½ iQ 9Q D} J
Thus the meaning of (4.13) is


Ž0 $ø / /

@ 1 and is the Klein-Gordon Lagrangian 5 F W 5 F

Þ

Example 4.3. Assume that  w

› g & g . A solution of the Klein-Gordon equation can be written as a Fourier ha integral hh ¶ ¶

º h ¶ ¶

0
]
Q @ s oc‚u v  L  Ê W  a  H v  L   Ê W ‡ a  D

ir
ž%ú
hr
ž%ú
r

Êg @ g H › g J
where

˜ r

Similarly we have a Fourier integral for ]

Q : ø

] Q @ s ’c‘u  º Ê )F Ê W v H Ê W v D J
 L  L a  L  L a
ø ´û ž ú ´û ž ú
˜ ir ir r

v v v
To quantize we replace with an operator Ê and with the adjoint operator Ê
ir
v hr
Ó

Q and ]
Q are Hermitian. The commutator relations (4.13) will be satisfied if
we ] require
and consider the above expansions as operator integrals. This implies that the operators
ð
ø

* v Ê
v Ê É , @  IF w p

the commutator relations

Ó
S ir r

*v Ê
v Ê É , @ *v Ê
v Ê É , @ O J Ó Ó

only one pair of operators

v v satisfying * v
v , @ o
* v
v , @ * v
v , @ O (or •
This is in complete analogy with the case of the harmonic oscillator, where we had

operators

v v satisfying *
,
v v @ 
* v
v , @ * v
v , @ O ). There is a big
Ó Ó Ó Ó

v v
Ó Ó Ó Ó

generated by o
Ê
Ê is
s s s s

L L L L
S

difference however.
infinite-dimensional.
L In our case the Heisenberg L
Lie algebra
Ó
 49

Exercises
@
D & be a Lagrangian on  with a metric defined on the space of
Þ Þ

4.1 Let
maps PU1  . '& Define the energy-momentum tensor by Þ
Ê

Ù M& @ Ž ⠍ â M& F M&

Þ

Ê S Ê
»{â â » »ãâ
ä S
Û ä

W …57 @ O if & satisfies the Euler-Lagrange equations.

(i) Show that this definition agrees with the one given for the Dirichlet action;
ãüqý
(ii) Prove that u ¶
ý
¶

4.2 Consider a system of ­ harmonic oscillators viewed as a finite set of masses

v
arranged on a segment *
à , , each connected to the next one via springs of length
ƒ . Write the Lagrangian Ö Mƒ describing this system. Show that the limit of Ö Mƒ when
ƒ goes to zero is equal to
 â &½ ]
Q â &½ ]
Q D J
â ] g F g â Q g, Q
e
þ þ

x `
where
are some positive constants and &½ ]
Q is the function which measures the
` g
- of the particle located at position Q at time ] •.
þ þ

4.3 Let be a Riemannian manifold of dimension and ¦ be a vector bundle

displacement

G¾
over ¾

® Ê Ù ¦  ®  @ Ê Ù } ¦ such that, for any
R
equipped with a Riemannian metric. Show that there exists a unique linear

Xr
® Ê Ù ¦ p
one has  v
½ . à Here @
is defined
 +à & +à
î
isomorphism I ¾ ¾ º

locally by extending via linearity the product `

;¾

¾
(à

w w v à. I
î 
Gº

î
º
î

ÊÙ
vol

Ù
#J to  ®  ¦ by the  formula
à Mà

w @ à metric
v ` is
à the inverse
w Êon v
àKextended
à Ã
   Ýà
Also G¾ Ã ¾ Ã

4`     4`
 à à 

4.4 Using the star-operator defined in the previous problem show that the DirichletÃ
action can be rewriten in the form

ž '& @  D & D &
I

D
where & is considered as a section of the bundlle Ù K & ÙÛ .
Ã
 à 
50 LECTURE 4. THE DIRICHLET ACTION
Lecture 5

Bosonic strings
( d@ c
From now on we stick with dimension
 of our QFT. This is where strings appear.
1g  ` SÜ j 1 o `Ü `
Our manifold will be a smooth 2-manifold with a pseudo-Riemannian metric Ê . It
could be the plane or a cylinder , or a torus

, or a sphere , or a g
compact Riemann surface Î of genus 
 . Of course each time we should specify

 `Ü 1
a metric on .

 @ 1 Ü *O
,
†
We shall begin with the case when is a cylinder
† (closed strings) or
1
&½ ]
ž
|

î
(an open string). We use the coordinate in the circle direction and the

map
]
coordinate (time) in the -direction. A map
Þ † †
can be considered as a

Þ &À ]
~1|  p
]    &½ ]
™ #

where  is the loop space of  , i.e. the space of smooth maps from a circle to  .
Þ

In the case of open strings  must be replaced with the space åæ  of paths in  .
We shall consider only closed strings, however occasionally we state the corresponding
results for open strings.
@ We shall @ also assume @ that 
@ 1  with the Lorentzian flat metric
 , where diag GF o
Ko
KJKJJ
KoB . We will write vectors
in the beginning
 in 
K

K
J 
J 4
J


K
J 
J 
J

K


J K
J 4
J

 

as 9Q
9Q `
JKJKJ4
Q Q and . denote by iQ Q the vector )FžQ ` Q general Q equal
Gß ß


Û{ä

`
Û{ä Û{ä
g target to

` consider the Dirichlet

ä
Later on we will
ä
of course consider more spaces

 
ä ä

ž '&Åç š @ Ù c  ð D &½ð g D~}  @ Ù c  ð ð â & â & D}  J (5.1)

. We action
„
Ê Ê Ê »ãâ
» â
Û

Here Ù is a certain parameter of a string (the string tension). It is equal to o7ˆ

c‚u w for
ä

º
open strings, where w is a certain other constant called the Regge slope. For closed
strings Ù
@ o7ˆ u w . We use the subscript to emphasize the dependence of the action
º
º
Ê
Ê
on .

ž &Åç š ) @ ž M&Åç b p
hº
<
º
 Ê
(5.2) Ê

51
52 LECTURE 5. BOSONIC STRINGS


¿;RSï [ k
where º is a diffeomorphism of . This means that the action is invariant with respect

ž M&Åç  ¼ š @ ž '&Åç š #J
to smooth reparametrizations of the maps. Also, for any , we have
Ê Ê
(5.3)

This means that the action is conformally invariant.

b @  ¼ \
It is known (see, for example, [Modern Geometry] by Dubrovin, Fomenko and

\
Novikov) that there exists a unique diffeomeorphism º such that º  Ê Ê

¿ )F o~
oB
, where
Ê
is a smooth function and is a flat metric given locally by the diagonal matrix

ž M&Åç š @ ž ¥& ç \ #

diag . By (5.2) and (5.3)

Ê < Ê

@
Gº (5.4)
D
We shall fix the metric on  by equipping 1 with the metric F Q and taking 
1°ˆ c‚u ‚m with the metric induced by the standard metric D ] g on g1 . Then we have `
two constraints on & . One comes from the Euler-Lagrange equation for the action
w

 and another comes from the vanishing

Lagrangian function for the action 
é A of the energy-momentum tensor. Since the
é A
is equal to

@ Ùc â & â & F â W & â W &

Þ

a a is Û
(5.5) Û

the Euler-Lagrange equation for the action 

Û Û

é A

â g F â Wg & @ O
} @ o
KJKJJ4
™(SJ
a
(5.6)

So our field '& satisfies the Klein-Gordon massless equation.

Û

The value of the energy-momentum tensor Ù at \ is equal to

Û
Ê
»ãâ

Ù \ @ Ù \ @ â & â W & @ ONç Ù \\ @ Ù @ co â & â & H â W & â W & @ O J

` ` a Û
Û
`` a a (5.7)
Û
Û
Û
Û

@ ] H Q
@ ] FÎQ J
To solve the wave equation (5.6) we introduce the light-cone coordinates

â
⠇ ì
 ì

‡  â @ cYo â W H â p
â @ c—o â F â W #J
Let denote the partial derivatives with respect to these coordinates. We have

‡ a  a
â â & @ OJ
Thus we can rewrite (5.6) in the form

‡  Û

& @ & ‡ ¥H &  pJ

This easily implies that a general solution of (??)EQ) can be written as as sum
‡ Gì ~ Gì
Û Û Û
 53

& H & 5 5ÿ
Using the boundary conditions, we see that, in the case of a closed string, the functions
‡ ƒ
~ , ¶
¸ÿ and ¶
¸ÿ are periodic with period w , so that we can use the Fourier
c‚u
õ
Û Û
¶ ¶

& ]
Q @ 4co Q H \ ‡ H ’ Ž  \ • o    L 
õ 8

expansion to write

‡ ¤ó ó
Û
(5.8a) Û
º Û ì w º
Û

& ]
Q @ co Q H \  H ’ Ž  \ • o    L 
ÿ 8

 z

~ ¤ó ó
Û
(5.8b) Û
º Û ì w º
Û
 z

@ and \
@ \ . We shall see in a moment a reason for the choice of
the constant g . € ` Also, since we want & to be real,
ó Q
where 
º Û º Û
ó

 @ 
 @ J
{z Û

 
º º Û º º Û

The field & ]

Q (resp. & ]
Q describes the “left-moving” modes (resp.
Û Û
‡ ~
Û Û
“right-moving” modes) of a closed string.

â & ]
Q @ Ž  
8

Note that

   L õ
‡ ó

⠇ & ]
Q @ Ž  
ÿ 8 J
(5.9a) Û
º
Û
{€ z

   L
~ ó


(5.9b) Û
º
Û
{€ z


It is clear that

Q @ \ g € & 9O
Q D Q
} @ o~
JKJJ4
™( Û {z Û

By analogy with (
@ o QFT the momentum field is defined to be
and can be interpreted as the center-of-mass coordinates.

  @ â ââ @ Ù â & J
Þ

& a Û Û

a Û

@ Ù  \ g € D & D MO
Q D Q @tc‚u Ù c \ @ c \ J
The expression

]
Û ó
¯
Û {z (5.10) w º Û ó º Û

is the total momentum coordinate of the string at ]

@ O . Now we can rewrite the
& ]
Q @ co Q H co g ‡ H ’ Ž  \ • o    L 
õ 8
(5.11a)
equations (5.11) in the form

‡ ó ó
¯ ì w º
Û Û Û Û

& ]
Q @ co Q H co g  H ’ Ž  \ • o    L 
ÿ 8
(5.11b)
 z

~ ó ó
¯ ì w º
Û Û Û  Û
z
54 LECTURE 5. BOSONIC STRINGS

 @ 1 Ü 
Remark 5.1. If we choose the Riemannian metric on instead of pseudo-Riemannian,
@ O
we will be
  able
!
to identify the cylinder
`
with the punctured complex plane

]
Q Y @   bai‡ L WB98 J
W W by means of the transformation
ô"
z

âBâ  = b
~¯ @ O
The Euler-Lagrange equation (3.7) gives
" "
Z ¾

& < @ c4o Q H c4o ’ g Á ÛH ’ ʎ  \ o Ê Ê

The equation of a string becomes

‡ " ó m " ó "

Û Û
(5.12a) ¯
Û
w º Û

& <¯ @ co Q H co ’ g Á ¯ H ’ ʎ  \ o Ê ¯ Ê
 r

" ó tm " ó "

~ (5.12b) ¯ w º Û
Û Û Û  r

The stress-tensor Ù can be rewritten in the new coordinates too. We have

Ù  @ â  & â  & @ Ž  @ â  â  Ž
»{â

˜ g
Ù & & @ ¯ ˜ g J
˜ ֘  ˜ ֘ 
"  "
Z Z Z Z
Û Û
Û {€ Û {€

This is a familiar expression from the conformal field theory.

@  \ g € â &   F D Q @ Ù c  \ g € â & â & H â W & â W & D Q J (5.13)

The Hamiltonian of our theory is equal to
Þ

a a a
. z {z
Û Û Û
Û Û Û

momentum tensor vanishes. Plugging in the expressions for & in terms of 

 ,
Observe that it vanishes on a string which satisfies the constraint that the energy-
º º
Û Û Û

@ co Ž   H  
we obtain

  
. (5.14)
{€
hº Û º
Û
º Û º
Û

Ù . Observe that we could simplify the sum by getting rid of but we don’t do it, since
ó
Now it is clear the introduction of the constant . It made our formulas not depend on

in a moment the coefficients 

 will become operators. g` º º

â & â & @ co Ù \\ HAÙ \ @ O

Using the new coordinates we can also rewrite the constraints (5.7) in the form

‡ ‡ ` Û
Û

â & â & @ co Ù \™\ F Ù \ @ O J

  ` Û
Û

â M& â M& @ â '& â M& @ O J

This immediately gives
~ ~ ‡ ‡

a a a a Û
Û
Û
Û
 55

@ Ù c  \ € â '&
â M & g ˜ W D Q @ co Ž
This can be restated in terms of the Fourier coefficients as follows:

  @ O (5.15a)
֘ Ù  a a⠍  L D @ o Ž  ˜  @
~ ~ º Û º
Û

@ c \ € â M&
Û Û

M& g ˜ W Q c  O J (5.15b)
{€

֘

  L
‡
 ˜  ‡

a a
º Û º
Û
Û {€ Û

@ \H \
Observe that

Ö Ö
Now we quantize & as in the previous lecture by taking  as operators in some
(5.16) . 

Hilbert space. Since we want & to be Hermitian we require

Û Û

 @ 
 @  J
Û
Ó Ó

 
º Û Gº º Û ¸º
Û Û

*   ]
Q #
]
Q w , @ Fž’  iQÍFÎQ w

We need

¾ 4S ß
Û ä äqÛ

Plugging in the mode expansions, we see that this is equivalent to the following com-
mutator relations

*Q
, @ ’
*
 , @ *
 , @ ›  \
(5.17a) Û
¯
ä

ß
Û{ä

˜ º

all other commutators between, Q


 are equal to zero.
˜ Û
˜‡ º
ä
(5.17b) º
Û
º
ä
XS ß
Û{ä

We can also quantize the expressions for

. In order they make sense º º
Û Û Û

Ö ˜ (byÖ ˜ analogy with the harmonic . 

oscillator) that in our representation  kills any state provided that • is large enough.
as operators in some Hilbert spaces we will require

Thus the sum  \   makes sense. Let us set, for any operators

˜ Ò ,
s

L
u | º º with n n
indices in an ordered set

@ 
if ’ { ; s Å

L
n n
s

L
n n (5.18) s n

L
, @ O,
otherwise. n

It is called the normal order for the composition of operators. Since * 

@ ˜
Ö ˜ › O in the @ formo Ž

º º
'
we can rewrite

֘ c  ˜  º Û º
Û

The situation with \ is more complicated since    do not commute. Of course

{€

  Ö  @ §
• H    so we can write
º Û º
Û

\ @ c \ \ H Ž \    H cYo ( F c Ž •
o
we know that º
Û
º Û º
Û
º Û

Ö Kº Û º


º Û º
6

`
ä | Û
56 LECTURE 5. BOSONIC STRINGS

(} @ F c }

o
\ \O
Here we get because when we sum with respect to , the contributions corre-
sponding to cancel each other. We shall deal with the last sum later. Now we
define
Ö Ö
and  by dropping out the infinite sum.

Similarly we define the operators  . The operators 

֘ ֘
are called the Vira-
֘
soro operators. Notice that
other.
and
֘ (resp.  and 
֘ Ö˜
) are adjoint of each
Ö˜
@ \ H \ H ( F c Ž [ •—J
The expression of the Hamiltonian operator is now straightforward:

Ö Ö .
‘ 
(5.19)

`
[Ž • @ oB @ o J
Since the last sum obviously does not make sense, we regularize it by setting

‘ )F F o c v

`
@ \ H \ F ( o Fc c

So, finally we get

Ö Ö . (5.20) 

\ @ co \ \ H Ž  
From now on

Ö   º Û º
6
º Û º

`
Û Û

\@ c \ \ H o Ž J
Ö     º Û º 
Û 6
º Û º
Û

`
Let us find the commutator relations between the operators  . First we use the
Ö
*
ï ( , @ *
ï, (zH ï *
( , HV*
ï, ( H ï *
™( , J
following well-known identity:

nÑo n o n o n o n ¬o

* Ê  Ê
 , @ Ê ~* Ê
 , H Ê  * Ê
, H
This gives

º
˜ 
º º
˜ º

 ˜    
º º º ¬º º º º º

* Ê
 , Ê H  * Ê , Ê Ê  Ê H Ê Ê  H
 
@  

˜    
º º

  ˜   HV* ó ˜  
 J  ˜   
¼

º º º º ¬º rS º º rS º º

i›úF ˜ Ê   Ê , 9›£F ˜ Ê   Ê
 } ó  
   
r S º º ¥r S º º

*
 , @ co Ž Ê  Ê  H 9›ŠF Ê  Ê pJ
Here we skip the upper index . This easily implies

֘ Ö Ê ˜ ‡ ˜ ‡
hrº Û º
Û
r º Û º
Û
 57

IF • › HA• @ O ¤'

*
 , @ i›£F •¥ 
if › Hݕ @ O J
Changing r to r in the first sum we obtain for

@ O we haveÖ ˜ a problem
Ö Ö˜‡
'

For › H• since Ê Ê ÀÊ is not defined. Since * Ê

Ê , @
F , we see that the difference
u {€ º Û º º Û º Û
Ñr

t
@ c   Û

*
, › \ H i› id
֘ Ö˜ Ö @ O,
n

for some scalar 9› . Using the Jacobi identity, we find that, for Hݕ§H ›

• FΛ ¥H i›úF •¥ ¥H F •¥ 9› @ O J
n r

@ o and › @ F • F o gives
n ir r n ir n

c c
Setting r

•ÍHVo7 @ •æH •¥ • F|F o •æHÔo7 oB J

n
n n

This shows that I i›

@ v › ” H=à › for some constants v
/à . We will fix the constants
n
v
v in some
when we consider the representation of the Lie algebra generated by
Hilbert space. ˜ ˜ Û

Û

Exercises
5.1 Let Vect M `  v   L v R
be the Lie algebra of complex vector fields on the circle. Each field
 U  U Ö @  L 
U  , where U  .
@ ’# i›£F •¥  .
is given by a convergent series u XW . Let

(i) Show that * 

{€

,
@ Ö Ö ˜ . Set Ö ‡ ˜
(ii) Let Vir Vect M  
` *
v #

àK @ *


) p
W

, , æ GF ;ß F ß o GF ß

where is a bilinear form on Vect M . Show that this defines a structure of a

o
`
æ *
,
ÅH æ *
,
ÅH æ *
7,
@ O J
Lie algebra on Vir if and only if satisfies o

v v v
o F ß o ß F o F ß

(iii) Let æ •—
›
@
. Show •—
› @ O unless •+H › @ O and
n
• —
F •¥ @ v • ” æH= Ö à/•  forÖ ˜ some
o o
v
à that
R . XW
o

(iv) Prove that two bilinear forms and w define isomorphic Lie algebras if and
only if æ •—
F •¥ F w •—
F •¥ is a linear function in • .
o o
o o

5.2 Let * ]
] , be the algebra of Laurent polynomials in one variable. For any ]
  @
v  Ê Ê ] Ê R 4` *@ ]
] , let Res   ] ) @ v . Define a bilinear form on * ]
] , by
W

&½ ] #
] ™ Res4` #J
u
d
ÉW
d U
U4` 4` W

a
58 LECTURE 5. BOSONIC STRINGS

&
&Å  
H &½
  ÅH &½  
@ O J
(i) Show that is skew-symmetric and satisfies
d } d } } d

@ Der * ]
] , @ * ]
] , be the Lie algebra of derivations of
* ]
] , . Show that  ` is a Lie algebra
U

`   D   v D   `@ a   D with
(ii) Let  hW W U

* D
#
D
àp , * D
D D   ,
&½  
™ #J
 
W W respect to the Lie bracket

d d d

] ] ] ]
@
p

@ K

7
o
Ö ˜ GF ] ˜ ‡Å` a O @ MO . Show thato
U þ
(iii) Let U

*
 , i›£F •¥  H o c i› ” FΛ   ‚J
֘ Ö Ö˜‡ ˜ 
þ
S

(iv) Show that any central extension of the Lie algebra Vect M with one-dimensional
`
center is isomorphic to the Lie algebra defined by the commutator relations as in
(iii).
Lecture 6

Fock space

Ì * v
à ,

 Ì
Let  be a Lie algebra over a field with the Lie bracket . Recall the construction

¿ ¿À * v
à , @
of the envelopping algebra   . It is an associative algebra over which is universal
v¿À ¿À àp ÝF ¿À Kà ¿À v
with respect to homomorphisms  n of associative algebras such that

@ Ù  ™ˆ Ò @ ~[  \ Ù   ) ˆ Ò

. It is constructed as the quotient of the tensor algebra

Û 

where Ò is the ideal generated by elements

v à F à @ v F * v
/à , , v where v
à R .
For example, if  is a commutative Lie algebra (i.e. *
/à ,
à
v O for all
/à R )   is
`à

isomorphic to the symmetric algebra Sym   ), i.e. a free commutative algebra generated

basis of  ). In general, Û  has a basis consisting of ordered products JJKJ

by the vector space  ( isomorphic to the polynomial algebra in variables indexed by a

JJKJ Ê
where  is an ordered basis of the vector space  .  œ  ` s s iÅ 
ú


L L  representation %
 Å

Recall that a linear
%
End , where the latter is equipped with a structure of a Lie
of  in a vector space is a homomorphism of
@ PF . We say that % is a  -module. By definition
algebra by setting *
l,
the Lie algebras  
n o n
<
o µo
<
n
%

module over   . This allows us to extend the terminology of the theory of modules
of the envelopping algebra, this is equivalent to equipping with a structure of a left

An example of a linear representation is the adjoint representation ad
 

over associative rings to modules over Lie algebras.

End  defined by ad  iQ
v @ * Q
v , . The fact that it is a linear @ O . representation follows
from the Jacobi identity * Q
* 
,, HV* š
B* b
QN,, Hd* N
B* Q
,6,
An ideal  in a Lie algebra  is a linear subspace such that *
/à ,
" " "

v R  for any v R
à R  , or, equivalently, a submodule in the adjoiont representation. An
example of an ideal in  is the commutator ideal * 
~, generated by the commutators
 and any 
* v
à ,
v
à R . A noncommutative Lie algebra without non-trivial ideals is called a
simple Lie algebra.
We will be mostly dealing with infinite-dimensional Lie algebras. An example
of such an algebra is a Heisenberg algebra. It is characterised by the condition that
its center (the set of elements commuting with all elements in the algebra) is one-
dimensional and coincides with the commutator. Let  be a Heisenberg algebra and let

59
60 LECTURE 6. FOCK SPACE

"
w
be a basis of its center  . We define a bilinear alternating form
/ on  by
* v
à , @ v
àK bJ "

Its kernel is equal to  w , and the induced bilinear form on ¯ 

@  ˆ  w is nondegenerate. For
example, if  is finite-dimensional,
 6
 à  @dc
j
¯  and ¯  has a basis 
KJKJJ4
 Ê<ç 
JKJKJ4
 Ê
` ` 
r

@ @ @
such that




 
 O
o ’
J
s S s s s  iÅ 

L  L L L 
r

Thus  is completely determined by the commutator relations

*
, @  NJ
 
s s "

L  L
S (6.1)

If  is infinite-dimensional, we assume additionally that  is m -graded, i.e.

So all Heisenberg Lie algebras of the same dimension are isomorphic.

@ 

  {€

where each linear subspace   is finite-dimensional, and

\
@  w
f*  
 ,   J
˜ ‡˜
,

@  \  
 @  ! \   J
Let


‡ 
|

*
 , @ *
 , @ O
It follows that

‡ ‡  
and the bilinear form on ¯  restricts to a non-degenerate alternating bilinear form on
each     . Thus we can choose a basis in  and a basis ÿ in
 such@ that  is @ determined
* b
, * N
, * N
, @ O bythesethearecommutator  õ ‡ as commutator
L L relations   L L with
in (6.1). Together
{€ {€

 Notice
L that  L
" " " $"
called the Heisenberg relations.
"
@ \ 
* * "

Ì
are maximal abelian Lie subalgebras of  . Consider a linear representation of
‡
in the

< oB @ v

oB @ O J
one-dimensional linear space defined by
"
  #
v x x ‡
Here RtÌ is a fixed parameter of the representation. Now we can define a linear
representation of the whole Lie algebra  by taking the induced representation:
% v @ Ind $%  k
@  & ('  %  Ì J =à

Recall that for any left module  over õ x an associative Ì õ -algebra the extension of
scalars of to a -algebra is a left -module *) defined as the quotient linear
 o
à
o n n n
 61

Ýdˆ7Ù Ù v v
space n à
à R
v R
› R 
, where" is the linear subspace spanned by tensors =à à
v w › ÝH Ù @ v<v à w › › F HÝÙ Óà › v
Û b

JKJKJ¥

KJKJJ Ì ] \ ] ` ] g , Û
*


KJKJJ
à à
with o Xn and multiplication " .
We can identify 
‡
with the algebra of polynomials in vari-
]L
ables corresponding to the basis "
` 
of . Similarly we identify  #
Ì *!JJKJ
]
] \
]
JKJKJ , J
4 ` `
(as a linear space) with the linear space of Laurent polynomials

@ H
j

However the multiplication is different:

\
] L ]  L ]  L ] L ] * \ ’

O
KJKJJ (6.2)
and any other pair of variables commutes. The Ì ] ] ] , -module corresponding
to the representation  is the quotient algebra of Ì ` * ] \
g ]
]
KJJKJ , modulo the ideal
generated by ] \ F
v
] x
]
KJKJJ . Let us first describe the `induced g
\ module % v as a
K
 
 
` g ] JJKJ ] Êof. Fordegree in Ì !
* 
J K
J 
J

] ] ] ` , is called K


J K
J J
any monomial ] 4` K ] write
linear space. A monomial s s r
ú

` g œ
] K ]
@ ] K ]
L œ L
Å  Å   Å
normally ordered if ú

JJKJ Ê and i’ L œ
JKJJ
L ’GÊ @ œ
 
KJJKJ

 Ê  for some permutation
s s
ú ú

ER ¥Ê .` Usingg the relations (6.2),` we can write `

Å  Å   Å Å
Å
where

] L K ] L @
] L KK ] L
H normally ordered monomials of degree less than J

ì

K ] and
ú ú r

œ the aboveœ theK normal ordering decomposition of the monomial ]

write it as n.o.d ] L œ of ] L g ] ` ] `
We call

Lœ ]L @
ú
. For example, the normal ordering decomposition
ú

]  g ] ` ] 4` ]  g ] ` ] ` H ]  g ] \ @
]  g ] ` ] K`
<H ]  g ] \ J

is equal to

Observe now that if in a normally ordered monomial ]

% v œ ] L thetoindex ’G Ê is positive,<o
the coset of ] K ] to in the induced module L is equal ] ] ] @ 
 K
 ú

and hence is L zero. L If the monomial is equal to ] KK ] ] \ then ] L K ] L ] \ ìL o

à à
ú ú ú

v ] KK ] oœ . This shows that % v has a basisL consisting L L œ L o o and ú ú

à

]space K
 
  + o
L œ Ì ] * L
, where
ú
’
à

J K
J J œ
’GÊ O . This allows us to identify with the linear %
of elements
œ v (à

K


J K
J J
à  

L œ L ] 4` ]  g , ` . We have an% isomorphism

ú


@
v Sym &  #J
of linear spaces


(6.3)

ð O
@ o do
The vector
 7à

is called the vacuum vector. The structure of a &  -module on is given by

% v
] K ]  ]  L K ]  L Ôo7 @ n.o.d ] K ]  ]  L K ]  L Ôo @
s s
ú ú
à s s
ú ú
=à

œ œ n.o.d KK  œ K ÿ ð O #J œ

% v ] ] ] L ]L s s
ú ú


Note that carries a natural grading œ definedœ by

K @ ’ HdJKJJ7H ’GÊ
j ó
ó

R  and ð O @ ,‰ +O . L œ L `
ú
ó j 

L  L ,‰ +
where -
62 LECTURE 6. FOCK SPACE

% v
% v Ð
Remark 6.1. More explicitly the representation
* Q
Q
KJJKJ ,
of  can be described as follows.

’ j

O
Ðj
O @
g  
v vW   * L W ’
Q Q ` Å 9Q Q ½ iQ L
We identify with the polynomial algebra W and assign to
, and to " the
, L L
the operator ¶ , to the operator ¯ ¯

%
¶ s
scalar operator id. Then ¶ id and hence we get a representation obvioulsy
¶

% v
isomorphic to .

There is an inner product on the space defined as follows. Let ¦ be any linear

we define the bilinear form in Ù 9¦ by
space over a field of characteristic 0 equipped with a symmetric bilinear form . First 

/.  ÔJJKJ
 9 à 
dJKJJ  @
K 

´à 9
Ä à ¸à Ä  9
Ä
 9
Ä

`
and then extend it to the whole Ù 9¦
` @  Ù  M¦ ` by ` requiring that Ù  9¦ and
Ù ˜ M¦  areKmutually orthogonal. Using the polarization process, we@ identify  9¦

with  9¦ equal to the subspace of symmetric tensors in Ù M¦
 
¦ . and then
restrict .   $

to  M¦ K  
to get a symmetric bilinear form sym . One can show that
 $ 

this inner product is non-degenerate if is non-degenerate. Recalling the polarization

  KJ JJ
JKJJ @ •—o – Ž

  JJKJ

  #
(6.4)
isomorphism we see that

 Lœ  L  œ   Lœ  œ  L 
 s s   s
 s 
sym  

where the sum is taken with respect to all permutations of • letters. Here we identify
Sym M¦ with the space of polynomials in a basis of ¦ . This defines a symmetric
bilinear form sym - on Sym 9¦ . Following the physics  L agreement we shall drop 10
`


fact, if we choose a positive definite hermitian form on ¦ we can complete the tensor
in this formula. A similar construction can be given for any hermitian bilinear form. In

algebra Ù M¦ with respect to the corresponding norm and obtain a Hilbert space Ù M¦ .
E

This space is called the Fock space associated to the unitary space ¦ . The completion
of the subspace Sym 9¦ is called the bosonic Fock space. Similarly we can restrict
ourselves with the exterior algebra ®Û 9¦ identified with the subspace of alternating
tensors in Ù 9¦ . Its completion is called the fermionic Fock space. We will deal with
it later.

taking ¦
@  the, where
We apply construction of the Fock space to the Heisenberg algebra over by W

 the  Lie algebra  with a linear basis o~


)• R=m
} @ o
KJJKJ4
™(
 is equipped with a structure of a unitary space.
Let us consider º
Û

*
 , @ › Ê \

with Lie bracket defined by commutator relations (5.17b):

˜ º
˜ ‡ Â6à   @ ( for all • @ O . Let
Û
º
ä
(6.5) XS ß
Û{ä

} @
  #
o~
JKJJ4
™(S
be a basis in   such that, for any •;j O ,
j '
Let  be the graded Heisenberg algebra over with W

* 
, @   bJ
Û

  ˜ ˜@ 2
"

@ 1 is viewed as an
S ß
Û ä Û{ä

Consider the direct sum of Lie algebras 
2
 , where 2
abelian Lie algebra. Let \ be a basis of . I claim that  is isomorphic to  . To see
 



Û
 63

this we define the linear map ¿
   as follows. 

 \ if • @ O ;
¿À   4 @ 35 L   if • @ O
} @ O ; Q ¤'
º
6

57 ` \ •if @ O ,
Q ¤' '
Û
(6.6) º
Û
º Û

and ¿À 
@ o . It is clear that this is an isomorphism of Lie algebras. We call  the
"
% v
oscillator algebra of @ 1 . Let be the linear representation of the subalgebra 


 L ’ O described above. We can extend the representation to the


" '
generated by

whole  by setting

š \ ð O
@ ðO

 
c

1
Û

\
R . Since  belongby to% thev ç center,
Û

for some c
Û , whereit defines @ the
KJJKrepresentation.
J
R 1 . ItWeis
Û

more natural to consider as a linear function on the ideal 2 of`  generated by \ ’s so


denote the obtained representation Ëc c ic c 

@ ¥ \ . We@ will be interested only in representations corresponding to v @ o c



% % o ç . Its vacuum state is denoted by ð . Recall that we can

Û

so that we set
that c c

%
Û

write any element of as

Û 
ic $c c
ic

ð ƒ7 k
@ ƒ Ê   Ê Ê JKJJ Ê ð p

   

@ 9ƒ  
KJJKJ
 ™ œisœ a tensor œ symmetric
hc  º Û º Û c

œ  in lower and upper indices



where ƒ
Û ŒŽŒŽŒ Û

 k
  1
ŒŽŒŽŒ

defining a linear map  ` #  @ with  ir r

œ  8
Û



K
J K
J 4
J
ŒŽŒŽŒ

 R m \ and set
Û
finite-dimensional support. It is called 

ð ƒ7 ¥ç#8 @ ƒB ç
KJJKJ4
 @ n ` n Ž n  ƒ Ê   Ê Ê JKJJ Ê ð pJ
a Lorentz polarization tensor. Fix hr r M|

   

` œœ  œ œ 
hc ic Ër r  º Û º Û  c
ŒŽŒŽŒ Û

`
  Û

Let us define the inner product in . We may assume that ð

%œ
ŒŽŒŽŒ
Û Û
ŒŽŒŽŒ

is of norm 1. hc c
Recall that we want the operators and to be adjoint to each other. Then

ð p
ðL @  L H :9 ð p
ð @ :9 J
º Û º Û


ð #
ð @  s 
 
º s º Û
  
º s
™Å
S s
  Å
S s

 L   ð form L an othonormal  basis L L

º Û c º c c c Gº Û c c

L in Minkowski
ä ä ä Û{ä Û{ä

So we see that 
\  ` ð  have squared norm equal to F • . Following the discussion above
Q º Û c

sense. In particular,

we can extend the inner product to the whole . Two different monomials in
the vectors º
%c


’s hc º Û
are orthogonal and L
ðð Ê JKJJ Ê ð ðð JKJJ 
@  

œ  `
º Û º Û  c r r

if all
KJKJJ
 are distinct. Weœ leave to the reader to deal with the general case. Note
r
@
` any §RSm \ , and ƒ Mƒ
KJJKJ
ƒ R 1 ,
r


ðð6ð ƒB ` ç ð6ð @ ð6𠃂ð6ð @ ƒ ƒ J

that, for r &| 


(6.7) ic Ër r
Û
Û
64 LECTURE 6. FOCK SPACE

%
¦ Ë @
Remark 6.2. One can define the Lie algebra   and the Fock space ic in a coordinate-

iQ
- ]
¦  R+¦
… 9
*
)
Ë • ] R;, m ¦
free way. Let be a vector space over a field equipped with a non-degenerate sym-
metric bilinear form  . An element of 

à
can be interpreted as a

 ]

9 à 9
finite linear combination of tensors . Consider the Lie algebra
with generators 9 à $"
, where " is central, satisfying the commutator relations
* ] ˜
 ]  , @ ›Ó 
  NJ
˜ ˜ ˜
9 à
9 à 9
9 "


(6.8) S

If ¦
@ 1 is the Euclidean vector space, by choosing an@ orthonormal basis in ¦ , we
see that … 9¦
@  . One defines the Fock space 9¦ Sym ]  Ë * ] ¦ . Its@


elements are finite linear combinations of tensors  JKJJ  4`

)• 4HS` JKJJMH • Ê
 D à
–

•Ej O
 JJKJ  R Sym Ê
9¦ . The @ inner
product  on M ¦ is defined ]  ` by extending
9 9 à


ú

œ *
D


9
9

 œ  ] ]˜
ú
Ë
] ] 4` defining ¦ to the symmetric
product. The Lie algebra … 9¦ ˜ has a representation ˜ in 4 9` ¦ by
9 à 9 à 9 9 à
the bilinear form on

› j O ˜ ] ˜ to be
D 9 à

 ˜ ]  ˜  JKJJ   @  ˜  ]  JK˜ JJ 

9 à 9 à

J
the adjoint of for and letting act by multiplication:

 ] ˜
˜ \ Ôo   œ    ˜  œ 
9 à
] 9
9
] ˜  ú
à 9 9 9
ú
à

9 à

\ Ôo  JKJJ  ]  @ Å \ #

Also we let act by


9 à 9
9 à 9


¦ Ë is a fixed linear % form. It is easy to see that, in the case ¦ @ 1 , we 
ú c

 œ
get a representation isomorphic to .

where c
ic

There is one more important requirement on the spaces . The Lie algebra
@ %
of the Poincarè group å of the Minkowski space
% 1  hc

representation in these spaces. Recall that å is the semi-product ` 4

 `

must have a linear
% and the orhogonal group@ ( F o~
oB . The Lie algebra ofof å theis@ translation v R
group
v v % @ 1 @
the direct
ò

 1— satisfying H
product of the abelian algebra
O . It has a set of generators 
’ o
KJJKJ4
™(
and the algebra of matrices – 
n



o S
(

Û{ä

L ’

s  » Å
ß

» commutator relations
 Û{ä ä ä[Û
ß

L ä

*
, @ O
’
@ o
KJKJJ4
™(S

and satisfying the



 L *  Ê
, @ ÊF Ê
s iÅ
(6.9a)

*
Ê , @  Ê  L F ÊL  F L  H Ê
s (6.9b) ß s ™ß s

 L    L  L     L  @ o L   @ o o (6.9c)
s ß s s ß s s

@ o.
ß ß

corresponds to the matrices ¦ F ¦ if ’ and F if ’

Define L the operators L  L 
s s s '
Here , where

@ Ž [ o
î Q FEQ FΒ ~ •   F  
 
Û{ä Û
¯
ä ä
¯ (6.10) Û
Gº Û º
ä
º
ä
º
Û

`
*
î <; , @ Fž’ ; H ’ =;
Then one checks that

¯ iß ¯
ß ¯
Û ä Û{ä Û ä
 65

* î
î ; , @ žF ’ ,; î H ’ =; î H ’ î =; FΒ î <; J a

ß
a

ß
a

ß
a
iß
a


 î is a representation of the
Û{ä ä Û Û ä ä Û Û ä

This shows that the correspondence

%
 (one has only to replace the commutators ¯

*
, with ’ *
, ).
Û Û{ä
Poincaré Lie algebra in the Fock spaces ¸Û hc ¸Û{ä

Here the operators Q F oB Q correspond to the matrices 1 in the natural linear


@ Q BW A F×Q W . We can inviewtheaspace ( F ~
o 

of vector fields in L as the vector fields
¯

% as a polynomial function
¯ s

state ð ƒ7 Cç 8 R
Û ä ä Û U
representation of @
> ?

on Sym & 
F
Û{ä  with values in Sym
Û ¶
¶

ä ¶ 1¶
ÿ U
ic


ic

such states. The translation part of the Lie algebra of the Poincaré group acts via the
so that the vector field acts naturally on F
Û{ä

operators . ¯
Û
Remark 6.3. Recall that an irreducible linear representation of the Poincaré group is
%
subgroup Ù . Since the latter is an abelian group the linear space decomposes into the
described by the following data. First one restricts the representation to the translation
@ % @
%

@ SO • R F o~,
where
E D Ù ‚
%
D %
R
]  F ] š
HG ] R
Ù J The Lorentz % D group B
o Ù
direct sum of eigensubspaces # F   9 ´9 49

set F R Ù ;

@ O is an orbit of acts. Leton be. Ittheisisotropy
! 
g
!!
easy to see that the

some F \ R
  ' 
subgroup of Ì g .
Ì
% @D . Now
. Then the restriction of the representation to
@ ˆ lifts to an action
defines an irreducible .

Ü % @ Ü % ˆ

¦  . There is a natural
representation of in the natural action of on
. A g Ì g .

æ š
< @ bundle action of on the space XY M¦ of sections
@
on the vector
D D , where acts on the product by
g A g A . .

` representation % is realized as an irreducible subrepresentation

Ê  9 Ê ËÊ 9
g

of XY M¦ .
of this bundle and the
%
vector ð . The translation group acts via the operators \ . This shows that ð is
For example, consider the irreducible representation which contains a vacuum
Ù
an eigenvector corresponding to the character =RV 1 . This shows that the fibres
 
c º Û c

of the vector bundle ¦ are one-dimensional and the group

@  • F o~
oB acts c
 

identically on the fibre over ÓR

@  F • 
o K

7
o —
 . áò a

representation consists of the orbit of determined by ð6ð ½ð6ð g (if the norm is positive
. Thus the data describing the
c Ì ò c

then the group SO • F o~

oB
@ SO • F oB is compact) and the trivial representation c c

of SO F • ~
o 

B
o . Physicists say that ð transforms like a scalar.
a –

c


We can define the similar space corresponding to right movers. Its vacuum
%
state is denoted by ð . Then we consider the tensor product  ~ % % . Its vacuum  hc
[à

state is ð ð .
c ic  ic
 ‡ à 
c c ~
Its vectors look like this

ƒ Ê   Ê é é       Ê JKJKJ Ê ð  JKJJ ð
  ‡

 œœ JJKœ œJ þ þ and œ Zœ ˜ Z  JKJJ  œ Z œ and the þ _ þ polarization tensor _

à  ~
 º Û º Û  c º ä º ä c

where 
Û ŒŽŒŽŒ Û ä ŒŽŒŽŒ ä
 ŒŽŒŽŒ   ŒŽŒŽŒ   

` ` ƒ @ 9ƒ Ê   Ê 4é` é     `
r r r

}
is symmetric in and 9 (resp. in andœ ) separately.
œ œœ þ þ Û ŒŽŒŽŒ Û

ä ŒŽŒŽŒ ä

%
One defines the norm on L similar to the norm on and then gets a non-
%
󌎌Žs Œ ŒŽŒŽŒ
r

degenerate inner product on

% %
. Finally we complete this space to get the
 ic

hc
à  ic
hc
66 LECTURE 6. FOCK SPACE

@JI º % % pJ
Fock space of the closed bosonic string theory
D E E
àE  ic
closed ic
a  ê

% %
֏
Finally, let us see the representation of the Virasoro algebra generated by the oper-

֘ @ o
(resp.  ) in the space hc (resp.  ic ). Recall that

\ c \ \ H ­
\ @ co \ \ H ­

ators

Ö Ö
3º Û º
Û
 3º Û º
Û


­ @ Ž  
­ @ Ž   J
where


 
º Û º º Û º
6 6

` `
Û Û

@ Ž  
@ Ž  
› @ O
֘  ˜ ֘  ˜  '
º Û º º Û º
6 6

` `
Û Û

The operators ­ and ­ are called level operators. It is easy to check that


­  Ê JKJKJ Ê ð @ HÔJJKJBH  Ê JJKJ Ê ð   ‡   ‡

œ  ` œœ 
º Û º Û  c hr r º Û º Û  c

and a similar formulaœ holds for ­ .

Recall that *
@ v =
H à
 , @ i›ñF •¥  H i› id. We have to find the constant 

9› › ” Ö * ˜ › .
Ö Ö˜‡ n

,
n
One applies
֘ Ö˜ to some ground states to compute these constants. Notice

@ c°o Ž   ð @ O
› j O
that

֘ ð  `/
c

(6.11) º Û º
Û
c


{€

\ ð @ co \ \ ð @ co ð6ð Åð6ð g ð
Ö
  
c (6.12) º Û º c c c
Û

@ co Ž @ co \ H \ ð @
Also

Ö 4` ð 
 4`#   ð ð #J   

4` ` 4`

c º Û º c Gº Û º º Û º c c º Û c
{€ Û Û Û Û

Here we used that the operators

Ö ˜ and
o Ö  ˜ are adjoint to each other. Thus

Åð Ö Ö ð @ Ö ð p
Ö ð @ ð6ð c g \ ð ð6ð g @ ð6ð ð ð6ð g @ ðð ½ðð g

      

` ` ` `  `

c c c c º Û º c c º Û c c
Û Û
¡

Åð * Ö
Ö ,“ð @ Åð Ö Ö @ ð6ð ½ð6ð g @
and we obtain
   

4` ` ` `
c c c c c
 67

½ð! c Ö \ H I o7 id ð @ ð6ð Åð6ð g orH oB g pJ

 

@ v Heà @ O . Also
c n c c n

This gives I oB

½ð Ö g Ö g ð @ ð6ð Ö g ð ð6ð g @ o ð6ð c g \ H

n


ð ð6ð g @
  

   ` 4`
c c c º Û º º Û º c
Û Û

ð6ð g \ ð ðð g H o ð6ð ð ðð g @Vc ð6ð ½ðð g H co (SJ

¡

 

 4` `
º Û º c º Û º c c
Û Û
¡

ð ðð g @ ½ð ð @ Åð H ð @
Here we used (6.7) and

ð6ð     

4` ` H ` ` 4` ` @ c ` ` ` @tc 4` @Vc (SJ

º Û º c c º º Û º Û º c c º hº Û º Û ß º c
ÛãÛ
Û Û Û Û Û


½ð ð ½ð ð   

` 4` ` 4` ` 4` ` 4`

c ß º º º º Û º Û º c c ß º º c ß ß
ÛãÛ ÛãÛ
Û Û Û Û ÛãÛ Û Û ÛãÛ

Åð * Ö g
Ö g ,“ð @ Åð Ö g Ö g ð @ ½ð Ö \ H c id ð @
Thus
     

 
c c c c c n c
¡

c ð6ð Åð6ð g H c @Vc ð6ð ½ð6ð g H co (+J c n c

This gives
c L@ K v H c à @ ¤ v @ ( , hence v @ ( ˆ<o c . Finally we obtain that
for all › ™

• R m , we% have the followingg` commutator relation for the Virasoro operators
n

acting in the space .

*
 , @ i›úF •¥  H o o c ( 9› ” FΛ   J
hc

֘ Ö Ö˜‡ ˜  (6.13) S

Exercises
@ 
iQ
-
6.1 Let   be a graded Heisenberg Lie algebra. Let Sp # be the symplectic

r % v )
group of linear automorphisms of  which preserve the alternating from
. Con-
% v
struct a linear projective representaion of Sp & in the space M
ible with the representation of  in
which is compat-

Ö ” ð Ê JJKJ Ê ð .
.

6.2 Compute the norm of the state c .


6.4 Let ½ •¥ denote the dimension of the



œ œ space of eigenvectors operator ­

6.3 Compute the norm of any state º Û º Û  c

with eigenvalue . Compute the generating function Tr ë

• ¯
ã @ of ~[ the \ level
Å •¥ 㠏 . Explain u ¯
the notation.
68 LECTURE 6. FOCK SPACE
Lecture 7

Physical states for bosonic string

The expression for the Hamiltonian provides the mass-squared formula . Recall that
in the special relativity theory the mass is defined as the negative of the norm of the
1 ÷ i
Q \
Q
Q
Q
moment vector in the Minkowski space-time. Let us explain it. We use the metric in
@ @ iQ \
ß D Q \g F D Q g F
D Q g F D Q g” Q \ ] ]
the space-time with coordinates
` g
defined by
” `
g
þ
. Here , where is time and þ is a constant equal to the speed of light.
To describe the motion we use the Lagrangian density

@ Fž› F|ð ß ð D ] @ › g y o F g D ]

Þ
9

g wg
„
þ þ þ

@
@

g þ

where ß w ß 9Q Q g Q ” and › is a constant called the mass. The energy and

U Š

` are equal to
U

a Lagrangian
@ â â Ø @ › g ˆ y o F g
’ @ o
c
Ÿ-

the moment for this Þ

L Q
þ

g
¯ þ

L
¦ @ L QØ F Ö @ › g ˆ y o F g J þ
9

L g
¯ þ

¦ g F g ð Nkð g @ › g ÷

We have
þ þ

O
@ M¦ ˆ g
ˆ ‘
g ˆ ‚
” ˆ K #

so if we set

` O
ãþ ãþ ãþ {þ
¯ ¯ ¯

@ F ð6ð ðð g

we obtain that

› g
where we use the Minkowski norm defined by the matrix diag * F o~
Ko
Ko
Ko , . The vector
O O
@
JKJKJ4
so we can
`

is called the total momentum vector. In our situation ½¯ ¯

69
70 LECTURE 7. PHYSICAL STATES FOR BOSONIC STRING

define the quantum mass-square operator by

 g@ F @ F \ \ dorHÔo \ \ #J do F o
à 7à à 7à

@g K g so that
¯ ¯ ¯ ¯ ó¡ hº Û º º  Û º
Û Û
Û Û Û Û
 ó

 g @ GF \ H ­ ÔorHdo )F \ H ­ #J
We shall scale the masses to assume that

Ö ð isÖ equal to
3à 7à  

Thus the mass-square of the ground state ð  ‡ à  ~

F ð Åð g @ g F gg HÔJJKJ F g J
c c

` had Virasoro constraints Ù @ O . It follows

c c ¥c ¥c 

Recall that in the pre-quantized theory we »{â

@
* 
 , @ @c •Ö ˜ \ H id,O forthisanywould
from (5.15) that the analogs of these constraints in the quantum string theory are the
conditions that /
@ O . Thus we have to require
element of the Fock space. However, because
@
/

Ö Ö Ö  O onlyÖ for positive › and Ö \ imply that

v for some 0
and similarly for the
>þ /

˜ Ö . We set
/ / / /
that

˜ @ R ž
@ O
› j O
\ F v @ O
operators 

D É / D !

@ R k
Ö ˜ º @ O
› j O
Ö \ F v @ O
/ /
phys hc (7.1a) ic
D  / D !

@ Ö ˜   Ö pJ
  hc  /  4/
phys hc (7.1b)
D D =à D
closed
(7.1c) ê h® 
phys a  phys hc phys ic

physical state & , we have

A state satisfying these conditions is called physical. Also for any state and a F

&Àð Ö  F @ Ö  '& ð F @ O
•Îj O  ²  ²

 with the sum @  \  belongs to the null-

D D D
Thus the intersection of
space of
D
. The elements of this space are called spurious
phys
Ö  states. The Hilbert space
phys spur |

@
which we want will be the quotient
ˆ J D closed D closed D T D closed

‡ in (6.9) commute with Virasoro operators, so that

phys spur phys

Note that the operators î defined Û{ä

the Poincaré Lie algebra acts in the spaces of physical states.

by generators N

)• R m with commutator relations
Remark 7.1. An abstract Lie algebra is called a Virasoro algebra if it can be defined
" $ó

*
 , @ 9›£F •¥  H › ” o FÎc ›   N
q* N
 , @ O J
˜ ˜‡ ˜ 
ó $ó ó " " $ó
S

It can be shown that any any Lie algebra obtained as a central extension with one-
dimensional center of the algebra of vector fields on a circle is isomorphic to a Vira-
v %
representation with highest weight and charge if acts as a scalar operator id «
soro algebra. A representation of the Virasoro algebra in a vector space is called a

and there exists a vector \ (called a highest weight vector) such that
þ " þ
9

\ @ O
› j O
\ \ @ v \J
˜
ó 9 Vó 9 9
 71

v
v% %
K
A universal representation with this property is called a Verma module and is denoted
$þ
by . It can be constructed by using a similar construction as the representations

{
…o @ O
› Ö j ˜ O › O ‡
we constructed for a Heisenberg algebra. One considers the subalgebra Vir

\ …o @ v

generated by the operators , then defines a one-dimensional representation

JKJKJ \ Û
• Ö Ö˜
Ö v  œ Ö  
by and finally takes the induced representation  Vir $à Vir
9
W . Its elements are linear combinations of monomials ú with positive

L
to a quotient of $þ % v
% K
’s. Any irreducible representation with highest weight and charge þ is isomorphic
. So, we see that each nonzero /
D
phys ic generates a R v
@ (
representation space ³ for the Virasoro algebra with highest weight and charge
þ
%
. Its highest weight vector is / . As we have seen before any physical state F
QP
\ with eigenvalue v , we obtain the
belongs to ³ .
Since all physical states are eigenvalues of
Ö
 g @ F× c \ F c ­ @tc ­ýF c v J
mass-formula for physical states:

Ö (7.2)
D closed

­eð @ O
Ö ˜  ð @ O
› j O
Ö \ ð @ co ð Åð g

Let us see which ground states in are physical. Since

  
c c c c

and @ the same is true for the right mode operators  , we see that the ground state
ð ð ð is physical if and only if Ö

  ‡ à 
c ~

ð ½ð g @tc v J
c c

(7.3) c

 gð @ F cv
For this vacuum state


v
c

We shall see from the next discussion that must be equal to o . Thus the vacuum vec-
tors have negative mass. Such states are called tachyons (they travel faster than light!).
The existence of such states will force us to abandon bosonic strings and consider su-

R ð ƒ
perstrings.
D
Let us look for / of level 1. Each such / has the form º Û c  . We
4`
phys ic
Û

@ O
› jVo
\ @ co ð ½ð g HÔo7
have
@ \ @ ƒ 0

Ö ` ` ֘ Ö
/ / / / /
º º c c
Û
Û
/

ƒ @ O
ð ½ð g @tc v F c J
Thus is physical if and only if

@ 9O
o~
O
KJJKJ
O , and then ƒ @ o
O
JKJJ4
O satisfies
(7.4)
v
c c

If j o , we may choose
Û
Û

(7.4) but ð6ð ðð

@ @ o
𠃂ð g F . This means that we have ghosts, i.e. states of negative
/
c

spaces with unitary inner product. This forces us to take

v o.
norm. This should be avoided since the the quantum mechanics deals only with Hilbert

72 LECTURE 7. PHYSICAL STATES FOR BOSONIC STRING

v o , we may take @ o~
O
KJJKJ
O and ƒ @ MO
ƒ
KJKJJ
ƒ so we have
( F o -dimensional space of physical states of positive norm` and no states of non-
If

c 

positive norm.
v @ o , wec may take @ o
F o
O
JKJJš
O and hence ƒ \ @ ƒ . This shows that
we have a ( F -dimensional space of states of positive norm and ` a one-dimensional
If c

ð @
ð ½ð g @ @
space of states of norm 0. The state
v @
O . Thus, if o ,Ö 4` contains
is spurious and is physical if
` a one-dimensional
c

c º Û

spurious states of norm 0. Factoring this space out we get a ( F -dimensional space
c
D
c c c space of Û
phys hc

ˆ , eachv element
Û
Û

o and no restriction on ( appears.

D D ²T D
phys ic spur ic of which can be represented by a state of phys ic

@
positive norm. So far, we find that
Let F R \ @ v #
whereo R \ #
R \ @ v pJ


o
D D D D

\
/ à / ²à  / / 

Ö thatL ­ weL see@ L ­ that .Ö Hence and applying L LÖ

u phys ic phys hc phys hc phys ic

Ö
à / +à   /

L L
Applying we see that .
/  /

F @ ƒ L   éL  
This implies

Ê œ   Ê é _ œ   _ þ œÊ K
J 
J J Ê œ_ 
J K
J J ð #
  

_


HdJJKJBH  @ œ Z HÔJœ JKJ‚H þZ .  Letœ us look at  theœ physical  states in

Û Û  ä ä º Û º Û  º ä º ä  c
ŽŒ ŒŽŒ ŽŒ ŒŽŒ

` i.e. ­ @ ­ ` @ . They˜ are of the form

ŒŽŒŽŒ ŒŽŒŽŒ D closed
where r r phys hc
/  / /

@ ƒ
of level 1,

ð pJ / 

4` `
º Û º c
ä
Û{ä

@ \ 9ƒ ™ @ ƒ @ O
› jtoJ (7.5)
We have

ð ð #

Ö` 4` Ö ˜
/   /

` 4` `
º Û º º Û º Û c c º c
Û ä
Û Û{ä Û{ä

@ ƒ @ O
› j o~J
Similarly,

ð #

Ö Ö ˜
 /   /

@ @ ð `½ð HÔo7 0
sothat`
(7.6) c º Û c

Also \
ä
Û{ä

Ö Ö g` g
/  / 4/

ð ½ð g @Vc v F c J
c

(7.7) c

ƒ @ ƒ @ OJ
In view of (7.5) and (7.6), we get

c c
Û ä
Û{ä Û{ä
/

@ ƒ ƒ @ ƒ ƒ J
The norm of the state is equal to
R s

@ O , we have no restriction on ƒ L andL hence we have physical states of negative

s ß ß
Û ä Û{ä
Û{ä Û{ä
If c
@ O . If v o , we may assume that @ o~
O
JKJKJ
O so that ƒ \™\ @ O
guarantees that is physical. But if we take ƒ
norm. So c
'
/ @ O for ’
@ O , we get a state of 
Û{ä
s
c
iÅ '

v @ oJ L
negative norm, a ghost. So

@ @ , and then the condition is ƒ \ @

ƒ ƒ \ O@ , weƒ
may
If c
'

KJJKJ4
™( thatF o so that o
F theo
norm
@’ Oassume O
JKJJšis
O equal
 ƒ g . We seeL that c

g
6

L` L ` L
to u
Û ä Û{ä
 73

}
{ c
all physical states are of nonnegative norm. The states of zero norm satisfy if ƒ @ O
9
ƒ@ c v ð @ ð @ ƒ
c @ ƒ O 4 ` ` ð Ö ` 4 ` ð
. Û{ä

Ö
   

ð Åð g ƒ @ ð ½ð g g ` @ O 4` F
ð ½ ð
Note that the states and 
4` `
º c c º Û º Û c ß º Û c c º Û º Û c
ä
are spurious. Since c ä Û ä Û
, these states are also physical because ä Û

c u c u ß . It is easy to see now that any physical state of norm 0

Û
is spurious
the space phys
Û and we can
D closed
hc
Û
( g F c ( F c ( @ ( ( F
Û factor it out. Thus we obtain that for any non-zero light-like c
is of dimension and all its elements

ƒ @ 9ƒ
¡
can be represented by physical states of positive norm.
( ( oB
( ( F Ÿ g` F F
For any c of norm 0, the space of solutions of (7.5) and (7.6) is the
o
direct sum of one-dimensional space of matrices with nonzero Û{ä trace, the

g`
-dimensional space of trace-less symmetric matrices and -dimensional
 g @ O
space of antisymmetric matrices. The corresponding physical states are called dilatons,

/
gravitons and anti-symmetric tensors. These are massless particles (i.e. ).
D
Let us go to the second level, i.e. consider the physical states in phys ic of the

@ ƒ
form

ð ½H g ð p
/  

@ ƒ . We have 4` ` 

º Û º c ß º Û c
ä

where ƒ
Û{ä Û
s s

L L @ \ H @dc 9ƒ H ð @ O

Ö` ` ` g
q/ / 

`
hº Û º º Û º c ß º Û c
ä
Û Û Û{ä Û

o
 ð ƒ
@ \ H c @ 9ƒ H c ð @ O

Ög g ` `
 4/ 

4`
c hº Û º º Û º c ß º Û c
Û Û
Û Û Û Û

o o
\ @ c°ð ½ð g ÎH c@ crð ½ð g H c J
Ö
/ / 4/
c c

H @ O
ƒ H c @ O

This implies

ƒ Û{ä
c
ä
(7.8)
ß
Û
Û
Û
c
Û
ß
Û

ð ½ð g @tc v F @ F c J
c R , where
c
¡

ƒ ƒ H @ ƒ \g \ F c Ž \ ƒ \g H Ž  \ ƒ g F \g H Ž \ g J
The norm of this state is equal to
R @
ß ß ™ß ß
Û{ä Û
Û{ä Û | ä | Û{ä | Û

Choose a system of coordinates in 1

@ ‚
O
KJKJJ4
O #
g @zc . Using ä Û ä Û

(7.8) we can eliminate ’s and ƒ \™\ so that

 þ þ
such that c

LR @ o ûŽ
ß

F c S  ` ƒ g H Ž 4 ` ƒ g J
 
s

L ` L6L ü L ` L
(7.9)
s
74 LECTURE 7. PHYSICAL STATES FOR BOSONIC STRING

Applying the Cauchy-Schwarz inequality, we obtain

{ Ž g ¹ ƒ g F ( cF S o Ž ` ƒ g @ Ž g ¹ ƒ g F ( cF S c ¤ Ž 4` ƒ g J (7.10)

R  

 L  L6L     L  L6L
s s
s s  s

Thus, if (
L c ¤ ,` all states are ofL non-negative` L ` norm. L If (ýj c ¤ , theL ` state c withcS ƒ @
@ @ @
o~
’ {to and ƒ O
’
O
’ has the norm equal to ( F o7 GF (VH ¤ ™ˆ L6L O .
 /

of level 2 and of positive norm,

s iÅ' ' Å

Now if we take L a physical state from

of negative norm. So we have ghosts. If (
/
@c then
¤ ,a
D

phys hc


@ O
’
{ o~
’ @ . The states
/à / D

state of zero norm must satisfy ƒ

closed
is an element of phys hc
s
Å ' Å

9
ƒ 6
ð ð ) ½H ð L @ 9ƒ H c ð6ð ¥H ð þ

Ö 4` ` Ög g

  þ   ¤þ 

4` ` 
º Û c c c 3S º Û º c c º Û c
ä
Û Û ä Û{ä Û

are spurious. It is easy to see that any norm 0 physical state is equal to a physical

norms. Thus we have shown that (

spurious state. So we can factor them out c ¤ . and obtain only the space with only postive
The proof that (
@Šc ¤ consists of analyzing states of the next level. We skip it. 

The result that

D closed
does not contain ghosts if and only if
v @ o~
( @ c ¤ is called
v @ o and ( @Uc ¤ agrees with the definition of[ the Hamiltonian
phys
the No Ghost Theorem. It was proven by R. Brower, P. Goddard and C. Thorn.
Observe that
@ \ H @ \ F ÷ g using the regularization of the sum ‘ • . So physical
Ö Ö O . g  
operator . u
states satisfy .
/
` invariant with
c
respect to SO such that its states represent all states of positive norm modulo spu-
D closed
Remark 7.2. One can show that one can choose a subspace in phys

c S
and the last coordinate & of the string. The group SO o
acts in the space in
¡
rious states. This is achieved by a “light-cone gauge” which consists of fixing the first
D closed

 c . It also leaves the homogeneous

Û D closed

‡
via its induced representation. Thus defines a linear representation of the group
ò parts invariant and hence defines a finite di-
¡
mensional representation in each space of given level. Elements of this space which

ple, the anti-symmetric tensors of level 1 define the adjoint representation of SO .

belong to an irreducible component are interpreted as elementary particles. For exam-
c
sentation of SO on the space  g 1 g .
÷
¡
The dilatons define the trivial representation and gravitons define the standard repre-
¡

Exercises

7.1 Find physical states of level 2 in the Fock space of a closed bosonic string.
D
7.2 By analyzing physical states of level 3 in ic finish the proof of the No Ghost
Theorem.
Lecture 8

BRST-cohomology

We shall discuss another approach to defining physical statesd which is called the BRST-
@
D

g O@
quantization. In this approachd
one introduces an operator in a Fock space of a
given string theory such that and
D phys
Ker  ˆ Im
d d
(8.1)

We shall start with reminding the definition of the cohomology group of a Lie
% @ ®  %
%
D
algebra  with coefficients in its linear representation . Let   +à be

be an analog of our Fock space. If 



Lie Gg for some Lie group g , then

the tensor product of the exterior algebra of the space  and the space . This will
D
can
@
be identified with the space À inv‡
of left-invariant smooth differential forms on g with
 @  % @ [ 
'¿ ï
values in the trivial vector bundle defined by the space . Let
 %

® B
be a basis of  and
%  L ‘
  \ ï
%
ï
%
D
be the dual basis of   . Let    £à 
L
so that .
Elements of  are linear combinations of the decomposable tensors

Ð ¿ JKJJ ¿ š
’ JKJJ ’ Ê
R % à 9 Ð 9

L v L @ @ ¿ and` to the contraction operator v@ @UT



Let us assign to ¿ theœ operator

 

M¿ L ¿ @ L Ð GF L o7 Ê ‡Å` ¿  L JKJvJ ¿ JKJKJ ¿ if ’ L ’“Ê for


J K
J J 
à à ±
T ± à#9 E à¿9

some and O otherwise. L œD the

Koperator
v IfL @  W Lie. Here L Jœ
Q L L isarethelocalexterior
L coordinates
such that  ià 

the differential Q and

r à Ã

à Q 
J K
J hg product with à Ã

L L L@ ¶
¶

`
on
ØfÙ Ú
g

v
v
v
v @ v
v @ O J
corresponding to the basis . We have
 !  ! É s !

L L L , L
s S s s

Here, for any associative algebra and Q

R

Q
@ Q H Q
n Xn


!  

The structure of  is determined by the constants Ê such that

It is called the anti-commutator or the Poisson bracket.
þ s

*
Ê, @ Ê Ê J L
L  L 
þ s

75
76 LECTURE 8. BRST-COHOMOLOGY

Ë @ š  . Define the BRST-operator in by D

L L @ Ž v Ë F co Ž Ê v v v Ê Ôo R End % pJ
Let
d s D

L L  Ê L L
à þ s à à
s

L L
@ O
Lemma 8.1.

@ v Ë , @ Ê v v v Ê to , where we skip the summation sign. g d

` L L @ vg v L L v v
à þ s à
Proof. Let n n

g` L Ë L Ë H L Ë Ë L @
We have
s s
à s à s
n

Ž v v MË Ë FÓË Ë @ Ž v v Ë J
L L_ _
s s

L
à s s à þ s

L
!
Ls
L L
! s

@ F co v v v Ê v H v v Ê v v Ê Ë J
Also
H s s

g g ` L a v v v Ê v a H L v v Ê v a v @L
=à þ s

`
n n n n

O unless ]
@ ’.
v v v v
Ê H Ê v v v v @ v v v v v v
Ê L H a a Ê L so that @ v v
s s
Using the anti-commutator relations we see that s s s s
In the latter case

H L @ F L co Ê v v L Ê H Ê v Ê v L L Ë @ L L F Ž v v Ê Ê Ë J
L s s s

` g g ` @ L v v @ v v L L! L L
þ s  þ s =à à¤þ s
n n n n
s

Here we used that Ê F Ê and @ Ê F Ê . This shows that g H g H

@ O . It remainsL to show
s s
þ s þ s

O . We have `v v `
n n n
n
g ` @ ¿ Ê ¿ v v v Ê v v v L that
n

d o @ g g Ê v v v v
¿ ¿ _ ˜ L Ê _ a ˜ _ a ˜ L Ê do~Jv v H v v v v
n

g@ g

L _ ˜ @ a L
_ a ˜
s s s
s à s =à

Z ] › ’ , the expression in theL bracket a is equal to zero. So

n
¡
' '
Å

gg @ ¿ L Ê ¿ Ê ˜ a v L v v Ê v Ê v a v ˜ H v Ê v a v ˜ v L v v Ê do7H
If r
s s
s 3à
n
¡

¿ Ê ¿ _ ˜ Ê v L v v Ê v _ v Ê v ˜ H v _ v Ê v ˜ v L v v Ê do7H
s
s s
=à

H ¿ LÊ ¿ ˜ v ˜ v v Ê v _ v v H v _ v v v ˜ v v Ê do7H s s

˜ _a a˜ a˜
s =à

¿ Ê ˜ ¿ _ ˜ L ˜ Ê _ a v ˜ H v _ v a v ˜ v L v ˜ v Ê do~J
v v v v v =à

L a
¿ ˜ ¿ ˜ Ê H ¿ ˜ Ê ¿ ˜ H ¿ ʘ ¿ ˜ @ O

Using the Jacoby identity

s s s

it is easy to see that each of L the four sums isL equalL to zero.
 77

%
Applying the previous lemma we can define the cohomology of the Lie algebra 


% @ Ker ˆ Im #J
with coefficients in as follows:
 d d
.

Example 8.1. Let  be an abelianÐ Lie algebra of dimension • . Its linear representation
is a module over@  
@ *
JKJJ
 , . Let % @ ï [ 1  with the action of 
defined by  Ë
¿   ` ¼ . Then• ï  
% can be identified with the space
– W

 L L forms of degree
¶ À

@ ¿ a iQ D Q JKJJ D Q J
¶
of smooth differential

Lœ L Lœ L
˜  

ŒŽŒŽŒ à Ã

@ Ž v Ë
The BRST-operator
d

L L
à

D L D g @ O and
  
% @  1  @ O
™•;j O J
coincides with the exterior derivative . We know that

@ Lie . Assume that is a complex semi-simple group and let

. . DR

@% be its trivial representation.

Example 8.2. Let  hg g

  
@ 
#

W Then

. W – . hg W

where is considered as a smooth manifold.

g
%
tation in the Fock space of bosonic string. The space ® 
In our situation we want to take for  the Virasoro algebra Vir and its represen-
D  is called the space of 

ghost fields.

cohomology. Instead of differential forms ¿

We will be dealing with a version of the BRST complex which uses the semi-infinite
@ JKJKJ ¿ we consider semi-infinite
forms. Let Ò \


K
J 
J 
J
\ 9’ and’ ` m ! \  be MÒ anym strictly L L of integers such that
A ú

the sets Ò m ! \ are finite. A semi-infinite form is a formal

decreasing sequence à Ã
T 6 T

@ ¿ ¿ JKJKJ¥J
expression of the form
/

@ ³ Ò m \ F ³ m n \  L 9Ò L mœ n \ is called the degree of  .
A

The number ­ T Ã T Ã /
&|

@ 

Let

ë
@ 9­
­ F o~
JKJKJ #J Its degree is equal to ­ . We extend the operators v Ê and
/ /

where Ò
v Ê to semi-definite forms in the obvious manner. Note that any form of degree ­ can
be obtained from ë by applying operators
/ v Ê v  . Also observe that
v Ê @ O
j 0
/ j /

‰,+
r
78 LECTURE 8. BRST-COHOMOLOGY

vÊ @ O

/
F j 0J
/

‰V+ 
r


Let Vir be the abstract Virasoro algebra with generators ó . We want to construct the
[ 8 g 9­ @ ­ *
w
representaion of Vir on the space of semi-infinite forms À of degree . Let ad
w i Q '
¿ - ¿À Q ,
M¿
be the coadjoint representation of  on   . It is defined by ad 
.
ó

L @ ¿ * L
 , @ ¿ ) i›úF •¥  @
Let us identify with the dual basis of Vir. We have

ad w  M¿
L ˜ @ L ˜ ‡˜
ó ó ó $ó
ó

L@
i›ŠF •¥ ¿  ˜ 9›ŠF •¥   ˜ i›úF •¥ ˜   J
ó

L ‡ L ‡ L
S S

@ 9’ÅF c •¥ ¿ 
This shows that

adw  M¿ ó

@ O , we can set L L
(8.2)

If • '

  ¿ó
¿ JKJKJ @ ʎ \ ¿ JKJJ š  '¿ JJKJ @ ó

L Lœ L L
A A ú
6
à à à à Ã

Ž ¿ JKJKJ ¿ ÿ 9’¥F c •¥ M¿  ¿ JKJJJ

Ê \ L
6
A

L œ @ L  Lõœ
ú ú ú
Ã
•
Ã

O . However it is not defined for •

Ã
@ O . We à Ã

easily check that, for •—

›
™•ÍH ›
Observe that the sum is finite because
@ O, '
¿'

*   p
 , @ • FΛ š  #J
˜ ‡˜
ó ó ó

So our problem is to define š \ such that all Virasoro commutators work. Next ob- ó

@ Ž i’½F c •¥ v  v @ Ž F •¥ v Ê v  Ê @ Ž • F v  Ê v Ê J
serve that

 
L L Ê Ê
ó

‡ ‡
ir ¥r

L
{€ {€ {€

@ Ž
vÊvÊ

We use this formula to set

š \
Ê
ó
r

Here
JKJJ Ê
denotes the normal order of a composition of operators defined by

{€

—0 Ë0

putting on ` the right the operator annihilating the vector ë and inserting the sign of /

we do nothing. Note that F Ê Ê

v v @ v Ê v ʞF o so changing v Ê v Ê to F v Ê v Ê differs from
the permutation which has been made. If no such operators occurs among the factors

the usual product. It is easy to see now that each   is well-defined.

Let us compute * š  #
š , . We have
ó

*   #
š ˜ , @ • FE’ i›£F < p* v  v
v v , J
ó ó

˜ ‡L L ˜‡
ó ó Å s
 79

@ O , ’ @ › H 
@ • H ’ . Then, its is easy to see that *!
v  v
6
K

v v •—
,
@ › O . Assume
Assumes
•—
› @ O , ’ @ › H 
@ •§H ’ . Then
' '

‡ L
 Å
Å

L
'

˜‡
' šÅ iÅ?'

*
v  v
6
K
v v
, @ v  v v v F v v v  v @
s
s s s

‡ L L v ˜v ‡ H v v v  ‡ v L L @ L v  v L J ‡ L L
s

L s s

L L L ‡L ‡L
*
v  v
6
K
v v
, @ F v v
Similarly we get
s

@ O
@ •æH ’
’ @ › ‡ HL L . Note˜ ‡ that @ •æH ˜ ’
‡ ’ @ L › H implies › @ F • .
s s

if •—
›
Thus, if ›
'
@ F • , we getiÅ ' šÅ Å ™Å

*   p
 , @ Ž • FˊF  i›úF < v  v F
'

˜ ‡˜‡
ó ó Å Å s
s

Ž 9›£F • FΒ v  v @ • FΛ v  v @ • FΛ š  #J s

˜‡ ‡L L ‡˜‡ ˜‡
s ó

Assume › —
• L
@ •
O and Fž› O . Then
'@ j
*   #
š  , @ c • F < GF • F < *
v v 


v  v
, J s s

­ we have  ‡  ‡
ó ó Å Å s s

Now, for any ’

such that ’ j ­

Å
Å 

* v v
v v , @ v o F v v v F v o F v v v @ v v F v v @ o7H
v v
F
v v J
s s s s s s

L if ’ L ­
Íj ­ , weL have L L L L LL
s s s s s

 iÅ

* v v
v v , @ F o7H
v v
F
v v J
Similarly,
s s s

LL
s s

L
*   #
š , @ Ž c • F  GF • F < o7H
v v
F
v  v 
@
Now
s s s

˜ ë! n ë  ‡  ‡
ó ó Å Å s
s

‡
c • š \ F oŸ~• ” F| 9­ g Fέ H • #J
¤ ¤ ó

*   #
š , @ • FE›   H F oB¤ Ÿ • ” H M­ g H ­ H •¤   J (8.3)
Finally

˜ ‡ ˜ @ o ) the central charge is equal ˜ to . Also,

ó ó ó
S

(i.e. take ­
 \ ð
If we fix the vacuum state
@ . Recall that` the representation of Vir in has the charge g` g¸ and the /

` ð ` is the eigenvector of Ö \ with eigenvalue v @ o . [ 8 ` g

ó / / D 

vacuumvector c


g 9­ by à D

o
We define the BRST operator (Bechi-Rouet-Stora-Tyutin) on À hc

@ v   F c • FE› k
v  v ˜ v 
Io~J
d

Ö ‡˜
à %à
80 LECTURE 8. BRST-COHOMOLOGY

( d@ c ¤ , then g @ O . d

Proof. Let  
@   . We have
Theorem 8.1. If
ó

@g o Ž v    v Ê  Ê H v Ê  Ê v    #J
g Ê !š n
¡

  v Ê @ • FE› v ˜ v  v Ê @ v Ê   FP c • F v Ê  J
We use that

‡˜  r

v    v Ê  Ê H v Ê  Ê v    @ v  v Ê *  
/ Ê ,bF| c • F v  v Ê   Ê J
Using this we get

 ¥r

Changing the index to HA• in the second sum, and applying (8.3), we get,
v    v Ê  Ê H v Ê  Ê v    @ v  v Ê *  
/ Ê,NF| • F )  Ê @
r r

‡
¥r

c ¤ • HA•¥ v  v  J
o c )F ” 
o v v V* 
, @
On the other hand,
@g v   v ˜ F c
Ö Ö˜ ˜ ˜ Ö Ö˜
à < à à

`
n

co • FE› v  v ˜ Ö  ˜ H c( • ” F •¥ v  v  J à

Finally,
‡  ¡

H @ F FEc › v 
v Ê v ˜ v Ê
 H
v Ê v ˜ v Ê
v   H
‡˜ Ö Ö
r

` g g ` ‡˜
%à à
n n n n

FÎc › v 
v ˜ v Ê v Ê
 H
v ˜ v Ê v Ê
v   @ F IFEc › Ê J

˜@Vc Ö Ö Ö ‡˜
r r

˜
%à à à

So H @ O if ¤ .( ‡ ‡
[ g g
` [ 8 g •¥ be the linear[ space of semi-infinite forms of@ degree • . It is clear that
hn n

Let ®
maps ® 8 •¥ to ® 8 g •×HÝo7  . Let Vir ç Ker  ˆ Im  ,
@ g [ 8 ¥
•
ð ® g . We set Vir ç \ be\ the subspace of Virç `
d D D D d d

where 
à à
.
d d à D D D

generated by the cosets of forms which do not contain ¿ and which are annihilated by
. .

š \ dorHdo
rel

\.
Ö % @ ‘[  \ %  with Â6à %  we set
ó =à 7à

[ Âà % 㠏 J
For any graded vector space

ã @ Ž


char
%
‘ \  j
 81

Ù
%  % R % ‘
Ù  @ • , then Âà %  @ Tr Ù ð %  and

% @  9 9 K9 ! j

% ã @ Tr ã #J
If such that

char 

% @ and Ù @ \ . Recall that D

Ö
\ v  JJKJ v  ð @ co ð Åð g H • HdJKJJBHA• Ê #J
We shall apply this to the case when ic

Ö œ 
ú 

`
Û Û ú c c

So it is easy to see that œ

char ™ k
Tr
@ ã @ ã &½ ã J
D ‡

1hœ W W
h 
A a 
ic

&½ ã @ Ê \ o F ã Ê pJ
where
l

We have already noticed that the representation of Vir in is reducible. Let us try
´|
D
hc

@ w w²w #

to decompose it into irreducible modules. First we write

D #D =à D

@
KJJKJ
÷ p
w²w @ \ . Let us assume that w @ O
w @ O . Let  ¥

hc hc ic

` moduleg for the representation of Vir with central charge and charac-
' ' Ê þ
where c ic c c c c c

ter (see the previous Lecture). The Verma module  ¥

K has the universal property
þ
denote the Verma
Ê Ê þ

such representations is a quotient of the Verma module. One can show that  ¥

þ Ê
with respect to all representations of Vir with central charge and character . Any

 JKJJ Ö  ð O , • j O . The grading of  

K is de-
Ê þ

 ¥
K Ö
• HdJKJJBHݕ Ê @ • . We  have
 Ê $þ

fined by taking  theœ subspace spanned L by the monomials as above with

is spanned by the elements ú
Ê þ
to be

` char  ¥
K ™
@ &½ ã J
4`
Ê þ

It is known that the Verma module  ¥

Ko7 is irreducible if O and irreducible and
unitary for ljo
+j o . Considering w²w as a representation of Vir with character
Ê Ê

@ F w g and central charge @ o . Comparing the characters, we find that

þ $Ê D 
hc

g`
Ê þ

w²w @  )F c°o ð w ð g
Ko7 #J
c

D
ic – c

v c S
The charge of the representation w is equal to
D

ð` w ð g . We have and the character is equal to

hc

g
c

char w )
@ ã É & gJ
h@œ W W h  c¹S
D a

ã É Ê &½ ã .
hc

The character of the irreducible module  ð w ð H

is equal to
g` o g œ h™‡ 4`
@h W W
c r
a

@ Ž  ÷  ™ cYð w ð g H
1c S #

This shows that

char w )
Ê \ g
D
hc ¯ hr c r
6
82 LECTURE 8. BRST-COHOMOLOGY

where
Ž   •¥ 㠏 @ &½ ã J

U

 ¯ U

@  GF co ð w²w ð g
oB ʎ \  g ÷  ™ co ð w ð g H
c1S #J
We conclude that
D =à
ic c ¯ hr c r
6

Let \ be the vacuum vector of  GF ð w ð

oB . Set
9

g ` go c

Ù @ Ê \  ÷  \
+R  crð w ð g H
cS p
 @ O
™•;j O J 

g Ö
6 à 9
¯ hr 8 8 c r G8

@ O . Then  Vir
) @ O for • @ O and
We have the following result due to I. Frenkel, Garland, and Zuckerman:
' D ¤'

Âà \ Vir
) @  ÷  o F cYo ð Åð g
Theorem 8.2. Assume c . rel hc

j D

oif F ð ½ð g is an integer and zero otherwise. g

. rel hc ¯ c

g`
Define the map Z
0 @ ï \ Vir
v by Z  @@ @
c

v @
¿O
™• H ¿› g { O JKJJJ We
have < )

• j )

•
D D

 , 
9 / à 9 /

O O , O 
 ` O 
 `
phys hc , where ic

Assume
4`  •E{ O .. Then @
v  v ˜ Ö v 
 O unless4•+` H @ › O and • or‡ ˜ › { @ ` O . 9 / /

› v \ v \ O v .\ If
› ‡ ˜ O @ ,
4v v `\ v ˜ v \v v  \ ˜Hto7
@ F v v \  ˜ vJ v ˜

3/  

O.
Thus
@ @
à Ã

•

 Ô/ /

So, › O and )F ‡ 4` ‡ Therefore,

4` 4` `
/  4/ /

` o \ \
we

< @ v \ Ö \ F c c
v v v \
@ v \ F  @ O J
obtain
d

Ö
/ à 9 / à 9 {/ à 9 / à 9 9

4
 `
This defines a map from
4
 `  \
)
#
J 4

Vir If Z R Im , then R
D ` 
phys
. rel
D
hc
9 d 9 D spur T
D

@ \ Vir
) #J
and we get an injective map
phys
D

Â6à =Ù @  ÷  o \F ð Åð and Ù rad @

 – . hc
phys rel

E Ù g to Virg`
g ™ and hence @
j D T ÉT D

O . \ Thus
™ is mapped isomorphically
On the other hand, phys ¯ ir c phys
D T D

Vir .
. ic  –
phys rel phys
D
. rel ic

Exercises
8.1 Show that the equivalent definition of the cohomology of a Lie algebra  with coef-

 % ï  bç %
ficients in a linear representation   can be given as follows. Let  be
 •bç % ° ï  bç % %

ï
the space of anti-symmetric -multilinear maps from  with coefficients in . Define

Ž GF o7 ¿À * Q
Q ,
Q
KJJKJ4
Q
KJKJJ4
Q K
JJKJ4
Q  ™H
 

‘¿ iQ
KJKJJ4
Q  @
the coboundary map S by the formula
s

L‡ L ` L
s E Es

` ‡Å` Ň `
hS
! s

L
 83

Ž o7
KJJKJ

KJKJJ4
) #J
 ‡Å` GF L ‡Å` š 9Q L M¿À 9Q ` Q L Q  ‡Å`
E

Check that g
D @ O andL ` set  bç % @ Ker D ð ï  šç % ™ ™ˆ Im D ð ï  šç % ™ #J
%
šç % @ trivial %
.
% 4` \ bç % @
8.2 Consider the
Ñ ›Ó  ˆ* 
~,
.
representaion of  in a vector space . Show that .
.
` .

@ %
% ˆ %
8.3 A central extension of a Lie algebra  with help of a vector space is a Lie algebra
 w containing as a central abelian subalgebra such that  w
g b ç % –  . Show that such
%.
central extensioncs can be classified by the space .   , where  acts trivially on

8.4 Prove that

g
. Vir ç 1— 1 . @ –

representation p
SR 1 . Show that the exists a constant such that H
v v g @ O.
d
8.5 Let be the BRST-operator defined for the Virasoro algebra with coefficients in a d
D 
ic c