Introduction to String Theory

ETH Zurich, HS11

Chapter 0
Prof. N. Beisert

0 Overview
String theory is an attempt to quantise gravity and unite it with the other
fundamental forces of nature. It combines many interesting topics of (quantum)
eld theory in two and higher dimensions. This course gives an introduction to the
basics of string theory.

0.1

Contents

1. Introduction
2. Relativistic Point Particle
3. Classical Bosonic String
4. String Quantisation
5. Compactication
6. Open Strings and D-Branes
7. Conformal Field Theory
8. String Scattering
9. String Backgrounds
10. Superstrings
11. AdS/CFT Correspondence

0.2

References

There are many text books and lecture notes on string theory. Here is a selection
of well-known ones:

classic: M. Green, J.H. Schwarz and E. Witten, Superstring Theory (2

volumes), Cambridge University Press (1988)

alternative: D. Lst, S. Theisen, Lectures on String Theory, Springer (1989).

standard: J. Polchinski, String Theory (2 volumes), Cambridge University
Press (1998)

basic: B. Zwiebach, A First Course in String Theory, Cambridge University

Press (2004/2009)

recent: K. Becker, M. Becker, J.H. Schwatz, String Theory and M-Theory: A

Modern Introduction, Cambridge University Press (2007)

online: D. Tong, String Theory, lecture notes,

http://arxiv.org/abs/0908.0333

...

Introduction to String Theory

ETH Zurich, HS11

Chapter 1
Prof. N. Beisert

1 Introduction
1.1

Denition

String theory describes the mechanics of one-dimensional extended objects in an

ambient space. Some features:

Strings have tension:

Strings have no inner structure:

vs.

Several pieces of string can interact:

Strings can be classical or quantum:

vs.

1.2

Motivation

Why study strings?

Extended Objects.

We know a lot about the mechanics of point particles. It is

natural to study strings next. Or even higher-dimensional extended objects like

membranes. . .

particle

string

These are objects are snapshots at xed time

volume of spacetime occupied by the object:

1.1

membrane

t.

Introduce the worldvolume as the

worldline

worldsheet

worldvolume

The worldsheet of a string is two-dimensional. In fact, there is a great similarity

between strings and static soap lms.

Quantum Gravity.
gravity (QG). (really?)

String theory oers a solution to the problem of quantum

Sketch without reference to quantum eld theory (QFT).

Classical gravity theories:

Newtonian Gravity (non-relativistic)

General Relativity (GR, relativistic, geometry of spacetime)

Nature is quantum mechanical, gravity must also be quantum. Need QG for: early
universe, black hole radiation. Field quantisation introduces quanta (particles):

electromagnetism: photon
strong nuclear forces: gluons
gravity: graviton

Particles interact through vertices (Feynman rules). Relatively simple rules for
standard model

S=

Einstein gravity has innitely many vertices

S=

In fact, can introduce additional couplings

+ (G2 + c4 )

+ ...

ck :

+ (G3 + c5 )

1.2

+ ...

Classically we do not need the

ck ,

but in QFT we do. Feynman loops generate

divergences, e.g.

= .

Need to sum up all competing processes:

(G2 + c4 )

+ G4

Divergence can be absorbed into

+ G6

c4 = G6 + c4,ren .

All well, but no good way to set renormalised

divergences require innitely many

ck .

c4,ren

to zero. Unfortunately,

Innitely many adjustable parameters, not

predictive! Only good prediction at low energies, densities: GR.

What does string theory have to do with it?
Quantum string theory turns out to contain gravitons. Moreover, generates no
divergences; nite! String theory has just a few coupling constants.
All well!?
Almost, there may be many more couplings elsewhere.

Unication.

String theory provides a unied description for all kinds of

fundamental forces of nature. (the correct one?)

Electromagnetic and weak forces combine into electroweak forces at suciently
2
high energies 10 GeV. Also with strong forces (Grand Unied Theory, GUT)?
Hints:

Charges of fermions appear to suggest larger gauge group:

SU(3) SU(2) U(1) SU(5), SO(10)?

Estimated GUT scale

1015 GeV

near Planck scale

1018 GeV.

Suggests unication

of all forces.

Wouldn't it be nice?

String theory describes gauge theories as well as gravity. In particular, group

sequence

SU(5), SO(10),

. . . appears.

Does it describe nature? So far no convincing derivation. Best option: Standard

Model (SM) among many(!) natures.

1.3

String/Gauge Duality.

Intricate relations between string and gauge theories

(used in SM).
In some cases gauge theory is string theory.
String theory is part of gauge theory, not just QG.

Treasure Chest.

String theory yields many interesting, novel, exceptional

structures, results, insights in physics and mathematics. Just to name a few:

supersymmetry, higher dimensions,

Many Unsolved Problems.

p-branes,

dualities, topological insights.

(despite 40 years of research)

How to match with nature?

How to nd direct/indirect evidence? (Susy?)
What is String Theory?
How to quantise gravity (otherwise)?

1.4

Introduction to String Theory

Chapter 2

ETH Zurich, HS11

Prof. N. Beisert

2 Relativistic Point Particle

Let us start slowly with something else: a relativistic particle. Here we will
encounter several issues of string theory, but in a more familiar setting. There are
many formulations, we will discuss several.

2.1

Non-Relativistic Actions

First: a free non-relativistic point particle

~x(t).

Action with resulting equations of

motion (e.o.m.):

Z
S[~x] =

dt 12 m~x (t)2 ,

~x(t) = 0.

Momentum and energy from Hamiltonian formulation:

S[~x]
= m~x (t),
~x (t)

p~(t) =

E(t) = H(t) =

p~(t)2
.
2m

Note: functional variation removes the integral.

Promote the above to a relativistic particle:

Z
S=

q
Z


2
2

dt mc c ~x dt mc2 + 21 m~x 2 + 18 mc2~x 4 .

Derive e.o.m.:

They imply collinearity

(c2 ~x 2 )~x + (~x ~x)~x = 0.

~x = ~x . Substitute to get c2~x = 0

hence

~x = 0.

Momentum and energy read

mc~x
,
p~ = p
c2 ~x 2

E=c

m2 c2 + p2 .

Fine, but is not manifestly relativistic: Non-relativistic formulation of a relativistic

particle. Want a manifestly relativistic formulation using 4-vectors X = (ct, ~

x)
and

P = (E/c, p~).

Momentum

Let us set

c=1

for convenience from now on.

is already a good 4-vector:

P 2 = E 2 + p~2 = m2 .
Mass shell condition

P 2 = m2

manifestly relativistic. But origin/role of

is quite distinct.

2.1

p~

and

Position

X m (t) = (t, ~x(t))

and action make explicit reference to time

(in a

particular Lorentz frame)

s 
2
dX(t)
.
dt m
dt

Z
S=

Note that Hamiltonian framework distinguishes between space and time:

Explicit reference to time derivatives.

2.2

Worldline Action

Notice: above action measures Lorentz-invariant proper time

path X (t) in spacetime (worldline)

Z
S = m

ds,

where

of the particle's

ds2 = dX 2 .

Proper time depends only on the location of the worldline, but not on a particular
Lorentz frame (denition of

t)

or parametrisation of the worldline (through

Let us assume an arbitrary parametrisation

curve parameter

X ( )

of the worldline through some

The proper time action reads (now dot denotes

Z
S=

t).

d/d )

s 
2
Z
p
dX( )
d m
= d m X 2 .
d

Nice manifestly relativistic formulation. Notice: 4 undetermined functions

instead of 3 undetermined functions

~x(t);

new function

X ( )

t( ).

Use this as starting point, derive equations of motion

= (X X)
X .
X 2 X
Implies collinearity

= cX
X

for all

with variable

straight.
Next derive momenta as derivatives of

w.r.t.

mX
.
P = p
X 2
2.2

c( ).

Meaning: worldline

P 2 = m2 !

Obey mass shell condition

Only three independent

X !
H = 0!

but four independent

Moreover naive Hamiltonian is strictly zero

Signals presence of constraint and gauge invariance:

Reparametrising

0 = f ( )

has no eect on physics.

Redundancy of description: worldline coordinate .

One linear dependency among the e.o.m. for X .

Gauge invariance eectively removes one X , e.g. time

t( ).

Situation inconvenient for Hamiltonian framework/QM.

Better to x a gauge, many choices, pick a convenient one.

Obtained a fully relativistic formulation, but packaged with complication of gauge

invariance. In fact, gauge invariance often considered a virtue: Symmetry!
Above worldline action has two further drawbacks:

Is non-polynomial; inconvenient for quantisation.

Does not work for massless particles

2.3

m = 0.

Polynomial Action

There is an equivalent action with an auxiliary variable

Z
S=

1 1 2
e X
2

e( )


21 em2 .

The resulting e.o.m. read

m2 e2 + X 2 = 0,

e X = 0.
eX

In combination they yield the same old equation for X . The momentum

1
conjugate to X reads P = e X , hence the equation of motion for e reduces to
P 2 = m2 . Momentum conjugate to e vanishes signalling a constraint.
Massless case m = 0 works at every step of above derivation, yields constant
P = e1 X as well as P 2 = 0. Notice: e not xed by e.o.m.; commonly gauge
freedom remains for massless case.

e has a nice geometrical interpretation: Einbein specifying a metric

g = e2 on the worldline. All terms in the action are in the right combination;
remain invariant under changing worldline coordinates (e transforms according to
e0 = e d 0 /d ).
Field

Here einbein

is a dynamical variable. Curiously, e.o.m. picks out metric induced

by ambient space. When substituting solution for

2.4

recover action of Sec. 2.2.

Various Gauges

We have freedom to x one of the coordinates

useful choices:

2.3

X ( )

at will. Some more or less

 Temporal Gauge. t( ) =

t( ) = .

or

Reduces to non-relativistic treatment of Sec. 2.1.

Spatial Gauge. z( ) = .

Works locally except at turning points of z( ).

Light Cone Gauge. x+ ( ) := t( ) + z( ) = .
Useful in some cases; prominent in string theory.

Proper Time Gauge. ds = d .

Fixes t( ) through integral

Z
t( ) =
S=
Constant Einbein. e = 0.
Action becomes trivial

d ;

q
1 + ~x ( 0 )2 .
d
0

deal with constraint.

In polynomial formulation, gauge xing may involve

is constant

e.

e.

Customary gauge choice

E.o.m. reduces to

= 0.
X
We replace dynamical variable

by a constant; we must remember its equation

of motion

X + m2 e2 = 0.
In gauge xed formulation it becomes constraint.

2.5

Quantisation

Quantisation can be done in several dierent ways. Depends on the choice of

classical formulation. Let us pick polynomial action discussed in Sec. 2.3. For the
Hamiltonian formulation it is best to also x a gauge; we will choose the einbein
to be constant. Momenta

associated to

P = e1 X,
Conventionally, a state

| i

and resulting Hamiltonian read:

H = 21 e P 2 + m2 .

is given by a wave function of position variables and

time

Z
| i =

d4 X (X, ) |Xi.

Slightly more convenient to immediately Fourier transform to momentum space

R 4 iP X

|Xi '

|P i
Z
| i = d4 P (P, ) |P i,

d Pe

Z
(P, ) =

d4 X eiP X (X, ).

Schrdinger equation reads

i = H = 2i e P 2 + m2 .
Obviously, solved by

(P, ) = exp 2i e(P 2 + m2 ) (P ).
2.4

Need to remember that system is constrained; wave function must vanish whenever
constraint not satised:

(P 2 + m2 ) (P, ) = 0.
(P, ) = (P ) are independent of . Makes perfect
sense: worldline coordinate unphysical. Schrdinger equation governing
-evolution replaced by constraint P 2 + m2 = 0 (governing t-evolution).
In eect, physical states

Fourier transform wave function back to position space

(X);

constraint becomes

the KleinGordon equation for spin-0 eld

( 2 + m2 )(X) = 0.

2.6

Interactions

Obviously, free particle is easy; eventually would like to include interactions. Let
us sketch how to add interactions with external potentials and with other particles:

Electrical and Gravitational Fields.

Coupling to electrical and gravitational

elds takes a very geometric form

Z
S=
A

1 1
e g (X)X X
2

potential for the electromagnetic eld


12 em2 + A (X)X .

F = A + A .

Likewise

is the

gravitational potential; takes the form of the metric of a curved spacetime.

These are xed external elds: Unaected by presence of particle, but inuence its

motion. Note: Fields are evaluated at dynamical position X ( ).

In quantum mechanics, one usually assumes weak interactions. Formally allows to
work with free quantum elds; interactions are introduced in perturbative fashion.
When free particle enters potential eld, it scatters o of it. Dominant
contribution from single scattering; multiple interactions suppressed. Only in rare
instances, potentials can be handled exactly.

Interactions Among Particles.

Local interactions: Several particles meet at

some spacetime point and split up, potentially into a dierent number of particles.
In worldline formulation achieved by introducing vertices where several particle
worldlines meet:

This is not standard treatment of interaction. Typically interaction of n elds:

n
term in QFT action. Our method is not very convenient, but it works as well.

2.5

Nevertheless instructive, mimics Feynman rules; It is the standard procedure for

string theory.

2.7

Conclusions

Seen many equivalent formulations of same physical system.

Had to deal with gauge invariance and constraints.
Dierent number of degrees of freedom (d.o.f.), but number of solutions (modulo
gauge) always the same.

Quantised the free relativistic particle.

Discussed interactions.
Not always most convenient path chosen; but treatment of string will be
analogous.

2.6

Introduction to String Theory

Chapter 3

ETH Zurich, HS11

Prof. N. Beisert

3 Classical Bosonic String

Mechanics of 1D extended object without inner structure.

Worldsheet coordinates: = (, ), time , space .

Embedding coordinates X ().
D-dimensional embedding Minkowski space, metric .

3.1

NambuGoto Action

Generalise insights from relativistic point particle:

worldline
action

worldsheet.

proper time

Area and Action.

'

length

Wick rotation

area.

t = iw.

Area of 2D euclidean surface:

| sin |
dA = d d |X 0 | |X|
p
= d d X 02 X 2 sin2
q
2
= d d X 02 X 2 (X 0 X)
p
= d2 det g ,
induced worldsheet metric

g = X X

(pull back).

Wick rotate back; string action (NambuGoto)

1
1
S=
A=
2
2
22

d2

det g .

Symmetries of action:

Lorentz: scalar products of Lorentz vectors X .

Poincar: X only through X .

Worldsheet Dieomorphisms: density d2 det g

0
reparametrisations 7 ().

Tension.

What is parameter

invariant under

Fundamental string length scale

quantum string.

2
Consider potential U : Time slice of action/area. Slice of length L: U L/ .
0
2
Constant force U = 1/2 = T is string tension. String not a spring or rubber
band!

3.1

Equations of Motion.

Vary action.

det g = det g g g .
p


det g g X = 0.

Use variation of determinant

Highly non-linear equations (g contains

X )!

How to solve? How to deal with?

Geometrically: stationary action; minimal area surface. Static soap lms! Mean
curvature zero. Saddle point everywhere, equal/opposite sectional curvatures.

3.2

Polyakov Action

Complication from non-linear e.o.m.. As for point particle, there is a polynomial

action with additional dynamical worldsheet metric

1
S=
22
E.o.m. for

d2

g :

det g 12 g ( X) ( X).

as above; e.o.m. for worldsheet metric

g:

( X) ( X) = 21 g g ( X) ( X).
Solution xes induced metric (up to local scale

f)

g = f () ( X) ( X).
f () cancels in action
invariance g () 7 f ()g ().
Arbitrary scale

3.3

and all e.o.m.. New redundancy:

Weyl

Conformal Gauge

E.o.m. for

linear, coupling to

makes non-linear.

Use gauge freedom: demand conformally at metric.

g () = f () .
Amounts to two equations

g () = 0, g () = g ().

Fixes almost all dieomorphisms.

Conformal transformations remain. Dieomorphisms preserving metric up to
scale.

May further set

f =1

by means of xing Weyl.

3.2

Action describes

free massless scalar particles in 2D

1
S=
22
E.o.m. simply

2X = 0

d2 21 ( X) ( X).

(harmonic wave equation)

= X 00 .
X
Do not forget e.o.m. for worldsheet metric

T := ( X) ( X) 12 ( X) ( X) = 0.
T

is energy-momentum tensor for

scalars. Trace absent

T = 0

construction (Weyl/conformal). Two remaining e.o.m. become

constraints

X X 0 = 0,

by

Virasoro

X X + X 0 X 0 = 0.

Forbids longitudinal waves along the worldsheet, no structure!

Conservation of

3.4

T :

impose constraints only on time slice.

Solution on the Light Cone

Solve harmonic wave equation. Light cone coordinates

L/R = ,

L/R

useful:

L/R = 21 ( ),

new worldsheet metric

d2 s = d 2 + d 2 = d L d R .
Now e.o.m. and Virasoro constraints read

L R X = 0,

(L/R X)2 = 0.

First equation solved by simple separation of variables

X ( L , R ) = XL ( L ) + XR ( R ).
D

left-movers

XL

plus

right-movers

XR .

Virasoro constraints

(XR,L )2 = 0

remove 1 left/right-mover. Two reparametrisations left:

conformal transformations
R 7 0R ( R ),

L 7 0L ( L ).

2D case special: innitely many transformations. removes another 1

left/right-mover.

constant shift
(D 2)

between

XL

and

XR .

left/right-movers remain (transverse).

3.3

3.5

Closed String Modes

So far worldsheet innitely extended in space and time; want nite spatial extent.

Closed String: circular topology.

Identify + 2 (other choices possible).
Open String: interval topology.
Boundary conditions at = 0, (later).
Periodic function

Fourier decomposition

X i
nL/R, exp(in L/R ).
2n
n6=0

XL/R
= 21 x + 12 2 p L/R +

Coecients

i/ 2n

chosen for later convenience.

L/R

2
Linear dependence in
okay: X = x + p

Reality of X : complex conjugate n = (n ) .

+ . . ..

Two kinds of parameters for solutions

Centre of mass motion x, p (conjugate variables

L/R,
String modes n
(left/right movers).

2 ).

Consider Virasoro constraints, substitute modes:

(L,R XL/R
)2 = 2

LL/R
exp(in L/R ) = 0
n

with Virasoro modes (drop L/R index)

Ln :=
Note that

1
2

0L = 0R = p/ 2. L0 = 0
p2 = M 2 ,

All Virasoro constraints

Ln = 0

nm m = 0.

constraint xes string mass

M2 =

4 X
m m
2 m=1

conserved by e.o.m.

L n = inLn .
Need to impose on initial data only.
String mass depends on mode amplitudes.

3.4

No modes excited: massless point particle.

Few/small excitations: light tiny particle.
Many/large excitations: big massive object.

Note: time-like modes

contribute negative

M 2.

Tachyons excluded by Virasoro

constraints.
Virasoro constraints non-linear, complicated.

3.6

Light Cone Gauge

Use conformal symmetry to solve Virasoro constraints.

Gauge Fixing.

Introduce as well light cone coordinates for spacetime

X =X X

D1

~
X

denotes components

1 . . . (D 2).

Gauge x!

+
L/R
1 2 +
.
XL/R
= x+
L/R + 2 pL/R
Virasoro constraint

~ L/R )2 2 p+ X = 0
( X
L/R
L/R

XL/R
()

Solution:

2(D 2)

Periodicity.
X

1
= 2 +
pL/R

arbitrary functions

All functions

~
X()

solved by

~ L/R ( 0 ) 2 .
d 0 X

~ L/R ( L/R ).
X

periodic! Furthermore: periodicity of

X+

and

requires

p+
L

p+
R

=p ,

~ R )2 ( X
~ L )2 = 0.
d ( X

0
Residual gauge freedom: Constant shift

XR () = XL ().

above residual constraints.

3.5

Corresponds to

String Modes.

Impose gauge xing on modes (n

n+ = 0,
Periodicity requires

n =

1 X

~ nm
~ m.
m
0+

~ 0L =
~ 0R , 0R,+ = 0L,+

6= 0)

and for

L
L

~ m

~m
=

m=1

R
R

~ m

~m

m=1

Resulting mass manifestly positive:

4 X
4 X
M = 2

~ m
~m = 2
|~
m |2
m=1
m=1
2

Benets: positivity, almost all constraints gone. Drawback: Lorentz symmetry not
manifest.

3.6

Introduction to String Theory

Chapter 4

ETH Zurich, HS11

Prof. N. Beisert

4 String Quantisation
We have seen that the classical closed string is described by

a bunch of harmonic oscillators

a relativistic particle

(x, p)

L/R

for the string modes;

describing the centre of mass.

Both systems coupled via Virasoro constraints.

We have two reasonable formulations:

Covariant formulation with D oscillators n per mode. Virasoro constraints

L
R
Ln = Ln = 0 and conformal symmetry.
Light cone formulation with D 2 oscillators
~ n per mode. No constraints, full
Poincar symmetry not manifest.

4.1

Canonical Quantisation

Derive commutation relations for variables

x, p, n .

Recall action in conformal gauge

1
S=
22

d2 21 X 2 X 02

Substitute closed string mode expansion (with free time)

X =

X
n

n ( ) exp(in)

Obtain tower of HO's (0 free particle)

1
S=
2

Z
d

X 
n

n n n2 n n

Canonical momentum and canonical commutator:

n = n ,
Match

[m
, n ] = i m+n .

with previous classical solution at

x =

0 ,

0
p =
,

=0
nL/R,

nn
n
= + ;
i 2
2

resulting commutators in original variables

[x , p ] = i ,

 L, L,   R, R, 
m , n = m , n = m m+n .

4.1

4.2

States

Compose space of states from free particle and oscillators.

Momentum eigenstates for free particle

HO vacuum

|0i

|qi.

and excitations for each mode/orientation.

Dene string vacuum state

|0; qi

p |0; qi = q |0; qi,

nL/R, |0; qi = 0

for

n > 0.

Problem: negative norm states

|0; qi = n .
|n, ; q|2 = h0; q|n n

|n, ; qi := n
|0; qi,

State not allowed by Virasoro constraints. General resolution: impose Virasoro

constraints.

4.3

Light Cone Quantisation

Continue covariant quantisation later. Fix light cone gauge instead; only physical
states.
Resulting commutators lead to positive denite states

 L,a L,b   R,a R,b 

m , n = m , n = m ab m+n .
Remember classical mass and residual constraint

4 X L
4 X R
L
M = 2

~ =

~R.
m=1 m m 2 m=1 m m
2

Operator ordering matters! A priori free to choose. Assume normal ordering plus
L/R
new constants a
:

4
M2 = 2

!
L
L

~ m

~m
aL

m=1

4
= 2

!
R
R

~ m

~m
aR

m=1

Combination measures string level

N :=

~ m
~m =

m=1

mNm

with

Nm :=

m=1

~ m
~ m.
m

Mass and constraint in terms of string level

M2 =

4.4

4
4
(N L aL ) = 2 (N R aR ).
2

String Spectrum

Mass depends on string level. Quantisation of string level

L
L
R
R
mass. Level matching: N a = N a .
Understand string states at each level; HO's.

4.2

quantisation of

Vacuum State.

Dene vacuum state

|0; qi

~ nL/R |0; qi = 0
Level zero:

N L = N R = 0.

for

n > 0.

Spin zero.

For physical state:

4a
M2 = 2 .

1
a. a 0?!
M = 2

aR = aL = a,
So far so good: spin-0 particle with
wave function

First Level.
implies

N =

(D 2)2

Spatial extent: HO

.
L
Lowest excited state has N = 1. Level matching and a
R
N = 1. One excitation
~ 1 each from left/right movers

states of mass

= aR

L,a R,b
1 |0; qi.
|ab; qi = 1

M = 21 1 a.

Spin under transverse rotations. Three combinations:

|(ab); qi := |ab; qi + |ba; qi

2ab
|cc; qi,
D2

|[ab]; qi := |ab; qi |ba; qi,

|1; qi := |cc; qi.
Transformation properties under

SO(D 2):

state

indices

Young tab.

|(ab); qi
|[ab]; qi
|1; qi

symmetric, traceless

anti-symmetric

singlet

Stabiliser (little group) for massive particle is

reps. into

SO(D 1)

spin

SO(D 1).

Can t these

SO(D 2)

reps.? No!

Only way out: massless particle; stabiliser

SO(D 2).

set

a = aR = aL = 1.

Three types of particles:

|(ab); qi:

massless spin-2 eld. okay as free eld.

WeinbergWitten: interactions are forbidden. except for gravitational

interactions: graviton!

|[ab]; qi: massless 2-form eld (KalbRamond).

B with 1-form gauge symmetry B =   .
|1; qi: massless scalar particle (dilaton).
dierent from string vacuum |0; qi.
What we have learned:

Graviton plus massless 2-form and scalar particles. Spatial extent

practically point-like.
a = aR = aL = 1.

Interacting string theory includes gravity!

4.3

is Planck scale.

Tachyon.

Revisit string vacuum

|q, 0i: M 2 = 4/2 < 0.

Tachyon!

Problem? Not really, compare Goldstone/Higgs mechanism:

Unstable vacuum at local maximum of a potential.

Let us ignore. Indeed tachyon absent for superstrings!

Physical ground state at local minimum. No tachyon!

Unclear if minimum exists. Where? What properties?

Higher Levels.
level

0
1
2
3

Levels zero and one work out. what about higher levels?
excitations

a
1
a
b
1
1
a
2
a
b
c
1
1
1
a
b
1
2
a
3
a
b
c
d
1
1
1
1
a
b
c
1 1 2
b
a
2
2
b
a
1
3
a
4

...

SO(D 2)

SO(D 1)

+
+
+

+
+ +
+
+ +
+

...

...

All higher levels combine into proper

SO(D 1)

...

reps.. Level matching: need to

square reps..

String describes collection of innitely many particle types.

Various vibration modes might correspond to elementary particles. Including
gravity.

intrinsic particle extent

is Planck scale

observed: Point-like particles!

few light particles; all others at Planck mass; 1 tachyon?!

very high excitations are long strings. mostly classical behaviour.
superheavy.

4.4

M  1/

Regge Trajectories.
level

Maximum spin (symmetric indices) increases linearly with

S = 2N

M2 =

2S 4a
,
2

called leading Regge trajectory:

0 = 2 is Regge slope.
2a is Regge intercept, spin

of massless particle.

Subleading trajectories for lower spins (indices anti-symmetric and in trace).

Qualitative similarity to hadron spectrum:

Regge trajectories for hadronic resonances

1/ is QCD scale ' 1 GeV
Intercept a 21 rather than a = 1.
another problem, see later.
qualitative description of QCD ux tubes.

4.5

Anomalies

In light cone gauge we have broken manifest

SO(D 2)

observed.

SO(D 1, 1)

Lorentz symmetry to a

subgroup.

Consequently spectrum of quantum strings organises manifestly into

SO(D 2)

multiplets.

Almost all multiplets t into

SO(D 1)

multiplets.
L
R

Mass assignments ll Poincar multiplets for

L
R
Poincar symmetry broken unless a = a =

a = a = 1.
1.

Anomaly: Failure of classical symmetry in quantum theory.

Sometimes okay, not here, want strings to propagate on Minkowski background
with intact Poincar symmetry.
So far only counting, more severe problem in algebra. Commutator

a
receives contributions from [ , ]:

[M

,M ] =


X
D2

24

n=1
vanishes if and only if

D = 26

and

[M a , M b ]



 
D2 1
1 n+ a
...
24
n

a = 1.

String theory predicts twenty-six

spacetime dimensions.

4.5

Shortcut derivation: reconsider nature of intercept

1
= 12 n
2 n

a. a

is sum of HO ground state

energies

a=

(D

2) 12 n

21 (D

2)

Sum divergent, black magic helps:

X
1
,
x
k
k=1

Analytical continuation

n.

n=1

n=1

(x) :=

-function

regularisation

a = 21 (D 2)(1) =

i.e.

1
(1) = 12

a=1

and

predicts

D2
.
24

D = 26!

Murky

derivation yields correct prediction.

4.6

Covariant Quantisation

In LC gauge Poincar symmetry is subject to anomaly, but can also keep Poincar
manifest: Covariant quantisation. See how spectrum arises in covariant approach.
Consider only L or R oscillators for simplicity.

Vacuum State. |0; qi dened as before.

Ln>0 |0; qi = 0

and

Satises

L0 |0; qi =

2 q 2
|0; qi.
4

State not annihilated by negative Virasoro modes. Instead

h0; q|Ln<0 = 0

hence

h0; q|(Ln n a)|0; qi = 0.

Impose Virasoro constraints for physical states

Ln>0 | i = 0,

One Excitation.

L0 | i = a| i,

| i
h |(Ln n a)| i = 0.

Generic ansatz for state

|; qi := 1 |0; qi.
Norm

potentially negative. Virasoro constraint implies

L1 |; qi = 1 0 |; qi =
Furthermore

L0 = a = 1

implies

q 2 = 0.

( q)

|0; qi = 0.
2

Then

q =0

removes negative norm

state(s). Remains:

D 2 states with positive norm.

State with = q is null. Does not

contribute to physics.

4.6

Two Excitations.

Generic ansatz

|0; qi + 2
|0; qi.
|, ; qi := 1
1
Impose constraints
Remains:

State

L0 , L1 , L2 = 0
of

to x

q 2 , , tr .

SO(D 1).

is positive denite.

= q + q
Null state for a = 1!
Ansatz for : = q q + .
for D = 26!
Ansatz for

Virasoro Algebra.

with

q = 0.

Negative norm for

Negative norm for

Algebra of quantum charges

[Lm , Ln ] = (m n)Lm+n +
Latter term is central charge of Virasoro,

Ln

D<1

or

1 < a < 2.

D > 26.

Null state

reads

c
m(m2 1)m+n .
12

c = D.

We are interested in primary

states

Ln>0 | i = 0,

L0 | i = a| i.

Can apply representation theory of Virasoro algebra

CFT.

Proper treatment (BRST) includes ghosts. Classical conformal algebra when

D = 26

and

a=1
[Lm , Ln ] = (m n)Lm+n .

Here: Conformal algebra is anomalous. Anomaly shifted to Lorentz algebra in

light cone gauge.

4.7

Introduction to String Theory

Chapter 5

ETH Zurich, HS11

Prof. N. Beisert

5 Compactication
We have seen that the closed string spectrum contains:

1 tachyonic scalar particle (wrong vacuum).

massless gravitons and few other particles.
tower of particles of increasing mass (inaccessible).

D = 26 dimensions, way
F 1/A 1/r24 not 1/r2 .
But:

5.1

too many! Gauss law: gravitational force:

KaluzaKlein Modes

Idea: Compactify 22 dimensions to microscopic size. Large distances only for 4

remaining dimensions. Small compact dimensions almost unobservable.

Compactify one dimension to a circle of radius

X 25 X 25 + 2R.
Quantum mechanical momentum quantised

P25 =
Eectively tower of massive particles

Zero mode

n=0

n
.
R

2
2
M25
= M26
+ n2 /R2 :

has original mass. Massless mode observable.

Higher modes are massive,

M ' 1/R.

For very small

unobservable.
Low-energy physics can be eectively four-dimensional.

5.2

Winding Modes

Peculiarity of strings on compact spaces: Winding.

5.1

R:

practically

X 25 =: X X + 2R. Need
X( + 2) = X() + 2Rm.


n
mR L/R
1
1 2
XL/R = 2 x + 2
2
+ modes.
R

Consider again one compact direction

periodicity:

to relax

Mass (for propagation in 25 non-compact dimensions)

4
M = 2 (N L/R a) +

mR
n
2
R

2
.

Level matching condition modied by winding

N L N R = nm.
L/R average formula for mass

M2 =

n2
m2 R 2
2
L
R
(N
+
N

2a)
+
+
.
2
R2
4

Winding also contributes mass. To hide innitely many modes:

, R

and

2 /R

small!

Decompactify circle as

R :

Winding modes become very heavy.

KK modes form become light and continuum.

NL =
6 N R exist
heavy at R .

Note: Also modes with

become innitely

Can also try to compactify circle

(new representations). Additional modes

R 0.
5.2

KK modes become very heavy.

Winding modes become light and form continuum.

R with
R and 2 /R.

Same as for
same for

role of

and

interchanged. Observe: spectrum the

Additional dimension remains observable at

R 0!

Dierent from regular point

particle with KK only.

5.3

T-Duality

Duality between small and large compactication radius. Can show at Lagrangian
level:

T-duality.

Start with action of 25-direction in conformal gauge

1
22
Action has global shift symmetry

d2 21 ab a X b X

X X + .

For winding we would need local

shift, let us make the symmetry local (gauge),

1
22

d2

1 ab
(a X
2

Aa Aa a 


a Ab .
+ Aa )(b X + Ab ) ab X

Aa and one local redundancy. Remove further d.o.f. by

. Done nothing
demanding Fab = a Ab b Aa = 0. through Lagrange multiplier X
(e.g. Aa = 0).

Added two d.o.f. in

Field

Aa

is algebraic, integrate out: E.o.m.

Aa = a X + ac cb b X
Substitute and obtain (up to boundary term)

1
22
Same as before, but with
Now can set

Aa = 0

b X.

d2 21 ab a X

instead of

X.

and obtain the duality relation

a X = ac cb b X,
For the standard solution

i.e.

0,
X = X

we nd the dual

X 0 = X.

n
+ mR + modes.
R
= x + mR + 2 n + modes.
X
R

X = x + 2

Duality interchanges

= 2 /R
RR

and

m n.

5.3

Eectively

R=

is minimum compactication radius. It is indeed a special

self-dual point. Duality between two models turns into enhanced symmetry.

R=

is minimum length scale in string theory: quantisation of spacetime in

quantum gravity.

5.4

General Compactications

So far compactied one dimension: Only circle or interval. Many choices and
parameters for higher compactications.

sphere

S n,

a
na
product of spheres S S
, dierent radii,
n
torus T , 3n 3 moduli (radii, tilts),
other compact manifolds.

Low-lying modes determined by manifold (bell).

Compactication determines observable spectrum.

Goal: nd correct manifold to describe SM.
Massless modes correspond to gauge symmetries.
Superstrings: CY 3-fold preserved 1 susy.

5.4

Introduction to String Theory

Chapter 6

ETH Zurich, HS11

Prof. N. Beisert

6 Open Strings and D-Branes

So far we have discussed closed strings. The alternative choice is open boundary
conditions.

6.1

Neumann Boundary Conditions

Conventionally

and we have to discuss

= 0, .

Start again in

conformal gauge

1
S=
22

d2 12 ab a X b X.

Variation including boundary terms due to partial integration

Z
1
S =
d2 ab a X b X
22
Z
1
=
d2 a ( ab X b X) . . .
22
Z


1
0
0
d
X()

X
()

X(0)

X
(0)
....
=
22
Boundary e.o.m. imply Neumann conditions (alternative later)

X 0 (0) = X 0 () = 0.
Virasoro constraints

X 0 X = X 02 + X 2 = 0
2 = 0.
imply that end points move at speed of light X
2
= 0 implies X = 0.)
soap lm: X

6.2

(no free ends of analogous

Solutions and Spectrum

Same equations in the string bulk, recycle solution.

Doubling Trick.
2

Map two copies of open string to closed string twice as long:

. Gluing condition X 0 = 0 at = 0, implies

nL = nR .

6.1

Left movers are reected into right movers at boundary. One copy of oscillators
and Virasoro algebra

X = x + 22 p +
(momentum

is doubled because

Quantisation.
a=1

X i


n exp(in L ) + exp(in R ) .
2n
n6=0

integration is halved)

Analogous to closed strings. Same anomaly conditions

(from bulk). Resulting spectrum (note dierent prefactor due to

M2 =

D = 26,

p).

1
(N a).
2

Only single copy of oscillators at each level.

level 0: singlet tachyon (of half mass).

level 1: massless vector: Maxwell eld.
level 2: massive spin-2 eld

...

Same as discussion for closed string without squaring!

Massless modes are associated to local symmetries:

of open string are spin-1 gauge elds,

of closed string are spin-2 gravitation elds.

String Interactions.

Open and closed strings interact:

Two ends of string can join.

Open strings must include closed strings. Dierent vacuum states

|0; qio

|0; qic

and

in same theory.

Opening of string can be suppressed. Closed string can live on their own.

String theory always contains gravity; May or may not include gauge eld(s).

6.3

Dirichlet Boundary Conditions

Now consider compactication for open strings. Almost the same as for closed
string. No winding modes because open string can unwind.

6.2

T-Duality.

What about applying T-duality? Introduce dual elds

X 0 = X,

0.
X = X

Boundary conditions translate to

25 = 0.
X

X 0 = 0 ( = 0, . . . , 24),
Dirichlet boundary condition for dual coordinate
25
choice of boundary e.o.m. X
= 0.

25 .
X

Corresponds to alternate

KK modes turn into winding modes:

25

Z
=

025

d X

Strings start and end at same

Z
=

d X 25 = 22 p25 =

.
x0 x0 + 2 R

22 n

= 2nR.
R

Note: No momentum

P25

because

position xed. Role of KK and winding exchanged.

Dirichlet condition modies oscillator relation:

nL,25 = nR,25 .
Although Dirichlet condition

25 = const.
X

appears unnatural, it has to be part of

string theory (on compact spaces).

D-Branes.

Take seriously.

At boundary can choose:

Neumann condition X 0 = 0 or
Dirichlet condition X = xed
for each direction

individually.

Geometrical picture: String ends conned to Dp-branes.

p + 1 dimensional (p, 1) submanifolds of spacetime.

Dirichlet conditions for D p 1 orthogonal directions.
Neumann conditions for p + 1 parallel directions.
D-branes can be curved (normal depends on position).
T-duality maps between Dp and D(p

1)

branes.

Pure Neumann conditions are spacetime-lling D-brane.

Strings propagate on backgrounds with D-branes:

spacetime bulk curvature governs string bulk propagation,

D-branes govern string end propagation.

Even more: Will continue discussion later.

6.3

6.4

Multiple Branes

Can have multiple branes of diverse types. Open strings stretch between two
branes.

Parallel Branes.
X

25

= 0, d

Simplest case: Two parallel planar Dp-branes located at

in non-compact Minkowski space

X = 22 p + modes,

X 25 =

Resulting (quantum) mass spectrum in

d
+ modes.

p+1

dimensions

d2
1
+ 2 (N a).
2
4
4

2
mass M = d/2 .

M2 =

Spin-1 particle at level-1 with

Vector massless at coincident branes.

Tachyon for

d < 2:

Multiple Branes.
There are

N2

Instability for nearby D-branes.

Consider now

parallel branes.

types of open string (and 1 closed): String vacua distinguished by

ChanPaton factors

|0; q; abio ,

a, b = 1, . . . , N.

with general mass formula

Ma2b

d2ab
1
= 2 4 + 2 (N a).
4

Consider vector particles at level 1 with mass

dab /22 .

Always N massless vectors. Gauge symmetry: U (1)N .

K coincident branes contribute K 2 massless vectors. Enhanced
U (1)K U (K).
Massive vectors indicate spontaneously broken symmetries.
Geometric picture of gauge symmetries:

Stack of

branes have local

U (N )

symmetry.

Separating branes breaks symmetry to

Creates

2K(N K)

U (K) U (N K).

massive vectors.

6.4

gauge symmetry

Can also produce

SO(N )

and

Sp(N )

symmetries: Unoriented strings, strings on

orientifolds (spacetime involution paired with orientation reversal).

Brane Worlds.

Can design many dierent situations.

Combine:

non-compact dimensions,
D-branes,
intersections of D-branes and non-compact dimensions,
orientifold action.

Consider physics:

along non-compact dimensions,

within D-branes.

Qualitative features:

Massless vectors indicate gauge symmetries.

Light vectors indicate spontaneous symmetry breaking.
Tachyons indicate instabilities of D-branes or spacetime.

String theory becomes framework analogous to QFT:

D-brane arrangements and compact directions (discrete),

moduli for D-branes and non-compact spaces (continuous).

Physics: Try to design the standard model at low energies.

Mathematics: Dualities relate various situations.

6.5

Introduction to String Theory

Chapter 7

ETH Zurich, HS11

Prof. N. Beisert

7 Conformal Field Theory

So far considered mostly string spectrum:

equations of motion (local),

closed/open periodicity conditions (global),
quantisation.

Quantum mechanics of innite tower of string modes

n .

Next will consider local picture on worldsheet: Fields

X().

Quantisation

Quantum Field Theory (QFT). Will need for string scattering.

Reparametrisation invariance:

worldsheet coordinates

articial,

gauge xing: conformal gauge,

worldsheet coordinates
dieomorphisms

meaningful,

residual conformal symmetry,

Conformal Field Theory (CFT).

CFT: QFT making use of conformal symmetry.

do not calculate blindly,

structure of nal results dictated by symmetry,
conformal symmetry: large amount, exploit!

Let us scrutinise conformal symmetry:

Central framework in string theory,

but also useful for many 2D statistical mechanics systems.

7.1

Conformal Transformations

Special coordinate transformation:

all angles unchanged,

denition of length can change,

Metric preserved up to scale

g0 0 0 (x0 ) =

Action on Coordinates.

dx dx
!
0
0 g (x) = f (x) g0 0 (x)
0
0
dx dx

Generally in

dimensions

Lorentz rotations x x ,

translations x x + a ,

scale transformations / dila(ta)tions x sx ,

2
conformal inversions (discrete) x x /x ,

7.1

conformal boosts (inversion, translation, inversion).

Conformal group:

SO(D, 2)

Action on Fields.

(rather: universal cover).

E.g. a free scalar

dD x 12 (x) (x).

Manifest invariance under Lorentz rotations & translations

0 (x) = (x + a).

Invariance under scaling

x0 = sx

requires

0 (x) = s(D2)/2 (sx).

Invariance under inversions

0 (x) = (x2 )(D2)/2 (1/x).

Similar (but more complicated) rules for:

(x) with dierent scaling 0 (x) = s (sx),

spinning elds , . . . ,
2
derivatives , , , . . . .

scalar eld

2D Conformal Symmetries.

QFT's in 2D are rather tractable. CFT's in 2D

are especially simple:

Conformal group splits SO(2, 2) ' SL(2, R)L SL(2, R)R

SL(2, R)L/R act on coordinates as (drop L/R)
0 =
L/R

are two translations,

a + b
,
c + d
L/R

= + 2 ;

are rotations and scaling,

L/R

are two

conformal boosts.

SL(2, R)L/R

extends to innite-dimensional Virasoro

L/R = L/R ( L/R ) =

L/R 1n
L/R
) .
n (

Boundaries typically distorted by Virasoro. Only subalgebra preserves

boundaries, e.g.

7.2

SL(2, R).

Conformal Correlators

In a quantum theory interested in

spectrum of operators (string spectrum),

7.2

probabilities,
expectation value of operators on states.

In QFT compute (vacuum) expectation values:

momentum eigenstates: particle scattering, S-matrix

h~q1 , ~q2 , . . .|S|~p1 , p~2 , . . .i = h0|a(~q1 )a(~q2 ) . . . S . . . a (~p2 )a (~p1 )|0i

position eigenstates: time-ordered correlation functions

h(x1 )(x2 ) . . .i = h0|T [(x1 )(x2 ) . . .]|0i

Correlator of String Coordinates.

Can compute a worldsheet correlator

using underlying oscillator relations

2
log exp(i2L ) exp(i1L )
2


2
log exp(i2R ) exp(i1R ) + . . .

h0|X (2 )X (1 )|0i =

Can reproduce from CFT? Scalar

of dimension

h(x1 )(x2 )i = F (x1 , x2 )

Correlator should be invariant!

Translation invariance

F (x1 , x2 ) = F (x1 x2 ) =: F (x12 ).

Just one vector variable.

Invariance under Lorentz rotations

F (x12 ) = F (x212 ).
Just a scalar variable.

Scaling invariance

h(x1 )(x2 )i = h0 (x1 )0 (x2 )i = s2 h(sx1 )(sx2 )i,

hence

F (x212 ) = s2 F (s2 x212 )

and

F (x212 ) =
Just a (normalisation) constant

N!

7.3

N
.
(x212 )

Logarithmic Correlator.
Constant correlator

F (x1 , x2 ) =
Note:
x212 =

= 0 correlator can
xL12 xR
12 and identify

= (D 2)/2 = 0.
D = 2 + 2, N = N2 /

Our scalar has scaling dimension

F (x1 , x2 ) = N ?!

Not quite: Take limit

N2
N2

N2 log x212 + . . . .
2 
(x12 )


be logarithmic. Still not there. Use LC coordinates

xL = exp(i L ),

xR = exp(i R ).

Why the identication?

2D conformal transformation,
closed string periodicity

+ 2 ,

but

xL/R

unique!

choose appropriate coordinates for boundaries.

L/R
String coordinates are functions of x
except for linear dependence on
= 2i log(xL xR ). Better choice of eld X /xL/R :

h0|L X (x2 )L X (x1 )|0i =

12 2

2
xL2 xL1

More manifestly conformal!

Wick Rotation.
(now

In this context: Typically perform Wick rotation

= i

real)

exp(i L ) = exp(
i) =: z,

exp(i R ) = exp(
+ i) =: z.

Cylindrical coordinates for (euclidean) string:

radius |z| is exponential euclidean

is angular coordinate (naturally

time

periodic).

Standard treatment: Euclidean CFT

z and z are complex

f (z, z) of complex z .

Worldsheet coordinates
Fields are functions

conjugates.

String coordinates are holomorphic functions

z ).
X(z, z) = X(z) + X(

Conformal transformations are holomorphic.

Employ powerful functional analysis: residue theorems.

Euclidean WS convenient and conventional. Could as well work on Minkowski

worldsheet, nothing lost!

7.3

Local Operators

We understand the basic string coordinate eld

X(z)

and

z ),
X(z, z) = X(z) + X(

z ).
X(

Basic objects in a CFT are local operators

7.4

Oi (z, z):

or better

products of elds

and derivatives

evaluated at the same point

(z, z)

n n X ,

on the worldsheet,

Oi = :. . .: implicit (no self-correlations),

O1 = :(X)2 :, O2 = :X X : :X X :, . . . .

normal ordered
for example

Local operators behave

classically as the sum of constituents,

quantum-mechanically as independent entities:
2
Virasoro charges (X) !

recall quantum eects in

Main task: classify local operators.

Descendants.

All local operators transform under shifts

(z,
z ) = (, )

as

O = O + O.
An operator

n n O

is called a descendant of

O.

Shifts are symmetries: No need to

consider descendants.

Weights.

Most local operators classied by weights

(z, z) (sz, sz)

or

.
(h, h)

Transformation under

(z, z) = (z, z)

O0 (z, z) = sh sh O(sz, sz),

+ zO).

O = (hO + zO) + (hO

Transformations are scaling and rotation, hence scaling dimension
spin

For unitary CFT: Both

(X)2 (2, 0).

h, h

Products of local operators

=h+h

and

.
S =hh
are real and non-negative. E.g. weights:

X (1, 0),

O = O1 O2 :

total weight is sum of individual weights classically;

weights usually not additive in quantum theory!

Note:

does not have proper weights, but

Quasi-Primary Operators.

A local operator with weights

quasi-primary if

O (z, z) =
for all

SL(2, C)

does.

dz 0
dz

h 

d
z0
d
z

h

(h, h)

O(z 0 , z0 ).

Mbius transformations

a
z + b
az + b
.
,
z0 =
cz + d
cz + d
boosts (z, z
) = (z 2 , z2 ) it must satisfy
z O + z2 O).

O = (2hzO + z 2 O) + (2h
z0 =

For innitesimal

Descendants of quasi-primaries are not quasi-primary.

Need to consider only quasi-primary operators.

7.5

is called

Primary Operators.

An operator is called primary if it satises the

quasi-primary conditions for all transformations

(z, z) (z 0 (z), z0 (
z ))

or

z )).
(z,
z ) = ((z), (

Innitesimally

O + O).

O = (h O + O) + (h
Note: Correlators are only locally invariant. Only a subclass of conformal
transformations (e.g. Mbius) leaves correlators globally invariant.

Example.

Operator

O = X

is primary;

hO1 O2 i
Invariance under

exact for

21 2
=
.
(z1 z2 )2

z = z 1n :

|n| 1

(Mbius),

|n| > 1

up to polynomials for

State-Operator Map.

= (1, 0).
(h, h)

1/(z1 z2 )2 ).

There is a one-to-one map between

quantum states on a cylinder

local operators (at

(small w.r.t.

R S1

and

z = 0).

Consider the conformal map

z = exp(+i),

z = exp(i),

State given by wave function at constant

, = i
.

= Im :

z plane.
corresponds to z = 0.
Local operator at z = 0 to excite asymptotic wave
Unit operator 1 corresponds to vacuum.

Time evolution is radial evolution in

Asymptotic time

7.4

function.

Operator Product Expansion

In a CFT we wish to compute correlation functions


O1 (1 )O2 (2 ) . . . On (n ) = F12...n .
Suppose

1 2 ;

then can Taylor expand

O1 (1 )O2 (2 ) =

X
1
(2 1 )n O1 (1 ) n O2 (1 ).
n!
n=0

Converts local operators at two points into a sum of local operators at a single
point. Classical statement is exact.

7.6

Quantum OPE.

Quantum-mechanically there are additional contributions from

operator ordering (normal ordering implicit). Still product of local operators can
be written as sum of some local operators

O1 (1 )O2 (2 ) =

X
i

i
(2 1 )Oi (1 ).
C12

More precise formulation with any (non-local) operators  . . .

 X i


O1 (1 )O2 (2 ) . . . =
C12 (2 1 ) Oi (1 ) . . . .
i

This statement is called Operator Product Expansion (OPE).

Cijk (2 1 )

are

called structure constants & conformal blocks. Sum extends over all local
operators (including descendants).
Idea: Every (non-local) operator can be written as an expansion in local operators.
Analog: Multipole expansion.
It works exactly in any CFT and is a central tool.

Higher Points.

Can formally compute higher-point correlation functions:

F123...n =

X
i

ci12 Fi3...n

Apply recursively to reduce to single point.

One-point function is trivial (except for unit operator

hOi i = 0,

h1i = 1.

Higher-point function reduced to sequence of

1)

Cijk :

vast simplication,
k
need only Cij for correlators in CFT,
hard to compute in practice,
result supercially depends on OPE sequence (crossing).

Lower Points.

Two-point function is OPE onto unity

Fij = hOi Oj i =

X
k

Cijk hOk i = Cij1 .

Three-point function determines OPE constants

Fijk = hOi Oj Ok i =

X
l

Cijl hOk Ol i =

X
l

Fkl Cijl .

Lower-point functions restricted by conformal symmetry:

Two-point function only for related operators.

No two-point or three-point conformal invariants. Can map triple of point to
any other triple of points.

7.7

Coordinate dependence of two-point function xed

Fij
Numerator

Nij
.
|i j |2i

depends on dimension, spin, level of descendant and operator

normalisation.

Coordinate dependence of three-point function xed

Fijk
with scaling weights

|i

j |ij |j

Nijk
k |jk |k i |ki

ij = i + j k .

Numerators

depend on dimension,

spin, level of descendant and operator normalisation.

Three-point functions exist for three dierent operators.

Normalise operators, then CFT data consists of

scaling dimensions, spins: spectrum,

coecients of three-point function: structure constants.

7.5

Stress-Energy Tensor

The Noether currents for spacetime symmetries are encoded into the conserved
stress-energy tensor

T =

1

1
(
X)

(
X)

(
X)

(
X)

2
42

Object of central importance for CFT/OPE! Trace is exactly zero: Weyl

symmetry. Remaining components

T =

TLL

=0
J

Current

TRR

translate to euclidean

1
(X)2 ,
2

Ignore string physical state condition

Conservation.

and

1
2.
T = 2 (X)

T = T = 0.

J(z) = (z)T (z)

for

z = (z).

Classical conservation

by means of e.o.m.. QFT: Conservation replaced by Ward identity:

J(z)O(w,
w)
= 2 2 (z w, z w)
O(w, w).

Current

conserved except at operator locations.

OPE: Integrate

over small ball around

1
2

d2 z . . .

|zw|<

R
R
z ( d2 z . . . = i dz . . .)
Z
1
dz J(z)O(w, w)
= O(w, w).

2i |zw|=

Evaluate integration over

Similarly for

T.

Consider only holomorphic part.

7.8

Stress-Energy OPE.

Derive OPE of

First consider translation

and

z = , O = O.

T (z)O(w, w)
= ... +

T.

Need simple pole to generate residue

O(w, w)

+ ...
zw

Further terms with higher poles and polynomials in  . . ..

O has holomorphic weight h. Consider scaling z = z ,

O = (hO + zO). Substitute and require following poles in OPE
Suppose

T (z)O(w, w)
= ... +

O(w, w)

hO(w, w)

+ ...
+
2
(z w)
zw

O is quasi-primary. Consider scaling z = z 2 ,

O = (2hzO + z 2 O). Substitute and require absence of cubic
Next suppose

T (z)O(w, w)
= ... +
Finally suppose

pole

0
hO(w, w)

O(w, w)

+
+
+ ...
(z w)3
(z w)2
zw

is primary. Leads to absence of higher poles

T (z)O(w, w)
=

hO(w, w)

O(w, w)

+
+ ...
2
(z w)
zw

Note that derivatives shift poles by one order.

Descendants are not (quasi-)primaries.
OPE of stress-energy tensor. Compute explicitly (Wick):

T (z)T (w) =

c/2
2T (w)
T (w)
+
+
+ ...
4
2
(z w)
(z w)
zw

Result applies to general CFT's. Virasoro algebra!

T
T
T
T

is a local operator,
has holomorphic weight

h=2

(classical),

is quasi-primary,
is not primary (unless

c = 0),

quartic pole carries central charge

Conformal transformations for

c = D.

almost primary:

c
T = z T + 2 z T + 3 z,
12
 0 2 

dz
c
T 0 (z) =
T (z 0 ) + S(z 0 , z) ,
dz
12

2  0 2
 3 0   0 1
dz
dz
3 d2 z 0
dz
0
S(z , z) =

3
2
dz
dz
2 dz
dz
Additional term

is Schwarzian derivative. Zero for Mbius transformations.

7.9

Introduction to String Theory

Chapter 8

ETH Zurich, HS11

Prof. N. Beisert

8 String Scattering
Compute a string scattering amplitude. Two methods:

worldsheet junction(s). string cylinders with cuts. integration over junctions.

vertex operators. integration over punctures locations.

8.1

Vertex Operators

State-operator map:

Which operator creates a string?

How to specify the momentum

q?

How to specify the string modes?

Solution is related to the operator

Momentum eigenstate: phase for translation

Compute OPE with stress-energy

T (z)O[q](w, w)
=
Primary operator with weights

O[q] = :exp(iq X ):.

Why?

exp(iq  ).

1 2 2
q O[q](w, w)

4
2
(z w)

O[q](w, w)

+ ...
zw

( 14 2 q 2 , 41 2 q 2 )!

non-trivial, non-integer weight,

2
quantum eect .

Consider two-point correlator


2
O1 [q1 ]O2 [q2 ] ' |z1 z2 | (q1 q2 ) .
In fact, zero mode

X = x + . . . contributes extra factor

Z
dD x exp(iq1 x + iq2 x) D (q1 + q2 ).

Hence compatible with primary of weight

( 41 2 q 2 , 14 2 q 2 )


D (q1 + q2 )
O1 [q1 ]O2 [q2 ] '
.
2 2
|z1 z2 | q1
8.1

Operator

O[q](z, z)

creates a string state at

(z, z).

Worldsheet location unphysical,

integrate:

Z
V [q] = gs
Can only integrate weight

(1, 1)

d2 z O[q](z, z).

primary operators. Hence:

M 2 = q 2 = 4/2 ; string tachyon!

intercept a = a
= 1 due to worldsheet integration.

mass

What about excited strings? Level-1 corresponds to

weight is

Z
[q] = gs

O[q].
d2 z X X

(1 + 41 2 q 2 , 1 + 14 2 q 2 ) = (1, 1)

for massless

q.

primary condition removes unphysical polarisations.

gauge d.o.f. are total derivatives.

Vertex operator picture:

CFT
R 2 vacuum is empty worldsheet (genus 0, no punctures).
R d z O[q](z, z) is string vacuum |0; qi (add puncture).
d2 z . . . O[q](z, z) are excited string states. Insertions of n X

n X correspond to
string oscillators n , insertions of
n .

8.2

Veneziano Amplitude

Consider

correspond to

n-point

Ok = O[qk ](zk , zk ))
Z
1
n2
An 2 hV1 . . . Vn i gs
d2n zhO1 . . . On i
gs

amplitude (with

simplest to use tachyon vertex operators,

can do others, but add complications (elds),
computation & result qualitatively the same.

Perform Wick contractions and zero mode integration

2q

Y

O1 . . . On D (Q)

j<k

|zj zk |

j qk

2
2
Integral invariant under Mbius transformations (qk = 4/ ). Map three punctures
to xed positions z1 = , z2 = 0, z3 = 1. Remaining integral for n = 4 strings

A4

gs2 D (Q)

2 q q
2 4

d2 z |z|
8.2

2q

|1 z|

3 q4

can be performed

A4 gs2 D (Q)

(1 2 s/4) (1 2 t/4) (1 2 u/4)

.
(+2 + 2 s/4) (+2 + 2 t/4) (+2 + 2 u/4)

Mandelstam invariants:

s = (q1 + q2 )2 ,
with relation

t = (q1 + q4 )2 ,

u = (q1 + q3 )2 ,

s + t + u = q12 q22 q32 q42 = 16/2 .

This is the Virasoro-Shapiro amplitude for closed strings. Corresponding

amplitude for open strings

A4 g s

(1 2 s) (1 2 t)
(+2 + 2 u)

was proposed (not calculated) earlier by Veneziano. Considered birth of string

theory (dual resonance model).
Amplitudes have many desirable features:

Poles at

s, t, u = (N 1)4/2

or

s, t = (N 1)/2 ,

virtual particles with string

mass exchanged.

J = 2N or J = N . Regge trajectory!
Soft behaviour at s . Even for gravitons!
Manifest crossing symmetry s t u or s t. Amazing!
Residues indicate spin

Not possible for QFT with nitely many particles.

8.3

String Loops

Result exact as far as

is concerned. Free theory in

0 !

However, worldsheet topology matters. String loop corrections for adding handles:
higher genus. Power of

Tree Level.
2 + n

or

gs

reects Euler characteristic of worldsheet.

Worldsheet is sphere or disk with

1 + n/2.

punctures. Euler characteristic

6 global conformal symmetries, integration over

8.3

n3

points.

One Loop.

Worldsheet is torus with

2n

shifts; integration over

integrations; result: elliptic & modular functions; feasible!

Two Loops.
6

n. 2
n 1 points.

punctures. Euler characteristic

moduli: integration over Teichmller space.

Worldsheet is 2-torus with

moduli, no shifts:

2(n + 3)

punctures. Euler characteristic

2 + n.

integrations. Hard, but can be done. No higher-loop

results available.

8.4

Introduction to String Theory

Chapter 9

ETH Zurich, HS11

Prof. N. Beisert

9 String Backgrounds
Have seen that string spectrum contains graviton. Graviton interacts according to
laws of General Relativity. General Relativity is a theory of spacetime geometry.
Strings can move in curved backgrounds.
How are strings and gravity related?

Should we quantise the string background?

Is the string graviton the same as the Einstein graviton?
Is there a backreaction between strings and gravity?

9.1

Graviton Vertex Operator

Compare graviton as string excitation and background. Assume momentum

polarisation

and

 .

Vertex Operator Construction.

Graviton represented by closed string state

L, R,
L, R,
|; qi =  1
1 + 1
1 |0; qi.
Corresponding vertex operator reads

+ X X
)eiqX :
O = :(X X
p
: det g g X X eiqX :.
Insertion into string worldsheet

Z
V =

d2 12  O .

Background Metric Construction.

Flat background with plane wave

perturbation

G (x) = +  eiqx + . . . .
Strings couple to background by replacement

1
S=
22
Same replacement

d2

det g g 12 G (X) X X .

in NambuGoto action.

Perturbation of metric same as vertex operator

S = S0

1
V + ...
22
9.1

Conclusion.

Graviton mode of string is the same as wave on background.

Quantum string on at space contains gravitons. Gravitons introduce curvature

and deform at background. String theory contains quantum gravity. Large
deformations away from at background represented by coherent states of
gravitons.
String theory can be formulated on any background. String quantisation probes
nearby backgrounds. Low-energy physics depends on classical background. Full
quantum string theory is background independent, contains all backgrounds as
dierent states (same as QG).

9.2

Curved Backgrounds

Consider strings on a curved background

G (x),

curious insight awaits. Action in

conformal gauge

For generic metric

Z
1
d2 12 G (X) X X .
S=
2
2
G, e.o.m. for X are non-linear.

Type of model called non-linear sigma model. String background called target
space. Metric eld
(Taylor expansion

G (x)
of G).

is sigma model coupling. Innitely many couplings

In most QFT's couplings are renormalised. Problem here:

Classical action has conformal symmetry.

Conformal symmetry indispensable to remove one d.o.f..
Renormalised coupling

G(x, )

depends on scale

New scale breaks quantum conformal invariance. Anomaly!

Renormalisation.

Compute the conformal anomaly. Background eld

quantisation:

add perturbations

Expansion of action

X0 of string e.o.m..
X = X0 + Y . Quantum eld Y .

Pick (simple) classical solution

S[X] = S[X0 ] + Y 2 + Y 3 + . . .

Value of classical action

S[X0 ]

at

in orders of

irrelevant.

No linear term in Y due to e.o.m. for X0 .

2
Order Y is kinetic term for quantum eld Y .
3
4
Order Y , Y , . . . are cubic, quartic, . . . interactions

+ ...

Use target space dieomorphisms s.t. locally

Z
S=

d2

Y Y + 13 2 R Y Y Y Y
2
9.2

R (x)

is target space curvature tensor.

Kinetic term and quartic vertex:

+ . . ..

At one loop we get tadpole diagram. Insert two-point correlator

2 R Y () Y ()hY ()Y ()i

but we know for

1 2
hY (1 )Y (2 )i ' log |1 2 |.

Not exact, but UV behaviour xed by conformal symmetry. Logarithmic

singularity responsible for renormalisation.

G
= = 2 R ,

Anomaly.

is running coupling, beta function

R = R .

Scale dependence breaks conformal symmetry: Trace of stress energy

tensor after renormalisation

T =

1
X X .
2
2

Anomaly of Weyl symmetry! (gauge xed already)

Conformal/Weyl symmetry is essential for correct d.o.f.. Remove by setting

= 0.

Einstein equation!

R = 0.
Quantum strings can propagate only on Einstein backgrounds. General relativity!

Spin-2 particles at level

Higher Corrections.
perturbative orders in

are gravitons.

There are corrections to the beta function from higher

+ 4

+ ...

= 2 R + 21 4 R R + . . . .
Also corrections from the expansion in the string coupling
Corrections to Einstein equations at Planck scale:

9.3

gs .

= 0.

9.3

Form Field and Dilaton

What about the other (massless) elds?

Two-form

and dilaton scalar

Two-form couples via antisymmetric combination

1
22

1
B (X) X X
2

1
=
22

Z
B.

In fact, canonical coupling of two-form to 2D worldsheet. Analogy to charged

particle in electromagnetic eld. String has two-form charge.
Dilaton couples to worldsheet Riemann scalar

1
4

d2

det g (X) R[g].

Interesting for several reasons:

Euler characteristic

of the worldsheet appears.

Not Weyl invariant.

Scalar can mix with gravity.
Can get away from 26 dimensions.

Low-Energy Eective Action.

of renormalised stress energy tensor

First discuss the various beta functions (trace

T)

1 p
G
B

det
g

X X

22
21 R[g]

g T =

with

= 2 R + 22 D D 41 2 H H ,
B

= 12 2 D H + 2 D H ,

= 21 2 D2 + 2 D D
Quantum string consistency requires

1 2
H H .
24

G = B = = 0.

Standard equations for

graviton, two-eld and scalar. Follow from an action

Z
S

d26 x

p


det g e2 R 21 H H + 4 .

String low-energy eective action. Encodes low-energy physics of string theory.

Further corrections from curvature and loops.
Trivial solution:

G = , B = 0, = 0

(at background). Can also use torus

compactication to reduce dimensions.

9.4

String Coupling.

= 0 is constant, then
Z
p
d2 det g R[g] .

Suppose

topological

Measures Euler characteristic

Set

i0

gs = e

= 2h 2

. Then action yields

dilaton coupling term is

of world sheet.

factors of

gs .

eiS ' ei0 = gs .

Expansion in worldsheet topology.

gs2
String coupling
eld

gs

+ gs0

+ gs2

determined through background: Asymptotic value

of dilaton

String Frame.

Notice unusual factor of

exp(2)

in

S.

Scalar degrees of freedom can mix with metric. Could as well dene

G0 = f ()G .
Remove

exp(2)

through suitable choice of

f.

Go from string frame to

Einstein frame. Standard kinetic terms for all elds.

Noncritical Strings.
D

We have seen earlier that

D 6= 26

breaks Weyl symmetry.

enters in eective action as worldsheet cosmological constant

S = . . . R 12 H H + 4 23 2 (D 26) .
Can have

D < 26,

but requires Planck scale curvature.

Dilaton Scaling.

Dilaton coupling to worldsheet is not Weyl invariant and has

unconventional power of

Consistent choice.
Moves classical Weyl breakdown to one loop. Cancel quantum anomalies of
other elds.

9.4

Open Strings

Open strings lead to additional states, elds and couplings.

Additional string states; e.g. massless vectors (photon):

|; qi = 1
|0; qi.

9.5

Additional vertex operators; e.g. photon

Z
V [, q]

d X exp(iq X)

Additional elds to couple to string ends.

Background couplings can be identied as for closed strings. Vertex operator has
same eect as background eld.
Coupling depends on string boundary conditions: Dp-brane.

Neumann Boundaries.

For all coordinates

conditions: couple a one-form gauge eld

X a , a = 0, . . . , p,

d X a Aa (X) =

end

with Neumann

to end of string

A.
end

natural coupling of a charged point-particle to gauge eld.

string end is a charged point-like object.

Gauge eld

Aa

exists only on Dp-brane. Okay since string ends constrained to Dp

brane.
Classical coupling of

respects Weyl symmetry. Quantum anomaly described by

beta function

aA 4 b Fab
Absence of conformal anomaly requires Maxwell
eective action

b Fab = 0.

dp+1 x 14 Fab F ab .

For planar Dp-brane can also include higher corrections in

Z
S

dp+1 x

Associated low-energy

BornInfeld action:

det(ab + 22 Fab ).

Leading order is Maxwell kinetic term. Corrections at higher orders in

Dirichlet Boundaries.
m = p + 1, . . . , D 1,

Coupling of Dirichlet directions

X m,

dierent.

X m xed, but X 0m can be used.

Dual eld Ym describes transverse
Dp-branes are dynamical objects!

Dp-brane displacement.

Beta function at leading order: massless scalar

a m m Ya
Eective action for higher orders: DiracBornInfeld action

Z
S

dp+1 x

det(gab + 22 Fab ).
9.6

Induced WS metric

gab = a Y b Y .

Embedding coordinates

for Dp-brane.

Combination of

Dirac action for

p-branes

and

BornInfeld action for gauge elds.

D-Branes in a Curved Background.

Z
S

dp+1 x e

Can even add eect of close string elds.

p
det(gab + 22 Fab + Bab ).

gab is induced metric from curved background.

Bab is pull back of 2-form eld B to Dp-brane.
combination 20 Fab + Bab is gauge invariant.
dilaton couples as prefactor like for closed string.

Coincident Branes.
to

For

coincident branes gauge group enlarges from

U (1)N

U (N ).

Gauge eld should couple via Wilson line

Z
T exp

A.
end

Resulting eective action at leading order is

Z
S

dp+1 x tr 14 (Fab )2 + 12 (Da Ym )2 + 14 [Ym , Yn ]2 .

YangMills, massless adjoint scalars, quartic interactions.

9.5

Two-Form Field of a String

We have seen that strings couple to various elds. A string also generates a eld
conguration. Analogy: charged point particle generates Coulomb potential.

Fundamental String.
1-brane.

Consider an innite straight string along

0, 1

directions:

Generates a two-form potential

B = (f 1 1)dx0 dx1
Interactions with metric

and dilaton

require

ds2 = f 1 ds22 + ds2D2 ,

The function

with

r2 = x22 + . . . + x2D1
f =1+

e2 = f 1 .

reads

gs2 N D4
.
rD4

This satises the low-energy eective string e.o.m.

function.
Note: Source at the location of the string (r

9.7

= 0).

because

is a harmonic

E.o.m. follow from combination of spacetime action and worldsheet coupling to

two-form

Z
H H +

D
Source term

D2 (r)

absorbed by worldsheet.

Charge of string measured by Gauss law via

sphere at xed

H .

Put

(D 3)-dimensional

r.
Z
H = N.

Q=
D3
Above string has

units of charge (quantised).

The fundamental string is not a D1-brane: Open strings do not end on it. It is the
string itself.
Solutions with more than one centre permissible.

Magnetic Brane.
(D 5)-brane.

Another solution of the string eective e.o.m. describes a

It uses a dual

(D 4)-form

potential

H = dB,

H = dC.

dened through

It carries magnetic charge

Z
H.

Q=
3
The source is located on the

(D 5)-brane(s).

compensates source.

9.8

The coupling of

(D 5)-branes

to

Introduction to String Theory

Chapter 10

ETH Zurich, HS11

Prof. N. Beisert

10 Superstrings
Until now, encountered only bosonic d.o.f. in string theory. Matter in nature is
dominantly fermionic. Need to add fermions to string theory.
Several interesting consequences:

Supersymmetry inevitable.
Critical dimension reduced from

D = 26

to

D = 10.

Increased stability.
Closed string tachyon absent. Stable D-branes.
Several formulations related by dualities.

10.1

Supersymmetry

String theory always includes spin-2 gravitons. Fermions will likely include spingravitini

3
2

supergravity. Spacetime symmetries extended to supersymmetry.

Super-Poincar Algebra.

Super-Poincar algebra is an extension of Poincar

algebra.
Poincar: Lorentz rotations

M ,

[M, M ] M,

translations

[M, P ] P,

Super-Poincar: Odd super-translation

[M, Q] Q,
N:

rank of supersymmetry

relates particles of

of dierent spin,

P .

QIm (a:

[Q, P ] = 0,

[P, P ] = 0.

spinor)

{QIm , QJn } IJ mn
P .

I = 1, . . . , N .

of dierent statistics,

and attributes similar properties to them. Symmetry between forces and

matter.
More supersymmetry, higher spin particles.

1): 16 Q's.
(spin 2): 32 Q's.

gauge theory (spin

gravity theory

Superspace.

Supersymmetry is symmetry of superspace. Add anticommuting

a
coordinates to spacetime x (x , I ). Superelds: expansion in yields various

elds

F (x, ) = F0 (x) + Im FmI (X) + 2 . . . + . . . + dim .

Package supermultiplet of particles in a single eld.

10.1

Spinors.

Spin(D 1, 1)

Representations of

(Cliord).

4
Complex spinors (Dirac) in (3 + 1)D belong to C . Can split into chiral spinors
2
2
2
2 ) = C2 .
(Weyl): C C . Reality condition (Majorana): Re(C C
Spinors in higher dimensions:

spinor dimension times

chiral spinors (Weyl) for

D D + 2.

for

even.

D = 0, 1, 2, 3, 4 (mod 8).
(MajoranaWeyl) for D = 2 (mod

real spinors (Majorana) for

real chiral spinors

8).

Maximum dimensions:

D = 10:
D = 11:

16

real chiral spinor with

real spinor with

32

components (gauge).

components (gravity bound).

Super-YangMills Theory. N = 1 supersymmetry in D = 10 Minkowski

space:

gauge eld

A :

8 on-shell d.o.f..

adjoint real chiral spinor

m :

8 on-shell d.o.f..

Simple action

Z
S

Supergravity Theories.

N
N
N
N

n D n .
d10 x tr 41 F F + mn
Four relevant models:

= 1 supergravity in 11D: M-Theory.

= (1, 1) supergravity in 10D: Type IIA supergravity.
= (2, 0) supergravity in 10D: Type IIB supergravity.
= (1, 0) supergravity in 10D: Type I supergravity.

Fields always 128+128 d.o.f. (type I: half, SYM only 8+8):

type

gr.

[4]

[3]

[2]

[1]

sc.

gravitini

spinors

IIA

(1,1)

(1,1)

IIB

(2,0)

(2,0)

(1,0)

(1,0)

SYM

(0,1)

M-theory has no 2-form and no dilaton: no string theory. Type IIA, IIB and I
have 2-form and dilaton: strings?!

10.2

GreenSchwarz Superstring

Type II string: Add fermions

Im

to worldsheet. Equal/opposite chirality: IIB/IIA

10.2

Action.

Supermomentum

d2
Z 

S
+

Im Jn .
= X + IJ mn

p
det g g
1

d d

dX + d d

Action has kappa symmetry (local WS supersymmetry). Only in

Note: fermions

D = 10!

have rst and second class constraints. Non-linear equations of

motion. In general dicult to quantise canonically. Conformal gauge does not

resolve diculties.

Light-Cone Gauge.

Convenient to apply light-cone gauge. Simplies

drastically: quadratic action, linear e.o.m.

Z
S
Bosons

~
X

with

~ R X
~ + 1 1 R 1 + 1 2 L 2
d L X
2
2
2

~ =0
L R X

Vector of transverse

SO(8): 8v

Left and right moving d.o.f.

Fermions

1 , 2

with

R 1 = 0

and

Real chiral spinor of transverse

Left and right moving d.o.f. in

IIB/IIA:

8s + 8s

Spectrum.

or

L 2 = 0

SO(8): 8s

or

8c .

Equal/opposite chiralities for

8s + 8c
1

and

2 ,

respectively.

Vacuum energy and central charge:

8 bosons and 8 fermions for L/R:

aL/R = 8(1) 8(1) = 0.

no shift

for

L0

constraint. Level zero is massless! No tachyon!

1
as
due to kappa).
2

c = 10 + 32 12 = 26 (fermions count
Super-Poincar anomaly cancels.
Expansion into bosonic modes
zero mode,

n > 0:

and fermionic modes

n . n < 0:

creation,

annihilation.

Zero modes and vacuum:

0 is c.o.m. momentum: ~q.

0 transforms the vacuum state:

(8s ) :
8v 8c
anti-chiral (8c ) : 8v 8s

chiral

|8v + 8c , qi
vacuum |8v + 8s , qi

vacuum

Spectrum at level zero: massless

Type IIA closed:

(8v + 8s ) (8v + 8c )

(IIA supergravity)

8v 8v + 8s 8c = (35v + 28v + 1) + (56v + 8v ),

8v 8s + 8v 8c = (56s + 8c ) + (56c + 8s ).
10.3

n = 0:

Type IIB closed:

(8v + 8c ) (8v + 8c )

(IIB supergravity)

8v 8v + 8c 8c = (35v + 28v + 1) + (35c + 28v + 1),

8v 8s + 8v 8s = (56s + 8c ) + (56s + 8c ).

Type I closed:

(8v + 8c ) (8v + 8c )

mod

Z2

(I supergravity)

(35v + 28v + 1) + (56s + 8c ).

Type I open:

10.3

8v + 8c

(SYM).

RamondNeveuSchwarz Superstring

There is an alternative formulation for the superstring: RNS. Manifest worldsheet

rather than spacetime supersymmetry!

Action.

Action in conformal gauge:

Z
S

d2

1
X R X
2 L

+ iL R L + iR L R .

action is supersymmetric.
fermions are worldsheet spinors but spacetime vectors.

Bosons as before. Fermions can be periodic or anti-periodic.

Ramond Sector. ( + 2) = () periodic.

Fermion modes n as for bosons.
Vacuum is a real 32-component fermionic spinor.
a = 12 8(1) + 12 8(1) = 0.
GSO projection: only chiral/anti-chiral states are

physical!

NeveuSchwarz Sector. ( + 2) = () anti-periodic.

Half-integer modes for fermions: n+1/2 .
Vacuum is a bosonic scalar.
a = 12 8(1) 14 8(1) = 21 .
GSO projection: physical states require 2n+1 .

String Models.

No tachyon!

IIB/IIA strings for equal/opposite chiralities in L/R sectors.

Independent choice for left/right-movers in closed string. Four sectors: NS-NS,

RN-S, NS-R, R-R. Independent vacua.

10.4

Superconformal Algebra.

(Left) stress-energy tensor and conformal

supercurrent:

TL = L X L X + 2i L L L ,
Superconformal algebra

Ln , Gr (2r

JL = L L X

is even/odd for R/NS):

[Lm , Ln ] = (m n)Lm+n + 18 cm(m2 1)m+n ,

[Lm , Gr ] = ( 21 m r)Gm+r ,
{Gr , Gs } = 2Lr+s + 12 c(r2 14 )r+s .
c=D

(conventional factor

Comparison.
related by

3
in
2

for super-Virasoro).

GS and RNS approach yield the same results. In light cone gauge:

SO(8)

triality

Compare features of both approaches:

fermions are spinors in

GS

RNS

target space

worldsheet

worldsheet supersymmetry

manifest

superconformal eld theory

target space supersymmetry

manifest

supergravity couplings

all

some (NS-NS)

spacetime covariant

Third approach exists: Pure spinors (Berkovits). Introduce auxiliary bosonic

spinor satisfying = 0. Shares benets of GS/RNS; covariant formulation.

10.4

Branes

Open superstrings couple to D-branes. Open string spectrum carries D-brane

uctuations.

massless:

N =1

Super-YangMills reduced to

(d + 1)D.

heavy string modes.

sometimes: scalar tachyon.

Stable Dp-Branes.

D-branes can be stable or decay. Open string tachyon

indicates D-brane instability.

D-branes in bosonic string theory are instable.

Dp-branes for IIB superstring are stable for
Dp-branes for IIA superstring are stable

p odd.
for p even.

10.5

T-duality maps between IIA and IIB.

Stability is related to supersymmetry. Boundary conditions break symmetry

SO(9, 1) SO(d, 1) SO(9 d).

p odd/even in IIB/IIA.
supersymmetries preserved for p even/odd in IIB/IIA.

Lorentz:

16 supersymmetries preserved for

no

Supersymmetry removes tachyon; stabilises strings.

Supergravity p-Branes.

D-branes are non-perturbative objects. Not seen

perturbatively due to large mass.

Stable Dp-branes have low-energy limit as supergravity solutions.

p-brane

supported by

(p + 1)-form,

gravity and dilaton.

IIB/IIA have dilaton and two-form (NS-NS sector).

IIB/IIA has forms of even/odd degree (R-R sector); relevant for stable
Dp-branes.

Features:

p-branes carry (p + 1)-form charge. charge prevents p-branes from evaporating.

charge density equals mass density.
16/32 supersymmetries preserved. 1/2 BPS condition.
Non-renormalisation theorem for 1/2 BPS: p-branes same at
weak/intermediate/strong coupling. BPS p-branes describe Dp-branes exactly.

Type-I Superstring.

Consider open strings on D9-branes.

Gravity and gauge anomaly cancellation requires:

gauge group of dimension

496.

some special charge lattice property.

Two solutions:

SO(32)

and

E 8 E8 .

Here:

SO(32).

Breaks 1/2 supersymmetry:

Type I.

Sometimes considered independent type of superstring.

Or: IIB, 16 D9 branes, space-lling orientifold-plane.

10.5

Heterotic Superstring

Two further superstring theories.

Almost no interaction between left and right movers. Exploit:

left-movers as for superstring: 10D plus fermions.

right-movers as for bosonic string: 26D (16 extra).

Heterotic string. 16 supersymmetries.

Anomaly cancellation requires gauge symmetry:

HET-O:

SO(32)

or

10.6

HET-E:

E8 E 8 .

Gauge group supported by

16

HET-E interesting because

E8

E5 = SO(10),

10.6

internal d.o.f..
contains potential GUT groups:

E4 = SU (5),

E3 = SU (3) SU (2).

Dualities

Dualities relate seemingly dierent superstring theories.

T-duality: time vs. space duality on worldsheet.

S-duality: analog of electro-magnetic duality.

Dualities considered exact because of supersymmetry. Tests.

A Unique Theory.

T-duality: IIA

Dualities related various superstrings:

IIB; HET-E

S-duality: HET-O

Type I; IIB

HET-O

IIB

Furthermore IIA and HET-E at strong coupling: 11D supergravity theory (with
membrane).
Suspect underlying 11D theory called M-theory. Superstring theories as various
limits of M-theory.

Mirror Symmetry.

Dualities applied to curved string backgrounds: Curved

spacetimes with

inequivalent metrics can have

equivalent string physics.

E.g.: T-duality between large and small circles.

Many examples for CalabiYau

manifolds.

String/Gauge Duality.

Some low-energy eective theories can become exact.

String physics at the location of a brane described exactly by corresponding YM

theory.
Example:

coincident D3-branes in IIB string theory. Eective theory:

Super-YangMills theory in 4D.

10.7

N =4

Introduction to String Theory

Chapter 11

ETH Zurich, HS11

Prof. N. Beisert

11 AdS/CFT Correspondence
Conjectured exact duality between string theory and CFT.

Remarkable!
Precise formulation of a string/gauge duality.
Holographic. Dierent number of spacetime dimensions.
5
Main example: AdS5 S string and N = 4 SYM.

11.1

Stack of D3-Branes

Consider 3-brane solution of IIB supergravity (4

1/2

ds = h

1/2

dx + h

with harmonic function

dy ,

y )

h(y) = 1 + N/|y|4 .

U (N ) N = 4

Now approach brane at

H5 = h dhdx + h2 (dhdx4 ),

IIB string theory background with stack of

physics described by

x k,

y = 0.

D3-branes. Low-energy brane

SYM.

Alternatively send

N .

Harmonic function limits to h(y) = N/|y|4 .

Background becomes AdS5 S 5 with 5-form ux.
S 5 at constant |y|. AdS5 combined from x and |y|.
Claims: AdS/CFT correspondence (Maldacena)

3-brane at boundary of

space.

U (N ) N = 4 SYM.
5
bulk AdS5 S strings.

Boundary physics described exactly by

Open string on boundary can probe

Precise matching of all observables in both models.

Map of coupling constants

11.2

(/R, gs )

with

(gYM , N ).

Anti-de Sitter Geometry

Anti-de Sitter space

AdS5

Non-brane modes decouple.

AdSd

is curved spacetime:

Constant scalar curvature.

Analogous to sphere and hyperbolic space
curvature

Euclidean

Minkowski

dS

AdS

11.1

Isometry group:

SO(d 1, 2).

Topology: Solid cylinder

d1

dimensions.

R Dd1

Boundary: Cylinder surface

Same as conformal group in

R S d1 .

time-like geodesics never reach boundary.

space-like geodesics reach boundary at innite distance.
light-like geodesics reach boundary in nite time. bulk and boundary interact
via massless elds.

11.3

N = 4 Super-YangMills

Maximally supersymmetric gauge theory in 4D. Dimensional reduction from

N =1

SYM in

D = 10.

Fields:

gauge eld,
4 adjoint Dirac fermions,
6 adjoint scalars.

Remarkable properties:

no running coupling,

= 0.

exact 4D superconformal symmetry; 4D (S)CFT.

...

11.4

Tests

Want to test AdS/CFT correspondence. Predictions:

String spectrum matches with spectrum of local operators.

String and gauge correlation functions match.

Problem: Strong/weak coupling duality.

Weakly coupled strings is strongly coupled gauge theory.

Weakly coupled gauge theory is strongly coupled strings.

11.2

Test BPS quantities, protected (independent of coupling).

Supergravity modes agree with BPS operators.

Supergravity correlators match with BPS correlators.

What about other quantities?

String and gauge theory appear integrable at large

N.

Integrability: Hidden symmetry to constrain dynamics.

Can compute observables eciently even at nite coupling.
Precise agreement found in all performed tests.

Other tests performed, e.g. Wilson loops vs. string area.

11.3

Introduction to String Theory

Problem Set 1

ETH Zurich, HS11

B. Schwab, Prof. N. Beisert

1. On the importance of quantum gravity (easy)

Let us get some intuition on the order of magnitudes:

a)

Consider a gravitational atom, an electron bound to a neutron by the gravitational

force. Electromagnetic dipole eects can be neglected. Perform a semiclassical calculation to determine the radius of the orbit of the electron (rst Bohr radius). Relate
this radius to an appropriate distance in physics.

b)

In natural units, where

~, G

and

are set to 1, a stellar black hole radiates like

kT = 1/8M .

a black body at a temperature given by

units (reinsert

G, ~

and

c)

Give the temperature in SI

and calculate the temperature of a black hole weighing

one solar mass.

2. Relativistic point particle (intermediate)

The action of a relativistic point particle is given by

Z
Srp =

ds
P

with the relativistic line element

ds2 = dX dX = c2 dt2 dx2 dy 2 dz 2

and

a (yet to be determined) constant. The path

can be parametrised by a parameter

between two points

The integral over the line element

X1

ds

and

X2

becomes an

integral over the parameter

Z2
Srp =

X X
.

()

a)
b)

Parametrise the path by the time coordinate

|~x |  c

to determine the value of the constant

t and take the non-relativistic limit

. Characterise the appearing terms.

Derive the equations of motion by varying the action in (). (You may set
from now on.) Hint: Calculate the canonically conjugate momentum

c)

Show that the form of the action is invariant under reparametrisations

c=1

rst.

0 = f ( ).

This is what we call manifestly invariant.

d)

Consider an electrically charged particle with charge

gauge eld

q.

In the presence of an external

there is an additional term in the action governing the interaction

between particle and eld given by

Sem

q
=
c

A (X) under
X to nd the

d A (X)

X
.

X .

Find the variation of

a variation of the path

S = Srp + Sem
P from above

equations of motion for the particle. Hint: Use

w.r.t.

to simplify the expression.

Vary the action

3. Polynomial action (intermediate  hard)

There is another way to write the action of a relativistic particle.
auxiliary eld called vierbein (or einbein in this case)

We introduce an

along the worldline of the

particle and rewrite the action in the form

Z
Spp =

d (e1 X 2 m2 e).

a)

Show that

b)

Derive the equations of motion by varying

c)

Show that
order in

.

Spp

is equivalent to

Srp

above by eliminating the einbein from the action.

Spp

with respect to

and

e.

Spp is invariant under innitesimal reparametrisations = ( ) to linear

First nd the correct transformation of X . The einbein transforms like

(can you derive it?)

e = (( )e).
d)

Reparametrisation invariance is a gauge invariance. Thus by xing a gauge we can

eliminate one degree of freedom. Assume a gauge in which

can be written like

e=

is constant. Show that

`
,
2 1

is the invariant length of the worldline for a path starting at X (1 ) and

ending at X (2 ). Hint: Meditate on the role of the einbein and on how to dene `.
where

Introduction to String Theory

ETH Zurich, HS11

Problem Set 2
B. Schwab, Prof. N. Beisert

(intermediate)
In this exercise we examine the classical symmetries of the Polyakov string
Z

T
SP =
d2 g g X X .
2

1. Symmetries of the classical string

We start with the global symmetries  Lorentz and translational symmetry  and proceed
to gauge symmetries  reparametrisation and Weyl symmetry.
a) Consider the transformation

X X + a
which is a combination of a Lorentz transformation and a translation, a.k.a. a Poincar
transformation. Using the Noether procedure show that in conformal gauge g =
the Noether currents corresponding to these symmetries are given by

P = T X ,

J
= P X P X .

b) Find and identify the conserved charges associated with Lorentz boosts and time

translations.

Hint

: For the Lorentz boost assume X 0 = t.

c) Show that the Polyakov string action is invariant under a reparametrisation

().

d) Show that the Polyakov string is also invariant under Weyl transformations: local

length-changing but angle preserving transformations of the metric g e2() g .

e) Consider an innitesimal Weyl transformation

and X = 0

g = 2g

and show that Weyl symmetry implies the vanishing of the trace of the worldsheet
energy-momentum tensor
T = 0.
Hint:

The variation of the determinant is given by

det g = det g g g .

(easy  intermediate)
The classical solution for the wave equation is given by

2. Classical spinning strings

X (, ) = XL ( + ) + XR ( ),
the constraints by

X X 0 = 0

and

X 2 + X 02 = 0.

It will be benecial to work in static gauge X 0 (, ) = R ( being the worldsheet time).

a) Show that

X 0 = R
X 1 = R cos() cos( )
X 2 = R cos() sin( )
can be written in the form of the general solution of the wave equation and that it
fulls the constraints. Calculate the energy P 0 = E and the angular momentum Jij
of the solution.
b) Show that

X 0 = R
X 1 = R cos() cos( )
X 2 = R cos(2) sin(2 )
can be written in the form of the general solution of the wave equation but does not
full the constraint equations.
c) (optional)

Closed strings can develop cusps. These points 0 on the string are
indicated by a singularity in the parametrisation

~
X
(0 , t) = 0.
t
Show that the string reaches the speed of light at a cusp. Moreover, show that cusps
move perpendicular to the direction of the string.
d) (advanced)

Explain why cusps form generically in 3 + 1 dimensions but not so in

higher dimensions.

Introduction to String Theory

Problem Set 3

ETH Zurich, HS11

B. Schwab, Prof. N. Beisert

1. Light cone tensors (easy)

We want to derive some relations between Lorentz tensors

a)

Starting from a Lorentz vector

(X+ , X , Xi ),
b)

and light cone tensors.

show that you can dene light cone coordinates

n .

in terms of two null vectors

Show that the equality of components of the Lorentz tensors

A1 2 ...n = B1 2 ...n

implies the equality of the light cone tensor components, i.e.

A++...+ = B++...+ ,

A++... = B++... ,

and give explicitly the light cone components

A... = B... ,

L++ , L+ , L+ , L

in terms of the

L .

components of a Lorentz 2-tensor

c)

...

Furthermore, show that the trace of a rank-2 light cone tensor

is given by

A = A+ A+ + Aii .

2. Light cone gauge and mode expansion (intermediate)

Using our newly found knowledge about light cone tensors, we will investigate the form
of the angular momentum generator. The mode expansion is given by

i X n in( )
i X
n in( +)
e
+
e
.
X (, ) = x0 + 2 p +
2 n6=0 n
2 n6=0 n
a)

In the earlier problem 2.1.a) you derived the angular momentum current

and its

conserved charge

2
0
d J
.

J =
0
Express

P0

in terms of the mode expansion.

Calculate

in terms of the mode

expansion. Hint: Contemplate the meaning of conserved quantity and use

d ein = 2n,0 .

b)

Express

J .

Ji

in terms of the above derived mode expansion for the Lorentz tensor

In a quantum eld theory, symmetry generators should be realised by hermitian

operators

(J ) = J .
Assume canonical commutation relations

[x , p ] = i ,

and show that

Ji

is not

hermitian. Hermiticise the generator.

3. Maxwell and Kalb-Ramond elds (intermediate)

Light cone gauge is not only useful in string theory to extract physical information. It
also is a valid gauge in other theories.

A .

First we will work on the Maxwell gauge eld

Then we will turn to the Kalb-Ramond eld

which will enter our description of

quantum string theory in due course. (If you feel condent enough you can skip parts
a) and b).

Otherwise work carefully through all the subproblems for maximal benet!

You may nd chapter 10 in Zwiebach  A rst course in String theory useful. Use the
language of dierential forms if you are familiar with it.)

a)

The Maxwell eld

A (x)

has a gauge symmetry

A0 = A + (x).
F

We dene the antisymmetric eld strength tensor

by

F = A A .
Show that

is gauge invariant and derive the equations of motion for

from the

action (leaving coupling constants aside)

SYM

1
=
4

dD x F F .

Rewrite the equations of motion in momentum space

b)

p .

We want to implement light cone gauge. Express the gauge transformation in mo-

(p), you can gauge away the

+-component of the light cone gauge eld (A+ , A , Ai ) and deduce that the equation

mentum space.

Show that, by a sensible choice of

of motion in momentum space drastically simplies in this gauge. Count the total
number of independent degrees of freedom of the gauged Maxwell eld.
The Kalb-Ramond eld

is an antisymmetric Lorentz tensor with the gauge symmetry

transformation

B =   .
We dene a eld strength and an action for

H = B + B + B
c)

and

Show that the gauge transformation of

by

SKR

1
=
12

dD x H H .

has a redundancy

0 =  +
under which

is invariant. Express the gauge transformations in light cone mo-

mentum space and show that you can gauge away the component
eective gauge transformation of

d)

Go through the steps in a), b) for

is generated by

and

and

+ ,

such that the

i .

 bearing in mind the result of c) 

and show that the Kalb-Ramond eld has only one independent degree of freedom in
four dimensions.

e)

by contracting the
In four dimensions, we can dene a dual eld H

eld strength H
with the totally antisymmetric tensor of fourth order

(advanced)

= H .
H
Using the result you found in the last part of this problem show that the dual eld
can be expressed by the derivative of a single scalar eld. What does this imply for
the Kalb-Ramond eld in four dimensions?

Introduction to String Theory

ETH Zurich, HS11

Problem Set 4
B. Schwab, Prof. N. Beisert

1. Virasoro algebra (intermediate)

In this exercise we want to investigate in detail the Virasoro algebra as it appears up in

light cone string theory. For simplicity we will only work with the left movers LLn as the
right movers LR
n commute with these and satisfy an identical algebra. The mode operators
ni (with i = 1, . . . , D 2) satisfy the algebra (we drop the L/R superscript)
j
[ni , m
] = m ij n+m ,

and the normal ordered Virasoro generators are given by

Ln =

1X i
1X i i
np pi +
.
2 p0
2 p<0 p np

An algebra g is a Lie algebra if its product [, ] : g g g (called the Lie bracket) is

antisymmetric [a, b] = [b, a] a, b g and satises the Jacobi identity

 
 

a, [b, c] + b, [c, a] + c, [a, b] = 0 a, b, c g.
a) Show that the commutator of two Virasoro generators with m + n 6= 0 is given by

[Lm , Ln ] =

1X i
i
i
p mp p+n
+ (m p)n+mp
pi
2 p

b) By relabelling the summands, rewrite the above result in the following form

[Lm , Ln ] = (m n)Ln+m .
Argue that the complete solution, including the terms n = m is given by

[Lm , Ln ] = (m n)Lm+n + C(m)m+n,0

where C(m) is a real valued, odd (C(m) = C(m)) function. The last term is called
the central extension of the Virasoro algebra. Determine C(m) up to two constants by
1
(D 2)(m3 m).
considering the Jacobi identity. The solution is given by C(m) = 12

2. Analytical continuation of the

-function (intermediate  hard)

P
1
In the lecture you have seen the peculiar result of the sum (1) =
n=1 n = 12 . In this
exercise we will try to understand where the result comes from by analytically continuing
the -function. The and functions of a complex variable z are given by

and

dt et tz1

(z) =

(z) =

X
1
nz
n=1

a) We start by regularising (1) using a small parameter . Show that we can write

P n

the zeta function like  (1) = 

in the limit  0. Argue that the
n=1 e
sum in this expression is convergent and give the solution. Expand the expression for
1
small  and show that the result is given by  (1) 12 12
+ O().

b) Show that for Re(z) > 1 you can write

Z
(z)(z) =
0

dt tz1
.
et 1

Conclude that it is possible to rewrite the integral to give



Z 1
1
1 1
t
1
1
1
z1
dt t
+
+

+
(z)(z) =
t
e 1 t 2 12
z 1 2z 12(z + 1)
0
Z
z1
dt t
+
.
et 1
1
c) (advanced)

The right hand side is well dened for Re(z) > 2 (why?). We know
that (z) has poles for z = 0, 1, 2, . . . with residues


 (1)n
.
Resz0 =n (z0 ) =
n!
Conclude that the values of (z) at z = 0 and z = 1 are

(0) =

1
2

and

(1) =

1
.
12

3. Poincar transformations (easy)

Poincar transformations x 7 x + a form a group whose product is dened as

T (1 , a1 )T (2 , a2 ) = T (1 2 , a1 + 1 a2 ). The inverse reads T (, a)1 = T (1 , 1 a).
Consider an innitesimal transformation with generators J and P

i
T (1 + , ) = 1 + J i P + . . . .
2
They dene the Lie algebra of the Poincar group. Show that

T (, a) T (1 + , ) T (, a)1 = T (1 + )1 ,  1 a .
How do J and P transform under T (, a)? What relations do you get when you take ,
a to be innitesimal as well?
2

Introduction to String Theory

Problem Set 5

ETH Zurich, HS11

B. Schwab, Prof. N. Beisert

(hard)
Since Lorentz invariance is obscured due to the light cone gauge it is not obvious that the
following commutator vanishes
1. Lorentz invariance in light cone gauge

[J i , J k ] = 0 .

This exercise sheet will be solely concerned with the calculation of this commutator and
its physical implications for bosonic string theory. In a previous sheet the generator was
determined to be given by (up to a doubling of the latter term due to left and right
movers)
J

X
1 i
i
i

i
= (p x0 + x0 p ) x0 p + i
(n
ni n
n ).
2
n=1

Furthermore, we have
L0
p = 2 +
2 p

and

2
Ln
p+

with Ln the Virasoro generators from the previous sheets. We depart here from the
denition of sheet 4 and account for the normal ordering ambiguity in L0 by dening

1 2 X i i
n n a
L 0 = 0 +
2
n=1

where a is the so called intercept. This means we have two free parameters to adjust
during this calculation: the intercept a and the dimension D of spacetime. There are
many subleties in this calculation. Careful checks after every step are recommended.

a)

Begin by calculating all the possible commutators between the zero modes p+ , p , pi ,

i
i
x
0 , x0 and the n , n modes using the given commutators
[p+ , x
0 ] = i,

[pj , xk ] = i jk ,

i
[m
, nj ] = mm+n ij .

Watch out for subtleties when it comes to commutators with p and . The

, n ] is the hardest here. However, you know its naive form already
commutator [m
from the last sheet. Remember that there is a normal ordering ambiguity for 0 !
Hint:

b)

Calculate the commutator. The expected result is

2
1 X 1 [i j]
i
j

[J , J ] = 2 p

0 p+ n=1 n n n




X
D2
1
D2
2
[i
j]
2 n+
2a
n n .
+ 2 + 2
(p ) n=1
12
n
12

For general values of a and D we say that the symmetry is anomalous because the
right hand side is not zero. However, we can make it vanish. What are the reasons
for the rst term and conditions for the second term in this expression to vanish?

Introduction to String Theory

ETH Zurich, HS11

Problem Set 6
B. Schwab, Prof. N. Beisert

(warm-up)
Already as early as 1921 T. Kaluza proposed a model to unify gravity and electromagnetism by introducing a (large) fth dimension. O. Klein greatly improved this ansatz
by introducing the idea of a compact fth dimension. In Kaluza-Klein theory the metric
tensor GM N in 4 + 1 dimensions is split into 3 elds: a metric eld g in 3 + 1 dimensions,
a U (1) gauge eld A and a scalar called the dilaton .
1. Kaluza-Klein theory

a)

Begin by considering a massless scalar eld in D dimensions. Argue that compactifying one spacelike dimension on a circle of radius R:
xD1 xD1 + 2R

will give you an innite tower of massive scalar elds in D 1 dimensions and one
massless scalar. Hint: Contemplate the kind of momenta you get in compact spaces.
Use the Fourier series. Write down the equations of motion of the scalar in D dimensions to see which D 1 dimensional modes are massive and which massless.
b)

How can you make this tower of massive scalar elds vanish or at least impossible to
detect at small energies?

c)

(optional) Now we return to KK theory. The metric has the following form
GM N =

1/3



g 2 A A A
.
A

5D gravity is of course invariant under reparametrisations. However after singling out

the fth dimension we can only have 4D reparametrisations and a transformation
x = x

x5 + f (x ) = x5 .

and

Show that this transformation implies a gauge transformation on the eld A ( which
is contained in G5 ).
We conclude by remarking that in KK theory the compactication of one dimension yields
one massless metric eld, one massless gauge eld and one massless scalar along with three
innite towers of massive modes. Let us now turn to string theory.

(intermediate)
When compactifying one dimension of string theory on a circle, new stringy eects appear
that cannot be seen in eld theory. One of these eects is winding of the string around
the compactied dimension. We compactify the coordinate D 2 on a circle (drop the
index D 2 for simplicity) and request that

2. T-duality: Self-dual radius

X D2 ( + 2, ) = X D2 (, ) + 22 w

where w = mR/2 is the winding. The left and right movers are then dened as usual for
all directions except for the compact dimension where
2
i X nL
1
exp(in L ),
XL = x + (p + w) L +
2
2
2 n6=0 n
1
2
i X nR
XR = x + (p w) R +
exp(in R ).
2
2
n
2 n6=0
a)

Derive the (D 1)-dimensional mass-squared M 2 = p2 of states in the presence of a

compact dimension in terms of the level operators N L and N R , the winding number m
and compact momentum p = n/R. Hint: Use the Virasoro constraints LL0 = LR0 = a.

b)

Show that the level matching constraint N L = N R does not hold for strings with
both winding number m and Kaluza-Klein momentum number n not equal zero.
What happens to these states at R ?

c)

Consider the mass formula for the cases

m = n = 0;

m = 0, n 6= 0;

m 6= 0, n = 0;

m = n = 1;

m = n = 1.

For which values of N L and N R does the spectrum (possibly) contain tachyonic and
massless states? What is their spin (scalar, vector, tensor) as viewed from D 1
non-compact spacetime dimensions?
d)

In the case of non-zero winding and non-zero compact momentum (m = n = 1 and

m = n = 1) show that there is a special radius R where some states become
massless. What happens for the other cases at this radius?

Nota bene: What happened here? After compactifying one dimension we were left with
two massless vector states with U (1) U (1) gauge group, that is one more than in KK
theory. Choosing the self-dual radius R produces four additional massless vectors which
combine with the generic ones to SU (2) SU (2) gauge elds. What we witness here is
symmetry enhancement at the selfdual point! These are true string theory eects that
cannot appear in KK theory.

Introduction to String Theory

ETH Zurich, HS11

1.

Problem Set 7
B. Schwab, Prof. N. Beisert

p-branes (intermediate)

The mechanics of a generic p-brane (in the sense of an object with p + 1 dimensional
worldvolume) can be described by the Dirac-type action
Z
p
S = T dp+1 det ,
where we have p + 1 coordinates with = 0, . . . , p. The matrix is the
the Minkowski metric onto the brane

pullback

of

X X
.

Show that this action is equivalent to the Polyakov-style action

Z
p


T
S=
dp+1 det g g X X (p 1) ,
2
where g is the dynamical worldvolume metric. Do these actions look familiar to you in
the cases p = 0, 1?
2. Stretched strings (intermediate)

An open string can stretch between two Dp-branes. In fact there are four possibilities
for a string to stretch between two branes, called sectors. Two of the sectors are strings
beginning and ending on the same branes denoted [11] and [22]. The sectors [12] and
[21] contain the cases where the string streched between the brane. The last two cases
are dierent because orientation matters. We will be interested in the case where the
endpoints of the string lie on two dierent branes.
a) Write down the mode expansion for a string stretched between two parallel Dp-branes

and interpret the result.

b) How does the distance between the branes aect the spectrum of the string? What

happens for coincident branes?

Hint:

Consider the mass-squared.

3. Orbifolds (intermediate  hard)

After having seen how a string behaves under compactication of one dimension, we
want to nd out how it behaves under restricting it to the half line xD1 0 by the
identication
xD1 xD1 .
Such a space is called orbifold in string theory. Again we abbreviate the relevant coordinate X() := X D1 (), and introduce an operator U acting as ( = 0, . . . D 2)

U X()U 1 = X( + ),

U X ()U 1 = X ( + ).

U is a symmetry of the orbifold theory, so only states invariant under U are physical.
a) How does U act on the modes x, p, nL and nR of the coordinate X ?
b) Dene the string vacuum |0; q , ri where = 0, . . . , D 2 and r is the momentum in

the folded dimension. We assume that

U |0; q , 0i = |0; q , 0i.

Give the action of U on |0; q a , ri and write down the ground states of the orbifold
theory.
c) What are the massless states of the orbifold theory?

Introduction to String Theory

Problem Set 8

ETH Zurich, HS11

B. Schwab, Prof. N. Beisert

(easy)
In this exercise we want to compute the closed string propagator
1. Two-point function

2
X (z, z)X (z , z ) = log |z z 0 |2
2

which is given by the dierence of the time-ordered and the normal ordered product of the
operators X (z, z) and X (z 0 , z0 ). Assume |z| > |z 0 |. Hint : You may ignore the eects of
the centre of mass coordinates x or use :p x : = x p .
(intermediate)
Consider conformal transformations z z 0 (z). Primary elds transform as tensors under
conformal transformations
2. Conformal transformations

O (z, z) =

z 0
z

h 

z0
z

h

O(z 0 (z), z0 (
z ))

a)

How does a primary eld transform under innitesimal transformations z 0 z +(z)?

b)

Show that the operator :eikX : for a single scalar eld X is primary by computing its
OPE with the stress-energy tensor. Determine the conformal weights h and h .

(intermediate)
The propagator that you calculated above and the two-point function
3. The complex logarithm


(z 0 , z0 ) = 2 2 (z z 0 , z z0 )
X (z, z)X

are related by the two derivatives. Show that

log |z|2 = 2 2 (z, z)
a)

. . . by considering the divergence theorem.

b)

. . . by regulating the singularity at z = 0.

(intermediate)
In QFT, fermions are represented by Grassmann-valued elds. In contradistinction to
ordinary numbers, Grassmann numbers 1 , 2 anticommute with each other

4. Free fermions

1 2 = 2 1 .

For instance, the Pauli exclusion principle is realised by the anticommuting Grassmann
numbers. We introduce a free real fermion eld in two dimensions with the action
1
S=
2

+
.

d2 z

The eld has the OPE

(z)(w) = (w)(z) =

1
+ ...
zw

and similar for . The strange looking minus sign is due to the anti-commutative nature
of Grassmann numbers. The stress-energy tensor for the free fermion reads
1
T (z) = ::.
2
a)

Show that is a primary eld with weight ( 12 , 0).

b)

Compute the OPE of two T 's and give the central charge of the theory.

Introduction to String Theory

Problem Set 9

ETH Zurich, HS11

B. Schwab, Prof. N. Beisert

1. Veneziano Amplitude (intermediate

The Veneziano amplitude1 led to the discovery

 hard)
of string theory. In this problem we will
attempt to calculate this amplitude. The open string tachyon vertex operator is given by
an integral over the boundary of the string
Z

V (k) = gs dx :exp(ik X (x)):

such that the four tachyon scattering amplitude A4 (k1 , k2 , k3 , k4 ) is given by the expression

A4 (k1 , k2 , k3 , k4 )
V1 . . . V4 gs
gs

4
Y


dxi :eik1 X(x1 ) : . . . :eik4 X(x4 ) : .

xi <xi+1 i=1

The ordering of insertions xi is due to ChanPaton factors.

a) The expectation value is computed using Wick's theorem and the two-point correlator


X (x)X (y) = 22 log |x y|.
A generic Wick contraction will have nij correlators between the pair of points (xi , xj ).
Count the combinatorial number of contractions thatPlead to a conguration specied
by the numbers nij . Hint: You need to split ni = j nij elds at point xi into nij
contractions to point xj .
b) Collect the combinatorial factors and expansion coecients of eiki X(xi ) . Then perform

the sum over nij 's. Hint: You should obtain the following integral expression for A4
(up to the momentum-conserving delta function which is more subtle)
!Z 4
X
Y
Y
2
A4 gs 26
ki
dxi
|xj xl |2 kj kl .
i

i=1

j<l

c) Show that the integral is invariant under the SL(2, R) Mbius transformation

xi
for on-shell momenta ki2 = 2 .

Hint:

axi + b
cxi + d

Use momentum conservation

ki = 0.

d) The integral given above is divergent because it has the non-compact Mbius group

as a symmetry. It thus contains an irrelevant factor of the group volume which is

innite. We divide by the latter and use the symmetry to set x1 = 0, x2 = x, x3 = 1
and x4 . Explain why the amplitude after the transformation reduces to
!Z
X
2
2
A4 gs 26
ki
dx |x|2 k1 k2 |1 x|2 k2 k3 + (k2 k3 ).
i

What is the integration range of x2 now? Why? What happened to the normalisation
in front of the integral?

1 For

the original paper see http://dx.doi.org/10.1007/BF02824451

e) The resulting integral is well known. It is in the form of the Euler beta function

B(a, b) =

dy y a1 (1 y)b1 =

Write down the solution for the amplitude.

s = (k1 + k2 )2 ,

Hint:

t = (k1 + k3 )2 ,

(a) (b)
.
(a + b)

Use the Mandelstam variables

u = (k1 + k4 )2

to simplify the result.

f ) Where does this amplitude have poles? What do these poles correspond to?

(intermediate)
The stress-energy tensor transforms under a nite conformal transformation z z 0 = f (z)
as
c
T (z) T 0 (z) = (f )2 T (z 0 ) + S(z 0 , z0 )
12
where
f (z) 3 f (z) 23 ( 2 f (z))2
0 0
S(z , z ) =
(f )2
2. Schwarzian derivative

is the Schwarzian derivative.

a) Show that the Schwarzian derivative reproduces the correct innitesimal transforma-

tion.

b) Show that the Schwarzian derivative has the correct property under successive con-

formal transformations.

c) Prove that for a transformation

f (z) =

az + b
cz + d

the Schwarzian derivative yields S(z, z) = 0. Why is this not surprising?

Introduction to String Theory

ETH Zurich, HS11

Problem Set 10
B. Schwab, Prof. N. Beisert

1. Regge behaviour and hard scattering limit (intermediate  hard)

We want to have a look at the features of the Veneziano amplitude

!
X


ki I(s, t) + I(t, u) + I(u, s)
A4 gs 26
i

where

I(s, t) =

(1 s2 ) (1 t2 )
(2 s2 t2 )

obtained on the last problem sheet by taking two interesting limits.

a) The rst limit we want to take is s with t xed. This is the so called

Regge

Why does this correspond to high energy and small angle scattering? Show
that in this limit I(s, t) reduces to
limit.

I(s, t) s1+ t (1 2 t).

Hint:

Use Stirling's approximation of the function (1 + x) xx ex 2x.

b) The second limit is the hard scattering limit s with t/s xed. Argue that

this corresponds to high energy and xed angle scattering. Show that the amplitude
reduces to


A4 exp s log(2 s) t log(2 t) u log(2 u) .

It might be easier to look at the integral expression of the amplitude and

attempt a saddle point approximation, but the limit can also be taken straightforwardly.
Hint:

2. Low-energy eective action (easy  intermediate)

In the string frame the low-energy eective action is given by

Z


p
1
1
S = 2 d26 X det G(X) e2 R 12
H H + 4 .
2

Here G is the metric, R the associated Ricci scalar, H = 3[ B] is the Kalb-Ramond

eld strength and is a scalar, the dilaton eld.
a) Show that the equations of motion of these elds are equivalent to the vanishing of

the functions

(G) =

2 R + 22

2
H H ,
4

2
H + 2 H ,
2
2
2
() = 2 + 2 H H .
2
24

(B) =

b) The kinetic energy term of the dilaton in the action seems to have the wrong sign.

Explain why this is not so.

Introduction to String Theory

Problem Set 11

ETH Zurich, HS11

B. Schwab, Prof. N. Beisert

1. Linear dilaton (intermediate)

The general worldsheet action for massless background elds is given by
Z


p
1

2
S=

det
g
(g
G
+

B
)
X

X
+

R
.
d

42
To consider a concrete example of a background for string theory we want to have a look
at the linear dilaton background where

G = ,

B = 0

and

= V X

with V a constant vector and R the worldsheet Ricci scalar.

a) Show that the functions dened on the last problem sheet vanish for V V =

(26 D)/62 .

b) Derive the worldsheet energy-momentum tensor

T (z) =

1
:X X : + V 2 X
2

of this theory and show that the central charge is given by

c = D + 62 V V .

2. (D)BI action (intermediate  hard)

In an attempt to solve the problem of the innite classical self-energy of a charged point
particle Born and Infeld proposed a non-linear generalisation of Maxwell's theory
Z
q
4
SBI d x det( + kF )
with k a constant. A further generalisation of this action appears in open string theory as
part of the world-volume action of a Dp-brane in the form of the Dirac-Born-Infeld action
Z
q
SDBI = TDp dp+1 det( X X + kF ).
In both cases, F is the Maxwell eld strength. For simplicity we'll have a look at the
BI-action.
a) Show that the Born-Infeld Lagrangian can be rewritten

LBI

q
1

2
2
= det( kF ) = exp tr log( k (F ) )
4

where (F 2 ) = F F .
b) Determine the equations of motion of the gauge eld from the BI action and expand

them in k to derive the leading order correction to the vacuum Maxwell eld equations.

c) (advanced)

Expand the DBI action to fourth order in k and show that the quadratic
term gives Maxwell's action. Repeat b) for the DBI action.
1

Introduction to String Theory

Problem Set 12

ETH Zurich, HS11

B. Schwab, Prof. N. Beisert

1. World-sheet supersymmetry

(intermediate)

An action with global worldsheet supersymmetry is given by

1
S=
42
The Grassmann-valued elds

-matrices

d2 ( X X + ).

are two-dimensional Majorana spinors. The

are

22

satisfying

{ , } = 2
= i 0 with


0 1
0
=
,
1 0

and the Dirac conjugate is

a)



0 1
=
.
1 0
1

Show that this action is invariant under

X =  ,
Why are the components of

b)

representation

Evaluate the commutators

and

[1 , 2 ]X

N =1

supersymmetry

= X .
Weyl spinors?
and

[1 , 2 ]

to show that the commutator of

two supersymmetry transformations amount to a translation along the world-sheet.

c)

Derive the Noether current (supercurrent) of supersymmetry transformations. Meditate on the relation between the supercurrent and the energy-momentum tensor.

2. Grande Finale: The super-Particle

A massless supersymmetric particle in

(intermediate)

D-dimensional

Minkowski spacetime is described

by the action

X X iX
+
i
2e
e

Z
S` =
where

a)

is the einbein and

its fermionic partner.

are 2D Majorana fermions.

X ( ), ( ) as well as e and . Find the correct

for e and by requiring that the action is invariant

Derive the equations of motion for

supersymmetry transformations

under (local!) supersymmetry transformations with

X = i
b)

Argue that you can set

e=1

and
and

= 0.

1
(X i ).
2e

Write down the gauged action. Important:

What are the constraint equations that follow from this procedure? Interpret them!

c)

Consider the (global) supersymmetry transformations (derive them from the ones
given in a)) and check that the commutator of two supersymmetry transformations
will result in a

translation by an amount