27.07.2014 Views

Lectures on String Theory

Lectures on String Theory

Lectures on String Theory

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Preprint typeset in JHEP style - HYPER VERSION<br />

<str<strong>on</strong>g>Lectures</str<strong>on</strong>g> <strong>on</strong> <strong>String</strong> <strong>Theory</strong><br />

Gleb Arutyunov a<br />

a Institute for Theoretical Physics and Spinoza Institute, Utrecht University<br />

3508 TD Utrecht, The Netherlands<br />

Abstract: The course covers the basic c<strong>on</strong>cepts of modern string theory. This<br />

includes covariant and light-c<strong>on</strong>e quantisati<strong>on</strong> of bos<strong>on</strong>ic and fermi<strong>on</strong>ic strings,<br />

geometry and topology of string world-sheets, vertex operators and string scattering<br />

amplitudes, world-sheet and space-time supersymmetries, elements of c<strong>on</strong>formal<br />

field theory, Green-Schwarz superstrings, strings in curved backgrounds, low-energy<br />

effective acti<strong>on</strong>s, D-brane physics.


– 1 –<br />

C<strong>on</strong>tents<br />

1. Introducti<strong>on</strong> to the course 2<br />

1.1 Historical remarks 2<br />

2. Relativistic particle 8<br />

3. Classical relativistic bos<strong>on</strong>ic string 13<br />

3.1 Nambu-Goto string 13<br />

3.2 The Polyakov acti<strong>on</strong> 16<br />

3.2.1 Symmetries 16<br />

3.2.2 Equati<strong>on</strong>s of moti<strong>on</strong> 16<br />

3.2.3 C<strong>on</strong>formal gauge 18<br />

3.2.4 Polyakov string in the first order formalism. 19<br />

3.3 Integrals of moti<strong>on</strong>. Classical Virasoro algebra. 21<br />

3.3.1 Soluti<strong>on</strong>s of the equati<strong>on</strong>s of moti<strong>on</strong> 24<br />

3.3.2 Poincaré symmetry. <strong>String</strong> tensi<strong>on</strong>. 25<br />

3.4 <strong>String</strong>s in physical gauge 28<br />

3.4.1 First order formalism 28<br />

3.4.2 Poiss<strong>on</strong> structure of the light-c<strong>on</strong>e theory 35<br />

3.4.3 Lorentz symmetry 37<br />

4. Quantizati<strong>on</strong> of bos<strong>on</strong>ic string 38<br />

4.1 Remarks <strong>on</strong> can<strong>on</strong>ical quantizati<strong>on</strong> 38<br />

4.1.1 Virasoro algebra 41<br />

4.1.2 Virasoro c<strong>on</strong>straints in quantum theory 43<br />

4.1.3 The spectrum 45<br />

4.1.4 Propagators 47<br />

4.1.5 Vertex operators. Tachy<strong>on</strong> scattering amplitude 48<br />

4.2 Quantizati<strong>on</strong> in the physical gauge 58<br />

4.2.1 Lorentz symmetry and critical dimensi<strong>on</strong> 59<br />

4.2.2 The spectrum 63<br />

4.3 BRST quantizati<strong>on</strong> 66<br />

5. Geometry and topology of string world-sheet 77<br />

5.1 From Lorentzian to Euclidean world-sheets 77<br />

5.2 Riemann surfaces 79<br />

5.3 Moduli space 84


– 2 –<br />

6. Classical fermi<strong>on</strong>ic superstring 92<br />

6.1 Spinors in General Relativity 92<br />

6.2 Superstring acti<strong>on</strong> and its symmetrices 99<br />

6.3 Superc<strong>on</strong>formal gauge and supermoduli 100<br />

6.4 Acti<strong>on</strong> in the superc<strong>on</strong>formal gauge 102<br />

6.5 Boundary c<strong>on</strong>diti<strong>on</strong>s 106<br />

6.6 Superc<strong>on</strong>formal algebra 107<br />

7. Quantum fermi<strong>on</strong>ic string 108<br />

7.1 Light-c<strong>on</strong>e quantizati<strong>on</strong> and superstring spectrum 111<br />

A. Dynamical systems of classical mechanics 119<br />

B. OPE and c<strong>on</strong>formal blocks 123<br />

C. Useful formulae 127<br />

D. Riemann normal coordinates 128<br />

E. Exercises 130<br />

1. Introducti<strong>on</strong> to the course<br />

1.1 Historical remarks<br />

<strong>String</strong> theory arose at the end of sixties in an attempt to describe the theory of<br />

str<strong>on</strong>g interacti<strong>on</strong>s. In 1969 Veneziano found a beautiful formula for the scattering<br />

amplitude of four particles. This amplitude comprised many features that physicists<br />

expected to be found in the theory of str<strong>on</strong>g interacti<strong>on</strong>s. It was realized very so<strong>on</strong> by<br />

Nambu and Susskind that the underlying dynamical object from which the Veneziano<br />

formula can be derived is a relativistic string. The fundamental difference of strings<br />

from the theory of point particles is that strings are extended, <strong>on</strong>e-dimensi<strong>on</strong>al,<br />

objects; when string moves through space and time it sweeps two-dimensi<strong>on</strong>al surface<br />

called the string “world-sheet”. The strings can be of two types: open – with topology<br />

of an interval and closed – with topology of a circle.<br />

Subsequent investigati<strong>on</strong>s revealed however severe difficulties to treat the string<br />

as the theory of str<strong>on</strong>g interacti<strong>on</strong>s. These difficulties are<br />

1. Existence of a “critical dimensi<strong>on</strong>”.<br />

2. Existence of a massless spin two particle which is absent in the hadr<strong>on</strong>ic world.


– 3 –<br />

If <strong>on</strong>e tries to c<strong>on</strong>struct the quantum mechanics of relativistic strings <strong>on</strong>e finds<br />

that mathematically c<strong>on</strong>sistent theory exists if and <strong>on</strong>ly if the dimensi<strong>on</strong> of spacetime<br />

where string propagates is 26. The number 26 was named the “critical dimensi<strong>on</strong>”.<br />

On the <strong>on</strong>e hand, it was pretty remarkable and unexpected to find an example<br />

of physical theory which puts restricti<strong>on</strong>s <strong>on</strong> space-time where it is defined. On the<br />

other hand, it was certainly not clear why a theory that shared at least some qualitative<br />

features with hadr<strong>on</strong>ic physics should exist in 26 dimensi<strong>on</strong>s <strong>on</strong>ly. A subsequent<br />

discovery of QCD (Quantum Chromo Dynamics) as the most appropriate candidate<br />

to describe the theory of str<strong>on</strong>g interacti<strong>on</strong>s led to a c<strong>on</strong>siderable loss of interest to<br />

string theory.<br />

In 1974, Scherk and Schwarz came up with a proposal to completely alter the<br />

view <strong>on</strong> string theory. They suggested to c<strong>on</strong>sider the massless spin two particle<br />

absent in the hadr<strong>on</strong>ic world as the gravit<strong>on</strong> – the quantum of the gravitati<strong>on</strong>al<br />

interacti<strong>on</strong>. Indeed, this particle neatly fits the properties of the gravit<strong>on</strong> – string<br />

theory predicts that this particle interacts according to the standard laws of General<br />

Relativity. Gravitati<strong>on</strong>al interacti<strong>on</strong>s have a natural scale, called the Planck mass,<br />

which is around 10 19 GeV. This is a huge number in comparis<strong>on</strong> with characteristic<br />

energies of hadr<strong>on</strong>ic physics, 100 − 200 MeV. Thus, according to their view, string<br />

theory could provide the unifying descripti<strong>on</strong> of all the particles and matter forces,<br />

including gravity and it operates <strong>on</strong> a new fundamental scale.<br />

Even if <strong>on</strong>e accepts that quantum mechanics of relativistic strings can be defined<br />

in the unusual number 26 of the space-time dimensi<strong>on</strong>s, another problem arises. Such<br />

string does not c<strong>on</strong>tain fermi<strong>on</strong>ic degrees of freedom and it predicts the existence of<br />

a particle with the negative mass squared: m 2 < 0. Such a particle, tachy<strong>on</strong>, is a<br />

source of instability and its existence indicates that either the theory is ill-defined<br />

or it is formulated around a “wr<strong>on</strong>g” ground state, or as physicists say, around a<br />

“wr<strong>on</strong>g” vacuum. Critical dimensi<strong>on</strong>, tachy<strong>on</strong> and absence of fermi<strong>on</strong>s were the<br />

puzzling features the string theory had to face.<br />

The status of string theory changed again with the discovery of supersymmetry.<br />

All universe is made of two fundamental types of particles: bos<strong>on</strong>s and fermi<strong>on</strong>s.<br />

Fermi<strong>on</strong>s c<strong>on</strong>stitute all the matter and bos<strong>on</strong>s mediate interacti<strong>on</strong>s of the matter<br />

particles. Supersymmetry is a new type of symmetry between bos<strong>on</strong>s and fermi<strong>on</strong>s<br />

(Wess and Zumino 1974). Many physicists hope today that supersymmetry could<br />

provide an underlying principle for unificati<strong>on</strong> of all interacti<strong>on</strong>s.<br />

The first success in incorporating supersymmetry in string theory was achieved<br />

in 1971 by Ram<strong>on</strong>d, who c<strong>on</strong>structed a string analogue of the Dirac equati<strong>on</strong> (the<br />

spinning string). Shortly afterwards, Neveu and Schwarz c<strong>on</strong>structed a new bos<strong>on</strong>ic<br />

string theory. They realized that the two c<strong>on</strong>structi<strong>on</strong>s were different facets of a<br />

single theory - an interacting superstring theory c<strong>on</strong>taining Neveu and Schwarz’s


– 4 –<br />

bos<strong>on</strong>s and Ram<strong>on</strong>d’s fermi<strong>on</strong>s. The supersymmetry of the two-dimensi<strong>on</strong>al string<br />

world sheet was recognized by Gervais and Sakita in 1971. This was advent of the<br />

NSR (Neveu-Ram<strong>on</strong>d-Schwarz) superstring.<br />

In 1972 Schwarz dem<strong>on</strong>strated the c<strong>on</strong>sistency of the superstring theory in 10 dimensi<strong>on</strong>s.<br />

Instead of 26 found for purely bos<strong>on</strong>ic string, the critical dimensi<strong>on</strong> for the<br />

NSR string appears to be 10. In 1977 Gliozzi, Scherk and Olive realized that further<br />

c<strong>on</strong>diti<strong>on</strong>s should be imposed <strong>on</strong> the spectrum (the GSO projecti<strong>on</strong> mechanism) of<br />

the NSR string which lead to both the so-called space-time supersymmetry (to compare<br />

with the world-sheet supersymmetry menti<strong>on</strong>ed above) and to the removal of<br />

tachy<strong>on</strong>. Thus, superstring theory has at least two advantages in comparis<strong>on</strong> with<br />

bos<strong>on</strong>ic strings: critical dimensi<strong>on</strong> 10 < 26 and the absence of tachy<strong>on</strong>. It also turned<br />

out that the GSO projecti<strong>on</strong> can be imposed in two different ways which lead to two<br />

different types of superstrings, called the Type IIA and Type IIB.<br />

<strong>String</strong> theories have a natural particle limit, when the length of string vanishes.<br />

In this limit superstrings give rise to the low-energy effective theories, known as<br />

supergravities. These theories can be defined in a way completely independent of<br />

string theory: they can be thought of as supersymmetric generalizati<strong>on</strong>s of the pure<br />

Einstein gravity. As is known, attempts to quantize gravity in the standard framework<br />

of quantum mechanics fail because gravity is a n<strong>on</strong>-renormalizable theory (there<br />

are infinitely many divergent graphs with any number of external legs and with an<br />

arbitrarily high index of divergence, cf. the course <strong>on</strong> Quantum Field <strong>Theory</strong>). Supersymmetric<br />

theories tend to be less divergent than n<strong>on</strong>-supersymmetric <strong>on</strong>es which<br />

gave initially a hope that supersymmetry could cure the n<strong>on</strong>renormalizable infinities<br />

of the quantum gravity. It seems that supergravities themselves are still not capable<br />

to solve the divergency problem 1 . Quite remarkably, there is a str<strong>on</strong>g evidence that<br />

the divergency problem of quantum (super)gravities is resolved by string theory.<br />

¢¡¢ ¡<br />

¢¡¢ ¡<br />

¢¡¢ ¡<br />

Resoluti<strong>on</strong> of the four-fermi interacti<strong>on</strong>. At high energies the weak force is<br />

mediated by a heavy bos<strong>on</strong>.<br />

1 For instance, it is unknown if the so-called N = 8 supergravity is finite or not.


– 5 –<br />

To get a better feeling why it happens it is c<strong>on</strong>venient to envoke an analogy<br />

with the theory of weak interacti<strong>on</strong>s. Trying to describe the decay of the neutr<strong>on</strong> by<br />

the Fermi-type Lagrangian c<strong>on</strong>taining the quartic, point-like, interacti<strong>on</strong> vertex <strong>on</strong>e<br />

finds irremovable ultraviolet divergencies at higher loop orders. Again, the reas<strong>on</strong><br />

is that the corresp<strong>on</strong>ding theory of Fermi-interacti<strong>on</strong>s in n<strong>on</strong>-renormalizable. The<br />

soluti<strong>on</strong> of this problem lies in a fact that at higher energies (more than 100 GeV),<br />

the pointlike vertex gets resolved and the interacti<strong>on</strong> is mediated by a heavy W-<br />

bos<strong>on</strong>. In the new theory <strong>on</strong>e has qubic vertices and this ultimately makes the<br />

theory renormalizable.<br />

Very similar phenomen<strong>on</strong> occurs in string theory. Expanding the Einstein-<br />

Hilbert acti<strong>on</strong> √ −gR <strong>on</strong>e gets more and more complicated point-like vertices which<br />

render the theory n<strong>on</strong>-renormalizable, analogously to the old Fermi theory. In opposite,<br />

in string theory all these vertices get dissolved by the exchange of the massive<br />

string states. <strong>String</strong> states form an infinite tower of particles of arbitrarily high mass<br />

and spin and all of them participate in the interacti<strong>on</strong> process. Resolving the ultraviolet<br />

divergency problem string theory appears to be a natural candidate for the<br />

theory of quantum gravity.<br />

There is a still an important questi<strong>on</strong> how to relate the superstring theory defined<br />

in a ten-dimensi<strong>on</strong>al space-time with a our c<strong>on</strong>venti<strong>on</strong>al four-dimensi<strong>on</strong>al physics.<br />

The basic approach to obtain four-dimensi<strong>on</strong>al theories is al<strong>on</strong>g the old ideas of<br />

Kaluza and Klein and it c<strong>on</strong>sists in compactifying of the ten-dimensi<strong>on</strong>al theory down<br />

to four dimensi<strong>on</strong>s. In this case, the four string coordinates remain uncompactified,<br />

while the other six are curled up and parametrize a compact space of a very small<br />

size (of the order of the Plank length). It appears that the internal space cannot be<br />

completely arbitrary – it must have vanishing Ricchi-curvature.<br />

One of the major obstacles to build a unified theory is the left-right asymmetry<br />

recorded in the present days experiments. A theory in which there is an asymmetry<br />

between the left and the right must c<strong>on</strong>tain chiral fermi<strong>on</strong>s. Chiral fermi<strong>on</strong>s are<br />

usually a source of anomalies, i.e., of breakdown of classical symmetries by quantum<br />

effects. Anomalies render a theory eventually inc<strong>on</strong>sistent. Only in some special,<br />

“rear” cases anomalies cancel (as it happens for instance for a standard generati<strong>on</strong><br />

of quarks and lept<strong>on</strong>s, cf. the course <strong>on</strong> the Standard Model). In higher space-time<br />

dimensi<strong>on</strong>s it becomes even more n<strong>on</strong>-trivial to achieve cancellati<strong>on</strong> of anomalies.


– 6 –<br />

?<br />

Type I<br />

Type IIB<br />

E 8 x E 8<br />

heterotic<br />

Type IIA<br />

SO(32) heterotic<br />

<strong>String</strong> theories<br />

Remarkably, in 1983, Alvarez-Gaumé and Witten dem<strong>on</strong>strated that anomaly cancelled<br />

out in the chiral Type IIB supergravity theory. This was nice, but the corresp<strong>on</strong>ding<br />

theory was still far from phenomenology. Shortly afterwards Green and<br />

Schwarz (1984) showed that cancellati<strong>on</strong> of anomalies in the open superstring theories<br />

singles out two gauge groups, namely SO(32) and E 8 × E 8 . The gauge groups<br />

SO(32) and E 8 × E 8 are realized in the so-called “heterotic string” which is a hybrid<br />

of the old d = 26 dimensi<strong>on</strong>al bos<strong>on</strong>ic string and the d = 10 superstring. It<br />

was shown that compactifying the theory down to four dimensi<strong>on</strong>s <strong>on</strong>e can obtain<br />

several generati<strong>on</strong>s of the chiral fermi<strong>on</strong>s. It was a first example of a string theory<br />

whose low-energy limit was not in an immediate c<strong>on</strong>flict with all known physics.<br />

One of the important questi<strong>on</strong>s is why there exists n<strong>on</strong> a unique but several<br />

c<strong>on</strong>sistent string theories. With a recent advent of the string dualities and D-branes<br />

an interesting new picture start to emerge (but still very far from being complete),<br />

according to which different superstring theories provide different descripti<strong>on</strong>s of<br />

the <strong>on</strong>e and the same theory valid in different regimes of the coupling c<strong>on</strong>stant<br />

parameters.<br />

Recommended literature:<br />

1. M. Green, J. Schwarz and E. Witten, “Superstring <strong>Theory</strong>”, volume I and II,<br />

Cambridge University Press, 1987.<br />

2. D. Lüst and S. Theisen, “<str<strong>on</strong>g>Lectures</str<strong>on</strong>g> <strong>on</strong> string theory”, Lecture Notes in Physics,<br />

Springer Verlag, 1989.<br />

3. ’t Hooft, “Introducti<strong>on</strong> to <strong>String</strong> <strong>Theory</strong>”, Lecture Notes, 2005.


– 7 –<br />

4. B. Zwiebach, “A first course in string theory”, Cambridge university press,<br />

2004.<br />

5. J. Polchinski, “<strong>String</strong> theory”, volume I and II, Cambridge university press,<br />

1998.


– 8 –<br />

2. Relativistic particle<br />

C<strong>on</strong>sider relativistic particle of mass m moving in d-dimensi<strong>on</strong>al Minkowski space:<br />

η µν = (−1, +1, +1, . . . , +1).<br />

Acti<strong>on</strong><br />

∫ s<br />

S = −α ds<br />

s 0<br />

Note that ∫ s<br />

s 0<br />

ds has maximum al<strong>on</strong>g straight lines, this explains the sign “-” in fr<strong>on</strong>t<br />

of the acti<strong>on</strong>.<br />

Embedding x µ ≡ x µ (τ):<br />

If x µ = (cτ, ⃗x) then<br />

Thus, the acti<strong>on</strong> is<br />

ds =<br />

√<br />

− dxµ<br />

dτ<br />

ds = √ c 2 − ⃗v 2 ,<br />

√<br />

dx µ<br />

dτ dτ ≡ dx<br />

−η µ dx ν<br />

µν<br />

dτ dτ dτ<br />

⃗v = d⃗r<br />

dτ<br />

∫ √<br />

τ1<br />

S = −αc 1 − ⃗v2<br />

τ 0<br />

c dτ 2<br />

The Lagrangian in the n<strong>on</strong>-relativistic limit<br />

√<br />

( )<br />

L = −αc 1 − ⃗v2<br />

c dτ = −cα 1 − ⃗v2 + . . . = −αc + α⃗v2<br />

2 2c 2 2c + . . .<br />

To get the standard kinetic energy <strong>on</strong>e has to identify<br />

α = mc<br />

In what follows we will work in units in which c = 1.<br />

The acti<strong>on</strong> is invariant under reparametrizati<strong>on</strong>s of τ:<br />

Let us show this<br />

δx µ = ξ(τ)∂ τ x µ as l<strong>on</strong>g as ξ(τ 0 ) = ξ(τ 1 ) = 0<br />

δ( √ 1<br />

−ẋ µ ẋ µ ) =<br />

2 √ 1<br />

−ẋ µ ẋ µ (−2ẋν δẋ ν ) = −√ −ẋµ ẋ µ ẋν ∂ τ (ξẋ ν ) =<br />

1<br />

[<br />

]<br />

= −√ ẋ ν 1<br />

ẋ −ẋµ ẋ µ ν ˙ξ + ξẋνẍ ν = −√ −ẋµ ẋ µ ẋν ẋ ν ˙ξ + ξ∂τ ( √ −ẋ µ ẋ µ )<br />

= √ −ẋ µ ẋ µ ˙ξ + ξ∂ τ ( √ −ẋ µ ẋ µ ) = ∂ τ (ξ √ −ẋ µ ẋ µ )


– 9 –<br />

Therefore, we arrive at<br />

δS = −m<br />

∫ τ1<br />

τ 0<br />

∂ τ (ξ √ −ẋ µ ẋ µ ) = −m<br />

[ξ √ ]<br />

−ẋ µ ẋ µ | τ=τ 1<br />

τ=τ 0<br />

= 0 ,<br />

i.e., the acti<strong>on</strong> is indeed invariant w.r.t. the local reparametrizati<strong>on</strong> transformati<strong>on</strong>s.<br />

The most elegant way to quantize a system is to use the Hamilt<strong>on</strong>ian formalism<br />

2 . Let us therefore try to develop the Hamilt<strong>on</strong>ian (can<strong>on</strong>ical) formalism for the<br />

relativistic particle. The can<strong>on</strong>ical momenta<br />

We notice that<br />

p µ = ∂L<br />

∂ẋ µ = −m ∂<br />

∂ẋ µ (√ −ẋ µ ẋ µ ) = m ẋµ √ −ẋµ ẋ µ<br />

p 2 ≡ p µ p µ = −m 2<br />

Thus, we see that the can<strong>on</strong>ical momenta are not independent, rather they obey the<br />

c<strong>on</strong>straint<br />

φ ≡ p 2 + m 2 = 0 (2.1)<br />

C<strong>on</strong>straints which follow just from the definiti<strong>on</strong> of the can<strong>on</strong>ical momenta without<br />

using equati<strong>on</strong>s of moti<strong>on</strong> are called primary c<strong>on</strong>straints. Mass-shell c<strong>on</strong>diti<strong>on</strong> for<br />

the point particle is a primary c<strong>on</strong>straint.<br />

In general the number of primary c<strong>on</strong>straints is equal to the number of zero<br />

eigenvalues of the Hessian matrix:<br />

H µν = ∂p µ<br />

∂ẋ ν =<br />

∂2 L<br />

∂ẋ µ ∂ẋ ν<br />

For the relativistic particle we have <strong>on</strong>ly <strong>on</strong>e eigenvector ẋ µ with zero eigenvalue<br />

(<br />

)<br />

∂p µ m<br />

∂ẋ ν ẋν = √ −ẋµ ẋ δ µ µν + m<br />

ẋµ ẋ ν<br />

ẋ ν = 0<br />

(−ẋ µ ẋ µ ) 3 2<br />

The inverse functi<strong>on</strong> theorem stats that absence of zero eigenvalues of the Hessian<br />

is a necessary c<strong>on</strong>diti<strong>on</strong> to be able to express the velocities ẋ µ via the can<strong>on</strong>ical<br />

momenta p µ . Dynamical systems with the Hessian of n<strong>on</strong>-maximal rank are called<br />

singular.<br />

C<strong>on</strong>straints which have vanishing Poiss<strong>on</strong> bracket:<br />

{φ i , φ j } = 0<br />

2 About the Hamilt<strong>on</strong>ian approach to dynamical systems of classical mechanics, the reader may<br />

c<strong>on</strong>sult appendix A.


– 10 –<br />

are called the first class c<strong>on</strong>straints. The mass-shell c<strong>on</strong>straint for the relativistic<br />

particle is of the first class.<br />

There is another acti<strong>on</strong> for the relativistic particle.<br />

characteristic features<br />

It has the following two<br />

• it does not c<strong>on</strong>tain square root<br />

• it admits generalizati<strong>on</strong> to the massless case<br />

Introduce e(τ) the auxiliary field <strong>on</strong> the world-line.<br />

S = 1 2<br />

∫ τ1<br />

τ 0<br />

dτ[ 1<br />

e<br />

( dx<br />

µ<br />

dτ<br />

dx<br />

) ]<br />

µ<br />

− em 2<br />

dτ<br />

Equati<strong>on</strong>s of moti<strong>on</strong>:<br />

for x µ d<br />

( 1<br />

)<br />

eẋµ = 0<br />

dτ<br />

for e(τ) − 1<br />

2e 2 ẋµ ẋ µ − 1 2 m2 = 0 =⇒ ẋ 2 + m 2 e 2 = 0<br />

The last equati<strong>on</strong> can be solved for e:<br />

which leads to<br />

d<br />

dτ<br />

e 2 = − 1 m 2 ẋ2<br />

[<br />

m<br />

]<br />

√ −ẋν ẋ ν ẋµ = 0<br />

which is nothing else as the old eom for x µ . Also if we substitute soluti<strong>on</strong> for e into<br />

the new acti<strong>on</strong> then this acti<strong>on</strong> reduces to the old <strong>on</strong>e:<br />

S = 1 ∫ τ1 [<br />

√<br />

m −ẋ<br />

2 ] ∫ τ1<br />

dτ √<br />

2 −ẋ<br />

2 ẋ2 −<br />

m m2 = −m dτ √ −ẋ 2 (2.2)<br />

τ 0<br />

τ 0<br />

Also<br />

p µ = 1 eẋµ =⇒ p 2 = 1 e 2 ẋ2 = −m 2<br />

but this time due to equati<strong>on</strong>s of moti<strong>on</strong> for e. Equati<strong>on</strong> of moti<strong>on</strong> for e is purely<br />

algebraic. The Hessian<br />

∂ 2 L<br />

∂ẋ µ ∂ẋ = 1 ∂<br />

ν e ∂ẋ µ ẋν = 1 e η µν<br />

is of maximal rank.<br />

The c<strong>on</strong>straint p 2 + m 2 = 0 does not follow from the definiti<strong>on</strong> of the can<strong>on</strong>ical<br />

momentum al<strong>on</strong>g, but <strong>on</strong>e has to use equati<strong>on</strong>s of moti<strong>on</strong>. C<strong>on</strong>straints which are<br />

satisfied as c<strong>on</strong>sequences of equati<strong>on</strong>s of moti<strong>on</strong> are called sec<strong>on</strong>dary.


– 11 –<br />

The acti<strong>on</strong> has local gauge symmetry which is now<br />

δx µ = ξẋ µ<br />

δe = ∂ τ (ξe)<br />

The first symmetry transformati<strong>on</strong> is generated by the c<strong>on</strong>straint p 2 + m 2 :<br />

δ η x µ = η{p 2 + m 2 , x µ } = 2ηp µ = η e ẋµ = ξẋ µ , η ≡ eξ .<br />

The sec<strong>on</strong>d transformati<strong>on</strong> for e is easily derived from e 2 = − 1 ẋ 2 . Thus, in our<br />

m 2<br />

new formulati<strong>on</strong> we have the set of fields (x µ , e) and reparametrizati<strong>on</strong> symmetry<br />

which acts <strong>on</strong> them and leaves the acti<strong>on</strong> invariant. This reparametrizati<strong>on</strong> freedom<br />

can be used to put e = 1 which results into the following equati<strong>on</strong><br />

m<br />

ẍ µ = 0<br />

This is not the end however, because there is eom for e which now reads as ẋ 2 = −1.<br />

Complete eoms are<br />

ẍ µ = 0 , ẋ 2 = −1<br />

Thus, the relativistic particle moves freely in Minkowski space over time-like geodesics.<br />

Space-like and light-like straight lines are excluded by the c<strong>on</strong>straint ẋ 2 = −1.<br />

In the case of the massless particle we can set e = 1 and get eoms<br />

ẍ µ = 0 , ẋ 2 = 0<br />

In both, the massive and massless cases, the c<strong>on</strong>straints are integral of moti<strong>on</strong>s: they<br />

are preserved in time due to the dynamical equati<strong>on</strong> ẍ µ = 0.<br />

Finally, we treat the relativistic particle in the so-called first order (the Hamilt<strong>on</strong>ian)<br />

formalism. To this end we have to represent the initial Lagrangian in the<br />

form<br />

L = p µ ẋ µ + L rest<br />

where p µ = 1 eẋµ and express in L rest the derivatives ẋ µ via p µ . In doing so we obtain<br />

the phase-space Lagrangian<br />

L = p µ ẋ µ − e 2 (p2 + m 2 )<br />

We clearly see that the auxiliary field e we introduced in our sec<strong>on</strong>d formulati<strong>on</strong><br />

plays here the role of the lagrangian multiplier to the c<strong>on</strong>straint p 2 + m 2 = 0. By<br />

using the gauge freedom we can fix the gauge e = 1 and the physical Hamilt<strong>on</strong>ian<br />

m<br />

becomes in this case<br />

H = 1<br />

2m (p2 + m 2 )


– 12 –<br />

which is in complete agreement with our previous discussi<strong>on</strong> (We actually have to<br />

identify θ = e). This Hamilt<strong>on</strong>ian 3 should be provided with the c<strong>on</strong>straint p 2 +m 2 = 0<br />

which is the eom for e.<br />

More generally, evoluti<strong>on</strong> of a singular system is governed by the Hamilt<strong>on</strong>ian<br />

H = H can + ∑ n<br />

χ n φ n<br />

Here {φ n } is an irreducible set of primary c<strong>on</strong>straints and H can is the can<strong>on</strong>ical<br />

Hamilt<strong>on</strong>ian:<br />

H can = p µ ẋ µ − L<br />

Only <strong>on</strong> the c<strong>on</strong>straint surface φ n = 0 the Hamilt<strong>on</strong>ian H coincides with H can . In<br />

our present case<br />

H can = m ẋνẋ ν<br />

√ −ẋµ ẋ − (−m√ −ẋ µ µ ẋ µ ) = 0<br />

and Hamilt<strong>on</strong>ian dynamics of the system is due to the mass-shell c<strong>on</strong>diti<strong>on</strong> <strong>on</strong>ly. The<br />

choice of the coefficients χ n (τ) in H is equivalent to the choice of the gauge.<br />

H = H can + χφ =<br />

We get the time-evoluti<strong>on</strong><br />

θ<br />

2m (p2 + m 2 ) , χ = θ<br />

2m<br />

dx µ<br />

dτ = { θ<br />

2m (p2 + m 2 ), x µ } = θ m pµ =<br />

θẋ µ<br />

√ −ẋµ ẋ µ<br />

Therefore ẋ 2 = −θ 2 . Choosing θ = 1 means that we identify the time variable with<br />

the proper time. This nicely illustrates the general point: in order to write down the<br />

evoluti<strong>on</strong> equati<strong>on</strong>s in a system with local gauge invariance <strong>on</strong>e has first to identify<br />

the “time” variable.<br />

Other gauge choices are possible. For instance, the static gauge c<strong>on</strong>sists in<br />

imposing the c<strong>on</strong>diti<strong>on</strong> t ≡ x 1 = τ. Equati<strong>on</strong> for p t ≡ p 1 allows us to determine e:<br />

δL<br />

= dt<br />

δp t dτ − ep t = 0 =⇒ e = 1 p t<br />

The physical Hamilt<strong>on</strong>ian dual to the world-line time τ coincides in this case with<br />

the momentum p t c<strong>on</strong>jugate to t: H = p t . It can be found from the eom for e:<br />

p 2 = m 2 = 0 =⇒ − p 2 t + ⃗p 2 + m 2 = 0 =⇒ p t = √ ⃗p 2 + m 2 .<br />

Note that here ⃗p = {p i } with i = 2, . . . , d.<br />

case.<br />

3 The gauge e = 1 m<br />

is a close analogue of the c<strong>on</strong>formal gauge to be c<strong>on</strong>sidered for the string


– 13 –<br />

This gauge choice leads to the comm<strong>on</strong> Hamilt<strong>on</strong>ian of the relativistic particle<br />

H = √ ⃗p 2 + m 2 .<br />

This exercise with the relativistic particle also shows how sensitive is the Hamilt<strong>on</strong>ian<br />

to the choice of the gauge. Fixing the gauge e = 1 we get the polynomial Hamilt<strong>on</strong>ian,<br />

while fixing the static gauge the Hamilt<strong>on</strong>ian appears to be a n<strong>on</strong>-linear square root.<br />

Finally, there is another type of gauge known as the light-c<strong>on</strong>e gauge. We introduce<br />

the light-c<strong>on</strong>e coordinates<br />

t = x + − x −<br />

x d = x + + x −<br />

p t = 1 2 (p+ + p − ), p d = 1 2 (p+ − p − )<br />

and denote the other “transverse coordinates” x i and p i with i = 2, . . . , d − 1 as ⃗x<br />

and ⃗p. Then the phase space Lagrangian becomes<br />

L = ẋ − p + − ẋ + p − + ˙⃗x⃗p − e 2 (−p− p + + ⃗p 2 + m 2 )<br />

The light-c<strong>on</strong>e gauge c<strong>on</strong>sists in choosing x + = τ. From the kinetic term of the<br />

Hamilt<strong>on</strong>ian it is clear that the variable x + is c<strong>on</strong>jugate to p − and therefore p − is<br />

the physical Hamilt<strong>on</strong>ian. It can be easily found from the equati<strong>on</strong> for e:<br />

The gauge-fixed Lagrangian becomes<br />

H = 1<br />

p + (⃗p2 + m 2 )<br />

L = ẋ − p + + ˙⃗x⃗p − p − = ˙⃗x⃗p − H = ˙⃗x⃗p − 1<br />

p + (⃗p2 + m 2 )<br />

The variable p + is can<strong>on</strong>ically c<strong>on</strong>jugate to x − .<br />

Notice that both in the static and in the light-c<strong>on</strong>e gauge the number of physical<br />

degrees of freedom is 2(d − 1). The auxiliary field e was solved in terms of physical<br />

fields. The physical phase space inherits the can<strong>on</strong>ical Poiss<strong>on</strong> bracket.<br />

In the next lecture we extend these different approaches to dynamics of relativistic<br />

particles to relativistic strings.<br />

3. Classical relativistic bos<strong>on</strong>ic string<br />

3.1 Nambu-Goto string<br />

Two-dimensi<strong>on</strong>al surface traced by string during its time evoluti<strong>on</strong> is called worldsheet.<br />

The acti<strong>on</strong> for a relativistic string should be a functi<strong>on</strong>al of a string trajectory,<br />

i.e. of the world-sheet.


– 14 –<br />

The Nambu-Goto acti<strong>on</strong><br />

∫ ∫<br />

S NG = −T dA = −T<br />

d 2 σ<br />

√<br />

−det<br />

( )<br />

∂X<br />

µ<br />

∂X ν<br />

∂σ α ∂σ η β µν<br />

(3.1)<br />

This we can write as<br />

[( )]<br />

−det<br />

Ẋµ Ẋ µ Ẋ µ X µ<br />

′<br />

Ẋ µ Ẋ µ X ′µ X µ<br />

′ = (ẊX′ ) 2 − Ẋ2 X ′2 . (3.2)<br />

Thus,<br />

∫<br />

S NG = −T<br />

√<br />

∫<br />

d 2 σ (ẊX′ ) 2 − Ẋ2 X ′2 = −T<br />

d 2 σ √ −Γ<br />

Here Γ = detΓ αβ , where<br />

Γ αβ = ∂Xµ<br />

∂σ α ∂X ν<br />

∂σ β η µν (3.3)<br />

is the metric induced <strong>on</strong> the string world-sheet.<br />

What is a local characteristic of the string world-sheet? C<strong>on</strong>sider a point <strong>on</strong> the<br />

world-sheet and the space of all vectors tangent to the surface is at this point. These<br />

vectors sweep two-dim vector space. The physical propagati<strong>on</strong> of the string requires<br />

that in these two-dim vector space there is a basis built over two vectors <strong>on</strong>e of them<br />

is time-like and another is space-like.<br />

Recall the standard definiti<strong>on</strong>s from the theory of special relativity. We have the<br />

invariant interval between two infinitezimal events:<br />

−ds 2 = η µν dx µ dx ν = −dx 0 dx 0 +<br />

3∑<br />

dx i dx i . (3.4)<br />

i=1<br />

• If ds 2 > 0 the interval is called time-like. In this case the different events which<br />

happen in the same space-point are always time-separated.<br />

• If ds 2 < 0 the interval is called space-like. In this case events which happen at<br />

the same time are space-separated.<br />

• If ds 2 = 0 the interval is light-like. Vectors v µ obeying the c<strong>on</strong>diti<strong>on</strong> v 2 =<br />

η µν v µ v ν = 0 are called light-like or null.<br />

Identify x 0 = t = τ. Then<br />

Ẋ µ Ẋ µ = −1 + ⃗v 2 ≤ 0 ⇐ time − like<br />

X ′µ X ′ µ = (X ′i ) 2 ≥ 0 ⇐ space − like


– 15 –<br />

This situati<strong>on</strong> should persist in any Lorentz frame. At any point <strong>on</strong> a string worldsheet<br />

<strong>on</strong>e should always be able to find two vectors: <strong>on</strong>e is time-like and another is<br />

space-like.<br />

C<strong>on</strong>sider S µ (λ) = ∂Xµ + λ ∂Xµ<br />

∂τ ∂σ<br />

must be space- or time-like as λ varies.<br />

S µ S µ = Ẋ2 + 2λẊX′ + λ 2 X ′2 ≡ y(λ) .<br />

Discriminant must be positive then there are two roots and therefore the regi<strong>on</strong>s of<br />

λ with time- and space-like vectors. Discriminant is<br />

(ẊX′ ) 2 − Ẋ2 X ′2 > 0 .<br />

This c<strong>on</strong>diti<strong>on</strong> guarantees the causal propagati<strong>on</strong> of string.<br />

Acti<strong>on</strong> is invariant under reparametrizati<strong>on</strong>s<br />

δX µ = ξ α ∂ α X µ ,<br />

ξ α = 0 <strong>on</strong> the boundary<br />

Two possibilities:<br />

• Open strings: 0 ≤ σ ≤ π<br />

• Closed strings: 0 ≤ σ ≤ 2π<br />

Equati<strong>on</strong>s of moti<strong>on</strong>:<br />

Variati<strong>on</strong><br />

δS<br />

δX µ = ∫ τ2<br />

= −<br />

+<br />

dτ<br />

τ 1<br />

∫ τ2<br />

τ<br />

∫ 1<br />

π<br />

0<br />

S =<br />

∫ π<br />

dτ<br />

0<br />

∫ π<br />

∫ τ2<br />

τ 1<br />

dτ<br />

( δL<br />

dσ<br />

0<br />

dσ<br />

dσ δL<br />

δẊµ δXµ | τ 2<br />

τ 1<br />

+<br />

• Open string boundary c<strong>on</strong>diti<strong>on</strong>s:<br />

• X µ (σ + 2π) = X µ (σ).<br />

Can<strong>on</strong>ical formalism. Momentum<br />

We see that<br />

∫ π<br />

0<br />

dσL<br />

∂ τδX µ + δL )<br />

δẊµ δX ∂ σδX µ<br />

(<br />

′µ δL δL<br />

∂ τ<br />

)<br />

+ ∂<br />

δẊµ<br />

σ<br />

δX ′µ<br />

∫ τ2<br />

τ 1<br />

(3.5)<br />

(3.6)<br />

δL<br />

δX ′µ δXµ | σ=π<br />

σ=0 (3.7)<br />

δL<br />

(τ, σ = π) = δL (τ, σ = 0) = 0<br />

δX ′µ δX ′µ<br />

P µ = δL = −T (ẊX′ )X ′µ − (X ′ ) 2 Ẋ µ<br />

√<br />

δẊµ (ẊX′ ) 2 − Ẋ2 X ′2<br />

P µ X ′ µ = 0 (3.8)<br />

P µ P µ + T 2 X ′2 = 0 (3.9)


– 16 –<br />

Exercise Show that the Hessian matrix has for each σ two zero eigenvalues corresp<strong>on</strong>ding<br />

to Ẋµ and X ′µ .<br />

Equati<strong>on</strong>s of moti<strong>on</strong> are very complicated:<br />

⎛<br />

⎞ ⎛<br />

⎞<br />

∂<br />

∂τ<br />

⎝ (ẊX′ )X ′µ − (X ′ ) 2 Ẋ µ<br />

√<br />

(ẊX′ ) 2 − Ẋ2 X ′2<br />

⎠ + ∂<br />

∂σ<br />

⎝ (ẊX′ )Ẋµ − (Ẋ)2 X ′µ<br />

√<br />

(ẊX′ ) 2 − Ẋ2 X ′2<br />

⎠ = 0 .<br />

Exercise Think how reparametrizati<strong>on</strong> invariance can be used to bring this equati<strong>on</strong><br />

to the simplest form.<br />

3.2 The Polyakov acti<strong>on</strong><br />

Introduce a word-sheet metric h αβ (σ, τ). C<strong>on</strong>sider the acti<strong>on</strong><br />

S p = − T ∫<br />

d 2 σ √ −hh αβ ∂ α X µ ∂ β X ν η µν (3.10)<br />

2<br />

Here h = deth αβ .<br />

3.2.1 Symmetries<br />

The reparametrizati<strong>on</strong> invariance<br />

δX µ = ξ α ∂ α X µ<br />

δh αβ = ξ γ ∂ γ h αβ + h αγ ∂ β ξ γ + h βγ ∂ α ξ γ<br />

δh αβ = ξ γ ∂ γ h αβ − h αγ ∂ γ ξ β − h βγ ∂ γ ξ α<br />

δ( √ −h) = ∂ α (ξ α√ −h)<br />

The Weyl invariance<br />

The Weyl invariance c<strong>on</strong>sists in rescaling the metric<br />

3.2.2 Equati<strong>on</strong>s of moti<strong>on</strong><br />

h αβ → e −2Λ(σ,τ) h αβ . (3.11)<br />

First we discuss the equati<strong>on</strong> of moti<strong>on</strong> for the intrinsic metric h αβ . This discussi<strong>on</strong><br />

amounts to the introducti<strong>on</strong> of the two-dimensi<strong>on</strong>al stress-energy tensor which is a<br />

resp<strong>on</strong>se of the acti<strong>on</strong> to the change of the metric<br />

∫<br />

δS p = −T d 2 σ √ −h T αβ δh αβ<br />

that is<br />

T αβ = − 1<br />

T √ δS p<br />

−h<br />

δh αβ


– 17 –<br />

Thus, eom for h αβ is<br />

Performing a variati<strong>on</strong> we obtain<br />

T αβ = 0 .<br />

T αβ = 1 2 ∂ αX µ ∂ β X µ − 1 4 h αβh γδ ∂ γ X µ ∂ δ X µ . (3.12)<br />

From here we immediately notice that the stress-energy tensor is traceless<br />

T α α = T αβ h αβ = 0<br />

This is a direct c<strong>on</strong>sequence of the Weyl symmetry.<br />

Due to the equati<strong>on</strong>s of moti<strong>on</strong> for the scalar fields X µ this tensor is covariantly<br />

c<strong>on</strong>served:<br />

∇ α T αβ = 0 .<br />

This can be easily derived as follows. C<strong>on</strong>sider a variati<strong>on</strong> of the acti<strong>on</strong>:<br />

∫ ( δL<br />

δS p = d 2 σ<br />

δh αβ δhαβ + δL )<br />

δX µ δXµ<br />

We see that <strong>on</strong> the equati<strong>on</strong>s of moti<strong>on</strong> for X µ , i.e. <strong>on</strong> δL = 0, we have<br />

δX<br />

∫<br />

µ<br />

δS p = −T d 2 σ √ ∫<br />

−h T αβ δh αβ = −2T d 2 σ √ −h T αβ ∇ α ξ β ,<br />

where <strong>on</strong> the r.h.s. we specified the variati<strong>on</strong> to be a diffeomorphism transformati<strong>on</strong>.<br />

The last expressi<strong>on</strong> can be integrated by parts and, die to δS p = 0, we c<strong>on</strong>clude that<br />

∇ α T αβ = 0. This derivati<strong>on</strong> is similar to the derivati<strong>on</strong> of the charge c<strong>on</strong>servati<strong>on</strong><br />

law in electrodynamics, the latter being a c<strong>on</strong>sequence of the gradient invariance.<br />

Let us show directly that ∇ α T αβ = 0. We have<br />

2∇ α T αβ = ∇ α ∂ α X µ ∂ β X µ + ∂ α X µ ∇ α ∂ β X µ − 1 2 ∇ β<br />

h γδ ∂ γ X µ ∂ δ X µ<br />

<br />

(3.13)<br />

Since ∇ α ∂ αX µ = 0 is eom for X µ we find<br />

2∇ α T αβ<br />

= ∂ α X µ ∇ α ∂ β X µ − 1 2 ∂ β<br />

h γδ ∂ γ X µ ∂ δ X µ<br />

<br />

=<br />

= ∂ αX µ h αδ ∇ δ ∂ β X µ − 1 2 ∂ β h γδ ∂ γ X µ ∂ δ X µ − h γδ ∂ γ X µ ∂ δ ∂ β X µ =<br />

= ∂ αX µ h αδ ∂ δ ∂ β X µ − ∂ αX µ ∂ sX µh αδ Γ s δβ − 1 2 ∂ β h γδ ∂ γ X µ ∂ δ X µ − h γδ ∂ γ X µ ∂ δ ∂ β X µ .<br />

Here the first term cancels with the last <strong>on</strong>e and we find<br />

2∇ α T αβ = −∂ α X µ ∂ s X µ h αδ Γ s δβ − 1 2 ∂ β h γδ ∂ γ X µ ∂ δ X µ<br />

Only symmetric part of h αδ Γ s δβ = 1 2 hαδ h sp ∂ δ h pβ + ∂ β h pδ − ∂ p h βδ<br />

<br />

in the indices α, s matters! This translates into symmetry of<br />

β, δ. Finally,<br />

Equati<strong>on</strong>s<br />

2∇ α T αβ = − 1 2 hαδ ∂ β h pδ h sp ∂ αX µ ∂ sX µ − 1 2 ∂ β h γδ ∂ γ X µ ∂ δ X µ = 0 . (3.14)<br />

T α α = 0 , ∇ β T αβ = 0<br />

are c<strong>on</strong>sequences of the Weyl and diffeomorphism symmetry, respectively. These<br />

two are gauge symmetries, and the relati<strong>on</strong>s above can be understood as the sec<strong>on</strong>d<br />

Noether theorem.


– 18 –<br />

3.2.3 C<strong>on</strong>formal gauge<br />

C<strong>on</strong>sider the c<strong>on</strong>formal Killing equati<strong>on</strong><br />

ξ γ ∂ γ h αβ + h αγ ∂ β ξ γ + h βγ ∂ α ξ γ = Λh αβ (3.15)<br />

and solve it assuming the c<strong>on</strong>formal gauge<br />

( ) −1 0<br />

h αβ = e 2φ 0 1<br />

(3.16)<br />

This equati<strong>on</strong> can be split as follows. First we take α = τ and β = σ and using<br />

h τσ = 0 we find<br />

h ττ ∂ σ ξ τ + h σσ ∂ τ ξ σ = 0 → ∂ σ ξ τ − ∂ τ ξ σ = 0<br />

Sec<strong>on</strong>d, we take α, β = τ and then α, β = σ. We get<br />

i.e.<br />

Subtracting <strong>on</strong>e from the other we get<br />

ξ γ ∂ γ h ττ + 2h ττ ∂ τ ξ τ = Λh ττ<br />

ξ γ ∂ γ h σσ + 2h σσ ∂ σ ξ σ = Λh σσ<br />

ξ γ h ττ ∂ γ h ττ + 2∂ τ ξ τ = Λ<br />

ξ γ h σσ ∂ γ h σσ + 2∂ σ ξ σ = Λ .<br />

∂ τ ξ τ − ∂ σ ξ σ = 0 .<br />

Thus, the c<strong>on</strong>formal Killing equati<strong>on</strong>s reduce in the c<strong>on</strong>formal gauge to<br />

or equivalently<br />

∂ σ ξ τ − ∂ τ ξ σ = 0<br />

∂ τ ξ τ − ∂ σ ξ σ = 0 (3.17)<br />

(∂ τ + ∂ σ )(ξ τ − ξ σ ) = 0<br />

(∂ τ − ∂ σ )(ξ τ + ξ σ ) = 0 (3.18)<br />

By using the world-sheet light-c<strong>on</strong>e coordinates σ ± = τ ± σ this can be reformulated<br />

as 4 ∂ + ξ − = 0 = ∂ − ξ + .<br />

4 The basic relati<strong>on</strong>s are ∂ τ = ∂ + + ∂ − , ∂ σ = ∂ + − ∂ − , ∂ ± = 1 2<br />

(∂ τ ± ∂ σ<br />

)<br />

.


– 19 –<br />

Thus,<br />

ξ + = ξ + (σ + ) , ξ − = ξ − (σ − )<br />

solve the c<strong>on</strong>formal Killing equati<strong>on</strong>s. This is a freedom (reparametrizati<strong>on</strong> + Weyl<br />

rescaling) which does not destroy the c<strong>on</strong>formal gauge c<strong>on</strong>diti<strong>on</strong>.<br />

The final remark c<strong>on</strong>cerns restricti<strong>on</strong>s <strong>on</strong> ξ ± following from the periodicity c<strong>on</strong>diti<strong>on</strong>s.<br />

Indeed, for the Weyl factor Λ we have<br />

ξ τ ∂ τ φ + ξ σ ∂ σ φ + 2∂ τ ξ τ = Λ(σ, τ) .<br />

The factor Λ must be periodic in σ because the metric h αβ is. Since φ is periodic<br />

the allowed soluti<strong>on</strong>s ξ ± of the c<strong>on</strong>formal Killing equati<strong>on</strong> are those which lead to<br />

periodic Λ. One can easily see that this implies that ξ ± must be periodic in σ.<br />

3.2.4 Polyakov string in the first order formalism.<br />

In the first order formalism the density L ≡ L(σ, τ) of the Polyakov Lagrangian takes<br />

the form<br />

L = P µ ∂ τ X µ + 1 (<br />

) ( )<br />

P<br />

2T γ ττ µ P µ + T 2 X µX ′ ′µ + γτσ<br />

P<br />

γ ττ µ X ′µ (3.19)<br />

The c<strong>on</strong>formal gauge c<strong>on</strong>sists in imposing the following two c<strong>on</strong>diti<strong>on</strong>s<br />

The gauged-fixed Lagrangian density is<br />

γ ττ = −1 , γ τσ = 0.<br />

L = P µ ∂ τ X µ − 1<br />

2T<br />

and the Hamilt<strong>on</strong>ian density is<br />

H = 1<br />

2T<br />

(P µ P µ + T 2 X ′ µX ′µ )<br />

(P µ P µ + T 2 X ′ µX ′µ )<br />

(3.20)<br />

The phase-space Lagrangian shows that the variables (P µ , X µ ) are can<strong>on</strong>ical, i.e. the<br />

corresp<strong>on</strong>ding Poiss<strong>on</strong> bracket is<br />

{X µ (σ, τ), X ν (σ ′ , τ)} = {P µ (σ, τ), P ν (σ ′ , τ)} = 0 , (3.21)<br />

{X µ (σ, τ), P ν (σ ′ , τ)} = η µν δ(σ − σ ′ ) . (3.22)<br />

The dynamics of the system in the c<strong>on</strong>formal gauge is governed by the Hamilt<strong>on</strong>ian<br />

∫<br />

H = dσ H = 1 ∫<br />

)<br />

dσ<br />

(P µ P µ + T 2 X ′<br />

2T<br />

µX ′µ<br />

Equati<strong>on</strong>s of moti<strong>on</strong><br />

Ẋ µ = {X µ , H} = 1 T P µ , (3.23)<br />

˙ P µ = {P µ , H} = T X ′′µ , (3.24)


– 20 –<br />

which result into<br />

We see that we have two c<strong>on</strong>straints<br />

We find the following Poiss<strong>on</strong> brackets<br />

Ẍ µ − X ′′µ = □X µ = 0<br />

C 1 = P µ P µ + T 2 X ′ µX ′µ , C 2 = P µ X ′µ (3.25)<br />

{C 1 (σ), C 1 (σ ′ )} = 4T 2 ∂ σ C 2 (σ)δ(σ − σ ′ ) + 8T 2 C 2 (σ)∂ σ δ(σ − σ ′ ) , (3.26)<br />

{C 1 (σ), C 2 (σ ′ )} = ∂ σ C 1 (σ)δ(σ − σ ′ ) + 2C 1 (σ)∂ σ δ(σ − σ ′ ) , (3.27)<br />

{C 2 (σ), C 1 (σ ′ )} = ∂ σ C 1 (σ)δ(σ − σ ′ ) + 2C 1 (σ)∂ σ δ(σ − σ ′ ) , (3.28)<br />

{C 2 (σ), C 2 (σ ′ )} = ∂ σ C 2 (σ)δ(σ − σ ′ ) + 2C 2 (σ)∂ σ δ(σ − σ ′ ) . (3.29)<br />

Instead of the c<strong>on</strong>straints C 1 and C 2 we can equally c<strong>on</strong>sider their linear combinati<strong>on</strong>s<br />

Their Poiss<strong>on</strong> algebra becomes<br />

T ++ = 1<br />

8T 2 (C 1 + 2T C 2 ) = 1<br />

8T 2 (P µ + T X ′ µ) 2 , (3.30)<br />

T −− = 1<br />

8T 2 (C 1 − 2T C 2 ) = 1<br />

8T 2 (P µ − T X ′ µ) 2 . (3.31)<br />

{T ++ (σ), T ++ (σ ′ )} = 1 (<br />

)<br />

∂ σ T ++ (σ)δ(σ − σ ′ ) + 2T ++ (σ)∂ σ δ(σ − σ ′ ) ,<br />

2T<br />

{T −− (σ), T −− (σ ′ )} = − 1 (<br />

)<br />

∂ σ T −− (σ)δ(σ − σ ′ ) + 2T −− (σ)∂ σ δ(σ − σ ′ ) ,<br />

2T<br />

{T ++ (σ), T −− (σ ′ )} = 0 . (3.32)<br />

C<strong>on</strong>straints T ++ and T −− Poiss<strong>on</strong> commute and form two independent Poiss<strong>on</strong> algebras!<br />

Now <strong>on</strong>e can easily find the evoluti<strong>on</strong> equati<strong>on</strong>s for T ++ and T −− . We have<br />

∂ τ T ++ = {T ++ (σ), H} = ∂ σ T ++ =⇒ (∂ τ − ∂ σ )T ++ = ∂ − T ++ = 0 ,<br />

∂ τ T −− = {T −− (σ), H} = −∂ σ T −− =⇒ (∂ τ + ∂ σ )T −− = ∂ + T −− = 0 ,<br />

Thus, we see that evoluti<strong>on</strong> equati<strong>on</strong>s imply that<br />

The Hamilt<strong>on</strong>ian itself is<br />

T ++ = T ++ (σ + ) , T −− = T −− (σ − ) . (3.33)<br />

H = 2T<br />

∫ 2π<br />

0<br />

dσ(T ++ + T −− ) ,<br />

i.e. it is a sum of zero modes of the left- and right-moving c<strong>on</strong>straints.


– 21 –<br />

3.3 Integrals of moti<strong>on</strong>. Classical Virasoro algebra.<br />

<strong>String</strong> has an infinite set of integrals of moti<strong>on</strong> (quantities which are c<strong>on</strong>served in<br />

time due to equati<strong>on</strong>s of moti<strong>on</strong>) which are c<strong>on</strong>structed with the help of T ±± .<br />

Note that T ±± themselves are not the good c<strong>on</strong>served quantities as they depend<br />

<strong>on</strong> time! However, if we define the Fourier comp<strong>on</strong>ents<br />

then 5<br />

dL m<br />

dτ<br />

= 2T<br />

= 2T<br />

= 2T<br />

∫ 2π<br />

0<br />

∫ 2π<br />

0<br />

∫ 2π<br />

0<br />

dσ<br />

dσ<br />

dσ<br />

L m = 2T<br />

∫ 2π<br />

0<br />

dσ e imσ− T −− (σ, τ)<br />

(<br />

)<br />

∂ τ e im(τ−σ) T −− (σ, τ) + e im(τ−σ) ∂ τ T −− (σ, τ)<br />

(<br />

)<br />

ime im(τ−σ) T −− (σ, τ) − e im(τ−σ) ∂ σ T −− (σ, τ)<br />

(<br />

)<br />

ime im(τ−σ) T −− (σ, τ) + ∂ σ e im(τ−σ) T −− (σ, τ) = 0<br />

(3.34)<br />

Thus, the Fourier comp<strong>on</strong>ents of the stress-energy tensor provide an infinite set of<br />

the c<strong>on</strong>served c<strong>on</strong>straints. Analogously, we define<br />

¯L m = 2T<br />

∫ 2π<br />

0<br />

dσ e imσ+ T ++ (σ, τ) ,<br />

which is also an integral of moti<strong>on</strong>. Note that in this derivati<strong>on</strong> we never used the<br />

c<strong>on</strong>straints T ++ = 0 = T −− .<br />

The Poiss<strong>on</strong> brackets of the L m and ¯L m generators are<br />

{L m , L n } = −i(m − n)L m+n ,<br />

{¯L m , ¯L n } = −i(m − n)¯L m+n , (3.35)<br />

{L m , ¯L n } = 0 .<br />

Z<br />

{L m , L n } = 4T 2 dσdσ ′ e −imσ−inσ′ {T −− (σ), T −− (σ ′ )}<br />

Z<br />

= −2T dσdσ ′ e −imσ−inσ′ ∂ σT −− (σ)δ(σ − σ ′ ) + 2T −− (σ)∂ σδ(σ − σ ′ <br />

) =<br />

Z <br />

= −2T dσ − T −− (σ)∂ σ e −i(m+n)σ′ + 2e −imσ T −− (σ)∂ σ e −inσ<br />

Z<br />

= −i(m − n) 2T dσe −i(m+n)σ T −− (σ) = −i(m − n)L m+n<br />

This is the so-called Wit algebra. This algebra acts <strong>on</strong> X µ (σ, τ):<br />

{L m , X µ } = 2T<br />

∫ 2π<br />

0<br />

dσ ′ e imσ′− {T −− (σ ′ ), X µ (σ)} =<br />

= − 1 2 eimσ− (Ẋµ − X ′µ ) = −e imσ− ∂ − X µ . (3.36)<br />

5 When finding the time dynamics of L m <strong>on</strong>e has to remember that the functi<strong>on</strong> L m has an<br />

explicit time-dependence and, therefore dL m<br />

dτ<br />

= ∂ τ L m + {L m , H}.


– 22 –<br />

Analogously,<br />

{¯L m , X µ } = 2T<br />

∫ 2π<br />

0<br />

dσ ′ e imσ′+ {T ++ (σ ′ ), X µ (σ)} =<br />

= − 1 2 eimσ+ (Ẋµ + X ′µ ) = −e imσ+ ∂ + X µ . (3.37)<br />

To summarize<br />

In particular,<br />

{L m , X µ } = −e imσ− ∂ − X µ , {¯L m , X µ } = −e imσ+ ∂ + X µ . (3.38)<br />

{¯L 0 − L 0 , X µ } = ∂ σ X µ (3.39)<br />

i.e. ¯L 0 −L 0 generates rigid σ-translati<strong>on</strong>s. The transformati<strong>on</strong>s we c<strong>on</strong>sider transform<br />

a soluti<strong>on</strong><br />

□X µ = 0<br />

into another soluti<strong>on</strong> of this equati<strong>on</strong>. Indeed, we have for instance<br />

∂ + ∂ −<br />

(<br />

e imσ− ∂ − X µ )<br />

= imσ − e imσ− ∂ + ∂ − X µ + e imσ− ∂ − ∂ + ∂ − X µ = 0<br />

as the c<strong>on</strong>sequence of ∂ + ∂ − X µ = 0.<br />

(P , X)<br />

Lm<br />

_<br />

L m<br />

_<br />

L m =0<br />

L m =0<br />

Fig. 1. The phase space of string. The Virasoro c<strong>on</strong>straints L m = 0 = ¯L m<br />

define a time-independent hypersurface <strong>on</strong> which the dynamics of string<br />

takes place. This hypersurface remains invariant under the acti<strong>on</strong> of L m ’s<br />

and ¯L m ’s.<br />

Even more generally, for any periodic functi<strong>on</strong> f we define<br />

L f =<br />

∫ 2π<br />

0<br />

dσf(σ + )T ++ .


– 23 –<br />

Then we see that<br />

0 =<br />

∫ 2π<br />

0<br />

( ) ∫ 2π ( )<br />

dσ∂ − f(σ + )T ++ = ∂ τ L f − dσ∂ σ f(σ + )T ++ .<br />

0<br />

Thus,<br />

∂ τ L f =<br />

∫ 2π<br />

0<br />

dσ∂ σ<br />

(<br />

f(σ + )T ++<br />

)<br />

= 0 .<br />

To summarize: we see that up<strong>on</strong> fixing c<strong>on</strong>formal gauge we are still left with<br />

gauge freedom. It corresp<strong>on</strong>ds to reparametrizati<strong>on</strong>s of the special type (soluti<strong>on</strong>s<br />

to the c<strong>on</strong>formal Killing equati<strong>on</strong>):<br />

σ + → ξ + (σ + ) , σ − → ξ − (σ − ) ,<br />

where ξ ± are two arbitrary functi<strong>on</strong>s periodic in σ.<br />

Finally, for the case of open string we define<br />

L m = 2T<br />

∫ π<br />

0<br />

)<br />

dσ<br />

(e imσ+ T ++ + e imσ− T −− . (3.40)<br />

The point is that there appears additi<strong>on</strong>al boundary terms which show that the old<br />

L m and ¯L m are not separately c<strong>on</strong>served. With our new definiti<strong>on</strong> of L m we obtain<br />

dL m<br />

dτ<br />

= 2T<br />

= 2T<br />

= 2T<br />

∫ π<br />

0<br />

∫ π<br />

0<br />

∫ π<br />

0<br />

)<br />

dσ<br />

(ime imσ+ T ++ + e imσ+ ∂ τ T ++ + ime imσ− T −− + e imσ− ∂ τ T −−<br />

)<br />

dσ<br />

(ime imσ+ T ++ + e imσ+ ∂ σ T ++ + ime imσ− T −− − e imσ− ∂ σ T −−<br />

)<br />

dσ<br />

(ime imσ+ T ++ − ∂ σ e imσ+ T ++ + ime imσ− T −− + ∂ σ e imσ− T −−<br />

)<br />

+ 2T e im(τ+π)( T ++ (π, τ) − e −2πim T −− (π, τ)<br />

( )<br />

− 2T e imτ T ++ (0, τ) − T −− (0, τ)<br />

The bulk term vanishes as before and we left with the boundary term<br />

dL m<br />

dτ<br />

) (<br />

)<br />

= 2T e<br />

(T im(τ+π) ++ (π, τ) − T −− (π, τ) − 2T e imτ T ++ (0, τ) − T −− (0, τ)<br />

Due to the open string boundary c<strong>on</strong>diti<strong>on</strong>s X ′µ (π, τ) = 0 = X ′µ (0, τ) we obviously<br />

have<br />

T ++ (π, τ) = T −− (π, τ) , T ++ (0, τ) = T −− (0, τ)<br />

Therefore, the boundary term vanishes and, therefore, L m are c<strong>on</strong>served quantities<br />

in the open string case.


– 24 –<br />

3.3.1 Soluti<strong>on</strong>s of the equati<strong>on</strong>s of moti<strong>on</strong><br />

Here we are going to discuss the soluti<strong>on</strong>s of the string equati<strong>on</strong>s off moti<strong>on</strong><br />

□X µ = 0<br />

which arises up<strong>on</strong> fixing the c<strong>on</strong>formal gauge.<br />

• Closed strings. We have<br />

X µ (σ, τ) = X µ L (τ + σ) + Xµ R<br />

(τ − σ)<br />

where<br />

X µ R (τ − σ) = 1 2 xµ + pµ<br />

(τ − σ) +<br />

i<br />

4πT<br />

∑<br />

√<br />

4πT<br />

n≠0<br />

n≠0<br />

1<br />

n αµ ne −in(τ−σ) (3.41)<br />

X µ L (τ + σ) = 1 2 xµ + pµ<br />

(τ + σ) +<br />

i ∑<br />

√ nᾱµ 1<br />

4πT<br />

ne −in(τ+σ) (3.42)<br />

4πT<br />

Since X µ (σ, τ) are real then (x µ , p µ ) are real as well and<br />

Let us define the zero modes as<br />

Oscillators obey the Poiss<strong>on</strong> relati<strong>on</strong>s<br />

α µ −n = (α µ n) † , ᾱ µ −n = (ᾱ µ n) †<br />

α µ 0 = ᾱ µ 0 = 1 √<br />

4πT<br />

p µ .<br />

{α µ m, α ν n} = {ᾱ µ m, ᾱ ν n} = −imδ m+n η µν ,<br />

{α µ m, ᾱ ν n} = 0 (3.43)<br />

{x µ , p ν } = η µν .<br />

The Virasoro c<strong>on</strong>straints become<br />

L m = 1 2<br />

∞∑<br />

n=−∞<br />

α µ m−nα nµ , ¯Lm = 1 2<br />

∞∑<br />

n=−∞<br />

ᾱ µ m−nᾱ nµ . (3.44)<br />

• Open strings Soluti<strong>on</strong> of the wave equati<strong>on</strong> with the open string boundary<br />

c<strong>on</strong>diti<strong>on</strong>s is<br />

X µ (σ, τ) = x µ + pµ<br />

πT τ +<br />

√ i ∑<br />

πT<br />

n≠0<br />

1<br />

n αµ ne −inτ cos nσ . (3.45)


– 25 –<br />

Oscillators obey the Poiss<strong>on</strong> algebra<br />

{α µ m, α ν n} = −imδ m+n η µν ,<br />

{x µ , p ν } = η µν . (3.46)<br />

If we define the zero mode as α µ 0 = 1 √<br />

πT<br />

p µ then the generators of the open<br />

string classical Virasoro algebra are realized as<br />

L m = 1 2<br />

∞∑<br />

n=−∞<br />

α µ m−nα nµ . (3.47)<br />

The hermiticity property of the n<strong>on</strong>-zero modes are α µ −n = (α µ n) † .<br />

3.3.2 Poincaré symmetry. <strong>String</strong> tensi<strong>on</strong>.<br />

The Noether currents of the Poincaré symmetry<br />

P α µ = −T √ −hh αβ ∂ β X µ (3.48)<br />

J µν<br />

α = −T √ −hh αβ (X µ ∂ β X ν − X ν ∂ β X µ ) (3.49)<br />

Here P α is a current corresp<strong>on</strong>ding to translati<strong>on</strong>al invariance X µ → X µ + a, where<br />

a is an arbitrary c<strong>on</strong>stant, and J µν is a current corresp<strong>on</strong>ding to Lorentz rotati<strong>on</strong>s<br />

X µ → Λ µ νX ν , where Λ µ ν is a (c<strong>on</strong>stant) matrix comprising the parameters of the<br />

Lorentz transformati<strong>on</strong>. We see that<br />

J µν<br />

α = X µ P α ν − X ν P α µ .<br />

Both currents are c<strong>on</strong>served due to equati<strong>on</strong>s of moti<strong>on</strong> of X µ and their τ-comp<strong>on</strong>ents<br />

integrated over σ define the c<strong>on</strong>served charges (for the open string case integrati<strong>on</strong><br />

rans from 0 to π)<br />

P µ =<br />

∫ 2π<br />

0<br />

dσP µ τ , J µν =<br />

∫ 2π<br />

0<br />

dσJ µν<br />

τ .<br />

Imposing the c<strong>on</strong>formal gauge and using the fundamental brackets for (X, P ) <strong>on</strong>e<br />

finds the following Poiss<strong>on</strong> algebra<br />

{P µ , P ν } = 0<br />

{P µ , J ρσ } = η µσ P ρ − η µρ P σ (3.50)<br />

{J µν , J ρσ } = η µρ J νσ + η νσ J µρ − η νρ J µσ − η µσ J νρ .<br />

Substituting soluti<strong>on</strong> for X µ <strong>on</strong>e finds<br />

P µ = T<br />

∫ 2π<br />

0<br />

dσẊµ = p µ ,


– 26 –<br />

i.e. the total mass momentum of string coincides with p µ .<br />

The total angular momentum of closed string in the c<strong>on</strong>formal gauge is is defined as<br />

J µν = T<br />

∫ 2π<br />

0<br />

Substituting the oscillator expansi<strong>on</strong> we get<br />

dσ(X µ Ẋ ν − X ν Ẋ µ ) . (3.51)<br />

where<br />

J µν = x µ p ν − x ν p µ<br />

} {{ }<br />

l µν +S µν + ¯S µν ,<br />

S µν = −i<br />

¯S µν = −i<br />

∞∑<br />

n=1<br />

∞∑<br />

n=1<br />

1 (<br />

α<br />

µ<br />

n<br />

−nαn ν − α−nα n) ν µ , (3.52)<br />

1 (ᾱµ<br />

n<br />

−nᾱn ν − ᾱ−nᾱ n) ν µ . (3.53)<br />

For the case of open string expressi<strong>on</strong>s are the same (again integrati<strong>on</strong> runs from 0 to<br />

π) except ¯S µν is absent. Here l µν is the angular momentum of string and S µν µν<br />

+ ¯S<br />

is its internal spin.<br />

Let us show that both P µ and J µν are invariant under the acti<strong>on</strong> of the Virasoro<br />

algebra. We have<br />

{L m , P µ } = {L m , p µ } = 0 (3.54)<br />

as L m does not c<strong>on</strong>tain x µ . Let us separate the zero mode part 6 of L m<br />

We first compute<br />

L m = α ρ 0α mρ + 1 2<br />

∑<br />

n≠0,m<br />

α ρ m−nα nρ .<br />

{L m , l µ } = {α ρ 0α mρ , x µ p ν − x ν p µ } = 1 √<br />

4πT<br />

α mρ {p ρ , x µ p ν − x ν p µ } =<br />

Sec<strong>on</strong>d, since S µν = − ∑ k≠0<br />

= α ν mα µ 0 − α µ mα ν 0 . (3.55)<br />

∑<br />

n,k≠0;n≠m<br />

i<br />

k αµ −k αν k<br />

we have<br />

{ 1 2 αρ m−nα nρ , − i k αµ −k αν k} = 0 ,<br />

∑<br />

{α0α ρ mρ , − i k αµ −k αν k} = α mα µ 0 ν − αmα ν µ 0 ,<br />

k≠0<br />

6 We assume here that m ≠ 0.


– 27 –<br />

therefore, {L m , l µν + S µν } = 0. Thus, Poincaré symmetry descends <strong>on</strong> the space of<br />

physical states. Physical states will be classified in terms of representati<strong>on</strong>s of the<br />

Poincaré group of d-dimensi<strong>on</strong>al Minkowski space.<br />

An important invariant of the Poincaré group is the square of the momentum P µ P µ .<br />

Indeed,<br />

{P µ P µ , P ν } = {P µ P µ , J ρλ } = 0 .<br />

It coincides with the mass: M 2 = −P µ P µ . Of course, P 2 is also invariant w.r.t. to<br />

the acti<strong>on</strong> of L m and ¯L m . C<strong>on</strong>sider now <strong>on</strong>e of the c<strong>on</strong>straints, L 0 = 0. Separating<br />

the zero mode<br />

we have<br />

L 0 = 1 2<br />

n=∞<br />

∑<br />

n=−∞<br />

From here we deduce the mass<br />

α n α −n = 1 2 α2 0 +<br />

p µ p µ + 8πT<br />

∞∑<br />

n=1<br />

α n α −n , α µ 0 = 1 √<br />

4πT<br />

p µ<br />

∞∑<br />

α n α −n = 0 .<br />

n=1<br />

M 2 = −p 2 = 8πT<br />

∞∑<br />

α n α −n .<br />

Mass is created due to internal excitati<strong>on</strong>s of string! From this expressi<strong>on</strong> it is not<br />

obvious that M 2 is n<strong>on</strong>-negative.<br />

n=1<br />

Another invariant of the Poincaré group is<br />

J 2 = 1 (<br />

J αβ J αβ + 2<br />

)<br />

2<br />

M P αJ αλ P β J 2 βλ .<br />

Checking Poincaré invariance of this expressi<strong>on</strong> is straightforward, as an intermediate step we note the relati<strong>on</strong>s<br />

{J µν , P ρ J ρσ } = η νσ J µρ P ρ − η µσ J νρ P ρ<br />

{P µ , J αβ J αβ } = −4J µρ P ρ .<br />

The terms J αβ J αβ and P αJ αλ P β J βλ separately Poiss<strong>on</strong> commute with J µν .<br />

C<strong>on</strong>sider an open string which has P i = p i = 0 for all i = 1, . . . , d − 1. By using reparametrizati<strong>on</strong> invariance fix the static<br />

gauge X 0 = τ. The angular moment l µν of this string is zero. Also P α J αλ P β J βλ = 0 and, therefore,<br />

J 2 = 1 2 J ij J ij ,<br />

where i, j = 1, . . . d − 1. We have<br />

J 2 = − 1 ∞X 1<br />

2 n,m=1 nm αi −n αj n − αj −n αi i<br />

n α −m αj m − <br />

αj −m αi m<br />

∞X 1<br />

=<br />

(α ∗ ∞X<br />

n<br />

n,m=1 nm<br />

αm)(α∗ m αn) − (α∗ n α∗ m )(αnαm) 1<br />

≤<br />

n,m=1 nm (α∗ n αm)(α∗ m αn) .


– 28 –<br />

It follows from the Schwarz inequality that<br />

(α ∗ n α m)(α ∗ m α n) = |(α ∗ n α m)| 2 ≤ (α ∗ n α n)(α ∗ m α m) ≤ nm(α ∗ n α n)(α ∗ m α m) .<br />

Therefore,<br />

∞X<br />

J 2 ≤<br />

n,m=1(α ∗ n α n)(α ∗ m α m) = 1 ∞X<br />

X ∞ 1<br />

α n α −n α m α −m =<br />

4<br />

n=−∞<br />

m=−∞<br />

(2πT ) 2 M 4<br />

because in the open string case<br />

∞X<br />

M 2 = πT α n α −n = 2πT X α n α −n .<br />

n=−∞ n>0<br />

Thus, for an open string moti<strong>on</strong> we found inequality<br />

J ≡<br />

pJ 2 ≤ 1<br />

2πT M 2 .<br />

Here the parameter<br />

α ′ = 1<br />

2πT<br />

is called a slope of the Regge trajectory. The functi<strong>on</strong> J = α ′ M 2 is a straight line in the (M 2 , J) plane whose slope is α ′ .<br />

C<strong>on</strong>sider a closed (pulsating) string soluti<strong>on</strong><br />

x = R cos σ cos τ , y = R sin σ cos τ , t = Rτ .<br />

We see that<br />

P 0 = 2πRT =⇒ T = P 0<br />

2πR ≡<br />

i.e. tensi<strong>on</strong> is energy per unit length.<br />

3.4 <strong>String</strong>s in physical gauge<br />

E<br />

2πR ,<br />

As we have seen up<strong>on</strong> fixing c<strong>on</strong>formal gauge we are still left with the gauge freedom.<br />

It corresp<strong>on</strong>ds to reparametrizati<strong>on</strong>s of the special type (soluti<strong>on</strong>s to the c<strong>on</strong>formal<br />

Killing equati<strong>on</strong>):<br />

σ + → ξ + (σ + ) , σ − → ξ − (σ − ) ,<br />

where ξ ± are two arbitrary functi<strong>on</strong>s (periodic in σ). This freedom can be further<br />

fixed leaving <strong>on</strong>ly physical excitati<strong>on</strong>s. This is achieved by imposing the so-called<br />

light-c<strong>on</strong>e gauge.<br />

3.4.1 First order formalism<br />

Introduce the light-c<strong>on</strong>e coordinates in the d-dimensi<strong>on</strong>al Minkowski space<br />

X ± = 1 √<br />

2<br />

(X 0 ± X d−1 ), X i , i = 1, . . . , d − 2 .<br />

C<strong>on</strong>sider the Polyakov acti<strong>on</strong> and introduce the light-c<strong>on</strong>e momenta c<strong>on</strong>jugate to<br />

the light-c<strong>on</strong>e coordinates<br />

P ± =<br />

∂L<br />

∂Ẋ± ,<br />

P i = ∂L<br />

∂Ẋi (3.56)


– 29 –<br />

Express now the velocities via the corresp<strong>on</strong>ding momenta<br />

Ẋ ± = 1<br />

T γ ττ (P ∓ − T γ τσ X ′± ) , Ẋ i = 1<br />

T γ ττ (−P i − T γ τσ X ′i ) .<br />

Note that the light-c<strong>on</strong>e indices are raised and lowered according to the rule<br />

P − = −P + , P + = −P − .<br />

Now we can c<strong>on</strong>struct the phase-space Lagrangian in the light-c<strong>on</strong>e coordinates:<br />

After explicit computati<strong>on</strong> we find<br />

L = P i Ẋ i + P + Ẋ + + P − Ẋ − + rest .<br />

L = P i Ẋ i + P + Ẋ + + P − Ẋ − + 1<br />

2T γ ττ (<br />

− 2P − P + + P i P i + T 2 X ′ iX ′i − 2T 2 X ′− X ′+ )<br />

+ γτσ<br />

γ ττ (<br />

P i X ′i + P − X ′− + P + X ′+ )<br />

. (3.57)<br />

The phase-space light-c<strong>on</strong>e gauge c<strong>on</strong>sists in imposing the following two c<strong>on</strong>diti<strong>on</strong>s<br />

1. Closed string<br />

2. Open string<br />

X + = p+<br />

2πT τ , P + = c<strong>on</strong>st ≡ 1<br />

2π p+ . (3.58)<br />

X + = p+<br />

πT τ , P + = c<strong>on</strong>st ≡ 1 π p+ . (3.59)<br />

This gauge choice is d<strong>on</strong>e to remove completely the gauge degrees of freedom (recall<br />

that in the c<strong>on</strong>formal gauge we still had a gauge freedom left which was generated<br />

by soluti<strong>on</strong>s of the c<strong>on</strong>formal Killing equati<strong>on</strong>).<br />

We further c<strong>on</strong>sider the closed string case in detail. We will derive and solve all<br />

the c<strong>on</strong>straints followed from the Lagrangian (3.57) in several steps.<br />

1. Varying the Lagrangian w.r.t. γ ττ and imposing the light-c<strong>on</strong>e gauge allows<br />

<strong>on</strong>e to solve for P − :<br />

P − = π (<br />

)<br />

P<br />

p + i P i + T 2 X iX ′ ′i . (3.60)<br />

Thus, equati<strong>on</strong> of moti<strong>on</strong> for γ ττ allows <strong>on</strong>e to determine P − .


– 30 –<br />

2. Varying w.r.t. γ τσ leads to determinati<strong>on</strong> of X ′− :<br />

X ′− = − 1<br />

P −<br />

P i X ′i = 2π<br />

p + P iX ′i . (3.61)<br />

If we integrate the last equati<strong>on</strong> over σ we obtain<br />

X − (2π) − X − (0) = 2π<br />

p + ∫ 2π<br />

0<br />

dσP i X ′i . (3.62)<br />

The closed string periodicity c<strong>on</strong>diti<strong>on</strong> requires the fulfillment of the following<br />

c<strong>on</strong>straint<br />

V =<br />

∫ 2π<br />

0<br />

dσP i X ′i = 0 . (3.63)<br />

This is the <strong>on</strong>ly c<strong>on</strong>straint which remains unsolved and it is known as the level<br />

matching c<strong>on</strong>diti<strong>on</strong>. We will impose it <strong>on</strong> physical states of the theory.<br />

3. Now we can determine the world-sheet metric γ αβ . Equati<strong>on</strong> of moti<strong>on</strong> for P +<br />

is<br />

0 = δL =<br />

δP Ẋ+ −<br />

P − γτσ<br />

+<br />

+ T γττ γ ττ X′+ ,<br />

which with our gauge choice gives<br />

γ ττ = −1 .<br />

4. Equati<strong>on</strong> of moti<strong>on</strong> for Ẋ− gives<br />

which gives<br />

0 = d δL δL<br />

−<br />

dt δẊ− δX = − d p +<br />

− dt 2π − δL δL<br />

=⇒<br />

δX− δX = 0 , −<br />

( ) γ<br />

τσ<br />

∂ σ<br />

γ P ττ − = 0 =⇒ ∂ σ γ τσ = 0 .<br />

For the closed string case this implies that γ τσ = γ τσ (τ) is an arbitrary functi<strong>on</strong><br />

of τ. The presence of this functi<strong>on</strong> signals a residual symmetry. Indeed, <strong>on</strong> the<br />

soluti<strong>on</strong>s of the level-matching c<strong>on</strong>straint V = 0 the ratio γτσ<br />

can be shifted by<br />

γ ττ<br />

an arbitrary functi<strong>on</strong> f(τ) of τ without affecting the Lagrangian.<br />

5. Varying w.r.t P − we find an evoluti<strong>on</strong> equati<strong>on</strong> for X − :<br />

0 = δL<br />

δP −<br />

= Ẋ− − P +<br />

T γ ττ = 0 =⇒ Ẋ− = π<br />

T p + (P iP i + T 2 X ′ iX ′i ) .


– 31 –<br />

Thus, the variable X − is not physical and can be solved from the following two<br />

equati<strong>on</strong>s we found above<br />

Ẋ − =<br />

These two equati<strong>on</strong>s can be rewritten as<br />

π<br />

T p + (P iP i + T 2 X ′ iX ′i ) (3.64)<br />

X ′− = 2π<br />

p + P iX ′i (3.65)<br />

∂ ± X − = 2πT<br />

p + (∂ ±X i ) 2 . (3.66)<br />

It is worth noting that two equati<strong>on</strong>s (3.64), (3.65) are compatible. Indeed, the<br />

σ-derivative of the first equati<strong>on</strong> must be equal the τ-derivative of the sec<strong>on</strong>d<br />

<strong>on</strong>e. One can see that this is indeed so due to equati<strong>on</strong>s of moti<strong>on</strong> for physical<br />

fields.<br />

Now we are ready to c<strong>on</strong>struct the gauge-fixed Lagrangian. Substituting soluti<strong>on</strong>s<br />

of all the c<strong>on</strong>straints and the gauge c<strong>on</strong>diti<strong>on</strong>s into eq.(3.57) we obtain the density<br />

L = P i Ẋ i − 1<br />

)<br />

2π p+ Ẋ − − p+<br />

2πT P − − γ τσ (τ)<br />

(P i X ′i − p+<br />

2π X′− . (3.67)<br />

Thus, the Lagrangian itself is<br />

L =<br />

∫ 2π<br />

0<br />

∫ 2π<br />

dσL = −p + ẋ − + dσ P i Ẋ i −H − γ τσ (τ)V . (3.68)<br />

0<br />

} {{ }<br />

defines Poiss<strong>on</strong> structure<br />

Here x − denotes the zero (c<strong>on</strong>stant) mode of the variable X − and the Hamilt<strong>on</strong>ian<br />

is<br />

H = 1 ∫ 2π<br />

)<br />

dσ<br />

(P i P i + T 2 X ′<br />

2T<br />

iX ′i .<br />

0<br />

We also see that γ τσ (τ) plays the role of the Lagrangian multiplier to the levelmatching<br />

c<strong>on</strong>straint V. Without loss of generality we will choose γ τσ = 0 which<br />

corresp<strong>on</strong>ds the c<strong>on</strong>formal gauge c<strong>on</strong>diti<strong>on</strong> discussed above.<br />

From the gauge-fixed Lagrangian we c<strong>on</strong>clude that our physical variables are (P i , X i ),<br />

where i = 1, . . . , d−2, and also (x − , p + ) and they have the following Poiss<strong>on</strong> brackets<br />

{X i (σ, τ), X j (σ ′ , τ)} = {P i (σ, τ), P j (σ ′ , τ)} = 0 ,<br />

{X i (σ, τ), P j (σ ′ , τ)} = δ ij δ(σ − σ ′ ) , (3.69)<br />

{p + , x − } = 1 .


– 32 –<br />

The physical Hamilt<strong>on</strong>ian and the Poiss<strong>on</strong> brackets look the same as the <strong>on</strong>es in the<br />

c<strong>on</strong>formal gauge, however, the important difference is that now they involve 2(d − 2)<br />

physical fields <strong>on</strong>ly plus two additi<strong>on</strong>al degrees of freedom (x − , p + ). First, from<br />

eq.(3.65) we find that the zero mode of X − evolves as<br />

ẋ − = 1<br />

p + H =⇒ x− (τ) = x − + H p + τ .<br />

It is this τ-independent mode x − which is c<strong>on</strong>jugate to p + . Sec<strong>on</strong>d, equati<strong>on</strong>s<br />

∂ ± X − = 2πT<br />

p + (∂ ±X i ) 2 (3.70)<br />

can be solved for the l<strong>on</strong>gitudinal oscillators αn − , ᾱn<br />

− with n ≠ 0 by substituting an<br />

expansi<strong>on</strong><br />

X − (τ, σ) = x − + p−<br />

}{{} 2πT<br />

τ + i<br />

p + H<br />

We find p − = 2πT<br />

p + H and<br />

√<br />

4πT<br />

∑<br />

n≠0<br />

1<br />

(<br />

αn − e −inσ− + ᾱn − e −inσ+) . (3.71)<br />

n<br />

α − n =<br />

ᾱ − n =<br />

√<br />

πT<br />

p +<br />

√<br />

πT<br />

p +<br />

∞<br />

∑<br />

m=−∞<br />

∑ ∞<br />

m=−∞<br />

α i n−mα i m , n ≠ 0, (3.72)<br />

ᾱ i n−mᾱ i m , n ≠ 0 . (3.73)<br />

These formulae give a complete soluti<strong>on</strong> for X − .<br />

Thus, the light-c<strong>on</strong>e gauge allows for the explicit soluti<strong>on</strong> of the Virasoro c<strong>on</strong>straints.<br />

The variables (P i , X i ) are physical excitati<strong>on</strong>s while X ± , P + were removed by the<br />

light-c<strong>on</strong>e gauge choice and by solving the c<strong>on</strong>straints. The variable P − plays the<br />

role of the Hamilt<strong>on</strong>ian for physical excitati<strong>on</strong>s! Equati<strong>on</strong>s of moti<strong>on</strong> for physical<br />

fields are the same as before<br />

Ẍ i − X ′′i = 0 , i = 1, . . . , d − 2.<br />

The variables (P i , X i ), where i = 1, . . . , d − 2, are called transversal, while X ± , P ±<br />

are l<strong>on</strong>gitudinal. The <strong>on</strong>ly c<strong>on</strong>straint which we were not able to solve explicitly is<br />

the level-matching c<strong>on</strong>straint V = 0. It is easy to check that<br />

{X i , V} = ∂ σ X i , {P i , V} = ∂ σ P i , (3.74)<br />

i.e. V generates the rigid σ-rotati<strong>on</strong>s. We also have the evoluti<strong>on</strong> equati<strong>on</strong>s<br />

{X i , H} = 1 T P i = ∂ τ X i , {P i , H} = T ∂ 2 σX i = ∂ τ P i . (3.75)


– 33 –<br />

One can also see that the τ- and σ-flows generated by H and V respectively commute<br />

with each other because<br />

{H, V} = 0 .<br />

Impositi<strong>on</strong> of the light-c<strong>on</strong>e gauge is possible because in the c<strong>on</strong>formal gauge the<br />

field X + satisfies the wave equati<strong>on</strong>: □X + = 0. The corresp<strong>on</strong>ding soluti<strong>on</strong> for X +<br />

is<br />

X + (τ, σ) = x + + p+<br />

2πT τ +<br />

√ i ∑<br />

4πT<br />

n≠0<br />

1<br />

(<br />

α n + e −inσ− + ᾱ n + e −inσ+) . (3.76)<br />

n<br />

Our gauge choice (3.58) is taken to be compatible with the form (3.76). Effectively,<br />

it means taken equal to zero all oscillators<br />

α + n = 0 = ᾱ + n , n ≠ 0<br />

and also x + = 0 or<br />

α + n = ᾱ + n = p+<br />

√<br />

4πT<br />

δ n,0 .<br />

One may w<strong>on</strong>der why <strong>on</strong>e cannot completely remove X + , i.e. to choose a gauge<br />

X + = 0. To understand this point <strong>on</strong>e has to remember that the infinitesimal<br />

c<strong>on</strong>formal transformati<strong>on</strong>s are of the form<br />

X → X + ∑ n<br />

a n e inσ− ∂ − X ,<br />

X → X + ∑ n<br />

ā n e inσ+ ∂ + X ,<br />

where a n , ā n are arbitrary c<strong>on</strong>stants. In other words these transformati<strong>on</strong>s can be<br />

written as<br />

X → X + ξ − (σ − ) , X → X + ξ + (σ + ) ,<br />

where ξ ± are arbitrary functi<strong>on</strong>s obeying <strong>on</strong>ly <strong>on</strong>e requirement: they must be periodic<br />

in σ. These functi<strong>on</strong>s can be used to remove all oscillator modes and the zero mode<br />

x + but they cannot remove<br />

p + τ = p+ 2 (τ − σ) + p+ 2<br />

(τ + σ)<br />

because the functi<strong>on</strong>s τ − σ and τ + σ are not periodic in σ.<br />

<strong>String</strong> in the light-c<strong>on</strong>e gauge can be treated in the standard framework of the Hamilt<strong>on</strong>ian reducti<strong>on</strong>.<br />

H = L 0 + ¯L 0 is invariant under the symmetry algebra <strong>on</strong> the surface L m = 0 = ¯L m :<br />

The Hamilt<strong>on</strong>ian<br />

{H, L m} = imL m , {H, ¯L m} = im ¯L m .<br />

The symmetry algebra itself is<br />

{L n, L m} = −i(m − n)L n+m , { ¯L n, ¯L m} = −i(m − n) ¯L n+m .


– 34 –<br />

Orbits of the Virasoro algebra<br />

+<br />

Gauge c<strong>on</strong>diti<strong>on</strong> for X<br />

C<strong>on</strong>strained surface<br />

_<br />

L m =L m =0<br />

Phase space (P,X)<br />

Fig. 2. The physical phase space is obtained by solving the Virasoro c<strong>on</strong>straints<br />

L m = 0 = ¯L m and reducing the acti<strong>on</strong> of the Virasoro algebra <strong>on</strong><br />

the c<strong>on</strong>strained surface by imposing the light-c<strong>on</strong>e gauge.<br />

One would like to reduce the dynamics of the system over the acti<strong>on</strong> of the symmetry algebra.<br />

Hamilt<strong>on</strong>ian reducti<strong>on</strong> c<strong>on</strong>diti<strong>on</strong>s<br />

L m = 0 = ¯L m<br />

In the framework of the<br />

corresp<strong>on</strong>d to fixing the moment map. The reduced phase space P is defined as a quotient space<br />

P = soluti<strong>on</strong>s of L m = 0 = ¯L m<br />

isotropy subalgebra<br />

,<br />

where the isotropy subalgebra is a subalgebra of the Virasoro algebra which leaves the surface L m = 0 = ¯L m invariant. In our present<br />

situati<strong>on</strong> this subalgebra coincides with the algebra itself and, therefore,<br />

P = soluti<strong>on</strong>s of Lm = 0 = ¯L m<br />

acti<strong>on</strong> of Virasoro<br />

.<br />

The acti<strong>on</strong> of the Virasoro algebra is factored out by imposing the light-c<strong>on</strong>e gauge, which simultaneously leads to solving the Virasoro<br />

c<strong>on</strong>straints. The transversal coordinates introduced above provide the descripti<strong>on</strong> of the reduced phase space.<br />

Mass of the string in the light-c<strong>on</strong>e gauge is computed as follows (recall that mass<br />

is a quadratic Casimir of the Poincaré group). Since we have found that p − = 2πT H<br />

p +<br />

we get for the mass<br />

The physical Hamilt<strong>on</strong>ian is<br />

H = 1 2<br />

Thus, we obtain<br />

M 2 = −p µ p µ = −(p i ) 2 + 2p + p − = −(p i ) 2 + 4πT H . (3.77)<br />

∞∑<br />

n=−∞<br />

(<br />

α<br />

i<br />

n α i −n + ᾱ i nᾱ i −n)<br />

=<br />

(p i ) 2<br />

4πT + ∞<br />

∑<br />

n=1<br />

( )<br />

α<br />

i<br />

n α−n i + ᾱnᾱ i −n<br />

i . (3.78)<br />

M 2 = 4πT<br />

∞∑<br />

n=1<br />

(<br />

α<br />

i<br />

n α i −n + ᾱ i nᾱ i −n)<br />

=<br />

2<br />

α ′<br />

∞<br />

∑<br />

n=1<br />

( )<br />

α<br />

i<br />

n α−n i + ᾱnᾱ i −n<br />

i . (3.79)<br />

This clearly shows positivity of M 2 , a property which was not obvious in the c<strong>on</strong>formal<br />

gauge.


– 35 –<br />

Finally, we can write the level-matching c<strong>on</strong>diti<strong>on</strong> in terms of transversal oscillators.<br />

We find<br />

V = 1 ∑ (<br />

)<br />

ᾱ i<br />

2T<br />

nᾱ−n i − αnα i −n<br />

i = 0 . (3.80)<br />

n≠0<br />

Thus, the level-matching c<strong>on</strong>diti<strong>on</strong> tells that the left- and right-moving oscillators<br />

c<strong>on</strong>tribute the same amount of energy.<br />

3.4.2 Poiss<strong>on</strong> structure of the light-c<strong>on</strong>e theory<br />

Using the Poiss<strong>on</strong> brackets of physical fields we can now establish the Poiss<strong>on</strong> relati<strong>on</strong>s<br />

between all quantities of interest. We first summarize the basic Poiss<strong>on</strong> relati<strong>on</strong>s<br />

for the closed string case<br />

{, } p + p − p j x j x − αm j αm<br />

−<br />

p + 0 0 0 0 1 0 0<br />

p − 0 0 0 − pi<br />

− p− 2πiT<br />

mα j 2πiT<br />

p + p + p + m mα − p + m<br />

p i 0 0 0 − δ ij 0 0 0<br />

x i 0<br />

x − − 1<br />

p i<br />

δ ij δ<br />

0 0 ij δ m<br />

p + 4πT<br />

p −<br />

0 0 0 0<br />

p +<br />

α i n 0 − 2πiT<br />

p + nα i n 0 − δij δ n<br />

αn − 0 − 2πiT nα − p + n 0 − αi n<br />

p +<br />

4πT<br />

0 − inδ ij δ n+m − i √ 4πT<br />

− α− n<br />

p +<br />

α i m<br />

p +<br />

α − m<br />

p +<br />

nα i p + n+m<br />

i √ 4πT<br />

p + mα i n+m {α − n , α − m}<br />

Tab. 1. Poiss<strong>on</strong> brackets of the light-c<strong>on</strong>e modes. The variable p − is<br />

essentially the Hamilt<strong>on</strong>ian: p − = 2πT H . The brackets involving ᾱ<br />

p +<br />

variables are the same.<br />

These relati<strong>on</strong>s are easy to derive. For instance,<br />

{p − , x − } = {2πT H p + , x− } = −2πT H 1<br />

(p + ) 2 {p+ , x − } = −2πT H<br />

(p + ) 2 = −p− p + .<br />

Also, <strong>on</strong>e has to remember that the variable α − n c<strong>on</strong>tains the zero mode α i 0<br />

α − n =<br />

√<br />

πT<br />

p +<br />

2αi 0α i n + . . . = pi α i n<br />

p + + . . .<br />

and, therefore,<br />

{x i , α − n } = 1<br />

p + {xi , p j α j n} = 1<br />

p + αi n .<br />

The most complicated bracket is {α − m, α − n }.


– 36 –<br />

Let us outline the computati<strong>on</strong> of this bracket.<br />

{α − m , α− n } = n p<br />

i α<br />

i<br />

m<br />

p +<br />

n p<br />

i α<br />

i<br />

m<br />

p +<br />

, pj α j o<br />

n<br />

p + +<br />

+ √<br />

πT<br />

p +<br />

√<br />

πT<br />

p +<br />

X<br />

k≠m,0<br />

α i m−k αi k , pj α j n<br />

p +<br />

n p<br />

i α<br />

i<br />

m<br />

p + , X<br />

α j o<br />

n−l αj −<br />

l<br />

l≠n,0<br />

√<br />

πT<br />

+ 2πT<br />

−im pi p i<br />

(p + ) 2 δ n+m −2i(m − n)<br />

(p + ) 2 pi α i n+m<br />

| {z } | {z }<br />

first bracket sec<strong>on</strong>d and third brackets<br />

√<br />

πT<br />

+ X<br />

p + α j o<br />

n−l αj =<br />

l<br />

l≠n,0<br />

√<br />

πT n p<br />

i α<br />

i<br />

n<br />

p + p + , X<br />

α j o<br />

m−l αj + πT n X<br />

l<br />

(p<br />

l≠m,0<br />

) 2 α i m−k αi k ,<br />

X<br />

α j o<br />

n−l αj l<br />

k≠m,0<br />

l≠n,0<br />

0<br />

1<br />

X<br />

@−i(m<br />

(p + ) 2 − n)<br />

α i n+m−k αi A<br />

k .<br />

k≠m+n,0<br />

(3.81)<br />

| {z }<br />

forth bracket<br />

Thus, we are getting<br />

0<br />

1<br />

{α − m , α− n } = −im pi p i<br />

(p + ) 2 δ n+m + 2πT<br />

∞X<br />

@−i(m<br />

(p + ) 2 − n) α i n+m−k αi A<br />

k ,<br />

k=−∞<br />

where due to α i 0 = √ pi we combined the sec<strong>on</strong>d and a third terms into <strong>on</strong>e sum. If m + n ≠ 0 then the first term vanishes and we<br />

4πT<br />

can rewrite the last formula as<br />

{α − m , α− n } = √<br />

4πT<br />

p +<br />

<br />

−i(m − n)α − <br />

m+n .<br />

If n = −m then in eq.(3.81) c<strong>on</strong>tributi<strong>on</strong> from the sec<strong>on</strong>d and the third term vanishes and we get<br />

0<br />

1<br />

{α − m , α− −m } = −im pi p i<br />

(p + ) 2 + 2πT @−2im X √ 0 √<br />

(p + ) 2 α i 4πT πT h p i ! 2<br />

−k αi A<br />

k ≡ −2im @<br />

p<br />

k≠0<br />

+ p + √ + X α i k −ki 1<br />

αi A .<br />

4πT<br />

k≠0<br />

Therefore,<br />

√ 0 √<br />

{α − 4πT πT<br />

m , α− −m } = −2im @<br />

p + p +<br />

1<br />

X ∞ α i k αi A<br />

−k .<br />

k=−∞<br />

It is therefore natural to define<br />

α − 0<br />

≡ √<br />

πT<br />

p +<br />

∞ X<br />

k=−∞<br />

α i k αi −k .<br />

With this definiti<strong>on</strong> we obtained a universal formula (valid for all indices m and n):<br />

{α − m , α− n } = √<br />

4πT<br />

p +<br />

<br />

−i(m − n)α − <br />

m+n .<br />

Also we c<strong>on</strong>clude that with this definiti<strong>on</strong><br />

p − = √ πT (α − 0 + ᾱ− 0 ) .<br />

Thus, <strong>on</strong>e finds the following result<br />

{α − m, α − n } =<br />

√<br />

4πT<br />

p + (<br />

−i(m − n)α<br />

−<br />

m+n<br />

)<br />

.<br />

If we introduce<br />

L n = p+<br />

√<br />

4πT<br />

α − n .<br />

we therefore find<br />

{L n , L m } = −i(n − m)L n+m<br />

which is the classical Virasoro algebra! Thus, in the light-c<strong>on</strong>e gauge the Virasoro<br />

algebra is carried over by the l<strong>on</strong>gitudinal oscillators α − n .


– 37 –<br />

3.4.3 Lorentz symmetry<br />

The light-c<strong>on</strong>e gauge manifestly breaks the d-dimensi<strong>on</strong>al Lorentz invariance of the<br />

theory. We have the generators of the Lorentz algebra which in terms of transversal<br />

oscillators become realized n<strong>on</strong>-linearly. For instance, the generator<br />

∞∑<br />

J i− = x i p − − x − p<br />

} {{ }<br />

i −i<br />

l i−<br />

n=1<br />

1 ( ∑ ∞<br />

α<br />

i<br />

n −n αn − − α−nαn) − i − i<br />

n=1<br />

1 (ᾱi<br />

n −n ᾱn − − ᾱ−nᾱn)<br />

− i<br />

is not anymore quadratic in oscillators because p − and α − are n<strong>on</strong>-trivial functi<strong>on</strong><br />

of the transversal oscillators.<br />

In spite of rather n<strong>on</strong>-linear realizati<strong>on</strong> <strong>on</strong>e can check that all the Poiss<strong>on</strong> relati<strong>on</strong>s<br />

involving J i− are still satisfied (this is also c<strong>on</strong>sequence of the fact that<br />

{L m , J µν } = 0 = {¯L m , J µν } , which means that the Poiss<strong>on</strong> bracket of J µν admits a<br />

reducti<strong>on</strong> <strong>on</strong> the c<strong>on</strong>straint surface L m = 0 = ¯L m ). In particular, from the Poincaré<br />

algebra we must have<br />

{J i− , J j− } = 0 .<br />

Let us check this explicitly. First we have<br />

{l i− , l j− } = {x i p − − x − p i , x j p − − x − p j } =<br />

= {x i p − , x j p − } − {x − p i , x j p − } − {x i p − , x − p j } + {x − p i , x − p j } .<br />

By using the Poiss<strong>on</strong> brackets from Table 1 we obtain<br />

{l i− , l j− } = pi<br />

p + p− x j − pj<br />

p + p− x i −p i x j p−<br />

p + + x− p − δ ij<br />

| {z }<br />

first bracket<br />

| {z }<br />

sec<strong>on</strong>d bracket<br />

+p j x i p−<br />

p + − x− p − δ ij<br />

| {z }<br />

third bracket<br />

.<br />

and the forth bracket vanishes. Thus,<br />

{l i− , l j− } = 0 .<br />

C<strong>on</strong>sider the Poiss<strong>on</strong> bracket of the internal spin comp<strong>on</strong>ents<br />

{S i− , S j− } = − X 1<br />

nm {αi −n α− n , αj −m α− m } =<br />

n,m≠0<br />

= − X 1 <br />

{α i <br />

−n<br />

nm<br />

, αj −m }α− n α− m + {αi −n , α− m }α− n αj −m + {α− n , αj −m }αi −n α− m + {α− n , α− m }αi −n αj −m<br />

=<br />

n,m≠0<br />

= iδ ij X α − √<br />

n α− −n 4πT X<br />

<br />

+ i<br />

n p<br />

n≠0<br />

+<br />

− 1 m αi −n+m αj −m α− n + 1 n αi −n αj n−m α− m + n − m<br />

<br />

nm<br />

αi −n αj −m α− m+n .<br />

m,n≠0<br />

Here the first first sum is zero (proved by changing n for −n). In the sec<strong>on</strong>d and the third summ<strong>on</strong>ds <strong>on</strong>e makes the change of<br />

summati<strong>on</strong> indices, n → n + m and m → m + n respectively. After this change these terms cancel exactly against the last <strong>on</strong>e.<br />

However, there are terms which are still left, these are the terms in the sec<strong>on</strong>d and third summ<strong>on</strong>ds c<strong>on</strong>taining zero modes, i.e. terms<br />

for which −n + m = 0 and also the term of the last summ<strong>on</strong>d for which m + n = 0 . Thus,<br />

{S i− , S j− } = − i<br />

√<br />

X 1 <br />

p + p i α j −n<br />

n<br />

− pj α i −n<br />

α − 4πT<br />

n + 2i X 1<br />

p<br />

n≠0<br />

+ α− 0<br />

n αi n αj −n . (3.82)<br />

n≠0<br />

Analogously, we will get the following c<strong>on</strong>tributi<strong>on</strong> from the left-moving modes Then we compute<br />

{l i− , S j− } = −i X 1<br />

n {xi p − − x − p i , α j −n<br />

| {z } | {z }<br />

α− n } = −i X 1<br />

n B<br />

α j n≠0<br />

n≠0 @<br />

−n p− αin<br />

p + + xi − 2πiT<br />

p + nαj −n α− n + 2πiT<br />

<br />

p + nαj −n α− n<br />

A B<br />

| {z }<br />

from A<br />

0<br />

− p i α j α − n<br />

−n<br />

p +<br />

| {z }<br />

from B<br />

1<br />

.<br />

C<br />

A


– 38 –<br />

Thus, we arrive at<br />

{l i− , S j− } + {S i− , l j− } = −2i p− X 1<br />

p + n αi n αj −n +<br />

i X 1 <br />

p<br />

n≠0<br />

+ p i α j −n<br />

n<br />

− pj α i <br />

−n α − n . (3.83)<br />

n≠0<br />

Now summing up equati<strong>on</strong>s (3.82) and (3.83) we obtain<br />

{l i− , S j− } + {S i− , l j− } + {S i− , S j− } = i 4√ πT<br />

p +<br />

<br />

α − 0 − α− 0 + ᾱ− X<br />

0<br />

2<br />

n≠0<br />

1<br />

n αi n αj −n . (3.84)<br />

At first glance this expressi<strong>on</strong> is n<strong>on</strong>-zero and it can not be compensated by the c<strong>on</strong>tributi<strong>on</strong> of left-moving modes<br />

{l i− , ¯S j− } + { ¯S i− , l j− } + { ¯S i− , ¯S j− } = i 4√ πT<br />

p +<br />

<br />

ᾱ − 0 − α− 0 + ᾱ− X<br />

0<br />

2<br />

n≠0<br />

1<br />

n ᾱi nᾱj −n . (3.85)<br />

However, we have to invoke the level-matching c<strong>on</strong>straint which simply tells that α − 0 = ᾱ− 0<br />

and makes both eqs.(3.84) and (3.85)<br />

separately vanish. Thus, we have indeed shown that the most n<strong>on</strong>-trivial relati<strong>on</strong> {J i− , J j− } = 0 is indeed satisfied.<br />

4. Quantizati<strong>on</strong> of bos<strong>on</strong>ic string<br />

4.1 Remarks <strong>on</strong> can<strong>on</strong>ical quantizati<strong>on</strong><br />

According to the standard principles of quantum mechanics can<strong>on</strong>ical quantizati<strong>on</strong><br />

c<strong>on</strong>sists in replacing the Poiss<strong>on</strong> brackets of the fundamental phase space variables<br />

by commutators<br />

{ , } → 1<br />

i [ , ] ,<br />

where is the Plank c<strong>on</strong>stant. Thus, we c<strong>on</strong>sider now X(σ, τ) and P (σ, τ) as the<br />

quantum mechanical operators which obey the following commutati<strong>on</strong> relati<strong>on</strong>s 7<br />

[X µ (σ, τ), X ν (σ ′ , τ)] = [P µ (σ, τ), P ν (σ ′ , τ)] = 0 ,<br />

[X µ (σ, τ), P ν (σ ′ , τ)] = iη µν δ(σ − σ ′ ) , (4.1)<br />

These commutati<strong>on</strong> relati<strong>on</strong>s induce the commutati<strong>on</strong> relati<strong>on</strong>s <strong>on</strong> the Fourier coefficients<br />

[α µ m, α ν n] = [ᾱ µ m, ᾱ ν n] = mδ m+n η µν ,<br />

[α µ m, ᾱ ν n] = 0 , (4.2)<br />

[x µ , p ν ] = iη µν .<br />

For the case of open string the modes ᾱ n are absent. In what follows we will work<br />

in units in which = 1, so that the commutati<strong>on</strong> relati<strong>on</strong>s read as<br />

[α µ m, α ν n] = [ᾱ µ m, ᾱ ν n] = mδ m+n η µν ,<br />

[α µ m, ᾱ ν n] = 0 , (4.3)<br />

[x µ , p ν ] = iη µν .<br />

7 Here the indices µ, ν run from 0 to d − 1.


– 39 –<br />

To restore , <strong>on</strong>e has to simply rescale the modes as α → 1 √<br />

~<br />

α, ᾱ → 1 √<br />

~ ᾱ.<br />

Let us take m > 0 and rescale a m = 1 √ m<br />

α m . Then using the hermiticity property<br />

the commutati<strong>on</strong> relati<strong>on</strong>s can be rewritten as<br />

[a µ n, a †ν<br />

m] = η µν δ n,m<br />

[ā µ n, ā †ν<br />

m] = η µν δ n,m<br />

which are the standard commutati<strong>on</strong> relati<strong>on</strong>s of two infinite sets of independent<br />

quantum harm<strong>on</strong>ic oscillators.<br />

Introduce the number operator for m’s mode<br />

Then we see that for m > 0<br />

From here we c<strong>on</strong>clude that<br />

N m =: α µ mα −m,µ :<br />

[N m , α m ] = −mα m ,<br />

[N m , α −m ] = mα −m .<br />

• Modes with m > 0 should be identified with the lowering operators<br />

• Modes with m < 0 should be identified with the raising operators<br />

C<strong>on</strong>structi<strong>on</strong> of the representati<strong>on</strong> of the can<strong>on</strong>ical commutati<strong>on</strong> relati<strong>on</strong>s is completed<br />

by introducing the ground state which satisfies the following properties<br />

α µ m|p ν 〉 = 0 ,<br />

ˆp µ |p ν 〉 = p µ |p ν 〉 . (4.4)<br />

The whole infinite-dimensi<strong>on</strong>al (Hilbert) space of states is obtained by acting <strong>on</strong> the<br />

ground state with creati<strong>on</strong> operators.<br />

This c<strong>on</strong>structi<strong>on</strong> brings us to the major problem of can<strong>on</strong>ical quantizati<strong>on</strong>.<br />

C<strong>on</strong>sider for m positive the following commutator<br />

[α 0 m, α 0 −m] = [α 0 m, (α 0 m) † ] = mη 00 = −m .<br />

Thus, we induce from here (let even a ground state carries zero momentum p µ )<br />

〈0|[α 0 m, (α 0 m) † ]|0〉 = 〈0|α 0 m (α 0 m) † |0〉 = ||(α 0 m) † |0〉|| 2 = −m < 0 .<br />

Thus, Minkowskian type of the target-space metric leads to the existence in the<br />

Hilbert space the states with negative norm. States with negative norm are sometimes<br />

called “ghosts” and they do not allow for probability interpretati<strong>on</strong> of the<br />

corresp<strong>on</strong>ding quantum-mechanical system.


– 40 –<br />

One can correctly anticipate that the problem with negative norm states arose<br />

because we did not take into account the Virasoro c<strong>on</strong>straints. The covariant approach<br />

to quantizati<strong>on</strong> c<strong>on</strong>sists in defining the subspace of physical states in the<br />

original Hilbert space which obey the Virasoro c<strong>on</strong>straints. One can further show<br />

that in a special dimensi<strong>on</strong> of space-time (d = 26) the negative norm states decouple<br />

from the physical Hilbert space.<br />

In the classical theory we have the c<strong>on</strong>straints L m = 0 = ¯L m . However, in<br />

quantum theory expressi<strong>on</strong>s for L m and ¯L m are quadratic in oscillators and might<br />

involve operators (quantum oscillators) which do not commute with each other! From<br />

all L m a c<strong>on</strong>straint which suffers from ordering ambiguity is L 0 as<br />

L 0 = 1 2<br />

∞∑<br />

n=−∞<br />

α µ nα −n,µ (4.5)<br />

and oscillators α n µ and α µ −n do not commute with each other. The standard way<br />

to deal with this ambiguity in quantum field theory is to use the normal ordering<br />

prescripti<strong>on</strong><br />

L m = 1 2<br />

∞∑<br />

n=−∞<br />

The normal ordering prescripti<strong>on</strong> means that<br />

: α m1 . . . α mk := α n1 . . . α np<br />

} {{ }<br />

all creati<strong>on</strong><br />

: α µ m−nα n,µ : . (4.6)<br />

α s1 . . . α sr<br />

} {{ }<br />

all annihilati<strong>on</strong><br />

in the operators are ordered in such a fashi<strong>on</strong> that all annihilati<strong>on</strong> operators are<br />

put <strong>on</strong> the right from all the creati<strong>on</strong> operators. The order of the creati<strong>on</strong> (or<br />

annihilati<strong>on</strong>) operators between themselves does not matter because these operators<br />

commute between themselves and therefore their expressi<strong>on</strong> does not have ordering<br />

ambiguity. In particular, for L 0 we have<br />

L 0 = 1 2 α2 0 +<br />

∞∑<br />

α −nα µ n,µ − a , (4.7)<br />

n=1<br />

where we include a so far unknown normal ordering c<strong>on</strong>stant a. As to the zero modes,<br />

the normal-ordering prescripti<strong>on</strong> here is<br />

: p µ x ν := x ν p µ .<br />

Since the ground state obeys ˆp µ |p〉 = p µ |p〉 it can be regarded as the usual quantummechanical<br />

eigenstate of the momentum operator ˆp µ = −i ∂<br />

∂x µ<br />

which is in the momentum<br />

representati<strong>on</strong> has the form of the plane-wave<br />

|p〉 ≡ e ipµ x µ<br />

|0〉 ,


– 41 –<br />

where <strong>on</strong> the r.h.s. p µ is not an operator and |0〉 denotes the zero-momentum ground<br />

state. Plane-waves are not square-integrable functi<strong>on</strong>s but they form a basis of a<br />

generalized Hilbert space and they are assumed to be normalized as<br />

〈p|p ′ 〉 = δ(p − p ′ ) .<br />

4.1.1 Virasoro algebra<br />

Let us now investigate the algebra of the operators L m in the quantum case. We<br />

first assume that the normal ordering c<strong>on</strong>stant a = 0. We start by computing<br />

∞∑<br />

∞∑<br />

)<br />

[α m, µ L n ] = 1 [α µ 2 m, : αpα ν n−p,ν :] =<br />

(mδ 1 2<br />

m+p α µ n−p + mα pδ µ m+n−p = mα µ n+m .<br />

Then we have<br />

p=−∞<br />

[L m , L n ] = 1 2<br />

p=−∞<br />

∞∑<br />

[: α pα µ m−p,µ :, L n ] .<br />

p=−∞<br />

We write down the normal ordering explicitly (to simplify the notati<strong>on</strong> we write the<br />

Lorentz summati<strong>on</strong> index <strong>on</strong> the same level)<br />

[L m , L n ] = 1 2<br />

= 1 2<br />

+ 1 2<br />

0∑<br />

[α pα µ m−p, µ L n ] + 1 2<br />

p=−∞<br />

0∑<br />

p=−∞<br />

∞∑<br />

p=1<br />

pα µ p+nα µ m−p<br />

} {{ }<br />

p=q−n<br />

∞∑<br />

[α m−pα µ p, µ L n ] =<br />

p=1<br />

+(m − p)α µ pα µ m−p+n<br />

(m − p)α µ m−p+nα µ p + pα µ m−pα µ n+p<br />

} {{ }<br />

p=q−n<br />

In the underbraced terms we make a change of summati<strong>on</strong> index p = q − n and get<br />

[L m , L n ] = 1 ( 0∑<br />

(m − p)α µ<br />

2<br />

pα µ m+n−p +<br />

+<br />

p=−∞<br />

∞∑<br />

(m − p)α m+n−pα µ p µ +<br />

p=1<br />

n∑<br />

(q − n)α q µ α µ m+n−q<br />

q=−∞<br />

+∞∑<br />

q=n+1<br />

(q − n)α µ m+n−qα µ q<br />

Without loss of generality we assume that n > 0. then we have<br />

[L m , L n ] = 1 ( 0∑<br />

(m − n)α µ<br />

2<br />

pα µ m+n−p +<br />

+<br />

p=−∞<br />

∞∑<br />

p=n+1<br />

(m − n)α µ m+n−pα µ p +<br />

n∑<br />

q=1<br />

q=1<br />

(q − n) α µ q α µ m+n−q<br />

} {{ }<br />

.<br />

not ordered!<br />

)<br />

.<br />

n∑<br />

)<br />

(m − q)α m+n−qα µ q<br />

µ .


– 42 –<br />

Using α µ q α µ m+n−q = α µ m+n−qα µ q + qδ m+n δ µ µ = α µ m+n−qα µ q + qdδ m+n , where d is the<br />

dimensi<strong>on</strong> of the Minkowskian space-time where string propagates. Thus, the algebra<br />

relati<strong>on</strong> is<br />

[L m , L n ] = 1 2<br />

∞∑<br />

(m − n) : α pα µ µ m+n−p : + d 2 δ n+m<br />

p=−∞<br />

n∑<br />

(q 2 − nq) .<br />

q=1<br />

Since<br />

n∑<br />

q=1<br />

q 2 = 1 n∑<br />

n(n + 1)(2n + 1) ,<br />

6<br />

q=1<br />

q = 1 n(n + 1) .<br />

2<br />

<strong>on</strong>e finds the final result<br />

[L m , L n ] = (m − n)L m+n + d 12 m(m2 − 1)δ m+n<br />

which is the famous Virasoro algebra. We see that it is different from the classical<br />

Virasoro (Wit) algebra by the presence of the central term.<br />

In a more general setting the algebra is written as<br />

[L m , L n ] = (m − n)L m+n + c<br />

12 m(m2 − 1)δ m+n ,<br />

where the c<strong>on</strong>stant term c is known as the central charge.<br />

If we introduce a normal ordering c<strong>on</strong>stant a by shifting the definiti<strong>on</strong> of L m as<br />

L m → L m − δ m,0 then the linear term in m in the central term changes<br />

[L m , L n ] = (m − n)L m+n +<br />

( c<br />

12 m3 + ( 2a − c ) )<br />

m δ m+n .<br />

12<br />

We see that the central term has an invariant meaning and cannot be removed for<br />

all L m by adjusting the normal ordering c<strong>on</strong>stant a.<br />

Finally, we comment <strong>on</strong> the relati<strong>on</strong> to semiclassics. If we restore the Plank the<br />

algebra relati<strong>on</strong>s take the form<br />

[L m , L n ] = (m − n)L m+n + 2 c<br />

12 m(m2 − 1)δ m+n .<br />

We see that <strong>on</strong>e can define the Poiss<strong>on</strong> bracket<br />

1<br />

{L m , L n } = lim<br />

~→0 i [L m, L n ] = −i(m − n)L m+n ,<br />

which coincides with the Wit algebra. The central term obviously vanishes in the<br />

semi-classical limit.


– 43 –<br />

4.1.2 Virasoro c<strong>on</strong>straints in quantum theory<br />

At first sight it seems that the natural analog of the classical equati<strong>on</strong>s L n = 0 = ¯L n<br />

in quantum theory is to require that a physical state should be annihilated by all<br />

Virasoro generators:<br />

L n |Φ〉 = 0 = ¯L n |Φ〉 , n ∈ Z .<br />

Due to the normal-ordering ambiguity in the definiti<strong>on</strong> of L 0 in quantum theory the<br />

classical c<strong>on</strong>diti<strong>on</strong>s L 0 = 0 = ¯L 0 are replaced now by<br />

(L 0 − a)|Φ〉 = 0 , (¯L 0 − ā)|Φ〉 = 0 , (4.8)<br />

where in fact the normal-orderings c<strong>on</strong>stants a and ā must be equal to each other 8 and<br />

L 0 , ¯L 0 are understood as the normal-ordered generators. It is easy to see, however,<br />

that eqs.(4.8) cannot be c<strong>on</strong>sistently imposed for all m. Indeed, if eqs.(4.8) would<br />

be satisfied for all m we would have<br />

[L n , L −n ]|Φ〉 = 2nL 0 |Φ〉 + d 12 n(n2 − 1)|Φ〉 ,<br />

i.e.<br />

0 =<br />

(<br />

2na + d )<br />

12 n(n2 − 1) |Φ〉 for any n .<br />

This is obviously not possible to satisfy unless |Φ〉 = 0. The physical reas<strong>on</strong> for<br />

impossibility to impose in quantum theory the same set of c<strong>on</strong>straints as in the<br />

classical <strong>on</strong>e is an anomaly. Because of the anomaly term the first-class Virasoro<br />

c<strong>on</strong>strains of the classical theory turn up<strong>on</strong> quantizati<strong>on</strong> into the c<strong>on</strong>straints of the<br />

sec<strong>on</strong>d class!<br />

From the experience with the quantum electrodynamics <strong>on</strong>e can try to impose <strong>on</strong>ly<br />

“half” of the c<strong>on</strong>straints, i.e.<br />

(L 0 − a)|Φ〉 = 0 ,<br />

L n |Φ〉 = 0 , n > 0 . (4.9)<br />

The c<strong>on</strong>jugate state then obeys 〈Φ|L −n = 0 for n > 0 and we see that 〈Φ|L n |Φ〉 for<br />

all n ≠ 0, i.e. expectati<strong>on</strong> values of L n vanish for all n<strong>on</strong>negative n.<br />

Let us recall that the mass operator is obtained from the c<strong>on</strong>straint L 0 − a = 0, We<br />

have<br />

∞∑<br />

M 2 = −p 2 = 4πT (−a + N) , N = α −nα µ n,µ .<br />

8 This follows from the c<strong>on</strong>straint (L 0 − ¯L 0 )|Φ〉 = 0.<br />

n=1


– 44 –<br />

Here N is the number operator. It turns out that all eigenvalues of the number<br />

operator N are n<strong>on</strong>-negative. Indeed,<br />

N =<br />

∞∑ (<br />

∑d−1<br />

− α−nα 0 n 0 +<br />

n=1<br />

i=1<br />

α i −nα i n<br />

We see the “-” sign coming from the time-like oscillators. However, the time-like<br />

oscillators themselves provide <strong>on</strong>ly n<strong>on</strong>-negative c<strong>on</strong>tributi<strong>on</strong> to N, because for any<br />

m > 0<br />

∞∑<br />

∞∑<br />

[N, a 0 −m] = − [α−nα 0 n, 0 a 0 −m] = − α−n[α 0 n, 0 a 0 −m] = mα−m<br />

0<br />

n=1<br />

since commutator of two time-like oscillators c<strong>on</strong>tributes with the negative sign.<br />

Thus, time-like creati<strong>on</strong> operators c<strong>on</strong>tribute positively to N.<br />

n=1<br />

)<br />

.<br />

Virasoro primaries, descendents and physical states<br />

Let us introduce the following useful definiti<strong>on</strong>s.<br />

1. States which are annihilated by all positively moded Virasoro operators and<br />

are eigenstates of the operator L 0 with an eigenvalue a are called Virasoro<br />

primaries. Number a is called a weight of the Virasoro primary.<br />

2. A Virasoro descendent of a given primary is a state that can be written as a<br />

finite linear combinati<strong>on</strong> of products of negatively moded Virasoro operators<br />

acting <strong>on</strong> the primary state.<br />

3. A state which is both primary and descendent is called a null state.<br />

If |Φ〉 is a primary state then L −1 |Φ〉 is its descendent. If N|Φ〉 = N Φ |Φ〉 then<br />

NL −1 |Φ〉 = (N Φ + 1)L −1 |Φ〉.<br />

There are two basis descendents with the number N Φ + 2, namely L −2 |Φ〉 and<br />

L −1 L −1 |Φ〉. The counting of descendents changes at N Φ + 3. Here the candidate<br />

descendents are<br />

L −3 |Φ〉 , L −2 L −1 |Φ〉 , L −1 L −2 |Φ〉 , L 3 −1|Φ〉 .<br />

The sec<strong>on</strong>d and the third states are not identical because the Virasoro operators do<br />

not commute. However, due to the Virasoro algebra there is <strong>on</strong>e relati<strong>on</strong> between<br />

the above states<br />

L −1 L −2 = [L −1 , L −2 ] + L −2 L −1 = L −3 + L −2 L −1 .<br />

Thus, there are <strong>on</strong>ly three descendents with number N Φ + 3.


– 45 –<br />

In general, for any fixed level N Φ + n, <strong>on</strong>e can choose an independent basis of<br />

descendents of the form<br />

L −n1 L −n2 · · · L −nk |Φ〉 , where n 1 ≥ n 2 ≥ . . . ≥ n k and<br />

k∑<br />

n i = n .(4.10)<br />

i=1<br />

It is a c<strong>on</strong>venti<strong>on</strong>al ordering of the Virasoro operators. Note that the number of<br />

descendents in this basis is equal to a number of partiti<strong>on</strong>s of integer n.<br />

For any given primary, there can be a linear combinati<strong>on</strong> of descendents (4.10)<br />

which vanishes. The vanishing of such combinati<strong>on</strong>s happens not due to the Virasoro<br />

algebra relati<strong>on</strong>s but rather due to specific properties of the primary state |Φ〉. For<br />

instance, <strong>on</strong>e can realize that for the zero-momentum ground (primary) state |0〉 the<br />

descendent L −1 |0〉 vanishes identically.<br />

An important property of the descendents is that they are all orthog<strong>on</strong>al to any<br />

primary. Indeed, a descendent |des〉 can be written as L −ni |χ〉 for some n i > 0 and<br />

some state |χ〉. Then for any primary state<br />

〈des|prime〉 = 〈χ|L ni |prime〉 = 0 ,<br />

since a primary |prime〉 is annihilated by all positively moded oscillators.<br />

Null states, which are both primary and descendents, corresp<strong>on</strong>d to pure gauge<br />

degrees of freedom. Any null state has a vanishing norm and it is orthog<strong>on</strong>al to any<br />

primary and to any descendent. If we alter a primary state by adding a null state,<br />

then the new primary state has the same inner products with all primary states as the<br />

original <strong>on</strong>e. Adding null states to primaries cannot change any physical expectati<strong>on</strong><br />

value. This motivates the following definiti<strong>on</strong> of a physical state:<br />

A physical state is an equivalence class of a primary<br />

state with the weight a = 1 modulo the null states.<br />

The choice a = 1 will be motivated by studying the quantizati<strong>on</strong> of strings in the<br />

physical (light-c<strong>on</strong>e) gauge. Note that here we talk about equivalent classes precisely<br />

because of the ambiguity created by the null states. Primary states which differ by a<br />

null state are physically indistinguishable. In the next secti<strong>on</strong> studying the spectrum<br />

of open string we will see that the null states are indeed resp<strong>on</strong>sible for the gauge<br />

degrees of freedom.<br />

4.1.3 The spectrum<br />

The classical strings cannot provide a reas<strong>on</strong>able particle physics because the masses<br />

of string states take c<strong>on</strong>tinuous values. Only the ground state is massless in the<br />

classical open string theory but because it carries no spin we are not able to identify<br />

it with phot<strong>on</strong>. Quantizati<strong>on</strong> procedure – this is what alters the nature of the classical


– 46 –<br />

string spectrum and makes it discrete. Simultaneously, states to be identified with<br />

phot<strong>on</strong> emerge in the quantum spectrum because of the downward shift of the M 2<br />

producing thereby the massless states with a proper spin labels.<br />

C<strong>on</strong>sider open strings. The Hamilt<strong>on</strong>ian is<br />

The basis vectors of the Hilbert space are<br />

H = L 0 − 1 = α ′ p 2 + N − 1 .<br />

|ψ〉 =<br />

∞∏ ∏25<br />

n=1 µ=0<br />

(α µ†<br />

n ) λ n,µ<br />

|p〉 ,<br />

are n<strong>on</strong>-negative integers. Generic state |ψ〉 is not physical. A physical state is a<br />

state which obeys the Virasoro c<strong>on</strong>straints (4.9) and is not a descendent.<br />

Let us look for some examples of physical states. The first <strong>on</strong>e is the ground state<br />

|p〉. The <strong>on</strong>ly n<strong>on</strong>-trivial c<strong>on</strong>straint is<br />

(L 0 − a)|p〉 = (α ′ p 2 − a)|p〉 = 0 .<br />

Since p 2 = −M 2 we get that the <strong>on</strong>-shell c<strong>on</strong>diti<strong>on</strong> for this state is M 2 = − a α ′ . Later<br />

<strong>on</strong> studying the light-c<strong>on</strong>e quantizati<strong>on</strong> we find that the normal-ordering c<strong>on</strong>stant<br />

a must be equal to <strong>on</strong>e. Thus, the mass-squared of the ground state is negative:<br />

M 2 = − 1 α ′ . The corresp<strong>on</strong>ding hypothetic particle moving faster than light is called<br />

tachy<strong>on</strong>.<br />

The next state to c<strong>on</strong>sider is ζ µ α µ −1|p〉. We have<br />

(L 0 − 1)ζ µ α µ −1|p〉 = (α ′ p 2 + N − 1)ζ µ α µ −1|p〉 = α ′ p 2 ζ µ α µ −1|p〉 = 0<br />

from which we deduce the <strong>on</strong>-shell c<strong>on</strong>diti<strong>on</strong> p 2 = 0, i.e. the corresp<strong>on</strong>ding particle<br />

is massless. Further c<strong>on</strong>diti<strong>on</strong> gives<br />

L 1 ζ µ α µ −1|p〉 = (α 0 α 1 + α −1 α 2 + · · · )ζ µ α µ −1|p〉 = √ 2α ′ ζ µ p µ |p〉 = 0 .<br />

Thus, for a physical state the momentum p µ and the polarizati<strong>on</strong> vector ζ µ must be<br />

related as ζ µ p µ = 0 which is nothing else as the Lorentz gauge c<strong>on</strong>diti<strong>on</strong>. All higher<br />

Virasoro modes L n , n ≥ 2 are automatically annihilate the state. We, however, have<br />

not described the physical state completely. The massless vector particle which is<br />

phot<strong>on</strong> must have d−2 independent polarizati<strong>on</strong>s while the Lorentz gauge lives d−1<br />

polarizati<strong>on</strong>s <strong>on</strong>ly. We should now recall that physical states are defined modulo the<br />

null states.<br />

C<strong>on</strong>sider a state (κ is any c<strong>on</strong>stant)<br />

|d〉 =<br />

κ √<br />

2α<br />

′ L −1|p〉 = κp µ α µ −1|p〉 , p 2 = 0 .


– 47 –<br />

This state has the same form as before with ζ µ = κp µ (l<strong>on</strong>gitudinally polarized<br />

phot<strong>on</strong>s). It is physical, because ζ µ p µ = κp 2 = 0. On the other hand, it is null as it<br />

appears to be a descendent of L −1 . Thus, the states<br />

ζ µ and ζ µ + κp µ<br />

should be identified. The reas<strong>on</strong> for this identificati<strong>on</strong> can be understood as follows.<br />

If ζ µ p µ = 0 then (ζ µ + κp µ )p µ = 0 as well, because p 2 = 0. As the result, this<br />

identificati<strong>on</strong> reduces the number of independent polarizati<strong>on</strong>s to d − 2, as it should<br />

be for a massless phot<strong>on</strong>. Indeed, for any vector ζ µ obeying ζ µ p µ = 0, <strong>on</strong>e can use<br />

the shift freedom ζ µ → ζ µ + κp µ to put, e.g., <strong>on</strong>e of the comp<strong>on</strong>ents of ζ µ to zero.<br />

4.1.4 Propagators<br />

Here we introduce the c<strong>on</strong>cept of propagator or, equivalently, the two-point Green<br />

functi<strong>on</strong>.<br />

C<strong>on</strong>sider first the right-moving fields of the closed string. Their propagator is<br />

defined as<br />

〈X R (τ, σ)X R (τ ′ , σ ′ )〉 = T ( X R (τ, σ)X R (τ ′ , σ ′ ) ) − : X R (τ, σ)X R (τ ′ , σ ′ ) : , (4.11)<br />

where T means time-ordering prescripti<strong>on</strong>. Thus,<br />

T ( X R (τ, σ)X R (τ ′ , σ ′ ) ) =<br />

{<br />

XR (τ, σ)X R (τ ′ , σ ′ ) , for τ > τ ′ ,<br />

X R (τ ′ , σ ′ )X R (τ, σ) , for τ < τ ′ .<br />

For τ > τ ′ we have<br />

〈X R (τ, σ)X R (τ ′ , σ ′ )〉 =<br />

1<br />

=<br />

2 xµ + pµ<br />

i X 1<br />

(τ − σ) + √<br />

4πT 4πT n αµ n e−in(τ−σ) 1<br />

2 xν + pν<br />

4πT (τ ′ − σ ′ i X 1<br />

) + √<br />

n≠0<br />

4πT m αµ m e−im(τ′ −σ ′ ) <br />

m≠0<br />

1<br />

− :<br />

2 xµ + pµ<br />

i X 1<br />

(τ − σ) + √<br />

4πT 4πT n αµ n e−in(τ−σ) 1<br />

n≠0<br />

Most of the terms cancelled and we are left with<br />

Thus,<br />

2 xν + pν<br />

4πT (τ ′ − σ ′ ) +<br />

〈X R (τ, σ)X R (τ ′ , σ ′ )〉 = 1<br />

8πT xµ p ν (τ ′ − σ ′ ) + 1<br />

8πT pµ x ν (τ − σ)<br />

i<br />

√<br />

4πT<br />

− 1<br />

8πT : xµ p ν : (τ ′ − σ ′ ) − 1<br />

8πT : pµ x ν : (τ − σ)<br />

− 1 X X 1<br />

4πT<br />

nm (αµ n αν m − : αµ n αν m :)e−in(τ−σ) e −im(τ′ −σ ′ ) .<br />

n≠0 m≠0<br />

X 1<br />

m αµ m e−im(τ′ −σ ′ ) : .<br />

m≠0<br />

〈X R (τ, σ)X R (τ ′ , σ ′ )〉 = 1<br />

8πT [pµ , x ν ](τ − σ) − 1 X X 1<br />

4πT n>0 m


– 48 –<br />

Thus, for τ > τ ′ <strong>on</strong>e obtains the propagators 9<br />

〈X R (τ, σ)X R (τ ′ , σ ′ )〉 = ηµν<br />

8πT<br />

〈X L (τ, σ)X L (τ ′ , σ ′ )〉 = ηµν<br />

8πT<br />

〈X R (τ, σ)X L (τ ′ , σ ′ )〉 = − ηµν<br />

8πT ln z ,<br />

where we made an identificati<strong>on</strong><br />

ln z −<br />

ηµν<br />

4πT ln(z − z′ ) ,<br />

ln ¯z −<br />

ηµν<br />

4πT ln(¯z − ¯z′ ) ,<br />

z = e i(τ−σ) , ¯z = e i(τ+σ) .<br />

Computati<strong>on</strong> for the case of open string is similar. For τ > τ ′ we have<br />

〈X(τ, σ)X(τ ′ , σ ′ )〉 = −i ηµν<br />

πT τ + 1<br />

πT<br />

Performing the sum <strong>on</strong>e finds<br />

[ (<br />

〈X(τ, σ)X(τ ′ , σ ′ )〉 = − ηµν<br />

log<br />

4πT<br />

+ log<br />

∞∑<br />

n=1<br />

e iτ − e −i(σ−σ′) e iτ ′) + log<br />

(<br />

e iτ − e −i(σ+σ′) e iτ ′) + log<br />

1<br />

n e−inτ+inτ ′ cos nσ cos nσ ′ .<br />

4.1.5 Vertex operators. Tachy<strong>on</strong> scattering amplitude<br />

(e iτ − e i(σ−σ′) iτ ′)<br />

e<br />

(<br />

e iτ − e i(σ+σ′) e iτ ′)] .<br />

Here we approach for the first time the questi<strong>on</strong> about string interacti<strong>on</strong>s. It is<br />

important to realize that the situati<strong>on</strong> here is different to what <strong>on</strong>e usually accounters<br />

in QFT. The interacti<strong>on</strong> of strings cannot be introduced by adding n<strong>on</strong>-linear terms<br />

to the string Lagrangian; in the latter case <strong>on</strong>e would obtain n<strong>on</strong>-linear interacting<br />

theory but still of a single string.<br />

Out<br />

split string<br />

<strong>on</strong> a mass−shell<br />

Out<br />

V<br />

Inserti<strong>on</strong><br />

of a local operator<br />

emmiting a particle<br />

In<br />

In<br />

Fig. 3. Open (closed) strings interact by means of joining and splitting.<br />

Emissi<strong>on</strong> of a point particle <strong>on</strong> mass-shell is represented by inserti<strong>on</strong> of a<br />

local vertex operator.<br />

9 Rigorous justificati<strong>on</strong> of these formulae requires an introducti<strong>on</strong> of an IR regularizati<strong>on</strong>, because<br />

correlati<strong>on</strong> functi<strong>on</strong>s of the massless field in two-dimensi<strong>on</strong>s suffer from IR divergencies.


– 49 –<br />

<strong>String</strong> interacti<strong>on</strong>s are introduced by allowing a topology of a wold-sheet to change.<br />

If a split string is <strong>on</strong>-shell it can be thought as an infinite collecti<strong>on</strong> of particles<br />

which form its spectrum. In the limiting case of single emitted particle the process<br />

of scattering can be viewed as applicati<strong>on</strong> to an initial state |In〉 of a local vertex<br />

operator V ≡ V (τ, σ) which depends <strong>on</strong> the emitted state and transforms |In〉 into<br />

the outgoing state |Out〉:<br />

|Out〉 = V |In〉 .<br />

Thus, to any physical state |Φ〉 from the spectrum <strong>on</strong>e can put in corresp<strong>on</strong>dence a<br />

local vertex operator V Φ .<br />

C<strong>on</strong>formal operators<br />

Operator A(τ) is called c<strong>on</strong>formal with the c<strong>on</strong>formal dimensi<strong>on</strong> ∆ if 10<br />

A ′ (τ ′ ) =<br />

( dτ<br />

dτ ′ ) ∆A(τ)<br />

.<br />

Under infinitezimal variati<strong>on</strong> τ → τ ′ = τ + ɛ(τ) <strong>on</strong>e gets<br />

With ɛ = −ie imτ<br />

δA(τ) = −ɛ dA<br />

dτ<br />

− ∆A<br />

dɛ<br />

dτ<br />

[L m , A(τ)] = e imτ (−i∂ τ + m∆)A(τ)<br />

If the operator A is expandable as A(τ) = ∑ A n e −inτ , for its Fourier modes the last<br />

relati<strong>on</strong> implies<br />

[L m , A n ] = (m(∆ − 1) − n)A n+m<br />

Thus, if A has c<strong>on</strong>formal weight ∆ = 1 then it’s zero mode commutes with all L m .<br />

Therefore, A 0 maps physical states into physical states:<br />

|Φ ′ 〉 = A 0 |Φ〉 .<br />

An alternative way to understand this is to notice that for ∆ = 1 we have<br />

[L m , A(τ)] = e imτ (−i∂ τ + m∆)| ∆=1 A(τ) = −i ∂ ( )<br />

e imτ A(τ) ,<br />

∂τ<br />

i.e. the r.h.s. is the total derivative. Thus, A 0 = ∫ dτA(τ) will transform as<br />

∫<br />

[L m , A 0 ] = −i dτ ∂ ( )<br />

e imτ A(τ) ,<br />

∂τ<br />

10 This transformati<strong>on</strong> law can be written as<br />

A(τ ′ )(dτ ′ ) ∆ = A(τ)(dτ) ∆ .<br />

For ∆ = 1 this implies that A(τ)dτ is a <strong>on</strong>e-form.


where the expressi<strong>on</strong> <strong>on</strong> the r.h.s. is specified by the periodicity/boundary properties<br />

of A(τ).<br />

Explicit example of a vertex operator<br />

C<strong>on</strong>sider the following operator acting in the Hilbert space of open string<br />

√<br />

V (k, τ, σ) = e 1 P ∞<br />

kµα µ −n<br />

πT n=1 e n inτ cos nσ e ik µ(x µ + pµ<br />

πT τ) e − √ 1 P ∞ kµα µ n<br />

πT n=1 n e−inτ cos nσ .<br />

At σ = 0 this simplifies to<br />

V (k, τ) = e 1 √<br />

πT<br />

P ∞<br />

n=1<br />

kµα µ −n<br />

e n<br />

inτ<br />

e ikµ(xµ + pµ<br />

πT<br />

} {{ τ)<br />

} e − 1 P<br />

√ ∞ kµα µ n<br />

πT n=1 n<br />

} {{ e−inτ<br />

} .<br />

V − V 0 V +<br />

} {{ }<br />

This operator is “almost” normal-ordered and can be c<strong>on</strong>cisely written as<br />

V (k, τ) = V − V 0 V +<br />

Here <strong>on</strong>ly zero modes entering V 0 are not normal-ordered. Indeed, according to our<br />

c<strong>on</strong>venti<strong>on</strong>s : p µ x ν := x ν p µ , the operator p µ should be always <strong>on</strong> the right from x µ<br />

which is not the case for V 0 .<br />

The normal-ordering of V 0 can be achieved with the help of the Baker-Campbell-<br />

Hausdorff formula 11<br />

Thus, we have<br />

: e ik µx µ e i k µp µ<br />

πT τ := e ik µx µ e i k µp µ<br />

e A e B = e A+B+ 1 2 [A,B]<br />

πT τ = e ik µx µ +i k µp µ<br />

πT<br />

Recalling that [x µ , p ν ] = iη µν we therefore find that<br />

τ− τ<br />

2πT k µk ν [x µ ,p ν ]<br />

e ik µx µ +i k µp µ<br />

πT τ = e iα′ k 2 τ<br />

: e ik µx µ e i k µp µ<br />

πT τ := e iα′ k 2 τ<br />

: V 0 :<br />

Thus, we obtain completely normal-ordered vertex operator<br />

V = e iα′ k 2τ V − : V 0 : V +<br />

We would like to investigate the transformati<strong>on</strong> properties of this operator under<br />

c<strong>on</strong>formal transformati<strong>on</strong>s. To this end we need to compute by using the Leibnitz<br />

rule<br />

e −iα′ k 2τ [L m , V ] = [L m , V − ] : V 0 : V + + V − [L m , : V 0 :]V + + V − : V 0 : [L m , V + ] .<br />

11 Suppose you forgot an exact coefficient in fr<strong>on</strong>t of the commutator term. Write the operator<br />

identity in the form e A e B = e A+B+α[A,B] with the forgotten coefficient α. Rescale A and B with a<br />

small parameter ɛ to get e ɛA e ɛB = e ɛA+ɛB+ɛ2 α[A,B] . Now expand both sides up to order ɛ 2 keeping<br />

the order of operators. You will find that fulfilment of the operator relati<strong>on</strong> at order ɛ 2 will require<br />

to fix α = 1 2 . – 50 –


– 51 –<br />

Assuming for definiteness that m > 0 it is not difficult to find 12<br />

[L m , V − ] = 1<br />

2 √ πT<br />

∞∑<br />

p=1<br />

[L m , : V 0 :] =: V 0 : (α mk)<br />

√<br />

πT<br />

[L m , V + ] = 1<br />

2 √ πT<br />

∑−1<br />

p=−∞<br />

e ipτ [<br />

V − (kα m−p ) + (kα m−p )V −<br />

]<br />

e ipτ [<br />

V + (kα m−p ) + (kα m−p )V +<br />

]<br />

=<br />

∞∑<br />

p=1<br />

e −ipτ<br />

√<br />

πT<br />

: V + (kα m+p ) :<br />

Here the first commutator is of particular importance because it still c<strong>on</strong>tains the<br />

terms which are not normal-ordered. Indeed, it can be written in the form<br />

[L m , V − ] = 1 √<br />

πT<br />

∞<br />

∑<br />

p=1,p≠m<br />

e ipτ : (kα m−p )V − :<br />

+ 1 √<br />

πT<br />

e imτ V − (kα 0 ) + 1<br />

2 √ πT<br />

m−1<br />

∑<br />

p=1<br />

e ipτ [kα m−p , V − ] .<br />

Further we get<br />

[kα m−p , V − ] =<br />

k2<br />

√<br />

πT<br />

e i(m−p)τ V −<br />

This leads to<br />

[L m , V − ] = 1 √<br />

πT<br />

∞<br />

∑<br />

p=1,p≠m<br />

e ipτ : (kα m−p )V − :<br />

+ 1 √<br />

πT<br />

e imτ V − (kα 0 ) + k2<br />

2πT<br />

m−1<br />

∑<br />

e imτ V − .<br />

p=1<br />

} {{ }<br />

(m−1)e imτ V −<br />

12 The calculati<strong>on</strong> is as follows:<br />

[L m, V − ] = 1 +∞ X<br />

[α µ m−n<br />

2<br />

αn,µ, V −] = 1 +∞ X<br />

[α µ m−n<br />

n=−∞<br />

2<br />

, V −]α n,µ + α µ m−n [αn,µ, V −] .<br />

n=−∞<br />

The two terms <strong>on</strong> the r.h.s. are computed separately, for instance<br />

1<br />

+∞ X<br />

[α µ m−n<br />

2<br />

, V −]α n,µ = 1 m−1 X<br />

[α µ 1 P<br />

m−n<br />

n=−∞<br />

2<br />

, e √ ∞p=1<br />

kν α ν −p<br />

e πT p<br />

ipτ 1<br />

∞X m−1<br />

]α n,µ =<br />

n=−∞<br />

2 √ X<br />

[α µ m−n<br />

πT<br />

, αν −p ]<br />

k ν<br />

p=1 n=−∞ | {z } p eipτ V − α n,µ<br />

n=m−p<br />

=<br />

1<br />

∞X<br />

2 √ e ipτ V − (k µ α m−p,µ ) .<br />

πT p=1<br />

The other terms are computed in a similar way.


– 52 –<br />

Plugging everything together we find<br />

e −iα′ k 2τ [L m , V ] =<br />

∞∑<br />

p=1,p≠m<br />

e ipτ<br />

√<br />

πT<br />

: (kα m−p )V − V 0 V + :<br />

} {{ }<br />

+ √ eimτ<br />

: V − V 0 (kα 0 )V + : + √ eimτ<br />

: V − [kα 0 , V 0 ]V + : +α ′ k 2 (m − 1)e imτ : V − V 0 V + :<br />

} πT<br />

{{ } πT<br />

+ √ 1<br />

∞∑ e −ipτ<br />

: V − V 0 V + (kα m ) : + √ : V − V 0 V + (kα m+p ) :<br />

} πT<br />

{{ } p=1 πT<br />

} {{ }<br />

Here the underlined terms are nicely combined with a single sum 13 with the range of<br />

summati<strong>on</strong> variable p form −∞ to +∞ and if we further take into account that<br />

we will get<br />

It remains to note that<br />

[kα 0 , V 0 ] =<br />

[L m , V ] = e imτ e iα′ k 2 τ<br />

∞∑<br />

p=−∞<br />

k2<br />

√<br />

πT<br />

V 0<br />

e −ipτ<br />

√<br />

πT<br />

: V − V 0 V + (kα p ) :<br />

+ α ′ k 2 (m + 1)e imτ e iα′ k 2τ : V − V 0 V + :<br />

} {{ }<br />

V<br />

−i∂ τ V = α ′ k 2 V + e iα′ k 2 τ<br />

∞∑<br />

p=−∞<br />

e −ipτ<br />

√<br />

πT<br />

: V − V 0 V + (kα p ) :<br />

With the account of this formula we obtain<br />

( )<br />

[L m , V ] = e imτ − i∂ τ + α ′ k 2 m V<br />

and, therefore, we c<strong>on</strong>clude that the operator V has the following c<strong>on</strong>formal dimensi<strong>on</strong><br />

∆:<br />

∆ = α ′ k 2 .<br />

In particular, for k 2 = 1 the c<strong>on</strong>formal dimensi<strong>on</strong> ∆ = 1 and the vertex operator<br />

α ′<br />

we discuss corresp<strong>on</strong>ds to emissi<strong>on</strong> of the tachy<strong>on</strong> with the mass m 2 = − 1 .<br />

α ′<br />

We also see that <strong>on</strong> the zero-momentum ground state<br />

V (k, 0)|0〉 = V − e ik µx µ |0〉 .<br />

13 It is c<strong>on</strong>venient to shift the summati<strong>on</strong> variable p for p → p − m.


– 53 –<br />

As we will discuss later <strong>on</strong>, <strong>on</strong>e can perform an analytic c<strong>on</strong>tinuati<strong>on</strong> τ → −iτ under<br />

which the exp<strong>on</strong>ent entering the vertex operator V − will transform as e iτ → e τ . Then<br />

in the limit τ → −∞ we see that V − → 1 and in the Euclidean picture<br />

lim V (k, 0)|0〉 =<br />

τ→−∞ eik µx µ |0〉 ,<br />

i.e. at τ → −∞ this vertex operator creates a particle with the momentum k µ .<br />

Normal-ordering the product of exp<strong>on</strong>ents<br />

C<strong>on</strong>sider the normal product : e X :: e Y :, where X and Y are two operators with the<br />

propagator 〈XY 〉. Clearly <strong>on</strong>e gets<br />

: e X :: e Y :=<br />

∞∑<br />

n,m=0<br />

: X n : : Y m :<br />

.<br />

n! m!<br />

To apply the Wick theorem we first calculate the number of ways we can pick up k<br />

X’s from X n n!<br />

, which is obviously . Analogously, the number of ways to pick<br />

k!(n−k)!<br />

up k Y ’s from Y m m!<br />

is . Now we have to pair (i.e. to form propagators) k fields<br />

k!(m−k)!<br />

X with k fields Y<br />

: X } .{{ . . X}<br />

:: Y } .{{ . . Y}<br />

: .<br />

k<br />

k<br />

The are k! ways to pair all the terms in the last expressi<strong>on</strong>. Thus, applicati<strong>on</strong> of the<br />

Wick theorem gives<br />

: e X :: e Y :=<br />

=<br />

∞∑<br />

n,m=0<br />

min(n,m)<br />

∑<br />

k=0<br />

Thus, we find<br />

∞∑<br />

n,m=0<br />

:<br />

min(n,m)<br />

∑<br />

k=0<br />

X n−k<br />

: Xn−k<br />

n!<br />

Y m−k<br />

(n − k)! (m − k)!<br />

Y m−k<br />

m!<br />

:<br />

:<br />

〈XY 〉k<br />

k!<br />

n! m!<br />

k!(n − k)! k!(m − k)! k! 〈XY 〉k =<br />

=<br />

∞∑ 〈XY 〉 k<br />

k=0<br />

: e X :: e Y :=: e 〈XY 〉+X+Y :<br />

k!<br />

∞∑<br />

n,m=k<br />

:<br />

X n−k<br />

Y m−k<br />

(n − k)! (m − k)! :<br />

It is now easy to see that the last formula can be generalized for the case of several<br />

vertex operators as follows<br />

∏<br />

: e X i<br />

:= e P i


– 54 –<br />

where we assume that τ 1 > τ 2 . By using the formula (4.12) we find<br />

〈V (k 1 , τ 1 )V (k 2 , τ 2 )〉 = e iα′ k 2 1 τ 1+iα ′ k 2 2 τ 2<br />

e k 1 k 2<br />

πT log(eiτ 1−e iτ 2) 〈0| : V (k 1 , τ 1 )V (k 2 , τ 2 ) : |0〉 .<br />

The vacuum expectati<strong>on</strong> value of the normal-ordered expressi<strong>on</strong> <strong>on</strong> the right hand<br />

side reduces to the c<strong>on</strong>tributi<strong>on</strong> of the zero modes <strong>on</strong>ly and we get<br />

〈V (k 1 , τ 1 )V (k 2 , τ 2 )〉 = e iα′ k1 2τ 1+iα ′ k2 2τ 2<br />

e 2α′ k 1 k 2 log(e iτ 1−e iτ2) 〈0|e i(kµ 1 +kµ 2 )x µ<br />

|0〉 .<br />

} {{ }<br />

δ(k 1 +k 2 )<br />

Thus, the two-point functi<strong>on</strong> is n<strong>on</strong>-zero <strong>on</strong>ly if k 1 = k = −k 2 . In this case we get<br />

We thus find that<br />

Four-tachy<strong>on</strong> scattering amplitude<br />

〈V (k, τ 1 )V (−k, τ 2 )〉 = eiα′ k 2 (τ 1 +τ 2 )<br />

(e iτ 1 − e<br />

iτ 2) 2α ′ k 2 .<br />

〈V (k, τ 1 )V (−k, τ 2 )〉 = ei∆(τ 1+τ 2 )<br />

(e iτ 1 − e<br />

iτ 2) 2∆ .<br />

Here we compute the scattering amplitude of four tachy<strong>on</strong>ic particles. This is the<br />

famous Veneziano amplitude which subsequently led to discovery of string theory.<br />

The amplitude is defines as<br />

A =<br />

∫ ∞<br />

0<br />

dτ 〈k 4 |V (k 3 , τ)V (k 2 , 0)|k 1 〉<br />

Here 〈k 4 | is understood in an unusual way 〈k 4 | = 〈0|e ikµ 4 x µ<br />

. Using the definiti<strong>on</strong> of<br />

the tachy<strong>on</strong>ic vertex operators and the formula (4.12) <strong>on</strong>e finds<br />

V (k 3 , τ)V (k 2 , 0) = e iα′ k 2 3 τ : V (k 3 , τ) :: V (k 2 , 0) :<br />

= e iα′ k 2 3 τ e −kµ 3 kν 2 〈X µ(τ)X ν (0)〉 : V (k 3 , τ)V (k 2 , 0) :<br />

Recalling the open string propagator at σ = σ ′ = 0:<br />

( )<br />

〈X(τ)X(0)〉 = − ηµν<br />

πT log e iτ − 1 .<br />

Thus we find<br />

A =<br />

∫ ∞<br />

Further simplificati<strong>on</strong> gives<br />

A =<br />

=<br />

=<br />

0<br />

∫ ∞<br />

0<br />

∫ ∞<br />

0<br />

∫ ∞<br />

0<br />

dτ e iα′ k 2 3 τ e 2α′ (k 2 k 3 ) log(e iτ −1) 〈k 4 |e i(kµ 2 +kµ 3 )x µ<br />

e i2α′ k µ 3 p µτ |k 1 〉<br />

dτ e iα′ k3 2τ (e iτ − 1) 2α′ (k 2 k 3 ) e 2iα′ (k 3 k 1 )τ δ ( ∑<br />

4 )<br />

k i<br />

i=1<br />

dτ e iα′ k3 2τ e 2iα′ (k 1 +k 2 )k 3 τ (1 − e −iτ ) 2α′ (k 2 k 3 ) δ ( ∑<br />

4 )<br />

k i<br />

i=1<br />

dτ e −iα′ k3 2τ e −2iα′ (k 3 k 4 )τ (1 − e −iτ ) 2α′ (k 2 k 3 ) δ ( ∑<br />

4 )<br />

k i .<br />

i=1


– 55 –<br />

Recall that for tachy<strong>on</strong>s <strong>on</strong> shell we have ki 2 = 1 . Performing now the Wick rotati<strong>on</strong><br />

α ′<br />

τ → −iτ and changing the integrati<strong>on</strong> variable τ for x = e −τ we find<br />

A =<br />

∫ 1<br />

0<br />

dx x 2α′ k 3 k 4<br />

(1 − x) 2α′ k 2 k 3<br />

,<br />

where for the sake of simplicity we omitted the δ-functi<strong>on</strong> which encodes the c<strong>on</strong>servati<strong>on</strong><br />

law of the momenta. Introducing the Mandelstam variables s = (k 1 + k 2 ) 2<br />

and t = (k 2 + k 3 ) 2 the last formula can be cast in the form<br />

A =<br />

∫ 1<br />

0<br />

dx x α′ s−2 (1 − x) α′ t−2 = Γ(α′ s − 1)Γ(α ′ t − 1)<br />

Γ(α ′ (s + t) − 2)<br />

In fact this functi<strong>on</strong> is known as the Euler beta-functi<strong>on</strong>. One of the interesting<br />

properties of the representati<strong>on</strong> of A in terms of the Euler beta-functi<strong>on</strong> is that the<br />

latter is explicitly symmetric under the interchange of s and t. Search of amplitudes<br />

with this symmetry property led Veneziano in 1960’s to this amplitude which was<br />

the starting point of modern string theory.<br />

It is interesting to analyze the Veneziano formula in more detail. The Γ-functi<strong>on</strong> has<br />

poles at n<strong>on</strong>-positive integers with residues<br />

Γ(x) → (−1)n<br />

n!<br />

1<br />

x + n<br />

as x → −n , n ≥ 0 .<br />

Thus, when α ′ s → 1 − n, n = 0, 1, . . . the amplitude behaves as<br />

A(s, t) → (−1)n<br />

n!<br />

1 Γ(α ′ t − 1)<br />

α ′ s − 1 + n Γ(α ′ t − 1 − n)<br />

Here the dependents of the variable t is polynomial because for n > 0 we have<br />

Γ(α ′ t − 1) Γ(w + n)<br />

Γ(α ′ = = (w + n − 1) · · · (w + 1)w<br />

}<br />

t −<br />

{{<br />

1 − n<br />

}<br />

) Γ(w)<br />

w<br />

= (α ′ t − 2)(α ′ t − 3) · · · (α ′ t − n − 1) ≡ P n (α ′ t) ,<br />

i.e. the r.h.s. is a polynomial of degree n. Thus, the scattering amplitude can be<br />

essentially written as<br />

∞∑ (−1) n P n (α ′ t)<br />

A(s, t) =<br />

n! n − 1 + α ′ s , P 0(α t) = 1 .<br />

n=0<br />

In scattering theory res<strong>on</strong>ances (or simply poles) of the scattering amplitude are<br />

interpreted as an exchange by intermediate particles whose masses are obtained from<br />

the c<strong>on</strong>diti<strong>on</strong> of having poles. In our case we see that the poles arise due to exchange<br />

by hypothetical particles whose masses are quantized as<br />

Mn 2 = −s = 1 (n − 1) .<br />

α<br />

′<br />

.


– 56 –<br />

It also follows from the scattering theory that appearance of a polynomial of degree<br />

n as the residue of the amplitude signals that the exchanged particles of the mass<br />

M 2 n carry spin up to the maximal value J max = n. Our later analysis of string in the<br />

physical gauge will reveal that the exchanged particles delivering the res<strong>on</strong>ances of<br />

the Veneziano amplitude are those from the spectrum of open string!<br />

Operator Product Expansi<strong>on</strong><br />

C<strong>on</strong>siderati<strong>on</strong> of the Veneziano amplitude shows that the object of primary interest<br />

in string perturbati<strong>on</strong> theory is a correletati<strong>on</strong> functi<strong>on</strong> of local operators<br />

〈O i1 (x 1 )O i2 (x 2 ) . . . O in (x n )〉 ,<br />

where O i (x) is a local operator. Here x ≡ (τ, σ) is a point <strong>on</strong> the two-dimensi<strong>on</strong>al<br />

world-sheet. It is important to understand the behavior of the correlati<strong>on</strong> functi<strong>on</strong><br />

when two operators are taken to approach each other. The technique to describe this<br />

limit in known as the Operator product Expansi<strong>on</strong> or OPE for short. The Operator<br />

Product Expansi<strong>on</strong> states that a product of two local operators can be approximated<br />

to arbitrary accuracy by a sum of local operators<br />

O i (x)O j (y) = ∑ k<br />

C k ij(x − y, ∂ y )O k (y) .<br />

Let us take as a local operator the stress tensor and try to work out the corresp<strong>on</strong>ding<br />

OPE. Introduce the short-hand notati<strong>on</strong> T ≡ T ++ and X µ ≡ X µ L . C<strong>on</strong>sider<br />

the comp<strong>on</strong>ent T −− of the stress tensor normalized as<br />

T ≡ T −− = 1 α ′ : ∂ − X µ ∂ − X µ := 1 α ′ : ∂ − X ν L∂ − X Lν : .<br />

We will also use the c<strong>on</strong>cise notati<strong>on</strong> z = e i(τ−σ) and w = e i(τ ′ −σ ′) .<br />

In what follows we c<strong>on</strong>sider the product of two stress tensors evaluated at two different<br />

points<br />

T (τ, σ)T (τ ′ , σ ′ ) = 1<br />

α ′2 : ∂ −X µ (z)∂ + X µ (z) :: ∂ − X ν (w)∂ + X ν (w) :<br />

and try to expand it over a basis of local operators. By using the Wick theorem, we get<br />

T (τ, σ)T (τ ′ , σ ′ ) = 1<br />

α ′2 : ∂ −X µ (z)∂ − X µ (z)∂ − X ν (w)∂ − X ν (w) :<br />

+ 4<br />

α ′2 〈∂ −X µ (z)∂ − X ν (w)〉 : ∂ − X µ (z)∂ − X ν (w) :<br />

+ 2<br />

α ′2 〈∂ −X µ (z)∂ − X ν (w)〉 〈∂ − X µ (z)∂ − X ν (w)〉 .


– 57 –<br />

We can now rewrite the r.h.s. by using the propagators introduces above<br />

T (τ, σ)T (τ ′ , σ ′ ) = 1<br />

α ′2 : ∂ −X µ (z)∂ − X µ (z)∂ − X ν (w)∂ − X ν (w) :<br />

+ 4<br />

α ′2 ∂x −∂ y − 〈Xµ (z)X ν (w)〉 : ∂ − X µ (z)∂ − X ν (w) :<br />

+ 2<br />

α ′2 ∂z −∂ w −〈X µ (z)X ν (w)〉 ∂ z −∂ w −〈X µ (z)X ν (w)〉 .<br />

A little computati<strong>on</strong> gives<br />

T (τ, σ)T (τ ′ , σ ′ ) = 1<br />

α ′2 : ∂ −X µ (z)∂ − X µ (z)∂ − X ν (w)∂ − X ν (w) :<br />

− 2 zw<br />

α ′ (z − w) 2 : ∂ −X µ (z)∂ − X µ (w) :<br />

+ ηµν η µν<br />

2<br />

z 2 w 2<br />

(z − w) 4 .<br />

It is further c<strong>on</strong>venient to redefine the stress tensor as follows<br />

so that<br />

Since<br />

we have ∂ ∂z = 1<br />

iz<br />

T (z) =<br />

T (τ, σ)<br />

z 2<br />

1<br />

T (z)T (w) = =<br />

α ′2 z 2 w : ∂ −X µ (z)∂ 2 − X µ (z)∂ − X ν (w)∂ − X ν (w) :<br />

− 2 1<br />

α ′ zw (z − w) : ∂ −X µ (z)∂ 2 − X µ (w) :<br />

+ ηµν η µν 1<br />

2 (z − w) . 4<br />

∂<br />

∂z<br />

∂ − = 1 ( ∂<br />

2 ∂τ − ∂ )<br />

= 1 ( ∂z<br />

∂σ 2 ∂τ − ∂z ) ∂<br />

∂σ ∂z = iz ∂ ∂z ,<br />

and, therefore, the last formula can be written as<br />

T (z)T (w) = = 1<br />

α ′2 : ∂ zX µ (z)∂ z X µ (z)∂ w X ν (w)∂ w X ν (w) :<br />

+ 2 1<br />

α ′ (z − w) : ∂ zX µ (z)∂ 2 w X µ (w) : + ηµν η µν<br />

2<br />

1<br />

(z − w) 4 .<br />

Expanding the r.h.s. around the point w = z, we will find the following most singular<br />

z → w c<strong>on</strong>tributi<strong>on</strong><br />

T (z)T (w) = d/2<br />

(z − w) 4 + 2<br />

(z − w) 2 T (w) + 1<br />

z − w ∂ wT (w) + . . . (4.13)<br />

The first term here reflects the appearance of the c<strong>on</strong>formal anomaly (a purely quantum<br />

mechanical effect). In the general setting the coefficient of this term is c/2,<br />

where c is the central charge. The coefficients 2 of the sec<strong>on</strong>d term coincides with<br />

the c<strong>on</strong>formal dimensi<strong>on</strong> of T .


– 58 –<br />

4.2 Quantizati<strong>on</strong> in the physical gauge<br />

Quantizati<strong>on</strong> of strings in the physical (light-c<strong>on</strong>e) gauge is perhaps the most straightforward<br />

way to obtain a restricti<strong>on</strong> <strong>on</strong> the space-time dimensi<strong>on</strong> as well as to understand<br />

the spectrum of physical excitati<strong>on</strong>s.<br />

Using the Poiss<strong>on</strong> brackets and the basic quantizati<strong>on</strong> rules we can easily get the<br />

table of the basic commutator relati<strong>on</strong>s of the light-c<strong>on</strong>e string theory.<br />

[ , ] p + p − p j x j x − αm j αm<br />

−<br />

p + 0 0 0 0 i 0 0<br />

p − 0 0 0 − i pi<br />

p + − i p−<br />

p + − 2πT mα j p + m − 2πT mα − p + m<br />

p i 0 0 0 − iδ ij 0 0 0<br />

x i 0 i pi<br />

iδ ij 0 0 i δij δ m<br />

p + 4πT<br />

x − − i i p−<br />

0 0 0 0 i α− p + m<br />

p +<br />

αn i 2πT<br />

0 nα i p + n 0 − i δij δ n<br />

0 nδ ij √<br />

δ 4πT<br />

4πT n+m p +<br />

αn − 2πT<br />

0 nα − p + n 0 − i αi n<br />

p +<br />

− i α− n<br />

p +<br />

i αi m<br />

p +<br />

nα i n+m<br />

− √ 4πT<br />

p + mα i n+m [α − n , α − m]<br />

Tab. 2. Can<strong>on</strong>ical structure of the light-c<strong>on</strong>e modes. The variable<br />

p − is essentially the Hamilt<strong>on</strong>ian: p − = 2πT H . The commutators<br />

p +<br />

involving ᾱ variables are the same.<br />

We would like to point out the following commutator<br />

√<br />

4πT<br />

[αn, i αm] − =<br />

p + nαi n+m .<br />

One of the most important commutators of the light-c<strong>on</strong>e theory is [αm, − αn − ]. It can<br />

be computed precisely in the same way as [L m , L n ] of the previous secti<strong>on</strong>. We find<br />

the same result as before except we have now <strong>on</strong>ly d − 2 transversal fields which<br />

c<strong>on</strong>tribute to the central charge term with the factor d − 2 instead of d<br />

[α − m, α − n ] =<br />

√<br />

4πT<br />

p +<br />

The normal ordering ambiguity<br />

(m − n)α− m+n + 4πT<br />

(p + ) 2 d − 2<br />

12 m(m2 − 1)δ m+n . (4.14)<br />

α − n → α − n −<br />

√<br />

4πT<br />

p + aδ n,0<br />

leads to the change as<br />

√<br />

4πT<br />

[αm, − αn − ] =<br />

p (m − + n)α− m+n + 4πT ( d − 2<br />

(p + ) 2 12 m3 + 2am − d − 2 )<br />

12 m δ m+n .


– 59 –<br />

4.2.1 Lorentz symmetry and critical dimensi<strong>on</strong><br />

Studying the classical string in the light-c<strong>on</strong>e gauge we realized that the generators<br />

J i− of the Lorentz symmetry become rather complicated functi<strong>on</strong>s of transversal<br />

oscillators<br />

J i− = x i p − − x − p i − i<br />

∞∑<br />

n=1<br />

1 ( ∑ ∞<br />

α<br />

i<br />

n −n αn − − α−nαn) − i 1 (ᾱi )<br />

− i<br />

n −n ᾱn − − ᾱ−nᾱ − n<br />

i .<br />

n=1<br />

We would like to ask a questi<strong>on</strong> whether we can use this expressi<strong>on</strong> in quantum<br />

theory regarding α n and ᾱ n as operators to define the quantum Lorentz generators?<br />

Due to the unusual can<strong>on</strong>ical structure of the light-c<strong>on</strong>e theory it is not obvious that<br />

c<strong>on</strong>sistent Lorentz generators should exist.<br />

In quantum theory we want to realize a unitary representati<strong>on</strong> of the Poincaré<br />

group and therefore, we require the Lorentz generators to be hermitian, i.e.<br />

(J µν ) † = J µν ,<br />

where we treat J µν as an operator acting in the Hilbert space. Also the Lorentz<br />

generators must be normal-ordered to have the well-defined acti<strong>on</strong> <strong>on</strong> the vacuum<br />

state. C<strong>on</strong>sider the following ansatz for the Lorentz generators J i− , which are the<br />

most intricate generators to be defined in quantum theory,<br />

J i− = 1 ∞∑<br />

2 (xi p − + p − x i ) − x − p i −i<br />

} {{ } n=1<br />

l i−<br />

1 ( ∑ ∞<br />

α<br />

i<br />

n −n αn − − α−nαn) − i − i<br />

n=1<br />

1 (ᾱi )<br />

n −n ᾱn − − ᾱ−nᾱ − n<br />

i .<br />

One can see that these generators are hermitian and normal-ordered so they can<br />

be c<strong>on</strong>sidered as candidates to realize the Lorentz algebra symmetry. The latter<br />

requirement is equivalent to<br />

[J i− , J j− ] = 0 .<br />

This is an equati<strong>on</strong> we would like to prove.<br />

First we discuss the orbital part. We have<br />

[l i− , l j− ] = [ 1 2 (xi p − + p − x i ) − x − p i , 1 2 (xj p − + p − x j ) − x − p j ]<br />

= 1 4 [xi p − , x j p − ] → i 4 (xj p i − x i p j ) p−<br />

p +<br />

+ 1 4 [xi p − , p − x j ] → i 4 (pi x j − x i p j ) p−<br />

p +<br />

−<br />

1 2 [xi p − , x − p j ] → − i 2 (x− p − δ ij − x i p j p−<br />

p + )<br />

+ 1 4 [p− x i , x j p − ] → −<br />

4<br />

i p −<br />

p + (pj x i − x j p i )<br />

+<br />

4 1 [p− x i , x − p j ] → −<br />

4<br />

i p −<br />

p + (pj x i − p i x j )<br />

−<br />

−<br />

−<br />

1 2 [p− x i , x − p j ] → i 2 ( p−<br />

p + pj x i − p − x − δ ij )<br />

1<br />

2 [x− p i , x j p − ] → − i 2 (xj p i p−<br />

p + − x− p − δ ij )<br />

1<br />

2 [x− p i , p − x j ] → − i 2 ( p−<br />

p + xj p i − x − p − δ ij ) .


– 60 –<br />

Here by arrow we indicated the explicit expressi<strong>on</strong> for the corresp<strong>on</strong>ding commutator. Reducing similar terms we get<br />

[l i− , l j− ] =<br />

4 i [pi , x j ] p−<br />

p + + 4<br />

i p −<br />

p + [pi , x j ] +<br />

2 i [x− , p − ]δ ij = 0 .<br />

Thus, the generators of the orbital part of the total momentum have vanishing commutator. Next we compute<br />

[l i− , S j− ] + [S i− , l j− ] = −2 p− X ∞ 1<br />

p +<br />

n=1 n α[i −n αj] n + (4.15)<br />

+ 1<br />

p +<br />

∞ X<br />

1<br />

n=1 n<br />

h<br />

− (α j −n α− n − α− −n αj n )pi + (α i −n α− n − α− −n αi n )pj i .<br />

Here and below we use the c<strong>on</strong>cise notati<strong>on</strong> α [i −n αj] n = α i −n αj n − αj −n αi n .<br />

Now we are in a positi<strong>on</strong> to study the most difficult commutator [S i− , S j− ]. We will do further analysis in several steps.<br />

1. First we c<strong>on</strong>sider the following commutator<br />

Therefore,<br />

This further gives<br />

[S i− , α − ∞ m ] = −i X 1<br />

n n [αi −n α− n − α− −n αi n , α− m ] (4.16)<br />

√ √<br />

∞X 1 4πT<br />

= −i −<br />

n=1 n p + nαi m−n α− n + αi −n [α− n , α− m ] −[α − 4πT<br />

<br />

−n , α− m ]αi n − p + nα− −n αi n+m .<br />

| {z }<br />

A<br />

[S i− , α − m ] = −i √<br />

4πT<br />

p +<br />

− i f(m)<br />

m αi m .<br />

X ∞ 1 <br />

− nα i <br />

m−n<br />

n=1 n<br />

α− n + αi −n (n − m)α− m+n + (n + m)α− m−n αi n − nα− −n αi n+m<br />

[S i− , α − m ] = −i √<br />

4πT<br />

+ i<br />

p +<br />

√<br />

4πT<br />

p +<br />

X ∞ <br />

α i −n α− m+n − αi m−n α− n + α − m−n αi n −α − <br />

−n αi n+m<br />

n=1<br />

| {z } | {z }<br />

A<br />

B<br />

X ∞ m<br />

n=1 n<br />

<br />

α i <br />

−n α− m+n − α− m−n αi n − i f(m)<br />

m αi m .<br />

The terms in the first line of the last equati<strong>on</strong> can be partially cancelled up<strong>on</strong> changing the summati<strong>on</strong> index and we find<br />

that<br />

[S i− , α − m ] = −i √<br />

4πT<br />

+ i<br />

p +<br />

√<br />

4πT<br />

p +<br />

m X<br />

n=1<br />

X ∞ m<br />

n=1 n<br />

<br />

− α i m−n α− n<br />

| {z }<br />

A<br />

+ α − m−n αi n<br />

| {z }<br />

B<br />

<br />

α i <br />

−n α− m+n − α− m−n αi n − i f(m)<br />

m αi m .<br />

<br />

Since we have<br />

mX<br />

α − m−n αi n =<br />

X 0 α − m−1<br />

k αi m−k ≡ X<br />

α − n αi m−n<br />

n=1<br />

k=m−1<br />

n=0<br />

we see that<br />

mX <br />

− α i mX<br />

m−n α− n + α− m−n αi n = − α i m−1<br />

m−n α− n + X<br />

α − n αi m−n<br />

n=1<br />

n=1<br />

n=0<br />

= α − m−1<br />

0 αi n − αi 0 α− m + X<br />

[α − n , αi m−n ]<br />

n=1<br />

Using the fact that P m−1<br />

n=1 (m − n) = 1 m(m − 1) we obtain<br />

2<br />

√ √<br />

[S i− , α − 4πT<br />

4πT<br />

m ] = i p + (αi 0 α− m − α− 0 αi m ) + i X ∞ m <br />

p +<br />

α i <br />

−n<br />

n=1 n<br />

α− m+n − α− m−n αi n<br />

4πT m(m − 1)<br />

+ i<br />

(p + ) 2 − f(m)<br />

!<br />

α i m<br />

2<br />

m<br />

. (4.17)<br />

Thus, under the acti<strong>on</strong> of the Virasoro operators α − m the spin comp<strong>on</strong>ents Si− transform in a n<strong>on</strong>trivial manner. Note that<br />

the r.h.s. of eq.(4.17) is normal-ordered.


– 61 –<br />

2. Analogously, we compute (m > 0)<br />

√ √<br />

[S i− , α − 4πT<br />

4πT<br />

−m ] = i p + (α− −m αi 0 − αi −m α− 0 ) − i X ∞ m <br />

p +<br />

α i <br />

−n<br />

n=1 n<br />

α− n−m − α− −m−n αi n<br />

4πT m(m − 1)<br />

+ i<br />

(p + ) 2 − f(m)<br />

!<br />

α i −m<br />

2<br />

m<br />

.<br />

In fact these formula is also obtained from eq.(4.17) by simply substituting m → −m.<br />

3. The next step c<strong>on</strong>sists in finding the commutator<br />

[S i− , α j −m ] = −i ∞ X<br />

= −i<br />

1 <br />

n=1 n<br />

√<br />

4πT<br />

p +<br />

α i −n [α− n , αj −m ] − [α− −n , αj −m ]αi n − α− −n δij δ n−m<br />

| {z }<br />

0 as i≠j<br />

X ∞ m<br />

n=1 n<br />

<br />

α i <br />

−n αj n−m − αj −n−m αi n .<br />

<br />

4. Finally we compute the commutator<br />

[S i− , α j m ] = i √<br />

4πT<br />

p +<br />

X ∞ m<br />

n=1 n<br />

<br />

α i <br />

−n αj m+n − αj m−n αi n .<br />

Substituting all our findings into the commutator [S i− , S j− ] we obtain<br />

[S i− , S j− ] =<br />

√<br />

4πT<br />

p +<br />

∞ X<br />

1<br />

n=1 n<br />

∞X<br />

−<br />

n=1<br />

√<br />

4πT X ∞ 1<br />

+<br />

p +<br />

m,n=1 n<br />

<br />

− α j <br />

−n α− 0 αi n + αj −n αi 0 α− n + αi −n α− 0 αj n − α− −n αi 0 αj n<br />

4πT n − 1<br />

(p + ) 2 − f(n)<br />

!<br />

2 n 2 α [i −n αj] n<br />

<br />

α j −m (αi −n α− m+n − α− m−n αi n ) − (αi −n αj n−m − αj −m−n αi n )α− m<br />

+ (α i −n α− n−m − α− −n−m αi n )αj m − α− −m (αi −n αj m+n − αj m−n αi n ) <br />

.<br />

We first analyze the last two lines of the equati<strong>on</strong> above, which we write as follows<br />

I ij =<br />

√<br />

4πT X ∞ 1 <br />

p +<br />

α j −m<br />

m,n=1 n<br />

αi −n α− m+n −(α i −n αj n−m<br />

−α j −m−n αi n )α− m<br />

| {z } | {z }<br />

A<br />

A<br />

+ α i −n α− n−m αj m − αj −m α− m−n αi n − α− −n−m αi n αj m −α − −m (αi −n αj m+n<br />

| {z } | {z }<br />

B<br />

B<br />

−α j m−n αi n ) <br />

.<br />

Up<strong>on</strong> changing the summati<strong>on</strong> variables the A-terms can be partially cancelled, the same is for the B-terms. We therefore obtain<br />

I ij =<br />

+<br />

√<br />

4πT X ∞ nX 1<br />

n<br />

p + −<br />

n=1 m=1 n αi −n αj n−m α− m +<br />

X 1<br />

<br />

m=1 n α− −m αj m−n αi n<br />

√<br />

4πT X ∞ <br />

p +<br />

α j <br />

−m−n αi n α− m + αi −n α− n−m αj m − αj −m α− m−n αi n − α− −m αi −n αj m+n<br />

.<br />

n,m=1<br />

One can recognize that in the sec<strong>on</strong>d line of the equati<strong>on</strong> above the first and the last terms are not normal-ordered. We c<strong>on</strong>sider the<br />

first sum, which is not normal-ordered, and try to bring it to the normal-ordered form:<br />

∞X 1<br />

∞<br />

n,m=1 n αj −m−n αi n α− m = X 1<br />

∞<br />

n,m=1 n αj −m−n α− m αi n +<br />

X 1<br />

n,m=1 n αj −m−n [αi n , α− m ]<br />

√<br />

∞X 1<br />

4πT<br />

=<br />

n,m=1 n αj −m−n α− m αi n + X ∞<br />

p +<br />

α j −m−n αi n+m<br />

n,m=1<br />

√<br />

∞X 1<br />

4πT<br />

=<br />

n,m=1 n αj −m−n α− m αi n + X ∞<br />

p + (k − 1)α j −k αi k ,<br />

k=2<br />

where in the last sum we made a substituti<strong>on</strong> k = m + n and then summed over m, n with the c<strong>on</strong>diti<strong>on</strong> m + n = k kept fixed; this<br />

resulted in the factor k − 1. Analogously, we achieve the normal-ordering of the sec<strong>on</strong>d sum<br />

∞X 1<br />

∞ √<br />

n,m=1 n α− −m αi −n αj m+n =<br />

X 1<br />

4πT<br />

n,m=1 n αi −n α− −m αj m+n + X ∞<br />

p + (k − 1)α i −k αj k .<br />

k=2


– 62 –<br />

Thus, our commutator takes the form<br />

I ij =<br />

+<br />

−<br />

√<br />

4πT X ∞ nX 1<br />

n<br />

p + −<br />

n=1 m=1 n αi −n αj n−m α− m +<br />

X 1<br />

<br />

m=1 n α− −m αj m−n αi n<br />

√<br />

4πT X ∞ 1 <br />

p +<br />

α j −m−n<br />

n,m=1 n<br />

α− m αi n + α i −n α− n−m αj m − α j −m α− m−n αi n − α i <br />

−n α− −m αj m+n<br />

| {z } | {z } | {z } | {z }<br />

A<br />

B<br />

A<br />

B<br />

4πT X ∞<br />

(p + ) 2 (n − 1)α [i −n αj] n .<br />

n=1<br />

Again we see that up<strong>on</strong> change of the summati<strong>on</strong> index the A-terms partially cancel (the same is for the B-terms) and we arrive at<br />

I ij =<br />

+<br />

−<br />

√<br />

4πT<br />

p +<br />

√<br />

4πT<br />

p +<br />

X ∞ nX 1 <br />

− α i <br />

−n<br />

n=1 m=1 n<br />

αj n−m α− m + α− −m αj m−n αi n<br />

∞ X<br />

n−1 X<br />

1<br />

n=1 m=0 n<br />

4πT X ∞<br />

(p + ) 2 (n − 1)α [i −n αj] n .<br />

n=1<br />

<br />

− α j <br />

m−n α− −m αi n + αi −n α− m αj n−m<br />

From here we find<br />

I ij =<br />

−<br />

√<br />

4πT X ∞ nX 1 <br />

p +<br />

− α i <br />

−n<br />

n=1 m=1 n<br />

αj 0 α− n + α− −n αj 0 αi n − αj −n α− 0 αi n + αi −n α− 0 αj n<br />

0<br />

1<br />

4πT X ∞ n−1 X n − m<br />

@<br />

A<br />

(p + ) 2 α [i −n<br />

n=1 m=1 n<br />

αj] n −<br />

4πT X ∞<br />

(p + ) 2 (n − 1)α [i −n αj] n .<br />

n=1<br />

| {z }<br />

1<br />

2 (n−1)<br />

Thus, the commutator of the internal spin comp<strong>on</strong>ents we are interested in acquires the form<br />

[S i− , S j− ] =<br />

√<br />

4πT<br />

p +<br />

∞X<br />

−<br />

n=1<br />

X ∞ 1 <br />

− α j −n<br />

n=1 n<br />

α− 0 αi n + αj −n αi 0 α− n + αi −n α− 0 αj n − α− −n αi 0 αj n<br />

− α i <br />

−n αj 0 α− n + α− −n αj 0 αi n − αj −n α− 0 αi n + αi −n α− 0 αj n<br />

!<br />

4πT<br />

f(n)<br />

(p + 2(n − 1) −<br />

) 2 n 2 α [i −n αj] n .<br />

The final step c<strong>on</strong>sists in commuting the factor α − 0<br />

We thus find<br />

[S i− , S j− ] = 2 2√ πT α − 0<br />

p +<br />

+ 1<br />

p +<br />

−<br />

∞∑<br />

n=1<br />

∞<br />

∑<br />

n=1<br />

1<br />

n<br />

<strong>on</strong> the left to compare with eq.(4.15).<br />

∞<br />

∑<br />

1<br />

n α[i −nαn j] +<br />

n=1<br />

]<br />

[(α −nα j n − − α−nα − n)p j i − (α−nα i n − − α−nα − n)p i j<br />

( )<br />

4πT f(n)<br />

2n − α<br />

(p + )<br />

2<br />

n<br />

−nα [i j]<br />

2 n .<br />

Finally, adding this expressi<strong>on</strong> to eq.(4.15) we arrive at<br />

(<br />

[J i− , J j− ] = 2 p − − 2 √ ) 1<br />

πT α0<br />

−<br />

+ 4πT<br />

(p + ) 2 ∞<br />

∑<br />

n=1<br />

∞<br />

∑<br />

p +<br />

n=1<br />

( [d − 2<br />

12 − 2 ]n + 1 n<br />

1<br />

n α[i −nα j]<br />

n +<br />

[<br />

2a − d − 2<br />

12<br />

] ) α [i −nα j]<br />

n + (α n → ᾱ n ).


– 63 –<br />

As for the case of classical string the first term here vanishes due to the level matching<br />

c<strong>on</strong>diti<strong>on</strong> α 0 = ᾱ 0 , while the sec<strong>on</strong>d sum is an anomaly which appears due to n<strong>on</strong>commutativity<br />

of the oscillators in quantum theory. Thus, for general values of a<br />

and d the theory is not Lorentz invariant: the quantum effects destroy the Lorentz<br />

invariance which was present at the classical level. However, for special values<br />

d = 26 , a = 1<br />

the anomaly term vanishes and the Lorentz invariance is restored!<br />

4.2.2 The spectrum<br />

In the light-c<strong>on</strong>e gauge the spectrum is generated by acting with transversal oscillators<br />

<strong>on</strong> the vacuum state. We first discuss the spectrum of open strings.<br />

The mass operator is the light-c<strong>on</strong>e gauge for open strings is<br />

M 2 = 1 α ′<br />

∞<br />

∑<br />

n=1<br />

( )<br />

α−nα i n i − a ,<br />

where, as was discussed in the previous chapter, the normal-ordering c<strong>on</strong>stant a<br />

should be equal to 1 in order to guarantee the Lorentz invariance of the light-c<strong>on</strong>e<br />

theory.<br />

The ground state |p i 〉 carries no oscillators and it has a mass<br />

α ′ M 2 |p i 〉 = −|p i 〉 =⇒ α ′ M 2 = −1.<br />

This is a tachy<strong>on</strong>.<br />

The first excited state is α i −1|p j 〉. It is a d − 2 comp<strong>on</strong>ent vector which transforms<br />

irreducibly under the transverse group SO(24). We see that<br />

α ′ M 2 (α i −1|p i 〉) = (1 − a)α i −1|p i 〉 = 0 ,<br />

i.e. this vector is massless.


– 64 –<br />

level α ′ mass 2 rep of SO(24) little rep of little<br />

group group<br />

0 −1 |0〉<br />

}{{}<br />

1<br />

1 0 α−1|0〉<br />

i<br />

} {{ }<br />

24<br />

SO(1, 24) 1<br />

SO(24) 24<br />

2 +1 α−2|0〉<br />

i<br />

} {{ }<br />

24<br />

α i −1α j −1|0〉<br />

} {{ }<br />

299 s +1<br />

SO(25)<br />

324 s<br />

3 +2 α−3|0〉<br />

i<br />

} {{ }<br />

24<br />

α i −2α j −1|0〉<br />

} {{ }<br />

276 a +299 s +1<br />

α i −1α j −1α k −1|0〉<br />

} {{ }<br />

2576 s +24<br />

SO(25)<br />

2900 s + 300 a<br />

Tab. 3. The spectrum of open bos<strong>on</strong>ic string up to level 3.<br />

In general, the Lorentz invariance requires that physical states transform irreducibly<br />

under the little Lorentz group which is<br />

• SO(d − 2) for massless particles<br />

• SO(d − 1) for massive particles (for tachy<strong>on</strong> SO(1, d − 2))<br />

For tachy<strong>on</strong> the little Lorentz group is n<strong>on</strong>-compact. Unitary representati<strong>on</strong>s of<br />

n<strong>on</strong>-compact groups are either trivial (i.e. <strong>on</strong>e-dimensi<strong>on</strong>al) or infinite-dimensi<strong>on</strong>al.<br />

Tachy<strong>on</strong> realizes the <strong>on</strong>e-dimensi<strong>on</strong>al representati<strong>on</strong>.<br />

Further analysis reveals that all states corresp<strong>on</strong>ding to higher levels are massive<br />

and that being the tensors of SO(24) they combine at any given mass level to representati<strong>on</strong>s<br />

of SO(25), the latter is the little Lorentz group for massive states. This<br />

is highly n<strong>on</strong>-trivial implicati<strong>on</strong> of the Lorentz invariance and it occurs <strong>on</strong>ly in the<br />

critical dimensi<strong>on</strong> and for a = 1!<br />

At level n the mass of the corresp<strong>on</strong>ding states is α ′ M 2 = n − 1. Am<strong>on</strong>g them there<br />

is always a symmetric traceless tensor of rank n. This is a state with maximal spin<br />

J max = n and, therefore, we have J max = n = α ′ M 2 + 1. In general states obey the<br />

inequality<br />

J ≤ α ′ M 2 + 1 .


– 65 –<br />

The values of J and M 2 are quantized and the last inequality implies that the states<br />

lie <strong>on</strong> the Regge trajectories.<br />

level α ′ mass 2 rep of SO(24) little rep of little<br />

group<br />

group<br />

0 −4 |0〉<br />

}{{}<br />

1<br />

SO(1, 24) 1<br />

1 0 α−1ᾱ i −1|0〉<br />

j<br />

} {{ }<br />

299 s+276 a+1<br />

SO(24) 299 s + 276 a + 1<br />

α i −2ᾱ j −2|0〉<br />

} {{ }<br />

299 s +276 a +1<br />

α i −1α j −1ᾱ k −1ᾱ l −1|0〉<br />

} {{ }<br />

299 s +1+299 s +1<br />

20150 s + 32175<br />

2 +4 SO(25)<br />

α i −2ᾱ j −1ᾱ k 1|0〉<br />

} {{ }<br />

(24)×(299+1)<br />

α i −1α j −1ᾱ k −2|0〉<br />

} {{ }<br />

(299+1)×(24)<br />

52026 + 324 s + 300 a + 1<br />

Tab. 4. The spectrum of closed bos<strong>on</strong>ic string up to level 2.<br />

Now we discuss the spectrum of closed strings. The mass operator for closed<br />

strings is<br />

M 2 = 2 ( ∑ ∞ ∞∑<br />

)<br />

α i<br />

α<br />

−nα i ′ n + ᾱ−nᾱ i n i − 2a<br />

n=1<br />

n=1<br />

In additi<strong>on</strong> <strong>on</strong>e has to impose the level-matching c<strong>on</strong>diti<strong>on</strong><br />

V =<br />

∞∑<br />

α−nα i n i −<br />

n=1<br />

∞∑<br />

ᾱ−nᾱ i n i = 0 ,<br />

which simply means that the excitati<strong>on</strong> (level) number of α-oscillators should be<br />

equal to the excitati<strong>on</strong> number of ᾱ-oscillators.<br />

The ground state is a tachy<strong>on</strong> which is scalar particle with α ′ M 2 = −4. The<br />

first excited state α−1ᾱ i −1|0〉 j is massless. It can be decomposed into irreducible<br />

n=1


– 66 –<br />

representati<strong>on</strong>s of the transversal (and simultaneously little) Lorentz group SO(24)<br />

as follows<br />

(<br />

α−1ᾱ i −1|0〉 j = α −1ᾱ [i −1|0〉<br />

j] + α<br />

} {{ }<br />

−1ᾱ (i j)<br />

−1 − 1 )<br />

24 δij α−1ᾱ k −1<br />

k |0〉 + 1<br />

} {{ } 24 δij α−1ᾱ k −1|0〉<br />

k .<br />

} {{ }<br />

276<br />

299<br />

singlet<br />

The massless excitati<strong>on</strong> of spin two transforming in representati<strong>on</strong> 299 of SO(24)<br />

was proposed to be identified with a gravit<strong>on</strong>, the quantum of the gravitati<strong>on</strong>al<br />

interacti<strong>on</strong>. To make this identificati<strong>on</strong> <strong>on</strong>e has to relate the string scale α ′ with the<br />

Planck scale G = M −2<br />

P<br />

, where G is the Newt<strong>on</strong> c<strong>on</strong>stant and M P is the Planck mass:<br />

α ′ = M −2<br />

P .<br />

Since the masses of the massive string modes are proporti<strong>on</strong>al to 1/α ′ = MP 2, these<br />

string excitati<strong>on</strong>s are extremely heavy due to the large value of MP 2 and, by this<br />

reas<strong>on</strong>, they do not show up at the energy scales of the Standard Model.<br />

As in the opens string case the higher massive states of closed string are combined<br />

at a given mass level into representati<strong>on</strong>s of the little Lorentz group SO(25). The<br />

relati<strong>on</strong> between maximal spin and the mass is now<br />

4.3 BRST quantizati<strong>on</strong><br />

J max = α′<br />

2 M 2 + 2 .<br />

The path integral approach proved to be a very useful tool for quantizing the theories<br />

with local (gauge) symmetries. The starting point is the Polyakov acti<strong>on</strong> and a new<br />

BRST (Becchi-Rouet-Stora-Tyutin) symmetry. We know that the induced metric<br />

Γ αβ = ∂ α X µ ∂ β X µ and the intrinsic metric h αβ are related classically through the<br />

c<strong>on</strong>diti<strong>on</strong> T αβ = 0. However, quantum-mechanically this is not the case.<br />

The basic idea is to define the path integral using the Polykov acti<strong>on</strong> and integrate<br />

over<br />

h αβ , X µ<br />

being c<strong>on</strong>sidered as independent variables:<br />

∫<br />

Z = Dh αβ (σ, τ)DX µ (σ, τ)e iSp[X,h]<br />

Due to the gauge invariance the last integral is ill-defined. This occurs because we<br />

integrate infinitely many times over physically equivalent, i.e. related to each other<br />

by gauge transformati<strong>on</strong>s, c<strong>on</strong>figurati<strong>on</strong>s. This can be understood looking at a much<br />

simpler example of the two-dimensi<strong>on</strong>al integral<br />

Z =<br />

∫ +∞ ∫ +∞<br />

−∞<br />

−∞<br />

dxdy e −(x−y)2 .


– 67 –<br />

It is divergent since the integrand depends <strong>on</strong> x − y <strong>on</strong>ly. Also <strong>on</strong>e sees that the<br />

group of translati<strong>on</strong>s<br />

x → x + a, y → y + a<br />

leaves the measure invariant.<br />

x<br />

Orbits x−y=c<strong>on</strong>st<br />

y<br />

x+y=0<br />

gauge slice<br />

intersect each orbit<br />

at <strong>on</strong>e point<br />

Fig. 4. Divergence caused by a presence of the translati<strong>on</strong>al symmetry. One<br />

integrates over the orbits of the gauge group while both the measure and<br />

the integrand are translati<strong>on</strong> invariant.<br />

Let us split the coordinates (x, y) as<br />

( x − y<br />

(x, y) = , y − x )<br />

} 2 {{ 2 }<br />

x−y<br />

2 + y−x<br />

2 =0 +<br />

( x + y<br />

, x + y )<br />

} 2 {{ 2 }<br />

shift by (a,a),<br />

a= x+y<br />

2<br />

This suggests to introduce new coordinates “al<strong>on</strong>g” the gauge orbit and “orthog<strong>on</strong>al”<br />

to it:<br />

(x, y) → (u, v) u = x − y, v = x + y,<br />

i.e. x = u+v and y = u−v . Then the integral takes the form<br />

2 2<br />

Z = 1 ∫ +∞<br />

(∫ +∞<br />

) √ ∫ π +∞<br />

e −u2 du dv = d(x + y) = √ π<br />

2 −∞ −∞<br />

2 −∞<br />

} {{ }<br />

√ π<br />

∫ +∞<br />

−∞<br />

da .<br />

} {{ }<br />

volume<br />

This example illustrates the basic idea to define the path integral – <strong>on</strong>e has to divide<br />

the original Z by the infinite volume of a symmetry group. Thus, our discussi<strong>on</strong><br />

suggests that a proper definiti<strong>on</strong> of the path integral in string theory should be<br />

∫<br />

1<br />

Z =<br />

Dh αβ (σ, τ)DX µ (σ, τ)e iSp[X,h] ,<br />

V Diff V Weyl<br />

where we divided over the infinite volumes of the reparametrizati<strong>on</strong> and Weyl groups.


– 68 –<br />

Let us illustrate this procedure <strong>on</strong> our simplified example. The integral can be rewritten by using the δ-functi<strong>on</strong>:<br />

Z +∞<br />

Z +∞<br />

Z = da e −(x−y)2 <br />

Z +∞<br />

Z +∞<br />

dxdy 2δ(x + y) = da e −z2 dz .<br />

−∞ −∞<br />

−∞ −∞<br />

Here inserti<strong>on</strong> of the δ-functi<strong>on</strong> can be regarded as the gauge-fixing c<strong>on</strong>diti<strong>on</strong>; it selects a single representative from each gauge orbit.<br />

What happens if we change the gauge fixing c<strong>on</strong>diti<strong>on</strong>? Suppose we take another slice “orthog<strong>on</strong>al” to the gauge orbits. Let s will be<br />

a free <strong>on</strong>e-dimensi<strong>on</strong>al parameter <strong>on</strong> this slice<br />

s : f(x, y) = 0 =⇒ (x(s), y(s))<br />

Chose now new coordinates<br />

Then<br />

where x − y = x ′ − y ′ and<br />

(x, y) = (x ′ , y ′ ) +(a, a)<br />

| {z }<br />

al<strong>on</strong>g s<br />

Z +∞<br />

Z +∞<br />

Z +∞<br />

Z +∞<br />

Z =<br />

e −(x−y)2 dxdy =<br />

e −(x′ −y ′ ) 2 dsda|J| ,<br />

−∞ −∞<br />

−∞ −∞<br />

∂(x, y)<br />

J =<br />

∂(s, a)<br />

is the Jacobian. Since x = x ′ + a and y = y ′ + a we have<br />

J =<br />

!<br />

∂x<br />

∂s<br />

∂y<br />

∂x<br />

∂a<br />

∂y =<br />

∂s ∂a<br />

!<br />

∂x<br />

∂s<br />

1<br />

∂y<br />

∂s 1 = ∂x<br />

∂s − ∂y<br />

∂s = ∂ ∂s (x′ − y ′ ) .<br />

Al<strong>on</strong>g s<br />

0 = ∂f ∂x<br />

∂x ∂s + ∂f ∂y<br />

∂y ∂s = ∂f ∂x ′<br />

∂x ∂s<br />

+ ∂f ∂y ′<br />

∂y ∂s ,<br />

which allows for a soluti<strong>on</strong> ∂y′<br />

∂s<br />

= − ∂f and ∂x′<br />

∂x ∂s = ∂f<br />

∂y ′ and, therefore,<br />

J =<br />

∂ ∂s (x′ − y ′ ) = ∂f<br />

∂x + ∂f<br />

∂y .<br />

The integral can be now written as<br />

Z +∞<br />

Z +∞<br />

Z = da ds e −(x′ −y ′ ) 2 ∂f<br />

<br />

−∞ −∞<br />

∂x ′ + ∂f ,<br />

∂y ′<br />

Here the integral over s is <strong>on</strong>e-dimensi<strong>on</strong>al, the functi<strong>on</strong>s x ′ and y ′ are the functi<strong>on</strong>s of s and it is independent of a. One can c<strong>on</strong>vert<br />

this <strong>on</strong>e-dimensi<strong>on</strong>al integral into a two-dimensi<strong>on</strong>al <strong>on</strong>e by substituting the δ-functi<strong>on</strong> with the gauge c<strong>on</strong>diti<strong>on</strong><br />

Z +∞<br />

da e −(x′ −y ′ ) 2 ∂f<br />

<br />

−∞<br />

∂x ′ + ∂f Z<br />

+∞<br />

=<br />

∂y ′ dxdy δ(f(x, y))e −(x−y)2 ∂f<br />

<br />

−∞<br />

∂x + ∂f<br />

.<br />

∂y<br />

The infinite volume arising up<strong>on</strong> integrating over a can be factored out and taking into account that<br />

<br />

∂f<br />

∂f<br />

∂a | a=0 =<br />

∂x + ∂f <br />

| a=0<br />

∂y<br />

we obtain the final and finite expressi<strong>on</strong><br />

Z +∞<br />

Z finite = dxdy δ(f(x, y)) ∂f<br />

e −(x−y)2<br />

−∞<br />

∂a a=0 Here ∂f<br />

∂a is known as the Faddeev-Popov determinant.<br />

a=0<br />

The first problem in realizing this approach for string theory is to find a measure<br />

for functi<strong>on</strong>al integrati<strong>on</strong> that preserves all symmetries of the classical theory<br />

(reparametrizati<strong>on</strong>s + Weyl symmetry).<br />

∫<br />

(δh, δh) =<br />

∫<br />

(δX, δX) =<br />

d 2 σ √ hh αβ h γδ δh αγ δh βδ<br />

d 2 σ √ hh αβ δX µ δX µ


– 69 –<br />

These scalar products define natural reparametrizati<strong>on</strong>-invariant and Poincaré invariant<br />

measures, however n<strong>on</strong>e of them is Weyl invariant.<br />

Let us first assume the simplifying situati<strong>on</strong> when all metric <strong>on</strong> a world-sheet M<br />

with a given topology are c<strong>on</strong>formally equivalent (i.e. they are related to each other<br />

by diffeomorphism and Weyl rescalings; this is the case when operator P † does not<br />

have zero modes). In this case by using reparametrizati<strong>on</strong>s we can bring the metric<br />

to the form<br />

h αβ = e 2φ g αβ ,<br />

where g αβ is a fiducial (reference) metric. Under reparametrizati<strong>on</strong>s and the Weyl<br />

rescalings the variati<strong>on</strong> of the metric can be decomposed as<br />

where P is the following operator<br />

δh αβ = (P ξ) αβ + 2˜Λh αβ , ˜Λ = Λ +<br />

1<br />

2 ∇ γξ γ ,<br />

(P ξ) αβ = ∇ α ξ β + ∇ β ξ α − ∇ γ ξ γ h αβ ,<br />

which maps vectors into traceless symmetric tensors. Then the integrati<strong>on</strong> measure<br />

can be written as follows<br />

∂(P ξ,<br />

Dh = D(P ξ)D(˜Λ) = D(ξ)D(Λ)<br />

˜Λ)<br />

,<br />

∣ ∂(ξ, Λ) ∣<br />

} {{ }<br />

Jacobian<br />

where in the last formula we changed the variables<br />

(P ξ, ˜Λ) → (ξ, Λ)<br />

for the price of getting a n<strong>on</strong>-trivial Jacobian. Here D(ξ) is the measure which gives<br />

up<strong>on</strong> integrati<strong>on</strong> an infinite volume of the diffeomorphism group and<br />

∂(P ξ, ˜Λ)<br />

∣ ∣ ∣∣∣∣ ∂(P ξ) ∂(P ξ)<br />

∣∣∣∣ ∂(P ξ)<br />

∣ ∂(ξ, Λ) ∣ = ∂ξ ∂Λ<br />

∂ ˜Λ<br />

∣ = 0<br />

∂ξ ∂ ˜Λ<br />

∣ = |detP |<br />

∂ξ<br />

∂ ˜Λ<br />

∂Λ<br />

∂ξ<br />

1<br />

In fact, we have<br />

δ(P ξ) αβ (σ)<br />

δξ γ (σ ′ )<br />

=<br />

(δ γ β ∇ α + δ γ α∇ β − h αβ ∇ γ )<br />

δ(σ − σ ′ )<br />

Thus,<br />

∫<br />

|detP | =<br />

DbDc e −i T 2 2 R )<br />

d 2 σd 2 σ ′√ h b αβ (σ)<br />

(δ γ β ∇ α+δα∇ γ β −h αβ ∇ γ σ δ(σ−σ′ )c γ (σ ′) .


– 70 –<br />

Here c α is called a ghost field, while b αβ is traceless and symmetric, it is called<br />

antighost field. The ghost c α corresp<strong>on</strong>ds to infinitezimal reparametrizati<strong>on</strong>s while<br />

b aβ corresp<strong>on</strong>ds to variati<strong>on</strong>s perpendicular to the gauge orbits. Both the ghost and<br />

the antighost fields are real. The last formula can be now written as<br />

∫<br />

|detP | = DbDc e − i R<br />

πα ′ d 2 σ √ h b αβ ∇ α c β .<br />

Thus, the total acti<strong>on</strong> is given now by the sum of the Polyakov acti<strong>on</strong> and the ghost<br />

acti<strong>on</strong>:<br />

S = − T ∫ (<br />

)<br />

d 2 σγ αβ ∂ α X µ ∂ β X µ + 4ib βγ ∇ α c γ<br />

2<br />

There are several subtle issues we have not touched so far<br />

• C<strong>on</strong>formal anomaly, i.e. possible dependence of the thrown away volume of<br />

the diffeomorphism group <strong>on</strong> the Weyl (scale) degree of freedom φ.<br />

• Reparametrizati<strong>on</strong>s which satisfy P ξ = 0, i.e. c<strong>on</strong>formal Killing vectors. We<br />

see that equati<strong>on</strong>s of moti<strong>on</strong> for c α are just c<strong>on</strong>formal Killing equati<strong>on</strong>s. Therefore,<br />

in order not to overcount the c<strong>on</strong>figurati<strong>on</strong>s which are related by a c<strong>on</strong>formal<br />

transformati<strong>on</strong> <strong>on</strong>e has to exclude integrati<strong>on</strong> over the zero modes of c α<br />

ghosts.<br />

• So far we assumed that all symmetric traceless deformati<strong>on</strong>s of the metric can<br />

be generated by reparametrizati<strong>on</strong>s. This is however not the case if P † has zero<br />

modes. These zero modes corresp<strong>on</strong>d to zero modes of the b ghosts.<br />

We can define the stress-energy tensor of the ghost fields<br />

∫<br />

δS gh = T d 2 σ √ −h T αβ δh αβ .<br />

Performing the variati<strong>on</strong> <strong>on</strong>e finds<br />

T gh<br />

αβ = i(b αγ∇ β c γ + b βγ ∇ α c γ − c γ ∇ γ b αβ − h αβ b γδ ∇ γ c δ ) .<br />

Here the last term vanishes <strong>on</strong> shell. In the deriving this expressi<strong>on</strong> we also used<br />

the tracelessness of b αβ . One can verify that this tensor is covariantly c<strong>on</strong>served<br />

∇ α T αβ = 0.<br />

In the world-sheet light-c<strong>on</strong>e coordinates σ ± the n<strong>on</strong>-vanishing comp<strong>on</strong>ents of<br />

the stress-tensor are<br />

T ++ = i ( 2b ++ ∂ + c + + (∂ + b ++ )c +) ,<br />

T −− = i ( 2b −− ∂ − c − + (∂ − b −− )c −) .


– 71 –<br />

Equati<strong>on</strong>s of moti<strong>on</strong> are<br />

This equati<strong>on</strong>s are supplemented by<br />

∂ − b ++ = ∂ + b −− = 0 ,<br />

∂ + c − = ∂ − c + = 0 .<br />

• by periodicity c<strong>on</strong>diti<strong>on</strong> for the closed closed string case<br />

b(σ + 2π) = b(σ) , c(σ + 2π) = c(σ) ;<br />

• by boundary c<strong>on</strong>diti<strong>on</strong>s for the open string case<br />

b ++ (σ) = b −− (σ) , c + (σ) = c − (σ) for σ = 0, π ,<br />

which follow by requiring the vanishing of the boundary terms arising up<strong>on</strong><br />

deriving equati<strong>on</strong>s of moti<strong>on</strong>.<br />

Note that for the closed string case b ++ and c + are left-moving waves, while b −− and<br />

c − are the right-moving <strong>on</strong>es. The can<strong>on</strong>ical anti-commutati<strong>on</strong> relati<strong>on</strong>s are<br />

{b ++ (σ, τ), c + (σ ′ , τ)} = 2πδ(σ − σ ′ ) ,<br />

{b −− (σ, τ), c − (σ ′ , τ)} = 2πδ(σ − σ ′ ) .<br />

For the closed string case the Fourier mode expansi<strong>on</strong>s look as<br />

c + (σ, τ) =<br />

c − (σ, τ) =<br />

b ++ (σ, τ) =<br />

b −− (σ, τ) =<br />

+∞∑<br />

n=−∞<br />

+∞∑<br />

n=−∞<br />

+∞∑<br />

n=−∞<br />

+∞∑<br />

n=−∞<br />

For the anti-commutati<strong>on</strong> relati<strong>on</strong>s this becomes<br />

{b m , c n } = δ m+n ,<br />

¯c n e −in(τ+σ) ,<br />

c n e −in(τ−σ) ,<br />

¯bn e −in(τ+σ) ,<br />

b n e −in(τ−σ) .<br />

{b m , b n } = {c m , c n } = 0<br />

and the same for the barred oscillators. The Virasoro generators are<br />

∞∑<br />

L gh<br />

m = (m − n) : b m+n c −n :<br />

¯L gh<br />

m =<br />

n=−∞<br />

∞∑<br />

n=−∞<br />

(m − n) : ¯b m+n¯c −n :


– 72 –<br />

The ghosts and anti-ghosts are c<strong>on</strong>formal fields. Indeed, it is easy to compute<br />

[L gh<br />

m , b n ] = (m − n)b m+n ,<br />

[L gh<br />

m , c n ] = −(2m + n)c m+n .<br />

Comparing this with the transformati<strong>on</strong> rule of the modes of a c<strong>on</strong>formal operator<br />

of dimensi<strong>on</strong> ∆<br />

[L m , A n ] = ( m(∆ − 1) − n ) A m+n<br />

we c<strong>on</strong>clude that b and c are indeed the c<strong>on</strong>formal fields of the c<strong>on</strong>formal dimensi<strong>on</strong><br />

∆ = 2 and ∆ = −1 respectively.<br />

Using the explicit expressi<strong>on</strong>s for the ghost generators L gh it is not difficult to<br />

compute the algebra<br />

where the central charge appears to be<br />

[L gh<br />

m , L gh<br />

n ] = (m − n)L gh<br />

m+n + c gh (m)δ m+n ,<br />

c gh (m) = 1<br />

12 (2m − 26m3 ) .<br />

Now if we introduce the total Virasoro generator as<br />

then it will satisfy the Virasoro algebra<br />

L m = L X m + L gh<br />

m − aδ m,0<br />

[L m , L n ] = (m − n)L m+n + c(m)δ m+n<br />

with<br />

c = d 12 (m3 − m) + 1 12 (2m − 26m3 ) + 2am .<br />

We see that the total central charge vanishes for d = 26 and a = 1. We again<br />

found the same values for the critical dimensi<strong>on</strong> and the normal-ordering c<strong>on</strong>stant<br />

as followed from the light-c<strong>on</strong>e approach! Here these c<strong>on</strong>diti<strong>on</strong>s <strong>on</strong> the theory follow<br />

from the requirement of vanishing of the total central charge.<br />

BRST operator<br />

The c<strong>on</strong>cept of the BRST operator is very general. In fact, the BRST operator can<br />

be associated to any Lie algebra and it is a useful tool to compute the Lie algebra<br />

cohomologies.<br />

C<strong>on</strong>sider a Lie algebra with generators K i satisfying the relati<strong>on</strong>s<br />

[K i , K j ] = f k ijK k .<br />

Introduce ghost and anti-ghost fields c i and b i satisfying the anti-commutati<strong>on</strong> relati<strong>on</strong>s<br />

{c i , b j } = δ i j , i = 1, . . . , dimK


– 73 –<br />

Introduce the ghost number operator U:<br />

U = ∑ i<br />

c i b i .<br />

The eigenvalues of this operator are integers ranging from 0 up to dimK.<br />

The BRST operator is defined as<br />

Q = c i K i − 1f k 2 ijc i c j b k . (4.18)<br />

First we compute a commutator<br />

[Q, U] = [c i K i − 1f k 2 ijc i c j b k , c m b m ]<br />

= −c m {c i , b m }K i − c m 1f k 2 ij{c i c j , b m }b k − 1f k 2 ijc i c j {b k , c m }b m<br />

= −c i K i + fijc k i c j b k − 1f k 2 ijc i c j b k = −Q .<br />

Thus, the BRST operator has the following commutator with U:<br />

[U, Q] = Q<br />

and as the result it increases the ghost number by <strong>on</strong>e:<br />

UQ|χ〉 = (QU + Q)|χ〉 = (N gh + 1)Q|χ〉 .<br />

Sec<strong>on</strong>d, compute the anticommutator<br />

{Q, Q} = {c i K i − 1 2 f k ijc i c j b k , c s K s − 1 2 f p mnc m c n b p }<br />

= c i c s K i K s − c s c i K s K i − 1 2 f k ijc i c j {b k , c s }K s − 1 2 f p mnc m c n {c i , b p }K i<br />

+ 1 4 f k ijf p mn{c i c j b k , c m c n b p } .<br />

It is easy to find<br />

f k ijf p mn{c i c j b k , c m c n b p } = 4f k ijf p km ci c j c m b p .<br />

Therefore, the expressi<strong>on</strong> we are interested in reduces to<br />

{Q, Q} = c i c j [K i , K j ] − 1 2 f k ijc i c j K k − 1 2 f k ijc i c j K k + f k ijf p km ci c j c m b p .<br />

Due to the algebra relati<strong>on</strong> [K i , K j ] = f k ijK k the first three terms in the last expressi<strong>on</strong><br />

cancel out and we are left with<br />

{Q, Q} = f k ijf p km ci c j c m b p .<br />

Due to the anti-commuting property of the ghosts the last expressi<strong>on</strong> can be rewritten<br />

as<br />

{Q, Q} = 1 3 (f k ijf p km + f k mif p kj + f k jmf p ki )ci c j c m b p .


– 74 –<br />

The expressi<strong>on</strong> in the bracket vanishes because this is the Jacobi identity written for<br />

structure c<strong>on</strong>stants of the Lie algebra<br />

f k ijf p km + f k mif p kj + f k jmf p ki = 0 .<br />

Thus, we found that the BRST operator is nilpotent, i.e. it its square is zero<br />

Q 2 = 1 {Q, Q} = 0 .<br />

2<br />

This is the fundamental property of the BRST operator. We also assume that Q is<br />

hermitian, i.e. Q † = Q.<br />

Let H k will be a Hilbert space of states with fixed ghost number U = k. An element<br />

|χ〉 ∈ H k is called BRST-invariant if<br />

Qχ = 0 (4.19)<br />

Clearly, any state of the form Q|λ〉, where |λ〉 is any state with the ghost number 14<br />

k − 1, is BRST-invariant because<br />

The state Q|λ〉 has zero norm because<br />

Q(Q|λ〉) = Q 2 |λ〉 = 0 .<br />

〈λ|Q † Q|λ〉 = 〈λ|Q 2 |λ〉 = 0 .<br />

The most important BRST-invariant states are those which can not be written in<br />

the form |χ〉 = Q|λ〉. We will regard two soluti<strong>on</strong>s of equati<strong>on</strong> (4.19) equivalent if<br />

for some λ.<br />

|χ ′ 〉 − |χ〉 = Q|λ〉<br />

In fact, we recognize that the BRST-operator mimics all the properties of the de-Rahm operator d which acts <strong>on</strong> the space of external<br />

(differential) forms <strong>on</strong> a manifold M. Indeed, it has a property that d 2 = 0. A differential form ω is called closed if dω = 0 and it is<br />

called exact if there is another form θ such that ω = dθ. The factor-space of all closed forms over all exact forms of a given degree n<br />

H n closed forms<br />

(M) =<br />

exact forms<br />

is called n-th cohomology group of the manifold M. In our present case the operator Q takes values in the Lie algebra and it defines<br />

cohomologies with values in an given representati<strong>on</strong> of the Lie algebra.<br />

Furthermore, the states with zero ghost charge are of special importance. Such<br />

a state must be annihilated by all b k . For such states the BRST operator reduces to<br />

Q|χ〉 = c i K i |χ〉 = 0 .<br />

14 As we found the BRST-operator increases the ghost number by <strong>on</strong>e.


– 75 –<br />

Thus, Q|χ〉 = 0 is equivalent to the c<strong>on</strong>diti<strong>on</strong> that K i |χ〉 = 0, i.e. it is invariant under<br />

the acti<strong>on</strong> of the Lie algebra. On the other hand, this state cannot be represented as<br />

|χ〉 = Q|λ〉 for some |λ〉 because the ghost number of |λ〉 should be equal −1 which<br />

is impossible.<br />

Let us apply this general c<strong>on</strong>structi<strong>on</strong> to the string case. The Lie algebra in this<br />

case is the Virasoro algebra and we supply it with the ghosts c m and ant-ghosts b m .<br />

The BRST operator is now<br />

Q =<br />

+∞∑<br />

−∞<br />

L X −mc m − 1 2<br />

∞∑<br />

(m − n) : c −m c −n b m+n : −ac 0 ,<br />

−∞<br />

where a is the normal-ordering ambiguity c<strong>on</strong>stant for L 0 . It turns out that this<br />

expressi<strong>on</strong> can be written as<br />

Q =<br />

+∞∑<br />

−∞<br />

:<br />

(<br />

L X −m + 1 2 Lgh −m − aδ m,0<br />

)c m :<br />

The ghost-number operator is<br />

U =<br />

+∞∑<br />

−∞<br />

: c −m b m :<br />

We would like to investigate the fulfilment of the relati<strong>on</strong> Q 2 = 0 in quantum theory.<br />

We find<br />

Q 2 = 1 +∞<br />

{Q, Q} =<br />

2<br />

∑<br />

n,m=−∞<br />

([L m , L n ] − (m − n)L m+n<br />

)<br />

c −m c −n .<br />

Here L m = L X m + L gh<br />

m − aδ m,0 is a total Virasoro operator. Thus, Q 2 = 0 for d = 26<br />

and a = 1 as the c<strong>on</strong>sequence of vanishing of the total central charge!<br />

Inverse statement is also true: from Q 2 = 0 it follows that the central charge of<br />

the Virasoro algebra vanishes. Indeed, we first note that<br />

From here<br />

L m = {Q, b m }<br />

[L m , Q] = [{Q, b m }, Q] = (Qb m + b m Q)Q − Q(Qb m + b m Q) = [b m , Q 2 ] = 0<br />

as Q 2 = 0. Therefore, we see that<br />

[L m , L n ] = [L m , {Q, b n }] = {[L m , Q], b n }<br />

} {{ }<br />

=0<br />

+ {Q, [L m , b n ]} = (m − n){Q, b m+n } = (m − n)b m+n .


– 76 –<br />

One can check that these objects are charges which corresp<strong>on</strong>d to new c<strong>on</strong>served<br />

currents<br />

J B + = 2c + (T X ++ + 1 2 T gh<br />

++)<br />

J + = c + b ++ .<br />

There is new and more fundamental fermi<strong>on</strong>ic symmetry present – it is BRST<br />

symmetry. Let λ be a grassman (anticomuting) parameter. The BRST transformati<strong>on</strong><br />

is defined as<br />

δY = [λQ, Y ]<br />

It is given by<br />

δX µ = λc + ∂ + X µ + λc − ∂ − X µ ,<br />

δc + = λc + ∂ + c + ,<br />

δb ++ = 2iλT ++ ,<br />

δT ++ = 0 .<br />

The ghost number operator<br />

U = 1 2 (c 0b 0 − b 0 c 0 ) +<br />

∞∑<br />

(c −n b n − b −n c n )<br />

n=1<br />

Here c n , b n for n > 0 are annihilati<strong>on</strong> operators.<br />

Zero-modes require special treatment. We have<br />

c 2 0 = b 2 0 = 0 , {c 0 , b 0 } = 1<br />

There is a two-dimensi<strong>on</strong>al representati<strong>on</strong> of this relati<strong>on</strong>s:<br />

c 0 | ↓〉 = | ↑〉 , b 0 | ↑〉 = | ↓〉<br />

c 0 | ↑〉 = 0 , b 0 | ↓〉 = 0 .<br />

The ghost numbers are U ↓ = −1/2 and U ↑ = 1/2.<br />

Physical states should have the ghost number −1/2. They are annihilated by b 0 .<br />

Indeed, c<strong>on</strong>sider<br />

c n |χ〉 = b n |χ〉 = 0 , n > 0 and b 0 |χ〉 = 0 .<br />

The c<strong>on</strong>diti<strong>on</strong> of the BRST invariance reduces to<br />

(<br />

0 = Q|χ〉 = c 0 (L 0 − 1) + ∑ c −n L n<br />

)|χ〉 .<br />

n>0<br />

We thus reproduced the c<strong>on</strong>diti<strong>on</strong>s for a physical state obtained in the old covariant<br />

quantizati<strong>on</strong> approach. Physical states of bos<strong>on</strong>ic string are cohomology classes of<br />

the BRST operator with the ghost number −1/2.


– 77 –<br />

5. Geometry and topology of string world-sheet<br />

5.1 From Lorentzian to Euclidean world-sheets<br />

<strong>String</strong> world-sheets and the target space both have Lorentzian signature. The c<strong>on</strong>necti<strong>on</strong><br />

to the theory of Riemann surfaces can be made by performing the Wick<br />

rotati<strong>on</strong> of both world-sheet and the target space metrics. In particular, for the<br />

world-sheet time τ this means τ → −iτ. Thus, <strong>on</strong> the string-world sheet this Euclidean<br />

c<strong>on</strong>tinuati<strong>on</strong> corresp<strong>on</strong>ds to<br />

σ ± = τ ± σ → −i(τ ± iσ) . (5.1)<br />

A word of cauti<strong>on</strong> is needed here: One cannot really prove that the theory of the<br />

Lorentzian world-sheets is equivalent to the theory of the Euclidean <strong>on</strong>es. However,<br />

treating the world-sheets as Euclidean provides by itself a c<strong>on</strong>sistent theory of interacting<br />

strings. The Euclidean versi<strong>on</strong> can be then used to learn as much as possible<br />

about its Lorentzian counterpart.<br />

We start with c<strong>on</strong>sidering a single closed string whose world-sheet has topology<br />

of a cylinder. Formula (5.1) suggests to introduce the complex coordinates<br />

w = τ + iσ , ¯w = τ − iσ .<br />

The Euclidean closed string covers <strong>on</strong>ly a finite interval of σ <strong>on</strong> the complex plane<br />

0 ≤ σ ≤ 2π and, therefore, <strong>on</strong>ly a strip of the 2dim plane.<br />

w<br />

Fig. 5. Mapping the cylinder (τ, σ) to a strip (w, ¯w) <strong>on</strong> the complex w-plane.<br />

One can further map the strip to the whole complex plane by using the c<strong>on</strong>formal<br />

map<br />

z = e w = e τ+iσ .<br />

Lines of c<strong>on</strong>stant τ are mapped into circles <strong>on</strong> the z plane and the operati<strong>on</strong> of time<br />

translati<strong>on</strong> τ → τ + a becomes the dilatati<strong>on</strong><br />

z → e a z .<br />

A procedure of identifying dilatati<strong>on</strong>s with the Hamilt<strong>on</strong>ian and circles about the<br />

origin with equal-time surfaces is called sometimes radial quantizati<strong>on</strong>. Mapping


¡<br />

¡<br />

£¢<br />

¢<br />

– 78 –<br />

from the cylinder to the plane cannot change the physical c<strong>on</strong>tent of the theory<br />

if the theory is c<strong>on</strong>formally invariant, which is the case of string theory in critical<br />

dimensi<strong>on</strong>.<br />

We note that the plane is a n<strong>on</strong>-compact manifold. However, <strong>on</strong>e can compactify<br />

it by adding a point at infinity. The corresp<strong>on</strong>ding compact surface arising in this<br />

way is the Riemann sphere. A metric <strong>on</strong> a plane can be transformed to a metric <strong>on</strong><br />

a sphere by a suitable choice of the c<strong>on</strong>formal prefactor. For instance <strong>on</strong>e can pick<br />

up the metric<br />

ds 2 =<br />

4dzd¯z<br />

(1 + |z| 2 ) 2<br />

The formula z = cot θ 2 eiφ defines a stereographic projecti<strong>on</strong> of the sphere <strong>on</strong>to the<br />

plane and under this projecti<strong>on</strong> the metric takes the form<br />

ds 2 = dθ 2 + sin 2 θdφ 2 ,<br />

i.e. it is the standard round metric <strong>on</strong> a sphere.<br />

north pole (outgoing string)<br />

¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦<br />

¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤<br />

¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦<br />

¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤<br />

¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦<br />

§¥§¥§ ¨¥¨¥¨ ¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤<br />

z−plane<br />

south pole (incoming string)<br />

¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤<br />

¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤<br />

¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤<br />

¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦<br />

¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦<br />

¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦<br />

¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦<br />

§¥§¥§ ¨¥¨¥¨<br />

¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤<br />

¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦<br />

¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦<br />

§¥§¥§ ¨¥¨¥¨ ¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤<br />

¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦<br />

§¥§¥§ ¨¥¨¥¨ ¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤<br />

¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦<br />

§¥§¥§ ¨¥¨¥¨ ¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤<br />

¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤<br />

¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦<br />

£<br />

¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦¥¦<br />

¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤<br />

¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤¥¤<br />

Fig. 6. Stereographic projecti<strong>on</strong> of a sphere <strong>on</strong>to z-plane. Asymptotic<br />

incoming and outgoing strings are mapped to the south and the north poles<br />

of the sphere respectively.<br />

Since cot π = 0 and cot(0) = ∞ the incoming and outgoing strings are mapped to<br />

2<br />

the south and the north poles of the sphere respectively.<br />

The example above can be generalized to general world-sheets corresp<strong>on</strong>ding<br />

to interacting strings. The crucial observati<strong>on</strong> is that c<strong>on</strong>formal invariance allows<br />

to c<strong>on</strong>sider compact world-sheets instead of surfaces with boundaries corresp<strong>on</strong>ding<br />

incoming and outgoing strings. The string boundaries are mapped to punctures <strong>on</strong> a<br />

compact Riemann surface.<br />

Under the Euclidean c<strong>on</strong>tinuati<strong>on</strong> the basic equati<strong>on</strong> □X = 0 transforms into<br />

with a general soluti<strong>on</strong><br />

∂ z ∂¯z X = 0<br />

X(z, ¯z) = X(z) + ¯X(¯z) ,<br />

i.e. the left and right-moving excitati<strong>on</strong> corresp<strong>on</strong>d now to analytic and anti-analytic<br />

fields <strong>on</strong> the complex z-plane.


– 79 –<br />

5.2 Riemann surfaces<br />

On a two-dimensi<strong>on</strong>al real manifold M the metric can be locally (i.e. in a given<br />

coordinate chart) defined by the line element<br />

Introducing the complex coordinates<br />

ds 2 = h 11 dx 2 + 2h 12 dxdy + h 22 dy 2<br />

z = x + iy ,<br />

¯z = x − iy<br />

the line element can be written as<br />

ds 2 = 2e φ |dz + µd¯z| 2<br />

with the following identificati<strong>on</strong>s<br />

(<br />

√<br />

)<br />

e φ = 1 h<br />

8 11 + h 22 + 2 h 11 h 22 − h 2 h 11 − h 22 + 2ih 12<br />

12 , µ =<br />

h 11 + h 22 + 2 √ .<br />

h 11 h 22 − h 2 12<br />

If h 11 = h 12 and h 12 = 0 then µ = 0 and the metric takes in this coordinate chart<br />

the form<br />

ds 2 = 2e φ |dz| 2 = 2e φ (dx 2 + dy 2 ) .<br />

The corresp<strong>on</strong>ding coordinate system is called isothermal or c<strong>on</strong>formal and the coordinates<br />

(x, y) define a c<strong>on</strong>formal map of a coordinate chart of a manifold to the<br />

Euclidean plane.<br />

A theorem of Gauss<br />

For any real two-dimensi<strong>on</strong>al orientable surface with a positive definite metric there<br />

always exists a system of isothermal coordinates (the theorem of Gauss). It is unique<br />

up to c<strong>on</strong>formal transformati<strong>on</strong>s. First, assume that we have already found a system<br />

of isothermal coordinates, i.e. the metric is locally in the form<br />

ds 2 = 2e φ |dz| 2 .<br />

Performing the coordinate transformati<strong>on</strong> with an analytic functi<strong>on</strong> of z:<br />

we get<br />

z → f(z)<br />

ds 2 → ds 2 = 2e φ |f(z)| 2 |dz| 2 ,<br />

i.e. we get a c<strong>on</strong>formally equivalent metric and, therefore, a new system of the<br />

isothermal coordinates.


– 80 –<br />

coordinate chart<br />

¡ ¢¡¢¡¢<br />

¡ ¢¡¢¡¢<br />

Fig. 6. Covering the Riemann surface with coordinate patches. Every patch<br />

is homeomorphic to an open domain of the Euclidean plane.<br />

Sec<strong>on</strong>d, in order to prove that isothermal coordinates exist, c<strong>on</strong>sider the so-called<br />

Beltrami equati<strong>on</strong><br />

∂w<br />

= µ(z, ¯z)∂w<br />

∂¯z ∂z .<br />

Suppose we solve this equati<strong>on</strong>, then<br />

Thus,<br />

|dw| 2 = |∂ z w + ∂¯z w| 2 = |∂ z w| 2 |dz + µd¯z| 2 = |∂ zw| 2<br />

2e φ ds2 .<br />

ds 2 =<br />

2eφ<br />

|∂ z w| 2 |dw|2 ≡ λ|dw| 2 ,<br />

i.e. w defines a system of isothermal coordinates. It is a mathematical theorem that<br />

for a sufficiently small coordinate patch and a differentiable metric a soluti<strong>on</strong> of the<br />

Beltrami equati<strong>on</strong> with ∂ z w ≠ 0 always exists.<br />

If two metrics are related by a diffeomorphism and a Weyl rescaling they are said<br />

to define the same c<strong>on</strong>formal structure. If a manifold M is covered by a system of<br />

c<strong>on</strong>formal (isothermal) coordinate patches U α , then <strong>on</strong> the overlaps the metrics are<br />

c<strong>on</strong>formally related, i.e. the transiti<strong>on</strong> functi<strong>on</strong>s <strong>on</strong> the overlaps U α ∩U β are analytic<br />

and the complex coordinates are globally defined. A system of analytic coordinate<br />

patches is called a complex structure and it is the same as a c<strong>on</strong>formal structure.<br />

A two-dimensi<strong>on</strong>al topological manifold endowed with a complex structure is<br />

called a Riemann surface. Thus, a Riemann surface is a complex manifold.<br />

Another way to understand that Riemann surface is a complex manifold is to<br />

note that in two dimensi<strong>on</strong>s the metric provides a globally-defined integrable complex<br />

structure<br />

I α β = √ hh αγ ɛ γβ<br />

such that I 2 = −1. The c<strong>on</strong>formal structure is c<strong>on</strong>formally-invariant and globally<br />

well-defined. The existence of an integrable complex structure is necessary and sufficient<br />

for an even-dimensi<strong>on</strong>al manifold to be complex.


– 81 –<br />

Recall the fundamental result from the theory of two-dimensi<strong>on</strong>al (real) manifolds.<br />

Any compact orientable c<strong>on</strong>nected two-dimensi<strong>on</strong>al manifold is homeomorphic to<br />

a sphere with handles. The number of handles, g, is called the genus, a topological<br />

invariant. Thus, every compact Riemann surface, being a compact orientable<br />

c<strong>on</strong>nected <strong>on</strong>e-dimensi<strong>on</strong>al manifold, has an associated genus.<br />

Gauss-B<strong>on</strong>net theorem<br />

Let M is a compact orientable two-dimensi<strong>on</strong>al manifold of genus g with a metric<br />

h αβ . Then<br />

∫<br />

1 √<br />

hR = χ(M) = 2 − 2g (5.2)<br />

4π M<br />

is a topological invariant and it coincides with the Euler characteristic<br />

χ(M) = 2 − 2g .<br />

Here we briefly discuss <strong>on</strong>e way to prove the Gauss-B<strong>on</strong>net theorem. Recall that<br />

in two dimensi<strong>on</strong>s the Riemann tensor has <strong>on</strong>ly <strong>on</strong>e independent comp<strong>on</strong>ent which<br />

is the scalar curvature:<br />

R αβγδ = R 2 (h αγh βδ − h αδ h βγ ) .<br />

Let us choose a system of isothermal coordinates. In this coordinate system the line<br />

element is ds 2 = 2e φ dzd¯z and, therefore, the metric has two comp<strong>on</strong>ents h z¯z = h¯zz =<br />

e φ . The Riemann tensor simplifies to<br />

R z¯zz¯z = −h z¯z R z¯z = − 1 2 (h z¯z) 2 R = e φ ∂ ¯∂φ .<br />

The scalar curvature is<br />

R = −2e −φ ∂ ¯∂φ =⇒ √ hR = −4∂ ¯∂φ = −△φ ,<br />

where △ is two-dimensi<strong>on</strong>al Laplacian (the Euclidean analogue of the □ operator).<br />

It is also useful to rewrite the integrati<strong>on</strong> measure in the complex coordinates<br />

dx ∧ dy = i dz ∧ d¯z ,<br />

2<br />

i.e. we get<br />

√<br />

hR d 2 x = −4 ∂2 φ<br />

∂z∂¯z dx ∧ dy = −2i ∂2 φ<br />

(d¯z<br />

∂z∂¯z dz ∧ d¯z = 2i ∂ )<br />

dz ∂φ<br />

∂¯z ∂z .


– 82 –<br />

The last formula can be understood in the sense of calculus of exterior differential<br />

forms ∂f = ∂f and ¯∂f = ∂f<br />

∂<br />

. In particular, the de Rahm operator d = dx + dy ∂ ∂z ∂¯z ∂x ∂y<br />

can be written as<br />

d = ∂ + ¯∂ ≡ dz ∂ ∂z + d¯z ∂ ∂¯z .<br />

Also <strong>on</strong>e has ∂∂ = ¯∂ ¯∂ = 0. Therefore, we can rewrite our formula as<br />

√<br />

hR = 2id(∂φ) .<br />

This shows that √ hR is locally a total derivative and therefore, we see again that<br />

eq.(5.2) cannot change under smooth variati<strong>on</strong>s of the metric, i.e. it is a topological<br />

invariant.<br />

On a compact Riemann surface M we c<strong>on</strong>sider an abelian differential Ω, which is<br />

a meromorphic differential form. It means that in a given coordinate patch (U α , z α )<br />

it can be written in the form Ω = f α dz α , where f α is a meromorphic functi<strong>on</strong> 15 .<br />

We can also suppose that the coordinate patches are chosen in such a way that<br />

every U α c<strong>on</strong>tains at most <strong>on</strong>e pole or <strong>on</strong>e zero of Ω. In a patch U α the metric is<br />

ds 2 = 2e φ α<br />

|dz α | 2 . Thus, <strong>on</strong> the intersecti<strong>on</strong>s of the patches U α ∩ U β we have<br />

e φ α<br />

e φ β<br />

= ∣ ∣∣ f α<br />

f β<br />

∣ ∣∣<br />

2<br />

.<br />

Thus, there exists a globally defined functi<strong>on</strong><br />

ϕ = eφ α<br />

|f α | 2 <strong>on</strong> U α for all α .<br />

This functi<strong>on</strong> is smooth except for singularities at zeros and poles of Ω. Since log |f α | 2<br />

is harm<strong>on</strong>ic outside zeros and poles of Ω we have<br />

√<br />

hR = 2id(∂ log ϕ) .<br />

Let M ɛ = M − ∪D k,ɛ , where D k,ɛ are small disks around the singularities of Ω.<br />

Then, by Stokes’s theorem we have<br />

∫<br />

M<br />

√<br />

∫<br />

hR = 2i lim d(∂ log ϕ) = −2i ∑<br />

ɛ→0<br />

M ɛ k<br />

∫<br />

lim ∂ log ϕ .<br />

ɛ→0<br />

∂D k,ɛ<br />

To evaluate the integrals over the circles we note that at a zero or pole of Ω the<br />

functi<strong>on</strong> ϕ is of the form ϕ = ψ/|z| 2m with a smooth functi<strong>on</strong> ψ without zeros and<br />

m is the order of zero or pole (m < 0 in the latter case). Therefore,<br />

∫<br />

lim ∂ log ϕ = lim ∂ log |z|<br />

ɛ→0<br />

∂D ɛ→0 k,ɛ<br />

∫|z|=ɛ<br />

−2m dz<br />

= −m lim<br />

ɛ→0<br />

∫|z|=ɛ z = −2πim .<br />

15 A functi<strong>on</strong> f(z) is called meromorphic if it does not have any other singularities except poles.


– 83 –<br />

Summing up we obtain<br />

∫<br />

√<br />

hR = −4π deg Ω ,<br />

M<br />

where deg Ω is defined as the difference of the number of zeros and the number of<br />

poles of Ω:<br />

deg Ω = # zeros − # poles .<br />

It is now the Poincaré-Hopf theorem that states that deg Ω = 2g − 2, where g is the<br />

genus of the Riemann surface. Thus, the Gauss-B<strong>on</strong>net theorem follows from the<br />

Poincaré-Hopf theorem.<br />

Finally we note that due to the identity<br />

ɛ αβ ɛ γδ = h αγ h βδ − h αδ h βγ<br />

the Euler characteristic can be rewritten in the form 16<br />

χ(M) = 1<br />

8π<br />

∫<br />

M<br />

ɛ αβ ɛ γδ R αβγδ .<br />

g<br />

1 2<br />

g<br />

g + g<br />

1<br />

2<br />

g=0<br />

+<br />

Fig. 7. If there are two surfaces of genera g 1 and g 2 then by removing from<br />

each surface a half-sphere we can glue the resulting surfaces into a surface<br />

of genus g 1 + g 2 and the Riemann sphere of genus zero.<br />

To illustrate the Gauss-B<strong>on</strong>net theorem, we compute the topological invariant<br />

eq.(5.2) for a sphere. We will take a model of a sphere which represent it as the<br />

complex plane (including the point at infinity) with the metric<br />

ds 2 =<br />

4dzd¯z<br />

(1 + |z| 2 ) 2 = 2eφ dzd¯z ,<br />

16 This is a n<strong>on</strong>-trivial characteristic class of the tangent bundle to M known as the Euler class.


– 84 –<br />

i.e the c<strong>on</strong>formal factor φ and the acti<strong>on</strong> of Laplacian <strong>on</strong> it are<br />

Therefore,<br />

φ = ln 2 − 2 ln(1 + |z| 2 ) =⇒ − △φ =<br />

∫<br />

1<br />

4π C<br />

dxdy √ hR = 1<br />

4π<br />

∫ ∞<br />

0<br />

8<br />

(1 + |z| 2 ) 2 .<br />

8<br />

2πrdr<br />

(1 + r 2 ) = 2 , 2<br />

which is indeed the Euler characteristic for sphere, a compact orientable manifold<br />

with genus g = 0. It is also known that <strong>on</strong> a torus, a manifold of genus g = 1, there<br />

exists the globally defined flat metric. Therefore, the Euler characteristic of torus<br />

is χ = 0. In fact, the Euler characteristic χ(g) for a Riemann surface of arbitrary<br />

genus g can be found by using the recurrent formula<br />

χ(g 1 + g 2 ) = χ(g 1 ) + χ(g 2 ) − χ(0) = χ(g 1 ) + χ(g 2 ) − 2<br />

and the fact that χ(1) = 0. This again leads to the formula χ(g) = 2 − 2g.<br />

5.3 Moduli space<br />

The moduli space of all the metrics is the same as the moduli space of Riemann<br />

surfaces and it is defined as the space of all metrics devided by diffeomorphisms and<br />

Weyl rescalings<br />

M g =<br />

all metrics<br />

diffeomorphisms × Weyl rescalings .<br />

The moduli space is finite-dimensi<strong>on</strong>al and it is parametrized by a finite number of<br />

complex parameters τ i called moduli. The dimensi<strong>on</strong> of the moduli space is another<br />

topological invariant and it depends <strong>on</strong> the genus g <strong>on</strong>ly.<br />

Complex geometry<br />

Since we have a system of well-defined complex coordinates <strong>on</strong> a Riemann surface<br />

we can c<strong>on</strong>sider general tensors<br />

V z...z¯z...¯z z...z¯z...¯z(z, ¯z) ,<br />

in particular, V z ∂ z and V ¯z ∂¯z are vector fields and V z dz and V¯z d¯z are <strong>on</strong>e-forms. All<br />

these tensors are <strong>on</strong>e comp<strong>on</strong>ent objects. The metric h z¯z and h z¯z can be used to<br />

c<strong>on</strong>vert all ¯z indices into z-indices. Tensors with <strong>on</strong>e type of indices (for example,<br />

z indices) are called holomorphic. Holomorphic tensors which depend <strong>on</strong> z variable


– 85 –<br />

<strong>on</strong>ly are called analytic. A holomorphic tensor with p lower and q upper indices has,<br />

by definiti<strong>on</strong>, the rank n = p − q:<br />

z } .{{ . . z}<br />

q<br />

V z } .{{ . . z(z, ¯z)<br />

}<br />

⇐ holomorphic tensor of rank n = p − q;<br />

p<br />

V z...z z...z(z) ⇐ analytic tensor;<br />

Under analytic coordinate transformati<strong>on</strong>s z → f(z) a holomorphic tensor of rank n<br />

transforms as follows<br />

( ) n ∂f(z)<br />

V (z, ¯z) →<br />

V (f(z),<br />

∂z<br />

¯f(¯z)) .<br />

Denote by V (n) the space of all holomorphic tensors of rank n. This space can be<br />

supplied with the scalar product<br />

∫<br />

(V (n)<br />

1 |V (n)<br />

2 ) =<br />

d 2 z √ h (h z¯z ) n (<br />

V (n)<br />

1<br />

) ∗V<br />

(n)<br />

2 , V (n)<br />

1 , V (n)<br />

2 ∈ V (n)<br />

and the associated norm ||V (n) || 2 = (V n |V (n) ). This scalar product is Weyl-invariant<br />

for n = 1 <strong>on</strong>ly.<br />

In the analytical coordinate system the Christoffel c<strong>on</strong>necti<strong>on</strong> has <strong>on</strong>ly two n<strong>on</strong>vanishing<br />

comp<strong>on</strong>ents<br />

Γ z<br />

zz = ∂φ ,<br />

Γ ¯z<br />

¯z¯z = ¯∂φ .<br />

This c<strong>on</strong>necti<strong>on</strong> allows to define two covariant derivatives<br />

∇ (n)<br />

z : V (n) → V (n+1) , ∇ (n)<br />

z T (n) (z, ¯z) = (∂ − n∂φ)T (n) (z, ¯z)<br />

∇ z (n) : V (n) → V (n−1) , ∇ z (n)T (n) (z, ¯z) = h z¯z ∇¯z T (n) (z, ¯z) = h z¯z ¯∂T (n) (z, ¯z) .<br />

These two differential operators defined <strong>on</strong> holomorphic tensors of a fixed rank n<br />

commute with the analytic coordinate transformati<strong>on</strong>s z → f(z). One can compute<br />

an adjoint of ∇ (n)<br />

z and find that<br />

(∇ (n)<br />

z ) † = −∇ z (n+1) .<br />

The elements of the complex geometry we introduced above allows <strong>on</strong>e to obtain<br />

some informati<strong>on</strong> about the moduli space. C<strong>on</strong>sider an arbitrary infinitesimal change<br />

of the metric<br />

δh αβ = Λh αβ + ∇ α V β + ∇ β V α + ∑ } {{ } } {{ }<br />

i<br />

Weyl<br />

diff<br />

δτ i<br />

∂<br />

∂τ i<br />

h αβ<br />

} {{ }<br />

moduli<br />

.


– 86 –<br />

The last term here reflects the dependence of the metric <strong>on</strong> moduli which cannot be<br />

compensated by diffeomorphisms and Weyl rescalings. We can split this variati<strong>on</strong><br />

into trace and traceless parts<br />

(<br />

δh trace<br />

αβ = Λ + ∇ γ V γ + 1 ∑ 2 hγδ i<br />

δh traceless<br />

αβ<br />

δτ i<br />

∂<br />

∂τ i<br />

h γδ<br />

)<br />

h αβ<br />

= ∇ α V β + ∇ β V α − h αβ ∇ γ V<br />

} {{ } γ + ∑ i<br />

Operator P<br />

( ∂<br />

δτ i h αβ − 1 ∂τ i 2 h αβh γδ ∂ )<br />

h γδ .<br />

∂τ i<br />

We see that we can always shift the Weyl rescaling parameter Λ as<br />

Λ → Λ − ∇ γ V γ − 1 2 hγδ ∑ i<br />

δτ i<br />

∂<br />

∂τ i<br />

h γδ<br />

so that the trace part of the variati<strong>on</strong> will transform as<br />

δh trace<br />

αβ = Λh αβ .<br />

In the complex coordinates we have h zz = h¯z¯z = 0 and therefore rewriting the<br />

variati<strong>on</strong> formulae in these coordinates we find<br />

δh z¯z = Λh z¯z ,<br />

δh zz = 2∇ (1)<br />

z V<br />

} {{ } z + ∑<br />

Operator P<br />

i<br />

δτ i µ i zz ,<br />

where µ i zz = ∂ τi h zz = h z¯z µ i¯z<br />

z . We see, in particular, that an infinitesimal change of<br />

h z¯z can always be written as a Weyl rescaling. Also the covariant derivative ∇ (1)<br />

z<br />

introduced above should be naturally identified with the operator P . Finally, we<br />

note that decompositi<strong>on</strong> into sum of two terms<br />

δh zz = 2∇ (1)<br />

z V z + ∑ i<br />

δτ i µ i zz<br />

is not orthog<strong>on</strong>al w.r.t. to the scalar product we introduced above. Denote by φ i zz a<br />

basis of the orthog<strong>on</strong>al complement of ∇ (1)<br />

z :<br />

(φ i zz|∇ (1)<br />

z V z ) = −(∇ z (2)φ i zz|V z ) for any V z ∈ V (1) .<br />

The last equati<strong>on</strong> is equivalent to<br />

∇ z (2)φ i zz = 0 =⇒ ¯∂φ<br />

i<br />

zz = 0 .<br />

Thus, the kernel (or, in other words, the space of zero modes) of the operator P † =<br />

∇ z (2)<br />

c<strong>on</strong>sists of global analytic tensors of the sec<strong>on</strong>d rank. Such tensors are of special<br />

importance and they are called quadratic differentials. Thus, the dimensi<strong>on</strong> of the


– 87 –<br />

moduli space is equal to the number of lineally independent quadratic differentials<br />

<strong>on</strong> a given Riemann surface of genus g. The theory of quadratic differentials was<br />

developed by Kurt Strebel (I will add more <strong>on</strong> Strebel theory in due course).<br />

The kernel of the operator ∇ (1)<br />

z is spanned by vectors from V (1) :<br />

∇ (1)<br />

z V z = h z¯z ∂V ¯z = 0 =⇒ ∂V ¯z = 0 .<br />

The globally defined vector fields which span a kernel of ∇ (1)<br />

z are called c<strong>on</strong>formal<br />

Killing vectors. They generate c<strong>on</strong>formal Killing group (or the group of c<strong>on</strong>formal<br />

isometries, i.e. globally defined diffeomorphisms which can be completely absorbed<br />

by the Weyl rescalings).<br />

Riemann-Roch theorem<br />

An important questi<strong>on</strong> of how many moduli for a Riemann surface of genus g exists<br />

is answered by the Riemann-Roch theorem. Define the index of ∇ (n)<br />

z as the number of<br />

its zero modes minus the number of zero modes of its adjoint ∇ z (n+1). The Riemann-<br />

Roch theorem states that<br />

ind∇ (n)<br />

z<br />

= dim ker∇ (n)<br />

z − dim ker∇ z (n+1) = −(2n + 1)(g − 1) = 1 2 (2n + 1)χ g .<br />

For n = 1 we therefore have<br />

#complex moduli − #c<strong>on</strong>formal Killing vectors = 3g − 3<br />

One can find the number of c<strong>on</strong>formal Killing vectors for a compact Riemann<br />

surface in an independent way. These are globally defined analytic vector fields whose<br />

norm is finite<br />

∫<br />

√<br />

||V || 2 = hhz¯z V z V ¯z < ∞ .<br />

M g<br />

Here V z = V n z n . For the case of sphere with metric ds 2 =<br />

there exists three independent c<strong>on</strong>formal Killing vectors<br />

Indeed, for the norm we have<br />

||V || 2 = 2π<br />

∂ z , z∂ z , z 2 ∂ z .<br />

∫ ∞<br />

0<br />

⎧<br />

8π<br />

8<br />

⎨ k = 0<br />

3<br />

rdr<br />

(1 + r 2 ) 4 r2k 4π<br />

= k = 1<br />

⎩<br />

3<br />

8π<br />

k = 2<br />

3<br />

4dzd¯z<br />

(1+|z| 2 ) 2<br />

<strong>on</strong>e finds that<br />

where k = 0, 1, 2 for the three vector fields in questi<strong>on</strong>. We see that if k ≥ 3 the<br />

integral becomes divergent, therefore, for instance, the field z 3 ∂ z has an infinite norm,


– 88 –<br />

i.e. it is not normalizable. The three c<strong>on</strong>formal Killing vectors are well behaved at<br />

∞ as can be seen by making c<strong>on</strong>formal transformati<strong>on</strong> w = 1/z under which<br />

−w 2 ∂ w , − w∂ w − ∂ w .<br />

The limit z → ∞ corresp<strong>on</strong>ds now to w → 0 and we see that these fields are wellbehaved<br />

at this point 17 . The three c<strong>on</strong>formal Killing vectors we found corresp<strong>on</strong>d to<br />

the Virasoro generators L 0 and L ±1 and they span (over the complex field) the Lie<br />

algebra sl(2, C). Therefore, sl(2, C) is the Lie algebra of analytic globally defined<br />

maps of the Riemann sphere <strong>on</strong>to itself.<br />

Thus, the Riemann sphere has three c<strong>on</strong>formal Killing vectors and, according to<br />

the Riemann-Roch theorem, no moduli. That means that all metrics <strong>on</strong> the sphere<br />

are c<strong>on</strong>formally equivalent, or, in other words, there is a unique Riemann surface of<br />

genus zero.<br />

One can count the number of c<strong>on</strong>formal Killing vectors for higher genus Riemann<br />

surfaces as well. To this end <strong>on</strong>e has to use the Ricci identity<br />

∇ z (n+1)∇ (n)<br />

z<br />

Let V (n) ∈ ker∇ (n)<br />

z . Then we have<br />

0 = (∇ (n)<br />

z<br />

− ∇ (n−1)<br />

z ∇ z (n) = 1 2 nR .<br />

V (n) |∇ (n)<br />

z V (n) ) = −(V (n) |∇ z (n+1) ∇(n) z V (n) ) =<br />

= − 1 2 (V (n) |∇ z (n+1) ∇(n) z V (n) ) − 1 2 (V (n) |∇ z (n+1) ∇(n) z V (n) )<br />

= − 1 2 (V (n) |∇ (n−1)<br />

z ∇ z (n) + 1 2 nR) − 1 2 (V (n) |∇ z (n+1) ∇(n) z V (n) )<br />

= 1 h<br />

(∇ (n)<br />

z<br />

2<br />

V (n) |∇ (n)<br />

z V (n) ) + (∇ z (n) V (n) |∇ z (n) V (n) ) − 1 2 nR(V (n) |V (n) i<br />

) .<br />

Therefore, for any vector from the kernel of ∇ (n)<br />

z<br />

the following equality is valid<br />

(∇ (n)<br />

z V (n) |∇ z (n) V (n) ) + (∇ z<br />

} {{ }<br />

(n)V (n) |∇ z (n)V (n) ) − 1 } {{ } 2 nR(V (n) |V (n) ) = 0 .<br />

n<strong>on</strong>−negative<br />

n<strong>on</strong>−negative<br />

C<strong>on</strong>sider the case of a torus g = 1. On a torus there is a globally defined flat metric<br />

ds 2 = dzd¯z which gives R = 0. Therefore, the equality above leads to two equati<strong>on</strong>s<br />

∂V (n) = ¯∂V (n) = 0 ,<br />

i.e. V (n) = c<strong>on</strong>st and, therefore, dim ker∇ z<br />

(n) = 1. Thus, there is a unique generator<br />

of c<strong>on</strong>formal isometries, it corresp<strong>on</strong>ds to the rigid U(1)×U(1) rotati<strong>on</strong>s of the torus.<br />

The Riemann-Roch theorem gives for g = 1<br />

#complex moduli − #c<strong>on</strong>formal Killing vectors = 3 − 3 = 0 ,<br />

} {{ }<br />

=1<br />

17 According to our general discussi<strong>on</strong> of Riemann surfaces the sphere requires at least two coordinate<br />

patches to make an atlas. Transformati<strong>on</strong> from <strong>on</strong>e patch to another is analytic and is given<br />

by w = 1/z.


– 89 –<br />

i.e. the torus is characterized by <strong>on</strong>e complex modulus τ.<br />

For g ≥ 2 there is theorem that states that the corresp<strong>on</strong>ding manifold always admits<br />

a metric with c<strong>on</strong>stant negative curvature. For n ≠ 0 this means that − 1 nR > 0 and,<br />

2<br />

therefore, dim ker∇ z (n)<br />

= 0. The Riemann surfaces with g ≥ 2 have no c<strong>on</strong>formal<br />

isometries and by the Riemann-Roch theorem that means that the number of complex<br />

moduli n = 1 is 3g − 3.<br />

For n = 0 we have dim ker∇ (0)<br />

z = 1, because the corresp<strong>on</strong>ding kernel is spanned by<br />

c<strong>on</strong>stants. The Riemann-Roch theorem implies then that<br />

dim ker∇ z (1) − dim ker∇ (0)<br />

z = (2n + 1) (g − 1) = g − 1 ,<br />

} {{ } } {{ }<br />

=1<br />

n=0<br />

i.e. dim ker∇ z (1) = g. The kernel of ∇z (1)<br />

is spanned by <strong>on</strong>e forms<br />

∇ z (1)ω z = h z¯z ¯∂ωz = 0 =⇒ ¯∂ω z = 0 .<br />

Thus we arrive at another important c<strong>on</strong>sequence of the Riemann-Roch theorem:<br />

<strong>on</strong> a Riemann surface of the genus g there exists precisely g linearly independent<br />

(globally defined) analytic <strong>on</strong>e-forms. These analytic differential forms are called<br />

abelian differentials of the first kind.<br />

The informati<strong>on</strong> we obtained by using the Riemann-Roch theorem is summarized<br />

in the Table below.<br />

g dim ker∇ z<br />

(n) dim ker∇ z (n+1)<br />

0 2n + 1 0<br />

1 1 1<br />

> 1 1 for n = 0 g<br />

0 for n > 0 (2n + 1)(g − 1)<br />

Moduli space of tori<br />

Here we would like to look more closely at the moduli space which describes c<strong>on</strong>formally<br />

n<strong>on</strong>-equivalent tori – the Riemann surfaces of the genus g = 1. The torus can<br />

be obtained by performing the following identificati<strong>on</strong> <strong>on</strong> the complex plane<br />

z ≡ z + nλ 1 + mλ 2 , n, m ∈ Z , λ 1 , λ 2 ∈ C .<br />

The parameters λ 1,2 are subject to c<strong>on</strong>formal transformati<strong>on</strong>s z → λz and, therefore,<br />

<strong>on</strong>ly their ratio τ = λ 2<br />

λ 1<br />

is scale-invariant. By using this freedom (the U(1)-rotati<strong>on</strong><br />

+ real rescaling) <strong>on</strong>e can always bring the parallelogram defining the torus up<strong>on</strong>


– 90 –<br />

gluing the opposite sides to the can<strong>on</strong>ical form depicted <strong>on</strong> Fig.9, which corresp<strong>on</strong>ds<br />

to Imτ > 0.<br />

z<br />

¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

Fig. 8. Defining the torus by factorizing the complex z-plane.<br />

The parameter τ takes values in the upper half-plane which is called Teichmüller<br />

space. The parameter τ itself is named the modular or Teichmüller parameter. The<br />

Teichmüller parameter is not yet a parameter describing the moduli space.<br />

+1<br />

0<br />

1<br />

Fig. 9. Can<strong>on</strong>ical representati<strong>on</strong> of the torus by parameter τ taking values<br />

in the Techmüller space which is identified with the upper half-plane.<br />

The reas<strong>on</strong> is that there are global diffeomorphisms which are not smoothly c<strong>on</strong>nected<br />

to the identity; they leave the torus invariant but they act n<strong>on</strong>-trivially <strong>on</strong> the<br />

Teichmüller parameter. They corresp<strong>on</strong>d to the so-called Dehn twists<br />

• λ 1 → λ 1 , λ 2 → λ 1 + λ 2 , which gives τ → τ + 1;<br />

• λ 1 → λ 1 + λ 2 , λ 2 → λ 2 , which gives τ →<br />

τ ; τ+1<br />

It turns out that these two transformati<strong>on</strong>s generate the group SL(2, Z). It is a group<br />

of matrices<br />

( ) a b<br />

,<br />

c d


– 91 –<br />

where a, b, c, d ∈ Z and ad − bc = 1. The acti<strong>on</strong> <strong>on</strong> the modular parameter τ is in<br />

the form of the fracti<strong>on</strong>al-linear transformati<strong>on</strong><br />

τ → τ ′ = aτ + b<br />

cτ + d .<br />

One can check that these transformati<strong>on</strong>s preserve the area of the parallelogram.<br />

Since two SL(2, Z)-matrices<br />

( )<br />

( )<br />

a b<br />

a b<br />

+<br />

−<br />

c d<br />

c d<br />

act <strong>on</strong> τ in the same way, the modular group of the torus is PSL(2, Z) = SL(2, Z)/Z 2 .<br />

i<br />

8<br />

¥¡¥¡¥<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

Fixed points of S and ST<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¦¡¦¡¦ ¥¡¥¡¥<br />

¦¡¦¡¦ ¥¡¥¡¥<br />

¨¡¨<br />

¤¡¤¡¤¡¤¡¤¡¤¡¤¡¤ §¡§<br />

£¡£¡£¡£¡£¡£¡£¡£<br />

¡¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¦¡¦¡¦ §¡§ ¨¡¨<br />

−1 −1/2 0 1/2 1<br />

Fig. 8. Fundamental domain M g=1 of the Teichimüller space describing the<br />

moduli space of c<strong>on</strong>formally n<strong>on</strong>-equivalent tori.<br />

The moduli space of c<strong>on</strong>formally n<strong>on</strong>-equivalent tori is then the quotient of the<br />

Teichmüller space of the modular group<br />

M g=1 =<br />

Teichmüller space<br />

modular group .<br />

One usually uses the following generators of the modular group<br />

T : τ → τ + 1 , S : τ → − 1 τ .<br />

Any element of SL(2, Z) is a compositi<strong>on</strong> of a certain number of S and T generators:<br />

SST ST T T ST....SST


– 92 –<br />

Any point in the upper-half plane is related by a modular transformati<strong>on</strong> to a point<br />

in the so-called fundamental domain F ≡ M g=1 of the modular group. It can be<br />

chosen as<br />

{<br />

M g=1 =<br />

− 1 2 ≤ Reτ ≤ 0, |τ|2 ≥ 1 ∪ 0 < Reτ < 1 }<br />

2 , |τ|2 > 1 .<br />

The modular group does not act freely <strong>on</strong> the upper-half plane because some of the<br />

modular transformati<strong>on</strong>s have fixed points. The point τ = i is the fixed point of S:<br />

τ → − 1 2πi<br />

and τ = e<br />

τ<br />

3 = − 1 + i √ 3 is the fixed point of ST . The existence of the<br />

2 2<br />

fixed points implies that M g=1 is not a smooth manifold, rather it has singularities<br />

of the orbifold type.<br />

Since it does not matter which fundamental regi<strong>on</strong> to choose in order to integrate<br />

over in the <strong>on</strong>e-loop path integral the integrand the corresp<strong>on</strong>ding string amplitude<br />

should be invariant under modular transformati<strong>on</strong>s. The requirement of the modular<br />

invariance is <strong>on</strong>e of the most important principles of string theory. In particular, it<br />

leads to str<strong>on</strong>g restricti<strong>on</strong>s <strong>on</strong> the possible gauge groups for the heterotic string.<br />

6. Classical fermi<strong>on</strong>ic superstring<br />

Introducti<strong>on</strong> of the world-sheet fermi<strong>on</strong>ic degrees of freedom requires understanding<br />

of how spinors <strong>on</strong> curved manifolds (world-sheets) are defined. The discussi<strong>on</strong> in the<br />

next paragraph is general and can be applied to a manifold of an arbitrary dimensi<strong>on</strong><br />

d.<br />

6.1 Spinors in General Relativity<br />

It is not straightforward to introduce spinors in General Relativity. If we have a<br />

tensor field T i 1...i p<br />

j 1 ...j q<br />

of rank (p, q) <strong>on</strong> a manifold M then under general coordinate<br />

transformati<strong>on</strong>s of the coordinates x i <strong>on</strong> M: x i → x ′i (x j ), this field transforms as<br />

follows<br />

T ′k 1...k p<br />

l 1 ...l q<br />

(x ′ ) = ∂x′k 1<br />

∂x · · · ∂x′k p<br />

i 1 ∂x i p<br />

∂x j 1<br />

∂x ′l 1 · · · ∂xj q<br />

∂x ′l q<br />

T i 1...i p<br />

j 1 ...j q<br />

(x) .<br />

Here tensor indices are acted with the matrices ∂x′i which form a group GL(d, R).<br />

∂x j<br />

This is a group of all invertible real d × d matrices. This group does not have spinor<br />

representati<strong>on</strong>s.


– 93 –<br />

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡<br />

d<br />

4<br />

e<br />

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡<br />

3<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡<br />

x<br />

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡<br />

2<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡<br />

e<br />

e<br />

e<br />

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡¡ 1<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡<br />

e<br />

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡¡<br />

¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢¡¢<br />

¡<br />

M<br />

Tangent plane at a point x<br />

Fig. 9. Vielbien e a α is a collecti<strong>on</strong> of d orthog<strong>on</strong>al vectors forming a basis<br />

of the tangent space at any point x <strong>on</strong> the d-dimeni<strong>on</strong>al manifold. In string<br />

theory M with d = 2 is the string world-sheet and x ≡ (σ, τ).<br />

On the other hand, spinors are objects which transform under the spinor representati<strong>on</strong>s<br />

of the Lorentz group. The Lorentz group in d-dimensi<strong>on</strong>s is SO(d − 1, 1)<br />

and it is not GL(d, R). At each point of the manifold there is an inertial frame. In<br />

this inertial frame the Lorentz transformati<strong>on</strong>s are well defined. One can think that<br />

the Lorentz transformati<strong>on</strong>s act in the flat Minkowski space tangent to the manifold<br />

M at any given point x. In the tangent plane we introduce a basis e a α(x), a = 1, . . . , d<br />

of orth<strong>on</strong>ormal vectors:<br />

(e a , e b ) ≡ h αβ e a αe b β = η ab ,<br />

where η αβ is the flat Minkowski metric. In fact, e a α is an invertible d × d x-dependent<br />

matrix which is called vielbein. The index α is called “curved” and its acted by<br />

the general coordinate transformati<strong>on</strong>s (diffeomorphisms) as the usual vector index,<br />

while the index a is called “flat” and its acted by the local (i.e. x-dependent) Lorentz<br />

transformati<strong>on</strong>s as we will see in a moment. The inverse matrix is e α a and it obeys<br />

e a αe α b = δa b . Because of this relati<strong>on</strong>, we also have that<br />

η ab e a αe b β = h αβ .<br />

There is no a preferred choice of the basis in the tangent space and <strong>on</strong>e orth<strong>on</strong>ormal<br />

set of tangent vectors can be transformed into the other by means of local Lorentz<br />

transformati<strong>on</strong>s<br />

e a α → Λ a be b α .<br />

Introducti<strong>on</strong> of the vielbein in favour of the metric introduces additi<strong>on</strong>al degrees<br />

of freedom. Indeed, the vielbein being d × d-matrix has d 2 comp<strong>on</strong>ents, while the<br />

metric h αβ has <strong>on</strong>ly d(d+1) comp<strong>on</strong>ents in d dimensi<strong>on</strong>s. On the other hand, if we<br />

2<br />

require that the theory we c<strong>on</strong>sider has the local Lorentz symmetry, then there are


– 94 –<br />

d(d−1)<br />

2<br />

local Lorentz transformati<strong>on</strong>s which should allow to remove the additi<strong>on</strong>al<br />

(unphysical) comp<strong>on</strong>ents of the vielbein:<br />

}{{} d 2 d(d − 1)<br />

−<br />

2<br />

vielbein } {{ }<br />

local Lorentz<br />

=<br />

d(d + 1)<br />

2 } {{ }<br />

metric<br />

Thus, the vielbein brings new degrees of freedom, but they can be removed by a<br />

new symmetry which is the local Lorentz transformati<strong>on</strong>s. On the other hand, the<br />

vielbein allows <strong>on</strong>e to introduce coupling of the gravitati<strong>on</strong>al degrees of freedom with<br />

spinors.<br />

Spinor representati<strong>on</strong>s<br />

(Pseudo-)orthog<strong>on</strong>al groups (SO(d − 1, 1)) SO(d) in additi<strong>on</strong> to the usual tensor<br />

representati<strong>on</strong>s have also spinor representati<strong>on</strong>s. These are not single-valued but<br />

rather double-valued representati<strong>on</strong>s of SO(d).<br />

C<strong>on</strong>sider, for instance, the group SO(3). This group has the so-called universal<br />

covering group which is SU(2). The relati<strong>on</strong> between them is as follows<br />

SO(3) → SU(2)/Z 2 .<br />

The group SO(3) is not simply c<strong>on</strong>nected, while SU(2) is. Here<br />

Z 2 =<br />

{( ) ( )}<br />

1 0 −1 0<br />

,<br />

0 1 0 −1<br />

is the center of SU(2), i.e. a set of matrices from SU(2) which commute with all other<br />

SU(2)-matrices. Due to existence of the discrete center Z 2 representati<strong>on</strong>s of SU(2)<br />

split into two different classes of integer and half-integer spin. Only representati<strong>on</strong>s<br />

with integer spin are those of (single-valid) SO(3). Representati<strong>on</strong>s of SU(2) with<br />

half-integer spin are spinor (double-valid) representati<strong>on</strong>s of SO(3). Indeed, rotati<strong>on</strong><br />

by the angle φ around the axes given by a fixed 3dim unit vector n = (n 1 , n 2 , n 3 ),<br />

n 2 i = 1, corresp<strong>on</strong>ds the transformati<strong>on</strong><br />

(<br />

g(φ, n) = exp − i ) ( cos<br />

φ<br />

2 φ n iσ i =<br />

− in 2 3 sin φ − (in<br />

2 1 + n 2 ) sin φ )<br />

2<br />

(−in 1 + n 2 ) sin φ cos φ + in 2 2 3 sin φ ,<br />

2<br />

where σ i are three Pauli matrices. One can easily verify that g(φ, n) ∈ SU(2).<br />

Rotati<strong>on</strong> by the angle φ = 2π is an identity transformati<strong>on</strong> in SO(3) but it is not<br />

the identity in SU(2). Indeed, <strong>on</strong>e can see that<br />

g(φ + 2π, n) = −g(φ, n) ,<br />

g(φ + 4π, n) = g(φ, n) .


– 95 –<br />

Another example is provided by the Lorentz group of the 4dim Minkowski spacetime,<br />

which is O(3, 1). The spinor representati<strong>on</strong> of O(3, 1) is realized <strong>on</strong> the space<br />

C 4 , which is called a space of 4-comp<strong>on</strong>ent spinors. 18 This representati<strong>on</strong> looks as<br />

( 1<br />

)<br />

g(ω) = exp<br />

2 γab ω ab ,<br />

where γ ab is the anti-symmetric product of the 4dim γ-matrices and ω ab = −ω ba are<br />

parameters of the Lorentz transformati<strong>on</strong>. [ ]<br />

It is important to note that in general<br />

d<br />

dimensi<strong>on</strong> d the spinor has 2 2 complex comp<strong>on</strong>ents.<br />

The spinor ¯ψ = ψ † γ 0 is called the Dirac c<strong>on</strong>jugate of ψ. Its importance is explained<br />

by the fact that the quantity ¯ψψ = ψ † γ 0 ψ is an invariant of O(3, 1).<br />

In any dimensi<strong>on</strong> <strong>on</strong>e can define the charge c<strong>on</strong>jugati<strong>on</strong> matrix C. Indeed, the<br />

Clifford algebra of γ-matrices transforms into itself under operati<strong>on</strong> of transpositi<strong>on</strong><br />

{γ a , γ b } t = {(γ a ) t , (γ b ) t } = 2η ab ,<br />

therefore by irreducibility of the corresp<strong>on</strong>ding representati<strong>on</strong> of the Clifford algebra<br />

there should exists a matrix C which intertwines the original and the transposed<br />

representati<strong>on</strong> of the algebra, namely:<br />

(γ a ) t = −Cγ a C −1 .<br />

Matrix C is called the charge c<strong>on</strong>jugati<strong>on</strong> matrix.<br />

Sometimes (depending <strong>on</strong> the dimensi<strong>on</strong> and signature od space-time) it is possible<br />

to define the noti<strong>on</strong> of Majorana spinor. Majorana c<strong>on</strong>jugate spinor is, by definiti<strong>on</strong>,<br />

ψ t C. The Majorana spinor is then the spinor for which the Dirac c<strong>on</strong>jugate is equal<br />

to the Majorana c<strong>on</strong>jugate:<br />

ψ † γ 0 = ψ t C .<br />

Spinor algebra in two dimensi<strong>on</strong>s<br />

18 The group O(3, 1) is not c<strong>on</strong>nected and it has four c<strong>on</strong>nected comp<strong>on</strong>ents, which however are<br />

not simply c<strong>on</strong>nected. The comp<strong>on</strong>ent which c<strong>on</strong>tains an identity coincides with SO(3, 1), which<br />

are the transformati<strong>on</strong>s preserving orientati<strong>on</strong> of the vierbein. The transformati<strong>on</strong>s which preserve<br />

the directi<strong>on</strong> of time are called orthochr<strong>on</strong>ous and they form the subgroup SO + (3, 1). The quotient<br />

group O(3, 1)/SO + (3, 1) is the Klein four-group Z 2 ×Z 2 , which is the semidirect product of SO + (3, 1)<br />

with an element of the discrete group {1, P, T, P T }, where P and T are the space inversi<strong>on</strong> and<br />

time reversal operators<br />

P = diag(1, −1, −1, −1) , T = diag(−1, 1, 1, 1) .<br />

The covering (or spin) group of SO + (3, 1) coincides with SL(2, C). One can show that SO + (3, 1) =<br />

SL(2, C)/{I, −I} ≡ PSL(2, C). Thus, SL(2, C) is the double-cover of SO + (3, 1).


– 96 –<br />

The two-dimensi<strong>on</strong>al Dirac matrices ρ a , a = 0, 1 obey the algebra<br />

{ρ a , ρ b } = 2η ab , η ab =<br />

A particular basis foe the Clifford algebra is given by<br />

ρ 0 =<br />

( ) 0 1<br />

, ρ 1 =<br />

−1 0<br />

( ) −1 0<br />

.<br />

0 +1<br />

( ) 0 1<br />

.<br />

1 0<br />

We also define a matrix ¯ρ<br />

¯ρ = ρ 0 ρ 1 =<br />

( 1 0<br />

)<br />

0 − 1<br />

being a 2dim analogue of the 4dim matrix γ 5 . The charge c<strong>on</strong>jugati<strong>on</strong> matrix can be<br />

taken to be C = ρ 0 . The Majorana spinor is then ψ † ρ 0 = ψ t C = ψ t ρ 0 , i.e. ψ = ψ ∗<br />

which simple means that the spinor is real. Thus, in 2dim Majorana spinor is just a<br />

spinor with real comp<strong>on</strong>ents.<br />

Finally, with the help of the vielbein (“zweibein” in 2dim) <strong>on</strong>e can define the<br />

“curved” ρ-matrices:<br />

ρ α = e α aρ a .<br />

They satisfy the following algebra<br />

{ρ α , ρ β } = 2h αβ .<br />

Spinors in 2dim have various interesting properties. One of them is the so-called<br />

spin-flip identity. If we have two Majorana spinors ψ 1 and ψ 2 , then the following<br />

identity is valid<br />

¯ψ 1 ρ α1 · · · ρ αn ψ 2 = (−1) n ¯ψ2 ρ αn · · · ρ α 1<br />

ψ 1 .<br />

It is proved as follows.<br />

¯ψ 1 ρ α1 · · · ρ α n<br />

ψ 2 = ( ¯ψ 1 ρ α1 · · · ρ α n<br />

ψ 2 ) t = −ψ t 2(ρ α n<br />

) t · · · (ρ α 1<br />

) t (ρ 0 ) t ψ 1<br />

= (−1) n ψ t 2Cρ αn C −1 · · · Cρ α 1<br />

C −1 Cψ 1 = (−1) n ¯ψ2 ρ αn · · · ρ α 1<br />

ψ 1 .<br />

Another identity is<br />

ρ α ρ β ρ α = 0 . (6.1)<br />

Indeed, <strong>on</strong>e has from the Clifford algebra that ρ a ρ α = 2 and, therefore<br />

ρ α ρ β ρ α = −ρ α ρ α ρ β + 2ρ β = −2ρ β + 2ρ β = 0 .


– 97 –<br />

Finally, we discuss the completeness c<strong>on</strong>diti<strong>on</strong> for the ρ-matrices. The matrices ρ a ,<br />

¯ρ and the identity matrix I form a basis in the space of all 2 × 2-matrices. Any<br />

2 × 2-matrix M can be expanded as<br />

M = q a ρ a + ¯q¯ρ + qI .<br />

The coefficients of this expansi<strong>on</strong> are found as follows<br />

q a = 1 2 Tr(Mρa ) , ¯q = 1 2 Tr(M ¯ρ) , q = 1 2 Tr(M) .<br />

Plugging this back we obtain<br />

2M = Tr(Mρ a )ρ a + Tr(M ¯ρ)¯ρ + Tr(M)I<br />

or, more explicitly,<br />

2M αβ = M γδ<br />

[<br />

(ρ a ) δγ ρ a αβ + ¯ρ δγ ¯ρ αβ + δ δγ δ αβ<br />

]<br />

.<br />

Since M is an arbitrary matrix from here we derive the following completeness relati<strong>on</strong><br />

(ρ a ) δγ ρ a αβ + ¯ρ δγ ¯ρ αβ + δ δγ δ αβ = 2δ αγ δ βδ .<br />

Spin c<strong>on</strong>necti<strong>on</strong><br />

We would like to introduce a new local symmetry which is the Lorentz symmetry.<br />

However, we have to guarantee that the the theory we are after should be invariant<br />

w.r.t. these local transformati<strong>on</strong>s. As for the case of any local gauge invariance,<br />

the local Lorentz invariance can be achieved by introducing q gauge field ωα a b (x) for<br />

SO(d − 1, d). Here a, b are SO(d − 1, d)-indices and α is the “curved” vector index.<br />

Under the local Lorentz transformati<strong>on</strong>s with the matrix Λ this field transforms as<br />

follows<br />

ω α → Λω α Λ −1 − ∂ α ΛΛ −1 .<br />

The gauge field of the local Lorentz symmetry is usually called the spin c<strong>on</strong>necti<strong>on</strong>.<br />

The spin c<strong>on</strong>necti<strong>on</strong> plays the same role for the “flat” indices as the Christoffel<br />

c<strong>on</strong>necti<strong>on</strong> plays for the “curved” <strong>on</strong>es. We have the following substituti<strong>on</strong> of the<br />

basic objects in the theory<br />

(<br />

) (<br />

)<br />

h aβ (x), Γ δ αβ(x) → e a α(x), ωα a b(x) .<br />

Introducti<strong>on</strong> of the spin c<strong>on</strong>necti<strong>on</strong> should not change the gravitati<strong>on</strong>al c<strong>on</strong>tent of<br />

the theory. This means that the spin c<strong>on</strong>necti<strong>on</strong> should not be a new independent<br />

field, rather it should be determined in terms of vielbein. The simplest and elegant<br />

way to do it is to notice that we have the covariant derivatives<br />

D α V β = ∂ α V β + Γ β αδ V δ<br />

D α V a = ∂ α V a + ω a α bV b


– 98 –<br />

The “flat” and “curved” indices of the vector are related as V a = e a αV α . This way of<br />

transforming “flat” to “curved” indices should be valid for any tensor, in particular,<br />

<strong>on</strong>e has to have<br />

D α V a = e a βD α V β .<br />

This is possible <strong>on</strong>ly if<br />

D α e a β = ∂ α e a β − Γ λ αβe a λ + ω a α be b β = 0 .<br />

This equati<strong>on</strong> can be solved for ω α expressing it through the vielbein:<br />

ω ab<br />

α = 1 2 eβa (∂ α e b β − ∂ β e b α) − 1 2 eβb (∂ α e a β − ∂ β e a α) − 1 2 eλa e γb (∂ λ e γc − ∂ γ e λc )e c α .<br />

Since the c<strong>on</strong>necti<strong>on</strong> is completely expressed via the dynamical vielbein and, by this<br />

means, is not an independent field, it is called sometimes composite.<br />

Spin manifolds<br />

On which manifolds <strong>on</strong>e can introduce spinors? This is rather n<strong>on</strong>-trivial questi<strong>on</strong>.<br />

Vielbein can always be introduces locally in a coordinate patch U α . It is quite<br />

rare that the vielbein can be also globally defined. In the latter case such manifolds<br />

are called parallelizable. Examples of parallelizable manifolds are Lie groups. On the<br />

other hand, two-sphere is not parallelizable because there is no globally defined vector<br />

field (not talking about the vielbein) which vanishes nowhere. Thus, the vielbein<br />

is defined locally and in the intersecti<strong>on</strong> of the coordinate patches U α ∩ U β <strong>on</strong>e has<br />

e (α) (x) = Λ (αβ) (x)e (β) (x) .<br />

Here Λ (αβ) (x) is the local Lorentz transformati<strong>on</strong> (i.e. a matrix from SO(d − 1, 1))<br />

which is called the transiti<strong>on</strong> functi<strong>on</strong>. In the regi<strong>on</strong> of triple intersecti<strong>on</strong> U α ∩ U β ∩<br />

U γ = U αβγ the transiti<strong>on</strong> functi<strong>on</strong>s should satisfy the following c<strong>on</strong>diti<strong>on</strong><br />

Λ (αβ) Λ (βγ) Λ (γα) = 1 .<br />

Now if we introduce locally a spinor field ψ (α) then passing from <strong>on</strong>e coordinate patch<br />

to another <strong>on</strong>e the field must transform according to<br />

ψ (α) (x) = ¯Λ (αβ) ψ (β) (x) .<br />

Here ¯Λ (αβ) is the SO(d − 1, 1) matrix in the spinor representati<strong>on</strong>. For spinorial<br />

transiti<strong>on</strong> functi<strong>on</strong>s ¯Λ it also makes sense to require that in the triple intersecti<strong>on</strong><br />

regi<strong>on</strong> the following relati<strong>on</strong> is satisfied<br />

¯Λ (αβ) ¯Λ(βγ) ¯Λ(γα) = ±1 . (6.2)<br />

However, since the spinor representati<strong>on</strong> is double-valued, instead of ¯Λ <strong>on</strong>e can<br />

equally use −¯Λ. Thus, to define spinors <strong>on</strong> a n<strong>on</strong>-parallelizable manifold <strong>on</strong>e has


– 99 –<br />

to pick up the signs for ¯Λ (αβ) such that relati<strong>on</strong> (6.2) is satisfied. When it is possible,<br />

the corresp<strong>on</strong>ding manifold M is said to admit the spin structure and it is called the<br />

spin manifold. Note that M may admit several inequivalent spin structures.<br />

<strong>String</strong> theory c<strong>on</strong>tains world-sheet fermi<strong>on</strong>s and therefore it can be defined <strong>on</strong><br />

spin-manifolds. It turns out that in 2 and 3dim any orientable manifold is the spin<br />

manifold. It is not so in 4dim and higher. Finally, we state without proof that <strong>on</strong> a<br />

Riemann surface of genus g there are 2 2g inequivalent spin structures.<br />

6.2 Superstring acti<strong>on</strong> and its symmetrices<br />

The superstring acti<strong>on</strong> is based <strong>on</strong> two multiplest of 2dim supersymmetry. The first<br />

<strong>on</strong>e is the matter multiplet (X µ , ψ µ , F µ )<br />

.<br />

Here X µ is a bos<strong>on</strong>ic field of the Polyakov string, ψ µ is the Majorana spinor in 2dim<br />

and F µ is the real scalar, which is an auxiliary field to guarantee the equality of<br />

the bososnic and fermi<strong>on</strong>ic degrees of freedom off-shell (i.e. without usage of the<br />

equati<strong>on</strong>s of moti<strong>on</strong>). Also the target space-time index µ is just a label so that for<br />

µ = 0, . . . , d−1 we have d matter multiplets. the sec<strong>on</strong>d multiplet is the supergravity<br />

multiplet ( )<br />

e a α, χ α , A .<br />

Here χ α is the gravitino, i.e. the Majorana spinor which is also a vector of the 2dim<br />

world-sheet. The field A is an auxiliary scalar field which is needed to guarantee the<br />

equality of the bososnic and fermi<strong>on</strong>ic degrees of freedom off-shell. We note that the<br />

kinetic term for the gravitino in any dimensi<strong>on</strong> is ¯χ α γ αβγ D β χ γ and it is absent in<br />

two dimensi<strong>on</strong>s (because there are <strong>on</strong>ly two ρ -matrices while the anti-symmetrized<br />

product γ αβγ requires at least three to exist). Finally we introduce e ≡ |dete a α| = √ h.<br />

The superstring acti<strong>on</strong> is a generalizati<strong>on</strong> of the bos<strong>on</strong>ic Polyakov acti<strong>on</strong> to include<br />

the include the world-sheet fermi<strong>on</strong>ic degrees of freedom in the supersymmetric<br />

way. It has the following structure<br />

S = − 1<br />

8π<br />

∫<br />

d 2 σe<br />

(h αβ ∂ α X µ ∂ β X µ + 2i ¯ψ µ ρ α ∂ α ψ µ − i¯χ α ρ β ρ α ψ µ (<br />

∂ β X µ − i 4 ¯χ βψ µ<br />

) ) .<br />

This acti<strong>on</strong> has five local symmetries<br />

1. Local supersymmetry. Let ɛ is the Majorana spinor. We c<strong>on</strong>sider it as the<br />

infinitezimal parameter of local supersymmetry transformati<strong>on</strong>s<br />

δ ɛ X µ = i¯ɛψ µ , δ ɛ e a α = i 2¯ɛρa χ α ,<br />

δ ɛ ψ µ = 1 (<br />

2 ρα ∂ α X µ − i 2 ¯χ αψ<br />

)ɛ µ , δ ɛ χ α = 2D α ɛ .


– 100 –<br />

Here D α ɛ = ∂ α ɛ − 1 2 ω α ¯ρɛ and ω α is the c<strong>on</strong>necti<strong>on</strong> with torsi<strong>on</strong>:<br />

ω α = ω α (e) + i 4 ¯χ α ¯ρρ β χ β ,<br />

where ω α (e) = − 1 e e αaɛ βγ ∂ β e a γ is the standard composite spin c<strong>on</strong>necti<strong>on</strong> (it has<br />

<strong>on</strong>ly <strong>on</strong>e-n<strong>on</strong>-trivial comp<strong>on</strong>ent in 2dim).<br />

2. Weyl invariance. Let Λ be a bos<strong>on</strong>ic local parameter Λ = Λ(σ, τ). The Weyl<br />

transformati<strong>on</strong>s are<br />

δ Λ X µ = 0 , δ Λ e a α = Λe a α ,<br />

δ Λ ψ µ = − 1 2 Λψµ , δ Λ χ α = 1 2 Λχ α .<br />

3. Super Weyl invariance. Let η be the Majorana spinor. Under the super Weyl<br />

transformati<strong>on</strong>s <strong>on</strong>ly the gravitino transforms as<br />

δ η χ α = ρ α η .<br />

The invarince of the acti<strong>on</strong> easily follows from the identity ρ α ρ β ρ α = 0.<br />

4. Local Lorentz symmetry. Let l be a bos<strong>on</strong>ic local parameter l = l(σ, τ). The<br />

local Lorentz transformati<strong>on</strong>s are<br />

δ l X µ = 0 , δ l e a α = lɛ a be b α ,<br />

δ l ψ µ = 1 2 l¯ρψµ , δ l χ α = 1 2 l¯ρχ α .<br />

5. Reparametrizati<strong>on</strong>s. Let ξ α be a bos<strong>on</strong>ic vector parameter ξ α = ξ α (σ, τ). The<br />

reparametrizati<strong>on</strong>s (diffeomorphisms) are<br />

δ ξ X µ = ξ β ∂ β X µ , δ ξ e a α = ξ β ∂ β e b α + e a β∂ α ξ β ,<br />

δ ξ ψ µ = ξ β ∂ β ψ µ , δ ξ χ α = ξ β ∂ β χ α + χ β ∂ α ξ β .<br />

6.3 Superc<strong>on</strong>formal gauge and supermoduli<br />

The gravitino field is the reducible representati<strong>on</strong> of the Lorentz group. To decompose<br />

it into irreducible representati<strong>on</strong>s <strong>on</strong>e can use the following trick:<br />

χ α = δαχ β β =<br />

(δα β − 1 )<br />

2 ρ αρ β χ β + 1 2 ρ αρ β χ β = 1 2 ρβ ρ α χ β + 1<br />

} {{ } 2 ρ αρ β χ β<br />

˜χ α<br />

Here ˜χ α part is called ρ–traceless because, due to the identity ρ α ρ β ρ α = 0 we get<br />

ρ α ˜χ α = ρ a e α a ˜χ α = ρ a ˜χ a = 0. Indeed the gravitino χ a transforms under local Lorentz


– 101 –<br />

transformati<strong>on</strong>s as the spin-vector: 19<br />

δ l χ a = −lɛ b<br />

a χ b + 1 2 l¯ρχ a .<br />

C<strong>on</strong>diti<strong>on</strong> ρ a χ a = 0 remains invariant under these transformati<strong>on</strong>s, i.e. ρ a δ l χ a = 0.<br />

Decompositi<strong>on</strong> of the gravitino into the ρ-trace and ρ-traceless part is decompositi<strong>on</strong><br />

into two irreducible representati<strong>on</strong>s of the Lorentz group corresp<strong>on</strong>ding to helicities<br />

±3/2 and ±1/2 respectively. This decompositi<strong>on</strong> is orthog<strong>on</strong>al w.r.t. the scalar<br />

product (φ|ψ) = ∫ d 2 σ ¯φ α ψ α .<br />

The local supersymmetry transformati<strong>on</strong> for the gravitino filed can be also decomposed<br />

into the traceless- and the trace-parts:<br />

Here we defined the operator<br />

δ ɛ χ α = 2D α ɛ = 2(Πɛ) α + ρ α ρ β D β ɛ<br />

} {{ }<br />

trace part<br />

(Πɛ) α = 1 2 ρβ ρ α D β ɛ , =⇒ ρ α (Πɛ) α = 0 .<br />

Locally <strong>on</strong>e can show that there always exists a spinor κ such that ˜χ α = ρ β ρ α D β κ.<br />

Comparing this with the supersymmetry transformati<strong>on</strong> for χ α we c<strong>on</strong>clude that κ<br />

can always be eliminated (locally!) by a supersymmetry variati<strong>on</strong>. The possibility<br />

to eliminate κ globally depends <strong>on</strong> the existence of a globally defined spinor ɛ which<br />

solves the equati<strong>on</strong><br />

(Πɛ) α = τ α<br />

for arbitrary τ α satisfying the c<strong>on</strong>diti<strong>on</strong> ρ α τ α =0. Global solvability of the last expressi<strong>on</strong><br />

relies <strong>on</strong> the absence of zero modes of the operator Π † : (Π † τ) = −2D α τ α .<br />

This equati<strong>on</strong> is the supercousin of the bos<strong>on</strong>ic equati<strong>on</strong><br />

(P ξ) αβ = t αβ<br />

whose global solvability relies <strong>on</strong> the absence of zero modes of P † . According to our<br />

discussi<strong>on</strong> of the bos<strong>on</strong>ic case it makes sense to call<br />

and also<br />

19 The element ɛ b<br />

a<br />

dim kerP † = moduli<br />

dim kerΠ † = supermoduli<br />

dim kerP = c<strong>on</strong>formal Killing vectors<br />

dim kerΠ = c<strong>on</strong>formal Killing spinors .<br />

( ) ( −1 0 0 1<br />

= η ac ɛ cb is the following matrix ɛa b =<br />

0 1 −1 0<br />

)<br />

=<br />

( ) 0 − 1<br />

.<br />

−1 0


– 102 –<br />

By using reparametrizati<strong>on</strong>s and local Lorentz transformati<strong>on</strong>s the zweibein can<br />

be brought to the form e a α = e φ δ a α which is locally always possible. The gauge<br />

e a α = e φ δ a α ,<br />

χ α = ρ α λ<br />

is called superc<strong>on</strong>formal gauge. In classical theory the Weyl symmetry and super<br />

Weyl symmetry can be used to eliminate the remaining gravitati<strong>on</strong>al degrees of<br />

freedom φ and λ. In quantum theory it will be possible in critical dimensi<strong>on</strong> <strong>on</strong>ly.<br />

6.4 Acti<strong>on</strong> in the superc<strong>on</strong>formal gauge<br />

In the superc<strong>on</strong>formal gauge the acti<strong>on</strong> becomes rather simple<br />

S = − 1 ∫<br />

d 2 σ<br />

(∂ α X µ ∂ α X µ + 2i<br />

8π<br />

¯ψ<br />

)<br />

µ ρ α ∂ α ψ µ . (6.3)<br />

The world-sheet indices are now raised and lowered with the help of the flat worldsheet<br />

metric η aβ and ρ α = δ α a ρ a . This acti<strong>on</strong> is invariant w.r.t. local reprametrizati<strong>on</strong>s<br />

and supersymmetry transformati<strong>on</strong>s which satisfy the requirement<br />

P ξ = 0 , Πɛ = 0 .<br />

We would like to check directly that the acti<strong>on</strong> (6.3) is invariant under the supersymmetry<br />

transformati<strong>on</strong>s<br />

δ ɛ X µ = i¯ɛψ µ ,<br />

δ ɛ ψ µ = 1 2 ρα ∂ α X µ ɛ<br />

δ ɛ ¯ψµ = − 1 2¯ɛρα ∂ α X µ<br />

provided the parameter ɛ satisfies the following equati<strong>on</strong><br />

ρ β ρ α ∂ β ɛ = 0 . (6.4)<br />

To check the invariance we perform the variati<strong>on</strong><br />

δ ɛ S = − 1 ∫<br />

d 2 σ<br />

(2∂ α X µ ∂ α (i¯ɛψ µ ) + i<br />

8π<br />

¯ψ<br />

)<br />

µ ρ α ∂ α (ρ β ∂ β X µ ɛ) − ¯ɛρ α ∂ α X µ ρ β ∂ β ψ µ .<br />

Now we integrate by parts the first term and write out the sec<strong>on</strong>d term more explicitly<br />

δ ɛ S = − 1 ∫ (<br />

d 2 σ − 2i□X µ ¯ɛψ µ + i<br />

8π<br />

¯ψ µ ρ α ρ β ∂ α ∂ β X µ ɛ<br />

}{{}<br />

η αβ<br />

+ i ¯ψ µ ρ α ρ β ∂ β X µ ∂ α ɛ − ¯ɛρ α ∂ α X µ ρ β ∂ β ψ µ<br />

)<br />

.


– 103 –<br />

Now we apply the spin-flip identity in the sec<strong>on</strong>d and the third term and get<br />

δ ɛ S = − 1 ∫ (<br />

d 2 σ − 2i□X µ ¯ɛψ µ + i¯ɛψ µ □X µ<br />

8π<br />

)<br />

+ i∂ α¯ɛρ β ρ α ψ µ ∂ β X µ − ¯ɛρ β ρ α ∂ β X µ ∂ α ψ µ .<br />

Finally, we integrate the last term by parts and get<br />

δ ɛ S = − 1 ∫<br />

d 2 σ 2i∂ α¯ɛρ β ρ α ψ µ ∂ β X µ .<br />

8π<br />

This vanishes as the c<strong>on</strong>sequence of eq.(6.4).<br />

We can write equati<strong>on</strong> (6.4) more explicitly<br />

) ( )<br />

)<br />

ρ β ρ 0 ∂ β ɛ =<br />

(ρ 0 ρ 0 ∂ 0 + ρ 1 ρ 0 ∂ 1 ɛ = − ρ 0 ρ 0 ∂ 0 + ρ 0 ρ 1 ∂ 1 ɛ =<br />

(∂ 0 + ¯ρ∂ 1 ɛ = 0 ,<br />

)<br />

)<br />

)<br />

ρ β ρ 1 ∂ β ɛ =<br />

(ρ 0 ρ 1 ∂ 0 + ρ 1 ρ 1 ∂ 1 ɛ =<br />

(ρ 0 ρ 1 ∂ 0 + ρ 1 ρ 1 ∂ 1 ɛ =<br />

(¯ρ∂ 0 + ∂ 1 ɛ = 0 .<br />

Since ¯ρ 2 = 1 the sec<strong>on</strong>d equati<strong>on</strong> is obtained from the first by multiplying with ¯ρ<br />

and by this reas<strong>on</strong> it is redundant. To analyze the first equati<strong>on</strong> it is c<strong>on</strong>venient to<br />

denote the comp<strong>on</strong>ents of any spinor as follows<br />

ψ =<br />

(<br />

ψ+<br />

ψ −<br />

)<br />

and ɛ =<br />

We thus see that the first equati<strong>on</strong> reduces to<br />

(<br />

ɛ+<br />

)<br />

.<br />

ɛ −<br />

(∂ 0 + ∂ 1 )ɛ + = ∂ + ɛ + = 0 , (∂ 0 − ∂ 1 )ɛ − = ∂ − ɛ − = 0<br />

One cab define the spinors with upper indices by using the following c<strong>on</strong>venti<strong>on</strong><br />

ψ − = ψ + , ψ + = −ψ − .<br />

With this c<strong>on</strong>venti<strong>on</strong> we obtain that comp<strong>on</strong>ents of the Majorana spinor which is a<br />

parameter of the supersymmetry transformati<strong>on</strong>s satisfy the equati<strong>on</strong>s<br />

∂ + ɛ − = ∂ − ɛ + = 0 =⇒ ɛ ± ≡ ɛ ± (σ ± ) .<br />

This equati<strong>on</strong>s should be c<strong>on</strong>tracted with the equati<strong>on</strong>s defining the c<strong>on</strong>formal Killing<br />

vectors, i.e. reparametrizati<strong>on</strong>s which do not destroy the c<strong>on</strong>formal gauge choice:<br />

∂ + ξ − = ∂ − ξ + = 0 =⇒ ξ ± ≡ ξ ± (σ ± ) .<br />

On-shell closer of the supersymmetry algebra


– 104 –<br />

C<strong>on</strong>sider first the commutator of two supersymmetry variati<strong>on</strong>s applied to a bos<strong>on</strong>ic<br />

field X µ :<br />

[δ ɛ1 , δ ɛ2 ]X µ = i¯ɛ 1 δ ɛ2 ψ µ − i¯ɛ 1 δ ɛ2 ψ µ = i 2(¯ɛ1 ρ α ɛ 2 − ¯ɛ 2 ρ α ɛ 1<br />

)<br />

∂α X µ = i¯ɛ 1 ρ α ɛ 2 ∂ α X µ ,<br />

where the last formula stems from the spin-flip property. We see that the commutator<br />

of two super-symmetry transformati<strong>on</strong>s generates a diffeomorphism transformati<strong>on</strong><br />

[δ ɛ1 , δ ɛ2 ]X µ = ξ α ∂ α X µ<br />

with the parameter ξ α = i¯ɛ 1 ρ α ɛ 2 . In fact, this is not an arbitrary diffeomorohism,<br />

rather it is a c<strong>on</strong>formal transformati<strong>on</strong>, because ξ α is nothing else as a c<strong>on</strong>formal<br />

Killing vector! Thus, bilinear combinati<strong>on</strong>s made of c<strong>on</strong>formal spinors<br />

C<strong>on</strong>sider now the commutator of two supersymmetry variati<strong>on</strong>s applied to a<br />

fermi<strong>on</strong> ψ µ :<br />

[δ ɛ1 , δ ɛ2 ]ψ µ = 1 2 ∂ α(δ ɛ1 X µ )ρ α ɛ 2 − 1 2 ∂ α(δ ɛ2 X µ )ρ α ɛ 1 = i 2 ∂ α(¯ɛ 1 ψ µ )ρ α ɛ 2 − i 2 ∂ α(¯ɛ 2 ψ µ )ρ α ɛ 1<br />

Therefore,<br />

[δ ɛ1 , δ ɛ2 ]ψ µ = i 2 (∂ α¯ɛ 1 ψ µ )ρ α ɛ 2 − i 2 (∂ α¯ɛ 2 ψ µ )ρ α ɛ 1 +<br />

+ i 2 (¯ɛ 1∂ α ψ µ )ρ α ɛ 2 − i 2 (¯ɛ 2∂ α ψ µ )ρ α ɛ 1 (6.5)<br />

C<strong>on</strong>straints<br />

In the original (gauge-unfixed) theory we have a world-sheet metric (vielbein) and<br />

the gravitino field. They are removed up<strong>on</strong> impositi<strong>on</strong> of the superc<strong>on</strong>formal gauge.<br />

However, before we fix the gauge, the metric and the gravitino have their equati<strong>on</strong>s<br />

of moti<strong>on</strong> which become the c<strong>on</strong>straints <strong>on</strong> the other fields of the theory after fixing<br />

the gauge. The stress tensor in now defined as<br />

T αβ = − 2π e<br />

δS<br />

e αa .<br />

We can also define the supercurrent as resp<strong>on</strong>se of the acti<strong>on</strong> for variati<strong>on</strong> of the<br />

gravitino field<br />

G α = −i 2π δS<br />

e δ ¯χ . α<br />

Analogously to what was in the bos<strong>on</strong>ic case the stress tensor T αβ will generate<br />

c<strong>on</strong>formal transformati<strong>on</strong>s, while the new object G α appears to be a generator of the<br />

supersymmetries. Equati<strong>on</strong>s of moti<strong>on</strong><br />

δe β a<br />

T αβ = 0 = G α


– 105 –<br />

are c<strong>on</strong>straints <strong>on</strong> the dynamics of our system. Using the acti<strong>on</strong> we find<br />

T αβ = 1 2 ∂ αX µ ∂ β X µ − 1 4 η αβ∂ γ X µ ∂ γ X µ + i 4 ¯ψρ α ∂ β ψ µ + i 4 ¯ψρ β ∂ α ψ µ<br />

G α = 1 4 ρβ ρ α ψ µ ∂ β X µ<br />

Note that G α is ρ-traceless, i.e.<br />

ρ α G α = 0 .<br />

This equati<strong>on</strong> is an analog of T α α = 0. Finally, by using equati<strong>on</strong>s of moti<strong>on</strong> <strong>on</strong>e can<br />

show that the stress tensor and the supercurrent are c<strong>on</strong>served<br />

∂ α T αβ = 0 , ∂ α G α = 0 .<br />

These c<strong>on</strong>servati<strong>on</strong> laws lead to existence of infinite number of c<strong>on</strong>served charges. To<br />

analyze the algebra of c<strong>on</strong>traints in more detail it is c<strong>on</strong>venient to use the world-sheet<br />

light-c<strong>on</strong>e coordinates σ ± . In the light-c<strong>on</strong>e coordinates the acti<strong>on</strong> becomes<br />

S = 1<br />

2π<br />

Equati<strong>on</strong>s of moti<strong>on</strong> are<br />

∫<br />

(<br />

)<br />

d 2 σ ∂ + X∂ − X + i(ψ + ∂ − ψ + + ψ − ∂ + ψ − ) .<br />

∂ + ∂ − X µ = 0 , ∂ − ψ µ + = ∂ + ψ µ − = 0 .<br />

Solving equati<strong>on</strong>s of moti<strong>on</strong> for fermi<strong>on</strong>s we get<br />

ψ µ + = ψ µ +(σ + ) , ψ µ − = ψ µ −(σ − ) .<br />

This, it appears that two comp<strong>on</strong>ents of the Majorana fermi<strong>on</strong> are left- and rightmoving<br />

fields <strong>on</strong> the world-sheet.<br />

are<br />

The comp<strong>on</strong>ents of the stress-tensor T +− = 0 = T −+ , while the other comp<strong>on</strong>ents<br />

T ++ = 1 2 ∂ +X∂ + X + i 2 ψ +∂ + ψ + ,<br />

T −− = 1 2 ∂ −X∂ − X + i 2 ψ −∂ − ψ − .<br />

The comp<strong>on</strong>ents of the supercurrent are<br />

G + = 1 2 ψ +∂ + X ,<br />

G − = 1 2 ψ −∂ − X .<br />

The c<strong>on</strong>servati<strong>on</strong> laws look in the light-c<strong>on</strong>e coordinates as<br />

∂ − G + = ∂ + G − = 0 , ∂ − T ++ = ∂ + T −− = 0 .


– 106 –<br />

Now we note that in additi<strong>on</strong> to the c<strong>on</strong>served charges generated by the stress tensor:<br />

∫<br />

Q ξ ± = dσ ξ ± (σ ± )T ±± (σ, τ)<br />

we will have the c<strong>on</strong>served charges generated by the supercurrent<br />

∫<br />

G ɛ ± = dσ ɛ ± (σ ± )G ± (σ, τ) .<br />

6.5 Boundary c<strong>on</strong>diti<strong>on</strong>s<br />

Varying the acti<strong>on</strong> to derive the equati<strong>on</strong>s of moti<strong>on</strong> we will get the following boundary<br />

term<br />

∫ (<br />

)<br />

dσ∂ σ ψ + δψ + − ψ − δψ − .<br />

For the case of closed string, to make this term vanishing <strong>on</strong>e has to impose the<br />

following c<strong>on</strong>diti<strong>on</strong><br />

)<br />

)<br />

(ψ + δψ + − ψ − δψ − (σ) −<br />

(ψ + δψ + − ψ − δψ − (σ + 2π) = 0 .<br />

Since the fermi<strong>on</strong>s ψ + and ψ − are independent this equati<strong>on</strong> implies that<br />

The following terminology is standard<br />

ψ + (σ) = ±ψ + (σ + 2π) ,<br />

ψ − (σ) = ±ψ − (σ + 2π) .<br />

• Periodic boundary c<strong>on</strong>diti<strong>on</strong>s in σ are called Ram<strong>on</strong>d boundary c<strong>on</strong>diti<strong>on</strong>s and<br />

they are denoted by the letter “R”.<br />

• Anti-periodic boundary c<strong>on</strong>diti<strong>on</strong>s in σ are called Nevew-Schwarz boundary<br />

c<strong>on</strong>diti<strong>on</strong>s and they are denoted by the letter “NS”.<br />

Universally, all fermi<strong>on</strong>ic quantities <strong>on</strong> the world-sheet have the following boundary<br />

c<strong>on</strong>diti<strong>on</strong>s<br />

ψ(σ + 2π) = e 2πiθ ψ(σ) ,<br />

where θ = 0 in the R-sector and θ = 1/2 in the NS sector.<br />

Boundary c<strong>on</strong>diti<strong>on</strong>s for ψ + and ψ − can be chosen independently, which gives in<br />

total four possibilities<br />

(R, R) , (NS, NS) , (R, NS) , (NS, R)<br />

The boundary c<strong>on</strong>diti<strong>on</strong>s for the two comp<strong>on</strong>ents of the supersymmetry parameter<br />

should be chosen in such a way as to make the variati<strong>on</strong> δX µ = i¯ɛψ µ periodic.


– 107 –<br />

As we will see the states in the R sector give the space-time fermi<strong>on</strong>s, while the<br />

states in the NS sector are the space-time bos<strong>on</strong>s. This further gives that (R,R) and<br />

(NS,NS) sectors are space-time bos<strong>on</strong>s while (R,NS) and (NS,R) are fermi<strong>on</strong>s.<br />

For the open string case we have that<br />

ψ + δψ + − ψ − δψ −<br />

must vanish at σ = 0 and σ = π. If we assume that ψ + = αψ − at the end point of<br />

string, then<br />

(α 2 − 1)ψ − δψ − = 0<br />

which allows for α = ±1. Thus, at each end of the string we should have ψ + = ±ψ − .<br />

We can always agree to choose ψ + (0, τ) = ψ − (0, τ) as it is a matter of c<strong>on</strong>venti<strong>on</strong>,<br />

then <strong>on</strong> the other hand of the string we have two possibilities<br />

6.6 Superc<strong>on</strong>formal algebra<br />

ψ + (π, τ) = ψ − (π, τ) (Ram<strong>on</strong>d) ,<br />

ψ + (π, τ) = −ψ − (π, τ) (Neveu − Schwarz) .<br />

In order to compute the Poiss<strong>on</strong> bracket between the comp<strong>on</strong>ents of the stress tensor<br />

and the supercurrent we need the fundamental Poiss<strong>on</strong> bracket for the fermi<strong>on</strong>s. The<br />

Dirac acti<strong>on</strong> leads to the following bracket<br />

{ψ µ +(σ), ψ ν +(σ ′ )} = −2πiδ(σ − σ ′ )η µν<br />

{ψ µ −(σ), ψ ν −(σ ′ )} = −2πiδ(σ − σ ′ )η µν .<br />

Using these brackets together with brackets between the bos<strong>on</strong>ic fields <strong>on</strong>e find the<br />

following Poiss<strong>on</strong> algebra of the c<strong>on</strong>straints<br />

(<br />

)<br />

{T ++ (σ), T ++ (σ ′ )} = −2π 2T ++ (σ ′ )∂ ′ + ∂ ′ T ++ (σ ′ ) δ(σ − σ ′ ) ,<br />

3<br />

)<br />

{T ++ (σ), G + (σ ′ )} = −2π(<br />

2 G +(σ ′ )∂ ′ + ∂ ′ G + (σ ′ ) δ(σ − σ ′ ) ,<br />

{G + (σ), G + (σ ′ )} = −iπT ++ (σ)δ(σ − σ ′ ) .<br />

This is the so-called N = 1 superc<strong>on</strong>fomal algebra in 2dim. Here N = 1 refers to<br />

the fact that supersymmetry transformati<strong>on</strong>s are performed with the help of <strong>on</strong>e<br />

Majorana spinor.<br />

The acti<strong>on</strong> of the supercurrent <strong>on</strong> the bos<strong>on</strong>ic and fermi<strong>on</strong>ic fields generate supersymmetry<br />

transformati<strong>on</strong>s (bos<strong>on</strong>ic field transforms into fermi<strong>on</strong>ic <strong>on</strong>e and vice-versa):<br />

{G + (σ), X µ (σ ′ )} = −π ψ µ +(σ)δ(σ − σ ′ ) ,<br />

{G + (σ), ψ µ (σ ′ )} = −iπ ∂ + X µ +(σ)δ(σ − σ ′ )


– 108 –<br />

One can also check that the world-sheet fermi<strong>on</strong> transforms under c<strong>on</strong>formal transformati<strong>on</strong>s<br />

as the c<strong>on</strong>formal field with weigh 1/2.<br />

C<strong>on</strong>sider closed strings. Using the mode expansi<strong>on</strong><br />

ψ + (σ, τ) = ∑ ¯bµ r e −ir(τ+σ) ,<br />

r∈Z+θ<br />

ψ − (σ, τ) = ∑<br />

r∈Z+θ<br />

where θ = 0 in the R-sector and θ = 1 2<br />

algebra of oscillators<br />

b µ r e −ir(τ−σ) ,<br />

{b µ r , b ν s} = −iη µν δ r+s ,<br />

{¯b µ r , ¯b ν s} = −iη µν δ r+s ,<br />

{b µ r , ¯b ν s} = 0 .<br />

The reality of the Majorana spinor implies that<br />

(b µ r ) † = b µ −r , (¯b µ r ) † = ¯b µ −r .<br />

in the NS-sector, we obtain the Poiss<strong>on</strong><br />

Introducing the modes of the stress tensor and the supercurrent<br />

L m = 1<br />

2π<br />

∫ 2π<br />

0<br />

dσe −imσ T −− ,<br />

G m = 1 π<br />

∫ 2π<br />

0<br />

dσe −irσ G − .<br />

Notice that the supercurrent G − satisfies the same boundary c<strong>on</strong>diti<strong>on</strong> as the fermi<strong>on</strong><br />

ψ − . Substituting the mode expansi<strong>on</strong> we get<br />

L m = 1 ∑<br />

α −n α m+n + 1 ∑ (<br />

r + m )<br />

b −r b m+r ,<br />

2<br />

2 2<br />

n∈Z<br />

G r = ∑ n∈Z<br />

α −n b r+n .<br />

These generators generate the classical super-Virasoro algebra<br />

{L m , L n } = −i(m − n)L m+n ,<br />

( 1<br />

)<br />

{L m , G r } = −i<br />

2 m − n G m+r ,<br />

{G r , G s } = −2iL r+s .<br />

7. Quantum fermi<strong>on</strong>ic string<br />

Can<strong>on</strong>ical quantizati<strong>on</strong> is again performed by substituting the Poiss<strong>on</strong> bracket for<br />

the (anti)-commutator: { , } pb → 1 [ , ]. Therefore, the anti-commutators for the<br />

i<br />

quantum fermi<strong>on</strong>ic fields are<br />

{ψ µ +(σ), ψ ν +(σ ′ )} = 2πδ(σ − σ ′ )η µν<br />

{ψ µ −(σ), ψ ν −(σ ′ )} = 2πδ(σ − σ ′ )η µν .<br />

r


– 109 –<br />

Anti-commutators of the modes are<br />

{b µ r , b ν s} = η µν δ r+s .<br />

We again see that we can split oscillators into creati<strong>on</strong> and annihilati<strong>on</strong> operators<br />

according to the sign of their index, namely<br />

• Oscillators with r > 0 are annihilati<strong>on</strong> operators ,<br />

• Oscillators with r < 0 are creati<strong>on</strong> operators .<br />

However, the modes with r = 0 which occur in the Ram<strong>on</strong>d sector <strong>on</strong>ly require<br />

special care. Indeed, in the bos<strong>on</strong>ic modes α µ 0 and ᾱ µ 0 corresp<strong>on</strong>d to the center of<br />

mass momentum of the string. Analogously, b µ 0 and ¯b µ 0 are distinguished from all the<br />

other modes, in particular, they form the Clifford algebra<br />

and analogously for ¯b µ 0.<br />

{b µ 0, b ν 0} = η µν<br />

The super-Virasoro generators are again defined as normal ordered expressi<strong>on</strong>s<br />

L m = 1 ∑<br />

: α −n α m+n : + 1 ∑ (<br />

r + m )<br />

: b −r b m+r : ,<br />

2<br />

2 2<br />

n∈Z<br />

G r = ∑ n∈Z<br />

α −n b r+n .<br />

r<br />

Only the generator L 0 is ambiguous doe to the undetermined normal ordering c<strong>on</strong>stant.<br />

Ignoring this c<strong>on</strong>stant for the moment we obtain the following answer for the<br />

quantum super-Virasoro algebra<br />

[L m , L n ] = (m − n)L m+n + d 8 m(m2 − 2ω)δ m+n ,<br />

( 1<br />

)<br />

[L m , G r ] =<br />

2 m − n G m+r ,<br />

{G r , G s } = 2L r+s + d (<br />

r 2 − ω )<br />

δ r+s .<br />

2 2<br />

Here ω = 0 for the R-sector and ω = 1 for the NS-sector. Both the R- and NSalgebras<br />

formally agree except the linear terms in anomalies. The linear term can<br />

2<br />

be changed by shifting the L 0 generator. Indeed, <strong>on</strong>e can see that if <strong>on</strong>e shifts<br />

L R 0 → L R 0 + d then both algebras have formally the same structure with ω = 1. Still<br />

16 2<br />

the R- and NS-algebras are very different. For instance, in the NS-sector the five<br />

generators L 1 , L 0 , L −1 , G 1/2 , G −1/2 form a closed superalgebra known as OSp(1|2). In<br />

the R-sector just adding to the generators L 1 , L 0 , L −1 the generator G 0 <strong>on</strong>e generates<br />

the whole infinite-dimensi<strong>on</strong>al algebra.


– 110 –<br />

The oscillator ground state is defined in both sectors as<br />

α µ m|0〉 = b µ r |0〉 = 0 , m, r > 0 .<br />

Here the dependence <strong>on</strong> the center of mass momentum is suppressed. In the Ram<strong>on</strong>d<br />

sector <strong>on</strong>e also has the zero mode b µ 0. The level number operator is<br />

where<br />

N = N (α) + N (b) ,<br />

N (α) =<br />

N (b) =<br />

∞∑<br />

α −m α m ,<br />

m=1<br />

∞∑<br />

r∈Z+θ>0<br />

rb −r b r .<br />

Note that the zero mode in the Ram<strong>on</strong>d sector does not c<strong>on</strong>tribute to the number<br />

operator! This leads to the fact the mass operator commutes with b µ 0: [b µ 0, M 2 ] = 0,<br />

i.e. the states |0〉 and b µ 0|0〉 have the same mass. These states are degenerate. On the<br />

other hand, all other oscillators α µ n, b µ r with n, r < 0 increase α ′ M 2 by 2n and 2r units<br />

respectively. This means that in the NS-sector the ground state is unique and it has<br />

Lorentz spin zero. In the R-sector the ground state is degenerate and since b µ 0 form<br />

the Clifford algebra the ground state is a spinor of the Lorentz group SO(d − 1, 1).<br />

This explains why in the NS-sector all the states are space-time bos<strong>on</strong>s, while in the<br />

R-sector they are all fermi<strong>on</strong>s. Indeed, all creati<strong>on</strong> operators have vector Lorentz<br />

index and by this reas<strong>on</strong> they cannot c<strong>on</strong>vert a space-time bos<strong>on</strong> into a space-time<br />

fermi<strong>on</strong> or vice versa. If we will write the Ram<strong>on</strong>d ground state as |a〉, where a is a<br />

SO(d − 1, d) spinor index, the b µ 0 act <strong>on</strong> it as the usual Γ-matrices<br />

b µ 0|a〉 = 1 √<br />

2<br />

(Γ µ ) a b|b〉 .<br />

Here Γ µ are the usual Γ-matrices of the d-dimensi<strong>on</strong>al Minkowski space and they<br />

satisfy the Clifford algebra {Γ µ , Γ ν } = 2η µν .<br />

We will not go into discussi<strong>on</strong> of the covariant quantizati<strong>on</strong> but will just state<br />

that c<strong>on</strong>sistency of the quantum theory will impose the following restricti<strong>on</strong>s <strong>on</strong> the<br />

c<strong>on</strong>stant a of the normal ordering ambiguity (for the Ram<strong>on</strong>d and Neveu-Schwarz<br />

sectors) and the dimensi<strong>on</strong> d of the target space-time:<br />

a NS = 1 2 , a R = 0 , d = 10 .<br />

The same result follows from the c<strong>on</strong>diti<strong>on</strong> of n<strong>on</strong>-anomalous Lorentz algebra in the<br />

light-c<strong>on</strong>e gauge. Instead of d = 26 found for bos<strong>on</strong>ic string, quantum fermi<strong>on</strong>ic<br />

string chooses to live in a ten-dimensi<strong>on</strong>al world.


– 111 –<br />

7.1 Light-c<strong>on</strong>e quantizati<strong>on</strong> and superstring spectrum<br />

As we have established impositi<strong>on</strong> of the superc<strong>on</strong>formal gauge does not completely<br />

remove the gauge (unphysical) degrees of freedom. The superc<strong>on</strong>formal transformati<strong>on</strong>s<br />

which do not destroy the superc<strong>on</strong>formal gauge choice are left. In order to<br />

remove the remaining unphysical degrees of freedom <strong>on</strong>e can try to fix the light-c<strong>on</strong>e<br />

gauge, similar to as was d<strong>on</strong>e for bos<strong>on</strong>ic string. We can also fix<br />

X + = α ′ p + τ<br />

in our fermi<strong>on</strong>ic theory and this choice will completely remove the reparametrizati<strong>on</strong><br />

invariance. However, the local supersymmetry transformati<strong>on</strong>s obeying the equati<strong>on</strong>s<br />

∂ + ɛ − = ∂ − ɛ + = 0<br />

are still left over. These transformati<strong>on</strong>s can be used in order to completely eliminate<br />

the fermi<strong>on</strong>ic field ψ + , where this time ψ ± refer to the target-space light-c<strong>on</strong>e<br />

comp<strong>on</strong>ents<br />

ψ ± = √ 1 (ψ 0 ± ψ d−1 ) .<br />

2<br />

This is equivalent to putting to zero the modes b + r for all r. After this gauge choice<br />

is d<strong>on</strong>e we can solve the super-Virasoro c<strong>on</strong>straints and find the l<strong>on</strong>gitudinal modes<br />

(remember that T = 1<br />

2πα ′ )<br />

∂ ± X − = 1 (<br />

)<br />

(∂<br />

α ′ p + ± X i ) 2 + iψ±∂ i ± ψ±<br />

i<br />

ψ − ± = 2<br />

α ′ p + ψi ±∂ ± X i .<br />

This shows that <strong>on</strong>ly the transversal comp<strong>on</strong>ents X i and ψ i are physical degrees of<br />

freedom. In terms of oscillators the previous equati<strong>on</strong>s read<br />

α − m =<br />

1<br />

(<br />

√ : αnα i i<br />

2α′ p + m−n : + ∑ r<br />

b − r = 2 ∑<br />

α<br />

α ′ p<br />

r−qb i i + q .<br />

q<br />

(m<br />

2 − r) : b i rb i m−r : −2aδ m<br />

)<br />

For the case of closed strings these expressi<strong>on</strong>s must be supplemented by the analogous<br />

<strong>on</strong>es for the left-moving modes. Here we also include a normal ordering c<strong>on</strong>stant<br />

a which is a = 1 in the NS sector and a = 0 in the R sector.<br />

2<br />

where<br />

The mass operator is<br />

( ∑<br />

α ′ MR 2 = 2<br />

M 2 = M 2 R + M 2 L ,<br />

n>0<br />

α i −nα i n + ∑ r>0<br />

)<br />

rb i −rb i r − a


– 112 –<br />

and similar for ML 2 . For the case of open string we have<br />

( ∑<br />

α ′ MR 2 =<br />

n>0<br />

α i −nα i n + ∑ r>0<br />

)<br />

rb i −rb i r − a .<br />

α ′ mass 2 rep of SO(8) little (−1) F rep of little<br />

group<br />

group<br />

NS − sector<br />

− 1 2<br />

|0〉 SO(9) −1 1<br />

0 b i −1/2|0〉<br />

} {{ }<br />

8 v<br />

SO(8) +1 8 v<br />

+ 1 α i 2<br />

}<br />

−1|0〉 , b i<br />

{{ }<br />

−1/2b j −1/2<br />

} {{ |0〉<br />

}<br />

8 v<br />

28<br />

SO(9) −1 36<br />

+1<br />

b i −1/2b j −1/2 bk −1/2|0〉<br />

} {{ }<br />

56 v<br />

+1 SO(9) 84 + 44<br />

α−1b i j −1/2<br />

} {{ |0〉<br />

}<br />

1+28+35 v<br />

, b i −3/2|0〉<br />

} {{ }<br />

8 v<br />

+1<br />

R − sector<br />

|a〉<br />

}{{}<br />

+1 8 s<br />

8 s<br />

0 SO(8)<br />

|ȧ〉<br />

}{{}<br />

8 c<br />

−1 8 c<br />

α−1|a〉<br />

i ,<br />

} {{ }<br />

b i }<br />

−1|ȧ〉<br />

{{ }<br />

+1 128<br />

8 c +56 c 8 s +56 s<br />

+1 SO(9)<br />

α−1|ȧ〉<br />

i ,<br />

} {{ }<br />

b i }<br />

−1|a〉<br />

{{ }<br />

−1 128<br />

8 s +56 s 8 c +56 c<br />

Tab. 5. The lowest levels of the open fermi<strong>on</strong>ic string spectrum.


– 113 –<br />

In the closed string case we have in additi<strong>on</strong> the c<strong>on</strong>diti<strong>on</strong> of level-matching<br />

which requires that for physical states<br />

M 2 L = M 2 R .<br />

Finally, we give expressi<strong>on</strong>s for the light-c<strong>on</strong>e acti<strong>on</strong><br />

S l.c = 1 ∫ ((Ẋi d 2 σ ) 2 − (X ′i ) 2 − 2i<br />

8π<br />

¯ψ<br />

)<br />

i ρ a ∂ α ψ i<br />

and the Hamilt<strong>on</strong>ian<br />

(7.1)<br />

H = (p i ) 2 + ∑ n>0<br />

(α i −nα i n + ᾱ i −nᾱ i n) + ∑ r>0<br />

r(b i −rb i r + ¯b i −r¯b i r) − 2a .<br />

Note that every sector, R and NS, has its own Hamilt<strong>on</strong>ian.<br />

Let us now analyze the closed string spectrum. We first discuss the right-moving<br />

part which (up to a mass rescaling by 2) is equivalent to the spectrum of open<br />

fermi<strong>on</strong>ic string.<br />

• NS-sector. The ground state is the oscillator vacuum |0〉 with α ′ M 2 = −a. The<br />

first excited state is b i −1/2 |0〉 with α′ M 2 = 1 − a. This is a vector of SO(d − 2),<br />

2<br />

where the critical dimensi<strong>on</strong> d = 10. Since the little Lorentz group for massless<br />

states in d-dimensi<strong>on</strong>s is SO(d − 2) this state must be massless which gives the<br />

normal ordering c<strong>on</strong>stant to be a = 1 . At the next level <strong>on</strong>e has the states<br />

2<br />

α−1|0〉 i and b i −1/2 bj −1/2 |0〉 with α′ M 2 = 1 . The number of these bos<strong>on</strong>ic states<br />

2<br />

is 8 + 28 = 36 = 9×8 , they are comprise an antisymmetric representati<strong>on</strong> of<br />

2<br />

SO(9), the little Lorentz group for massive states.<br />

• R-sector. The Ram<strong>on</strong>d ground state is a spinor of SO(9, 1). The dimensi<strong>on</strong><br />

of the Dirac spinor in d = 10 is 2 d 2 = 2 5 = 32, i.e. it has 32 complex or 64<br />

real comp<strong>on</strong>ents. In ten dimensi<strong>on</strong>s it is possible to impose both Majorana<br />

and Weyl c<strong>on</strong>diti<strong>on</strong>s 20 which reduce the number of independent comp<strong>on</strong>ents<br />

to 64 = 16. On shell the number of comp<strong>on</strong>ents is further reduced by two<br />

2×2<br />

because the Dirac equati<strong>on</strong> Γ µ ∂ µ ψ relates half of the comp<strong>on</strong>ents to the other<br />

half (which satisfies the Klein-Gord<strong>on</strong> equati<strong>on</strong>). The 8 remaining comp<strong>on</strong>ents<br />

can be viewed as the comp<strong>on</strong>ents of the Majorana-Weyl spinor of SO(8), the<br />

latter being the little Lorentz group for massless states in d = 10. Indeed, the<br />

spinor of SO(8) should have<br />

(2 8 2 complex comp<strong>on</strong>ents)/(Majorana × Weyl) = 32/4 = 8<br />

20 In general, for the groups SO(p, q) the Majorana and Weyl c<strong>on</strong>diti<strong>on</strong>s can be simultaneously<br />

imposed if and <strong>on</strong>ly if p − q = 0 mod 8. For Minkowski space, p = d − 1, q = −1, this gives<br />

d = 2 + 2n and for Euclidean space, p = d, q = 0, this gives d = 2n.


– 114 –<br />

real comp<strong>on</strong>ents. Thus, the Ram<strong>on</strong>d ground state is massless.<br />

It turns out that the group SO(8) has three inequivalent representati<strong>on</strong>s of<br />

dimensi<strong>on</strong> 8: two of them are spinor representati<strong>on</strong>s and another is the vector<br />

<strong>on</strong>e. Spinor representati<strong>on</strong>s are comm<strong>on</strong>ly denoted by 8 s and 8 c and the<br />

corresp<strong>on</strong>ding representati<strong>on</strong> bases are depicted as<br />

|a〉 and |ȧ〉 .<br />

The vector representati<strong>on</strong> is 8 v and the basis is |i〉.<br />

The first excited level c<strong>on</strong>sist of states α i −1|a〉 and b i −1|a〉 and their chiral partners<br />

with α ′ M 2 = 1. Once again, for d = 10, all the massive light-c<strong>on</strong>e states<br />

can be uniquely assembled into representati<strong>on</strong>s of SO(9), the little Lorentz<br />

group for massive states.<br />

GSO projecti<strong>on</strong><br />

It turns out that fermi<strong>on</strong>ic string with all the states we found in the R and NS<br />

sectors is inc<strong>on</strong>sistent. This can seen, for instance, from the fact that the 1-loop<br />

amplitude is not modular invariant. In order to c<strong>on</strong>struct a c<strong>on</strong>sistent modular<br />

invariant theory <strong>on</strong>e should truncate the string spectrum in a specific way. This<br />

truncati<strong>on</strong> is known as the GSO (Gliozzi-Scherk-Olive) projecti<strong>on</strong>. It restores the<br />

modular invariance, removes from the theory the tachy<strong>on</strong> and, in additi<strong>on</strong>, provides<br />

the space-time supersymmetry of the resulting string spectrum. Below we will use<br />

an inverse argument to motivate the GSO projecti<strong>on</strong> – we will show that it allows<br />

to achive a spectrum which exhibits space-time supersymmetry.<br />

Looking at the massless states in the Ram<strong>on</strong>d sector we see that <strong>on</strong>e has two SO(8)<br />

spinors 8 s and 8 c . On the other hand, the massless states of the NS sector comprise<br />

a vector 8 v . If we project <strong>on</strong>e of the two spinors out then there will be match of NS<br />

bos<strong>on</strong>ic (8) and R fermi<strong>on</strong>ic (also 8) degrees of freedom. These massless vector and<br />

the massless spinor is indeed a c<strong>on</strong>tent of the N = 1 super Yang-Mills theory in ten<br />

dimensi<strong>on</strong>s.<br />

One has also to get rid of tachy<strong>on</strong> which is in the NS sector. This all can be achieved<br />

if <strong>on</strong>e first defines an operator<br />

∞∑<br />

G = (−1) F , F = b i −rb i r − 1<br />

r= 1 2<br />

and then requires that all allowed states should have G = 1:<br />

G|Φ〉 = |Φ〉.


– 115 –<br />

α ′ mass 2 rep of SO(8) little (−1) F (−1) ¯F rep of little<br />

group<br />

group<br />

(NS, NS) − sector<br />

−2 |0〉 L × |0〉 R SO(9) −1 −1 1<br />

0 b i −1/2|0〉 L × b j −1/2<br />

} {{ }<br />

|0〉 R SO(8) +1 +1 1 + 28 + 35 v<br />

} {{ }<br />

8 v 8 v<br />

(R, R) − sector<br />

+1 |a〉<br />

}{{} L × |b〉<br />

}{{} R<br />

+1 1 + 28 + 35 s<br />

8 v 8 v<br />

−1 |ȧ〉<br />

}{{} L × |ḃ〉 R<br />

}{{}<br />

−1 1 + 28 + 35 c<br />

8 c 8 c<br />

0 SO(8)<br />

|ȧ〉<br />

}{{} L<br />

8 c<br />

× |b〉<br />

}{{} R<br />

8 s<br />

−1 +1 8 v + 56 v<br />

|a〉<br />

}{{} L × |ḃ〉 R<br />

}{{}<br />

+1 −1 8 v + 56 v<br />

8 s 8 c<br />

(R, NS) − sector<br />

× +1 +1 8 c + 56 c<br />

× −1 +1 8 s + 56 s<br />

+1<br />

+1<br />

|a〉<br />

}{{} L b i −1/2|0〉<br />

} {{<br />

R<br />

}<br />

8 s 8 v<br />

0 SO(8)<br />

|ȧ〉<br />

}{{} L b i −1/2|0〉<br />

} {{<br />

R<br />

}<br />

8 c 8 v<br />

(NS, R) − sector<br />

¯bi −1/2 |a〉 L<br />

} {{ }<br />

× |a〉<br />

}{{} R<br />

+1 8 c + 56 c<br />

8 v 8 s<br />

0 SO(8)<br />

¯bi −1/2 |a〉 L<br />

} {{ }<br />

8 v<br />

× |ȧ〉<br />

}{{} R<br />

8 c<br />

−1 8 s + 56 s<br />

Tab. 6. The lowest levels of the closed fermi<strong>on</strong>ic string spectrum.


– 116 –<br />

we get<br />

Since a general state in the NS sector has the form<br />

|Φ〉 = α i 1<br />

−n1 · · · α i N<br />

−nN b j 1<br />

−r1 · · · b j M<br />

−rM |0〉<br />

G|Φ〉 = (−1) M−1 |Φ〉.<br />

Thus, all states with M even are projected out, in particular, the tachy<strong>on</strong>. This<br />

removes the tachy<strong>on</strong> and all the states with half-integer α ′ M 2 . Indeed,<br />

α ′ M 2 =<br />

r M<br />

∑<br />

i=1<br />

r i − 1 2 }{{}<br />

a<br />

and for M even the sum of half-integers is always an integer α ′ M 2 is half-integer<br />

number.<br />

In the Ram<strong>on</strong>d sector the operator G is defined as follows<br />

G = (−1) F = b 1 0 · · · b 8 0(−1) P ∞<br />

n=1 bi −n bi n .<br />

The transversal zero modes b i 0 form the Clifford algebra {b i 0, b j 0} = δ ij . Thus, these<br />

operators can be represented by SO(8) γ-matrices Γ i which have size 16 by 16. These<br />

matrices act <strong>on</strong> the 16-dimensi<strong>on</strong>al Majorana (i.e. real) spinor whose comp<strong>on</strong>ents<br />

can be thought to combine two Weyl projecti<strong>on</strong>s, which are precisely |a〉 and |ȧ〉:<br />

( ) |a〉8s<br />

|ψ〉 = .<br />

|ȧ〉 8c<br />

In the Majorana representati<strong>on</strong> these matrices Γ i can be taken in the block-diag<strong>on</strong>al<br />

form as<br />

( ) 0 γ<br />

Γ i i<br />

=<br />

(γ i ) t .<br />

0<br />

The fact that Γ i obey the standard Glifford algebra {Γ i , Γ j } = 2δ ij implies that 8×8<br />

real matrices γ i satisfy the following algebra<br />

γ i (γ j ) t + γ j (γ i ) t = 2δ ij .<br />

If we introduce the standard Pauli matrices<br />

σ 1 =<br />

0 1<br />

1 0<br />

, σ 2 =<br />

<br />

<br />

0 −i<br />

1 0<br />

, σ<br />

i 0<br />

3 =<br />

0 −1<br />

then the matrices γ i can be defined as<br />

γ 1 = −iσ 2 ⊗ σ 2 ⊗ σ 2 , γ 2 = iI ⊗ σ 1 ⊗ σ 2 ,<br />

γ 3 = iI ⊗ σ 3 ⊗ σ 2 , γ 4 = iσ 1 ⊗ σ 2 ⊗ I ,<br />

γ 5 = iσ 3 ⊗ σ 2 ⊗ I , γ 6 = iσ 2 ⊗ I ⊗ σ 1 ,<br />

γ 7 = iσ 2 ⊗ I ⊗ σ 3 , γ 8 = I ⊗ I ⊗ I .


– 117 –<br />

The matrix γ i cab be understood as carrying the matrix indices a and ȧ: γaȧ i . The chiral and anti-chiral representati<strong>on</strong>s of SO(8) are<br />

c<strong>on</strong>structed then with the help of matrices<br />

γ ij<br />

s = 1 <br />

γ i (γ j ) t − γ j (γ i ) t ,<br />

2<br />

γ ij<br />

c = 1 <br />

(γ i ) t γ j − (γ j ) t γ i ,<br />

2<br />

The operator Γ 9 = b 1 0 · · · b0 8 is the chirality operator (the analog of the γ 5 -matrix<br />

in 4dim.) and it projects out <strong>on</strong>e of the two Weyl comp<strong>on</strong>ents of the Ram<strong>on</strong>d<br />

ground state |ψ〉. We see that G anti-commutes with any mode b −n : {G, b i −n} = 0<br />

and, therefore, the eigenvalues of G in the Ram<strong>on</strong>d sector are ±1, depending <strong>on</strong> their<br />

chirality, if we define G|a〉 = |a〉 and G|ȧ〉 = −|ȧ〉. Further, a general state in the R<br />

sector is<br />

|Φ〉 a = α i 1<br />

−n1 · · · α i N<br />

−nN b j 1<br />

−m1 · · · b j M<br />

−mM |a〉<br />

or<br />

We therefore find that<br />

|Φ〉ȧ = α i 1<br />

−n1 · · · α i N<br />

−nN b j 1<br />

−m1 · · · b j M<br />

−mM |ȧ〉<br />

G|Φ〉 a = (−1) M (−1) P i δ m i ,0<br />

|Φ〉 a ,<br />

G|Φ〉ȧ = −(−1) M (−1) P i δ m i ,0<br />

|Φ〉ȧ .<br />

The GSO projecti<strong>on</strong> c<strong>on</strong>sists in leaving the states which have either G = 1 or G = −1.<br />

To c<strong>on</strong>struct the spectrum of the closed superstring we have to tensor left and<br />

right-moving states (such that the level matching c<strong>on</strong>straint ML 2 = M R 2 is satisfied)<br />

and then impose the GSO projecti<strong>on</strong>. Here we have to distinguish four different<br />

sectors<br />

(R, R) , (NS, NS) , (R, NS) , (NS, R)<br />

} {{ } } {{ }<br />

space−time bos<strong>on</strong>s<br />

space−time fermi<strong>on</strong>s<br />

The GSO projecti<strong>on</strong> is imposed separately for the left- and right-moving modes. In<br />

the NS sector <strong>on</strong>e keeps the states with<br />

G = (−1) F = +1 , Ḡ = (−1) ¯F = +1 .<br />

In the Ram<strong>on</strong>d sector there are essentially two possibilities which lead to supersymmetric<br />

and tachy<strong>on</strong>ic-free spectrum. One of them is to take G = Ḡ = 1. The<br />

massless spectrum is<br />

]<br />

]<br />

Bos<strong>on</strong>s :<br />

[(1) + (28) + (35) v +<br />

[(1) + (28) + (35) s<br />

]<br />

]<br />

Fermi<strong>on</strong>s :<br />

[(8) c + (56) c +<br />

[(8) c + (56) c .<br />

In total there are 128 bos<strong>on</strong>ic and 128 fermi<strong>on</strong>ic states. The GSO projecti<strong>on</strong> imposed<br />

in this way defines the so-called Type IIB superstring and its massless spectrum is


– 118 –<br />

that of Type IIB supergravity in ten dimensi<strong>on</strong>s. In particular, 35 v are <strong>on</strong>-shell<br />

degrees of freedom of the gravit<strong>on</strong>, two 28 are two anisymmetric tensor fields and<br />

35 s is a rank four antisymmetric self-dual tensor. In additi<strong>on</strong> <strong>on</strong>e has two real scalars.<br />

The fermi<strong>on</strong>ic degrees of freedom are two spin-3/2 gravitinos, 56 c , and two spin-1/2<br />

fermi<strong>on</strong>s. The presence of two gravitinos indicates that the corresp<strong>on</strong>ding theory<br />

has N = 2 supersymmetry. Since the graviti<strong>on</strong>s and the fermi<strong>on</strong>s are of the same<br />

chirality this theory is chiral.<br />

One can make another choice of the GSO projecti<strong>on</strong> by requiring that G = −Ḡ = 1.<br />

This gives the following massless spectrum<br />

]<br />

]<br />

Bos<strong>on</strong>s :<br />

[(1) + (28) + (35) v +<br />

[(8) v + (56) v<br />

]<br />

]<br />

Fermi<strong>on</strong>s :<br />

[(8) c + (56) c +<br />

[(8) s + (56) s .<br />

There are <strong>on</strong>-shell degrees of freedom of the gravit<strong>on</strong> (35 v ), antisymmetric rank three<br />

tensor (56 v ), an antisymmetric rank two tensor (28), <strong>on</strong>e vector 8 v ) and <strong>on</strong>e real<br />

scalar, which is called dilat<strong>on</strong>. The fermi<strong>on</strong>ic degrees of freedom comprise two spin-<br />

3/2 gravitinos and two spin-1/2 fermi<strong>on</strong>s. Graviti<strong>on</strong>s and fermi<strong>on</strong>s are of opposite<br />

chirality. Thus, this theory has N = 2 supersymmetry also but it is n<strong>on</strong>-chiral. This<br />

string theory is called Type IIA, and the corresp<strong>on</strong>ding supergravity is Type IIA<br />

supergravity.


– 119 –<br />

Appendices<br />

A. Dynamical systems of classical mechanics<br />

To motivate the basic noti<strong>on</strong>s of the theory of Hamilt<strong>on</strong>ian dynamical systems c<strong>on</strong>sider<br />

a simple example.<br />

Let a point particle with mass m move in a potential U(q), where q = (q 1 , . . . q n )<br />

is a vector of n-dimensi<strong>on</strong>al space. The moti<strong>on</strong> of the particle is described by the<br />

Newt<strong>on</strong> equati<strong>on</strong>s<br />

m¨q i = − ∂U<br />

∂q i<br />

Introduce the momentum p = (p 1 , . . . , p n ), where p i = m ˙q i and introduce the energy<br />

which is also know as the Hamilt<strong>on</strong>ian of the system<br />

H = 1<br />

2m p2 + U(q) .<br />

Energy is a c<strong>on</strong>served quantity, i.e. it does not depend <strong>on</strong> time,<br />

dH<br />

dt = 1 m p iṗ i + ˙q i ∂U<br />

∂q i = 1 m m2 ˙q i¨q i + ˙q i ∂U<br />

∂q i = 0<br />

due to the Newt<strong>on</strong> equati<strong>on</strong>s of moti<strong>on</strong>.<br />

Having the Hamilt<strong>on</strong>ian the Newt<strong>on</strong> equati<strong>on</strong>s can be rewritten in the form<br />

˙q j = ∂H<br />

∂p j<br />

,<br />

ṗ j = − ∂H<br />

∂q j .<br />

These are the fundamental Hamilt<strong>on</strong>ian equati<strong>on</strong>s of moti<strong>on</strong>. Their importance lies<br />

in the fact that they are valid for arbitrary dependence of H ≡ H(p, q) <strong>on</strong> the<br />

dynamical variables p and q.<br />

The last two equati<strong>on</strong>s can be rewritten in terms of the single equati<strong>on</strong>. Introduce<br />

two 2n-dimensi<strong>on</strong>al vectors<br />

( )<br />

(<br />

∂H<br />

p<br />

x = , ∇H =<br />

q<br />

)<br />

∂p j<br />

∂H<br />

∂q j<br />

and 2n × 2n matrix J:<br />

J =<br />

( ) 0 −I<br />

I 0<br />

Then the Hamilt<strong>on</strong>ian equati<strong>on</strong>s can be written in the form<br />

ẋ = J · ∇H , or J · ẋ = −∇H .


– 120 –<br />

In this form the Hamilt<strong>on</strong>ian equati<strong>on</strong>s were written for the first time by Lagrange<br />

in 1808.<br />

Vector x = (x 1 , . . . , x 2n ) defines a state of a system in classical mechanics. The<br />

set of all these vectors form a phase space M = {x} of the system which in the present<br />

case is just the 2n-dimensi<strong>on</strong>al Euclidean space with the metric (x, y) = ∑ 2n<br />

i=1 xi y i .<br />

The matrix J serves to define the so-called Poiss<strong>on</strong> brackets <strong>on</strong> the space F(M)<br />

of differentiable functi<strong>on</strong>s <strong>on</strong> M:<br />

{F, G}(x) = (∇F, J∇G) = J ij ∂ i F ∂ j G =<br />

n∑<br />

j=1<br />

( ∂F<br />

∂p j<br />

∂G<br />

∂q j − ∂F<br />

∂q j ∂G<br />

∂p j<br />

)<br />

.<br />

Problem. Check that the Poiss<strong>on</strong> bracket satisfies the following c<strong>on</strong>diti<strong>on</strong>s<br />

{F, G} = −{G, F } ,<br />

{F, {G, H}} + {G, {H, F }} + {H, {F, G}} = 0<br />

for arbitrary functi<strong>on</strong>s F, G, H.<br />

Thus, the Poiss<strong>on</strong> bracket introduces <strong>on</strong> F(M) the structure of an infinitedimensi<strong>on</strong>al<br />

Lie algebra. The bracket also satisfies the Leibnitz rule<br />

{F, GH} = {F, G}H + G{F, H}<br />

and, therefore, it is completely determined by its values <strong>on</strong> the basis elements x i :<br />

which can be written as follows<br />

{x j , x k } = J jk<br />

{q i , q j } = 0 , {p i , p j } = 0 , {p i , q j } = δ i j .<br />

The Hamilt<strong>on</strong>ian equati<strong>on</strong>s can be now rephrased in the form<br />

ẋ j = {H, x j } ⇔ ẋ = {H, x} = X H .<br />

A Hamilt<strong>on</strong>ian system is characterized by a triple (M, {, }, H): a phase space<br />

M, a Poiss<strong>on</strong> structure {, } and by a Hamilt<strong>on</strong>ian functi<strong>on</strong> H. The vector field X H<br />

is called the Hamilt<strong>on</strong>ian vector field corresp<strong>on</strong>ding to the Hamilt<strong>on</strong>ian H. For any<br />

functi<strong>on</strong> F = F (p, q) <strong>on</strong> phase space, the evoluti<strong>on</strong> equati<strong>on</strong>s take the form<br />

dF<br />

dt = {H, F }


– 121 –<br />

Again we c<strong>on</strong>clude from here that the Hamilt<strong>on</strong>ian H is a time-c<strong>on</strong>served quantity<br />

dH<br />

dt<br />

= {H, H} = 0 .<br />

Thus, the moti<strong>on</strong> of the system takes place <strong>on</strong> the subvariety of phase space defined<br />

by H = E c<strong>on</strong>stant.<br />

In the case under c<strong>on</strong>siderati<strong>on</strong> the matrix J is n<strong>on</strong>-degenerate so that there<br />

exist the inverse<br />

J −1 = −J<br />

which defines a skew-symmetric bilinear form ω <strong>on</strong> phase space<br />

ω(x, y) = (x, J −1 y) .<br />

In the coordinates we c<strong>on</strong>sider it can be written in the form<br />

ω = ∑ j<br />

dp j ∧ dq j .<br />

This form is closed, i.e. dω = 0.<br />

A n<strong>on</strong>-degenerate closed two-form is called symplectic and a manifold endowed<br />

with such a form is called a symplectic manifold. Thus, the phase space we c<strong>on</strong>sider<br />

is the symplectic manifold.<br />

Imagine we make a change of variables y j = f j (x k ). Then<br />

ẏ j = ∂yj<br />

}{{} ∂x k<br />

A j k<br />

ẋ k = A j k J km ∇ x mH = A j km ∂yp<br />

kJ ∂x m ∇y pH<br />

or in the matrix form<br />

ẏ = AJA t · ∇ y H .<br />

The new equati<strong>on</strong>s for y are Hamilt<strong>on</strong>ian if and <strong>on</strong>ly if<br />

AJA t = J<br />

and the new Hamilt<strong>on</strong>ian is ˜H(y) = H(x(y)).<br />

Transformati<strong>on</strong> of the phase space which satisfies the c<strong>on</strong>diti<strong>on</strong><br />

AJA t = J<br />

is called can<strong>on</strong>ical. In case A does not depend <strong>on</strong> x the set of all such matrices form<br />

a Lie group known as the real symplectic group Sp(2n, R) . The term “symplectic


– 122 –<br />

group” was introduced by Herman Weyl. The geometry of the phase space which<br />

is invariant under the acti<strong>on</strong> of the symplectic group is called symplectic geometry.<br />

Symplectic (or can<strong>on</strong>ical) transformati<strong>on</strong>s do not change the symplectic form ω:<br />

ω(Ax, Ay) = −(Ax, JAy) = −(x, A t JAy) = −(x, Jy) = ω(x, y) .<br />

In the case we c<strong>on</strong>sidered the phase space was Euclidean: M = R 2n . This is not<br />

always so. The generic situati<strong>on</strong> is that the phase space is a manifold. C<strong>on</strong>siderati<strong>on</strong><br />

of systems with general phase spaces is very important for understanding the<br />

structure of the Hamilt<strong>on</strong>ian dynamics.<br />

Dynamical systems with symmetries<br />

Let g(t) will be a <strong>on</strong>e-parametric group of transformati<strong>on</strong>s of the phase space:<br />

x → g(t)x. This group does not need to coincide with the <strong>on</strong>e generated by the<br />

Hamilt<strong>on</strong>ian H. The acti<strong>on</strong> of this group is called Hamilt<strong>on</strong>ian if there exists a<br />

functi<strong>on</strong> C such that<br />

d<br />

dt g(t)x| t=0 = J · ∇C .<br />

The flow of any functi<strong>on</strong> F under the <strong>on</strong>e-parameter group generated by C is then<br />

δF ≡ d dt F (g(t)x)| t=0 = ∇F · d<br />

dt g(t)x| t=0 = (∇F, J · ∇C) = {F, C}<br />

If we take F = H, we get<br />

δH ≡ {H, C} .<br />

Thus, if Ċ = {H, C} = 0, i.e. if C is an integral of moti<strong>on</strong>, then it generates the<br />

symmetry transformati<strong>on</strong>s which leave the Hamilt<strong>on</strong>ian invariant. Infinitezimally,<br />

the symmetry transformati<strong>on</strong>s are realized as<br />

δF = {F, C} .<br />

There could be several <strong>on</strong>e-parametric groups which are <strong>on</strong>e-parametric subgroups of<br />

a n<strong>on</strong>-abelian Lie G, the latter being the symmetry of the Hamilt<strong>on</strong>ian. Accordingly,<br />

there are the integrals of moti<strong>on</strong> C i , i = 1, . . . , dim G. Since C i are integrals of<br />

moti<strong>on</strong>, from the Jacobi identity<br />

{{C i , C j }, H} + {{H, C i }, C j } + {{C j , H}, C i } = 0<br />

we c<strong>on</strong>clude that {{C i , C j }, H} = 0, i.e. {C i , C j } is an integral of moti<strong>on</strong>. If <strong>on</strong>e can<br />

chose the functi<strong>on</strong>s C i is such a way that they form the Lie algebra of G under the<br />

Poiss<strong>on</strong> bracket:<br />

{C i , C j } = f k ijC k ,<br />

then the corresp<strong>on</strong>ding acti<strong>on</strong> of G <strong>on</strong> the phase space is called Poiss<strong>on</strong>. Here f k ij<br />

are the structure c<strong>on</strong>stants of the Lie algebra of G.


– 123 –<br />

B. OPE and c<strong>on</strong>formal blocks<br />

Every c<strong>on</strong>formal (primary) operator enters in the Operator Product Expansi<strong>on</strong> with<br />

its c<strong>on</strong>formal family (descendants). C<strong>on</strong>tributi<strong>on</strong> of the entire c<strong>on</strong>formal family<br />

associated to a primary operator O into the OPE is called the c<strong>on</strong>formal block.<br />

C<strong>on</strong>formal blocks are completely fixed by c<strong>on</strong>formal symmetry. As an example, let<br />

us show how to find the c<strong>on</strong>formal block associated to the scalar primary operator O<br />

of c<strong>on</strong>formal dimensi<strong>on</strong> ∆ O , which arises in the OPE of two scalar operators A and<br />

B of c<strong>on</strong>formal dimensi<strong>on</strong>s ∆ A and ∆ B , respectively.<br />

Assume the Operator Product Expansi<strong>on</strong><br />

A(x)B(0) =<br />

1<br />

(x 2 ) 1 2 (∆ A+∆ B −∆ O )<br />

∞∑<br />

k=0<br />

1<br />

k! Λk (x, ∂)O(0) + . . . (B.1)<br />

Here dots indicate the c<strong>on</strong>formal families of other primary operators. We assume<br />

that all primary operators are orthog<strong>on</strong>al w.r.t. to the two-point functi<strong>on</strong>s<br />

〈O(0)O(y)〉 =<br />

C O<br />

.<br />

(y 2 ) ∆ O<br />

The three-point functi<strong>on</strong>s are fixed by c<strong>on</strong>formal symmetry. In particular,<br />

〈A(x)B(0)O(y)〉 =<br />

C ABO<br />

(x 2 ) 1 2 (∆ A+∆ B −∆ O ) (y 2 ) 1 2 (∆ B+∆ O −∆ A ) ((x − y) 2 ) 1 2 (∆ A+∆ O −∆ B ) .<br />

Plugging the OPE into the tree-point functi<strong>on</strong> we get<br />

〈A(x)B(0)O(y)〉 =<br />

1<br />

(x 2 ) 1 2 (∆ A+∆ B −∆ O )<br />

∞∑<br />

r=0<br />

1<br />

r! Λr (x, −∂ y )〈O(0)O(y)〉.<br />

Thus, compatibility of the 3-point functi<strong>on</strong> with the OPE results into<br />

C ABO<br />

C 0<br />

1<br />

(y 2 ) 1 2 (∆ B+∆ O −∆ A ) ((y − x) 2 ) 1 2 (∆ A+∆ O −∆ B ) = ∞<br />

∑<br />

k=0<br />

1<br />

1<br />

k! Λk (x, −∂ y ) .<br />

(y 2 ) ∆ O<br />

Taking into account that e −x∂ y<br />

is the shift operator acting as<br />

the last relati<strong>on</strong> may be written as<br />

C ABO<br />

C 0<br />

k=0<br />

e −x∂ y<br />

f(y) = f(y − x) ,<br />

∑ 1 1<br />

k! (y 2 ) (−x · ∂ 1 2 (∆ B+∆ O −∆ A<br />

y) k 1<br />

)<br />

(y 2 ) = ∑ ∞<br />

1<br />

1<br />

1<br />

2 (∆ A+∆ O −∆ B ) k! Λk (x, −∂ y ) .<br />

(y 2 ) ∆ O<br />

It suggests to define<br />

Λ k (x, ∂ y ) = C ABO<br />

C 0<br />

Q a,b<br />

k<br />

(x, ∂ y),<br />

k=0


– 124 –<br />

where the operator Q k (x, ∂ y ) is defined by the relati<strong>on</strong><br />

and a, b are given by<br />

(y 2 ) a (x · ∂ y ) k (y 2 ) b = Q k (x, ∂ y )(y 2 ) a+b (B.2)<br />

a = − 1 2 (∆ B + ∆ O − ∆ A ), b = − 1 2 (∆ A + ∆ O − ∆ B ).<br />

The explicit form of the operator Q k (x, ∂ y ) is found by the Fourier transform. Indeed,<br />

we have<br />

(y 2 ) a = 2 2a+d π d/2 Γ(a + d)<br />

∫<br />

2<br />

1 e −ip·y<br />

(B.3)<br />

Γ(−a) (2π) d (p 2 ) a+d/2<br />

and similar for the others. Substituting the Fourier transform of every functi<strong>on</strong> of<br />

y 2 <strong>on</strong>e gets<br />

1 Γ(a + d)Γ(b + d)<br />

∫<br />

2 2<br />

Γ(−a − b)<br />

π d/2 Γ(a + b + d) Γ(−a)Γ(−b)<br />

2<br />

∫<br />

e −ipy<br />

= dp<br />

(p 2 Qa,b<br />

)<br />

a+b+d/2 k<br />

(x, −ip),<br />

where Q a,b (x, −ip) is defined by<br />

k<br />

Q a,b<br />

k<br />

(x, ∂ y)e ipy = Q a,b<br />

k (x, −ip)eipy ,<br />

and we have used the change of variables p → p − q.<br />

e −ipy<br />

From here <strong>on</strong>e gets that Q a,b (x, −ip) is given by the integral<br />

k<br />

dpdq<br />

(p 2 ) a+d/2 (q 2 ) b+d/2 (−ixq)k (B.4)<br />

Q a,b<br />

1 Γ(a + d<br />

k<br />

(x, −ip) = )Γ(b + d)<br />

∫<br />

2 2<br />

Γ(−a − b)<br />

π d/2 Γ(a + b + d) Γ(−a)Γ(−b) (p2 ) a+b+d/2<br />

2<br />

(−ixq) k<br />

dq<br />

((p − q) 2 ) a+d/2 (q 2 ) . b+d/2<br />

Thus, the problem is reduced to evaluati<strong>on</strong> of the integral<br />

∫<br />

(−ixq) k<br />

I(α 1 , α 2 ) = dq<br />

((p − q) 2 ) α 1 (q2 ) . α 2<br />

(B.5)<br />

One has<br />

I(α 1 , α 2 ) = Γ(α ∫<br />

1 + α 2 ) 1<br />

∫<br />

dtt α1−1 (1 − t) α2−1 (−i) k (xq) k<br />

dq<br />

Γ(α 1 )Γ(α 2 ) 0<br />

[(q − tp) 2 + t(1 − t)p 2 ] α 1+α 2<br />

= Γ(α ∫<br />

1 + α 2 ) 1<br />

∫<br />

dtt α1−1 (1 − t) α2−1 (−i) k (xq + txp) k<br />

dq<br />

Γ(α 1 )Γ(α 2 ) 0<br />

[q 2 + t(1 − t)p 2 ] α 1+α 2<br />

= Γ(α ∫<br />

1 + α 2 ) 1<br />

[k/2]<br />

∑<br />

∫<br />

dtt α1−1 (1 − t) α2−1 (−i) k Ck<br />

2m (txp) k−2m (xq) 2m<br />

dq<br />

Γ(α 1 )Γ(α 2 )<br />

[q 2 + t(1 − t)p 2 ] α 1+α 2<br />

,<br />

0<br />

m=0


– 125 –<br />

where Ck 2m<br />

k!<br />

= . There appear <strong>on</strong>ly even powers 2m since for odd powers<br />

(2m)!(k−2m)!<br />

the internal integral is zero. Thus, we now need to evaluate the integral<br />

∫<br />

L =<br />

dq (xq)2m<br />

[q 2 + h] γ .<br />

Under the integral the symmetric product q i1 ...q i2m<br />

may be substituted for<br />

q i1 ...q i2m =<br />

(q 2 ) m<br />

d(d + 2)...(d + 2(m − 1)) (δ i 1 i 2<br />

...δ i2m−1 i 2m<br />

+ permutati<strong>on</strong>s),<br />

(B.6)<br />

where <strong>on</strong> the r.h.s. all n<strong>on</strong>-trivial permutati<strong>on</strong>s are present. The total number of<br />

this permutati<strong>on</strong>s is (2m)! . Therefore, under the integral <strong>on</strong>e may substitute<br />

2 m m!<br />

(xq) 2m = x i 1<br />

...x i 2m<br />

q i1 ...q i2m =<br />

(x 2 ) m (q 2 ) m (2m)!<br />

d(d + 2)...(d + 2(m − 1)) 2 m m! .<br />

Now we have<br />

∫<br />

dq<br />

(q2 ) m<br />

[q 2 + h] = πd/2<br />

γ Γ(d/2)<br />

∫ ∞<br />

0<br />

dq 2 (q2 ) m+d/2−1<br />

[q 2 + h] γ ,<br />

Performing the change of variables t =<br />

∫<br />

h , we arrive at<br />

q 2 +h<br />

dq<br />

(q2 ) m<br />

∫ 1<br />

[q 2 + h] = πd/2<br />

γ Γ(d/2) hm+d/2−γ dtt γ−m−d/2−1 (1 − t) m+d/2−1<br />

= πd/2<br />

Γ(d/2)<br />

Thus, we evaluate the integral L<br />

∫<br />

L =<br />

dq (xq)2m<br />

[q 2 + h] γ =<br />

0<br />

Γ(γ − m − d/2)Γ(m + d/2)<br />

h m+d/2−γ .<br />

Γ(γ)<br />

(x 2 ) m (2m)! π d/2 Γ(γ − m − d/2)Γ(m + d/2)<br />

=<br />

h m+d/2−γ<br />

d(d + 2)...(d + 2(m − 1)) 2 m m! Γ(d/2)<br />

Γ(γ)<br />

d/2 (2m)! Γ(γ − m − d/2)<br />

= π (x 2 ) m h m+d/2−γ .<br />

4 m m! Γ(γ)<br />

Finally, <strong>on</strong>e gets<br />

I(α 1 , α 2 ) = π d/2 Γ(α 1 + α 2 )<br />

Γ(α 1 )Γ(α 2 )<br />

×<br />

∫ 1<br />

0<br />

[k/2]<br />

∑<br />

m=0<br />

dtt α 1+k−2m−1 (1 − t) α 2−1 (2m)!<br />

4 m m!<br />

Ck<br />

2m (−ixp) k−2m (−1) m<br />

Γ(α 1 + α 2 − m − d/2)<br />

(x 2 ) m (t(1 − t)p 2 ) m+d/2−α 1−α 2<br />

,<br />

Γ(α 1 + a 2 )


– 126 –<br />

where we have substituted γ = α 1 + α 2 and h = t(1 − t)p 2 . This is simplified to<br />

I(α 1 , α 2 ) =<br />

×<br />

π d/2<br />

Γ(α 1 )Γ(α 2 )<br />

∫ 1<br />

0<br />

[k/2]<br />

∑<br />

m=0<br />

Thus, the final formula reads as<br />

I(α 1 , α 2 ) =<br />

Ck<br />

2m (2m)!<br />

4 m m! (−ixp)k−2m (−x 2 ) m (p 2 ) m+d/2−α 1−α 2<br />

dtt k−m+d/2−α 2−1 (1 − t) m+d/2−α 1−1 Γ(α 1 + α 2 − m − d/2).<br />

π d/2 [k/2]<br />

∑<br />

Γ(α 1 )Γ(α 2 )<br />

m=0<br />

C 2m<br />

k<br />

(2m)!<br />

(−ixp)<br />

(− k−2m 1 ) m<br />

m!<br />

4 x2 p 2 (p 2 ) d/2−α 1−α 2<br />

× Γ(k − m + d/2 − α 2)Γ(m + d/2 − α 1 )<br />

Γ(α 1 + α 2 − m − d/2).<br />

Γ(k + d − α 1 − α 2 )<br />

For Q a,b (x, −ip) <strong>on</strong>e therefore finds<br />

k<br />

Q a,b<br />

k (x, −ip) = 1<br />

Γ(a + b + d/2)<br />

×<br />

Γ(−a − b)<br />

Γ(−a)Γ(−b)<br />

[k/2]<br />

∑<br />

m=0<br />

C 2m<br />

k<br />

Γ(k − m − a)Γ(m − b)<br />

Γ(a + b + d/2 − m).<br />

Γ(k − a − b)<br />

By using the relati<strong>on</strong> (a) −m = (−1)m<br />

(1−a) m<br />

Q a,b<br />

k (x, ∂ y) =<br />

[k/2]<br />

1 ∑<br />

(−a − b) k<br />

m=0<br />

(<br />

(2m)!<br />

(−ixp) k−2m − 1 ) m<br />

m!<br />

4 x2 p 2<br />

it can be further simplified to give<br />

(<br />

k!(−a) m (−b) k−m<br />

(x · ∂ y ) k−2m − 1 ) m<br />

m!(k − 2m)!(−d/2 − a − b + 1) m 4 x2 ∆ y .<br />

Further summati<strong>on</strong> gives the c<strong>on</strong>formal block of the scalar field and it is performed<br />

by changing the order of the summati<strong>on</strong> and the shift of the summati<strong>on</strong><br />

variable:<br />

∞∑ 1<br />

k! Qa,b k (x, ∂ y) =<br />

Since<br />

k=0<br />

∞∑<br />

∞∑<br />

m=0 k=2m<br />

=<br />

∞∑<br />

m=0<br />

1<br />

(−a − b) k<br />

we finally get<br />

∞∑ 1<br />

k! Qa,b k (x, ∂ y) =<br />

k=0<br />

=<br />

m=0<br />

(<br />

(−a) m (−b) k−m<br />

(x · ∂ y ) k−2m − 1 ) m<br />

m!(k − 2m)!(−d/2 − a − b + 1) m 4 x2 ∆ y<br />

(<br />

(−a) m<br />

− 1 ) m ∑ ∞<br />

(−b) k+m (x∂ y ) k<br />

m!(−d/2 − a − b + 1) m 4 x2 ∆ y .<br />

(−a − b) k+2m k!<br />

(−b) k+m<br />

(−a − b) k+2m<br />

=<br />

k=0<br />

(−b) m (−b + m) k<br />

(−a − b) 2m (−a − b + 2m) k<br />

∞∑<br />

(<br />

(−a) m (−b) m 1<br />

1F 1 (−b + m; −a − b + 2m; x∂ y ) − 1 ) m<br />

(−µ − a − b + 1) m m! (−a − b) 2m 4 x2 ∆ y ,


– 127 –<br />

where 1 F 1 is a degenerate hypergeometric functi<strong>on</strong>:<br />

C. Useful formulae<br />

1F 1 (α, β; x) =<br />

∞∑<br />

k=0<br />

(α) k<br />

(β) k<br />

x k<br />

k! . (B.7)<br />

In discussi<strong>on</strong> of the central charge of the Virasoro algebra we encounter the sum<br />

S n =<br />

n∑<br />

q 2 .<br />

q=1<br />

Here we present a general method of computing this and similar sums. The method<br />

is based <strong>on</strong> c<strong>on</strong>sidering the generating functi<strong>on</strong><br />

so that<br />

For x < 1 we have<br />

℘(x) =<br />

∞∑<br />

S n x n =<br />

n=1<br />

℘(x) =<br />

S n = 1 n!<br />

∞∑<br />

n=1 q=1<br />

∞∑<br />

S n x n ,<br />

n=1<br />

( ∂ n ℘<br />

∂x n )<br />

| x=0 .<br />

n∑<br />

q 2 x n =<br />

∞∑<br />

q=1<br />

q 2<br />

∞<br />

∑<br />

n=q<br />

x n =<br />

∞∑<br />

q=1<br />

q 2 x q<br />

1 − x .<br />

We further notice that<br />

∂ 2<br />

∂x 2<br />

∞<br />

∑<br />

q=1<br />

x q =<br />

∞∑<br />

q(q − 1)x q−2 = 1 x 2<br />

q=2<br />

∞<br />

∑<br />

q=2<br />

q 2 x q − 1 x 2<br />

∞<br />

∑<br />

q=2<br />

qx q = 1 x 2<br />

∞<br />

∑<br />

q=1<br />

q 2 x q − 1 x 2<br />

∞<br />

∑<br />

q=1<br />

qx q .<br />

Thus,<br />

∞∑ ( ∂<br />

q 2 x q = x 2 2<br />

x<br />

∂x 2 1 − x + 1 ∂<br />

x 2 ∂x<br />

q=1<br />

and, therefore, we obtain the generating functi<strong>on</strong><br />

Finally we compute<br />

1<br />

( ∂ n ℘<br />

)<br />

= 1 n! ∂x n n!<br />

x<br />

)<br />

1 − x<br />

= x2 + x<br />

(1 − x) 3<br />

℘(x) = x2 + x<br />

(1 − x) 4 = 1<br />

(1 − x) 2 − 3<br />

(1 − x) 3 + 2<br />

(1 − x) 4 .<br />

( )<br />

2 · 3 · · · (n + 1) 3 · 4 · · · (n + 2) 4 · 5 · · · (n + 3)<br />

− 3 + 2 .<br />

(1 − x) n+2 (1 − x) n+3 (1 − x) n+4<br />

The last formula results into<br />

S n = (n + 1)<br />

(1 − 3 2 (n + 2) + 1 )<br />

3 (n + 2)(n + 3)<br />

= 1 n(n + 1)(2n + 1) .<br />

6


– 128 –<br />

D. Riemann normal coordinates<br />

C<strong>on</strong>sider a Riemannian manifold M of dimensi<strong>on</strong> n with coordinates x i , 1 = 1, . . . , m.<br />

The geodesic equati<strong>on</strong> is<br />

ẍ i + Γ i jk(x)ẋ j ẋ k = 0 .<br />

Let us c<strong>on</strong>sider two points p and q with coordinates x i and x i + δx i respectively. We<br />

assume that these points are closed so there is a unique geodesic c<strong>on</strong>necting them.<br />

A parameter t <strong>on</strong> the geodesic can be chosen proporti<strong>on</strong>al to the length of the<br />

arc c<strong>on</strong>necting these two points (the natural parameter). The soluti<strong>on</strong> x i (t) can be<br />

chosen so that x i (0) ≡ x i and x i (1) = x i + δx i . The tangent vector to the geodesic<br />

at t = 0 is defined by ξ i = ẋ i (0). Then equati<strong>on</strong> for the geodesic can be solved<br />

perturbatively by assuming the following expansi<strong>on</strong><br />

x i (t) = x i + c i 1t + c i 2t 2 + · · · .<br />

Plugging this into the geodesic equati<strong>on</strong> <strong>on</strong>e finds<br />

where<br />

x i (t) = x i + ξ i t − 1 2 Γi j 1 j 2<br />

(x)ξ j 1<br />

ξ j 2<br />

t 2 − 1 3! Γi j 1 j 2 j 3<br />

(x)ξ j 1<br />

ξ j 2<br />

ξ j 3<br />

t 3 − · · · ,<br />

Γ i j 1 j 2 j 3<br />

= ∂ j1 Γ i j 2 j 3<br />

− Γ l j 1 j 2<br />

Γ i lj 3<br />

− Γ l j 1 j 2<br />

Γ i j 3 l .<br />

Here all the quantities are evaluated at x i . At t = 1 we have x i (1) = x i + δx i so that<br />

x i + δx i = x i + ξ i − 1 2 Γi j 1 j 2<br />

(x)ξ j 1<br />

ξ j 2<br />

− 1 3! Γi j 1 j 2 j 3<br />

(x)ξ j 1<br />

ξ j 2<br />

ξ j 3<br />

− · · ·<br />

Hence, the point x i + δx i is parameterized by the tangent vector ξ i . We can therefore<br />

take ξ i as the new coordinate system <strong>on</strong> our manifold. The coordinates ξ i are<br />

called Riemann normal coordinates. They are used to define a map of an open<br />

neighbourhood U of the zero-vector 0 ∈ T x M to the manifold M:<br />

exp x :<br />

U ∈ T x M → M<br />

which is well-defined diffeomorphism of U <strong>on</strong> its image. Now we are parameterizing<br />

the points of the curved manifold with tangent vectors. The image of the geodesic<br />

through a point x in the Riemann normal coordinates is just a straight line.<br />

How metric will look in the new coordinate system? Up<strong>on</strong> changing the coordinates<br />

form x i to ξ i the metric transforms in the standard way<br />

We have<br />

g ij(ξ) ′ = ∂(xk + δx k ) ∂(x l + δx l )<br />

g<br />

∂ξ i ∂ξ j kl (x + δx)<br />

∂(x k + δx k )<br />

∂ξ i = δ k i − Γ k ijξ j + · · ·


– 129 –<br />

and also<br />

g kl (x + δx) = g kl<br />

(x i + ξ i − 1 )<br />

2 Γi j 1 j 2<br />

ξ j 1<br />

ξ j 2<br />

+ · · · = g kl (x) + ∂ k g ij ξ k + · · ·<br />

Thus, at the linearized level we find that<br />

)<br />

g ij(ξ) ′ =<br />

(δ i k − Γ k inξ<br />

)(δ n j l − Γ l jmξ<br />

)(g m kl (φ) + ∂ p g kl ξ p + · · ·<br />

)<br />

= g ij (x) +<br />

(∂ n g ij − Γ k ing kj − Γ k jng ki ξ n + · · ·<br />

} {{ }<br />

D ng ij<br />

We see that<br />

g ′ ij(ξ) = g ij (x) + D k g ij (x)ξ k + · · ·<br />

(D.1)<br />

It is important to realize that the coordinates ξ k depend <strong>on</strong> x because they define<br />

the tangent vector to the manifold at the point x. Under reparametrizati<strong>on</strong>s of<br />

x → x ′ the object ξ k (x) transforms as a vector! The expansi<strong>on</strong> (D.1) is covariant<br />

under general coordinate transformati<strong>on</strong>s x → x ′ because it includes the tensorial<br />

quantities <strong>on</strong>ly.<br />

In the case of the minimal c<strong>on</strong>necti<strong>on</strong> (c<strong>on</strong>necti<strong>on</strong> compatible with the metric)<br />

we have D k g ij = 0. This, the expansi<strong>on</strong> of the metric start from the quadratic order<br />

in ξ i . Extending this calculati<strong>on</strong> to higher orders in ξ <strong>on</strong>e finds<br />

g ′ ij(ξ) = g ij (x) − 1 3 R ik 1 jk 2<br />

ξ k 1<br />

ξ k 2<br />

− 1 3! D k 1<br />

R ik2 jk 3<br />

ξ k 1<br />

ξ k 2<br />

ξ k 3<br />

+ · · ·<br />

Also we recall the transformati<strong>on</strong> property of the Christoffel c<strong>on</strong>necti<strong>on</strong><br />

(<br />

Γ k′ ∂xk′<br />

p ′ q ′(ξ) =<br />

∂x k<br />

∂x p<br />

Γ k pq<br />

∂x p′<br />

∂x q<br />

∂x ′q′ + ∂2 x k<br />

∂x p′ ∂x q′ )<br />

,<br />

where x i ≡ φ i and x ′i ≡ φ i + π i . Expanding the r.h.s. of the last equati<strong>on</strong> in ξ i it is<br />

easy to see that expansi<strong>on</strong> does not c<strong>on</strong>tain the c<strong>on</strong>stant piece because the c<strong>on</strong>stant<br />

terms in the bracket cancel against each other, in other words, in the Riemann normal<br />

coordinates we have<br />

Γ k′<br />

p ′ q ′(ξ) = O(ξ)<br />

In the Riemann normal coordinate system die to the vanishing of Γ k pq at ξ = 0 the<br />

geodesic equati<strong>on</strong> takes a form of the free moti<strong>on</strong> ¨λ = 0. This is coordinate system<br />

corresp<strong>on</strong>ds to the rest frame of a freely falling observer.


– 130 –<br />

E. Exercises<br />

Exercise 1. Show that the Hessian matrix associated to the Nambu-Goto Lagrangian<br />

has for each σ two zero eigenvalues corresp<strong>on</strong>ding to Ẋµ and X ′µ .<br />

Exercise 2. How reparametrizati<strong>on</strong> invariance can be used to bring the equati<strong>on</strong><br />

⎛<br />

⎞ ⎛<br />

⎞<br />

∂<br />

⎝ (ẊX′ )X ′µ − (X ′ ) 2 Ẋ µ<br />

√<br />

⎠ + ∂ ⎝ (ẊX′)Ẋµ − (Ẋ)2 X ′µ<br />

√<br />

⎠ = 0 .<br />

∂τ<br />

(ẊX′ ) 2 − Ẋ2 X ′2 ∂σ<br />

(ẊX′ ) 2 − Ẋ2 X ′2<br />

to the simplest form?<br />

Exercise 3. The Polyakov string. Prove that equati<strong>on</strong>s of moti<strong>on</strong> for the fields<br />

X µ imply c<strong>on</strong>servati<strong>on</strong> of the two-dimensi<strong>on</strong>al stress-energy tensor<br />

∇ µ T µν = 0<br />

Exercise 4. Show that T αβ = 0 implies that the end points of the open string<br />

move with the speed of light.<br />

Exercise 5. N<strong>on</strong>-relativistic string.<br />

• C<strong>on</strong>sider a string in equilibrium <strong>on</strong> the x-axis between (0, 0) and (L, 0) and<br />

suppose that the infinitesimal parts of the string can move <strong>on</strong>ly in the y-<br />

directi<strong>on</strong>. Derive the Lagrangian with µ the mass density and T the string<br />

tensi<strong>on</strong>.<br />

• Derive the equati<strong>on</strong> of moti<strong>on</strong> from this Lagrangian and keep explicit attenti<strong>on</strong><br />

to the boundary c<strong>on</strong>diti<strong>on</strong>s.<br />

• Analyze the boundary terms. What must you impose in order to have a stati<strong>on</strong>ary<br />

acti<strong>on</strong>?<br />

• C<strong>on</strong>struct the momentum functi<strong>on</strong> P .<br />

• Calculate the time derivative of the momentum (c<strong>on</strong>sider the boundary c<strong>on</strong>diti<strong>on</strong>s).<br />

What do you c<strong>on</strong>clude?<br />

• Fourier transform the x-coordinate and solve the eom. Do this for both boundary<br />

c<strong>on</strong>diti<strong>on</strong>s.


– 131 –<br />

Exercise 6. Show that the Polyakov acti<strong>on</strong> is invariant under reparametrizati<strong>on</strong>s<br />

δX µ = ξ α ∂ α X µ<br />

δh αβ = ∇ α ξ β + ∇ β ξ α<br />

δ( √ h) = ∂ α (ξ α√ h)<br />

Exercise 7. Show that the Weyl invariance implies the tracelessness of the stressenergy<br />

tensor T αβ .<br />

Exercise 8. Show that the Gauss-B<strong>on</strong>net term<br />

χ = 1 ∫<br />

d 2 σ √ hR<br />

4π<br />

is topological, i.e. it vanishes under smooth variati<strong>on</strong>s of the world-sheet metric h αβ .<br />

Take into account that in 2dim the Ricci tensor is proporti<strong>on</strong>al to Ricci scalar and<br />

also<br />

δ( √ (<br />

hR) ∼ R αβ − 1 )<br />

2 h αβR δh αβ .<br />

Exercise 9. Let S(q, t; q 0 , t 0 ) be the acti<strong>on</strong> of the classical path between (q 0 , t 0 )<br />

and (q, t). Show that<br />

∂S<br />

∂q = p(t) ,<br />

where p(t) is the c<strong>on</strong>jugate momentum of q at time t. Show that<br />

∂S<br />

∂t<br />

= −H(q,<br />

∂S<br />

∂q ) ,<br />

where H is the Hamilt<strong>on</strong>ian. Suppose that H(q, p) = p2<br />

+ V (q) and define<br />

2m<br />

ψ(q, t) = e i ~ S(q,t;q 0,t 0 ) .<br />

Show that the schrödinger equati<strong>on</strong> approximately holds for ψ,<br />

i ∂ψ (q,<br />

∂t = H −i ∂ )<br />

ψ + O() .<br />

∂q<br />

This is of course related to Dirac’s idea that the phase of the wave functi<strong>on</strong> is proporti<strong>on</strong>al<br />

to the classical acti<strong>on</strong>.


– 132 –<br />

Exercise 10.<br />

• Show that the c<strong>on</strong>straints<br />

have the following Poiss<strong>on</strong> brackets<br />

C 1 = P µ P µ + T 2 X ′ µX ′µ , C 2 = P µ X ′µ<br />

{C 1 (σ), C 1 (σ ′ )} = 4T 2 ∂ σ C 2 (σ)δ(σ − σ ′ ) + 8T 2 C 2 (σ)∂ σ δ(σ − σ ′ ) ,<br />

{C 1 (σ), C 2 (σ ′ )} = ∂ σ C 1 (σ)δ(σ − σ ′ ) + 2C 1 (σ)∂ σ δ(σ − σ ′ ) ,<br />

{C 2 (σ), C 1 (σ ′ )} = ∂ σ C 1 (σ)δ(σ − σ ′ ) + 2C 1 (σ)∂ σ δ(σ − σ ′ ) ,<br />

{C 2 (σ), C 2 (σ ′ )} = ∂ σ C 2 (σ)δ(σ − σ ′ ) + 2C 2 (σ)∂ σ δ(σ − σ ′ ) .<br />

• Define the linear combinati<strong>on</strong>s<br />

T ++ = 1<br />

8T 2 (C 1 + 2T C 2 ) = 1<br />

8T 2 (P µ + T X ′ µ) 2 ,<br />

T −− = 1<br />

8T 2 (C 1 − 2T C 2 ) = 1<br />

8T 2 (P µ − T X ′ µ) 2 .<br />

and show that their Poiss<strong>on</strong> algebra is<br />

{T ++ (σ), T ++ (σ ′ )} = 1 (<br />

)<br />

∂ σ T ++ (σ)δ(σ − σ ′ ) + 2T ++ (σ)∂ σ δ(σ − σ ′ ) ,<br />

2T<br />

{T −− (σ), T −− (σ ′ )} = − 1 (<br />

)<br />

∂ σ T −− (σ)δ(σ − σ ′ ) + 2T −− (σ)∂ σ δ(σ − σ ′ ) ,<br />

2T<br />

{T ++ (σ), T −− (σ ′ )} = 0 .<br />

Exercise 11. For the closed string case define<br />

L m = 2T<br />

¯L m = 2T<br />

∫ 2π<br />

0<br />

∫ 2π<br />

0<br />

dσ e imσ− T −− (σ, τ)<br />

dσ e imσ+ T ++ (σ, τ) .<br />

Show that for any integer m the generators L m and ¯L m are time-independent.<br />

Exercise 12. Compute the Poiss<strong>on</strong> brackets of the c<strong>on</strong>straints L m , ¯L m . What<br />

kind of c<strong>on</strong>straints they are, i.e. the first or the sec<strong>on</strong>d class?<br />

Exercise 13. It is known in curved space-time that we can transform the metric<br />

locally in the neighborhood of a point x µ = 0 to the following form g µν (x) = η µν −


– 133 –<br />

1<br />

R 3 µσνρx σ x ρ , this are Riemann normal coordinates. So that the deviati<strong>on</strong> of g from<br />

the flat metric η is <strong>on</strong>ly sec<strong>on</strong>d order in x (this is a coordinate system of an observer<br />

in free fall). Suppose we have a coordinate system x with metric expanded around<br />

0,<br />

g µν (˜x) = g µν (0) + ∂ σ g µν (0)˜x σ + 1 2 ∂ σ∂ ρ g µν (0)˜x σ ˜x ρ .<br />

We want to transform this using a coordinate transformati<strong>on</strong> ˜x → x = x(˜x) to<br />

Riemann normal coordinates. We expand the coordinate transformati<strong>on</strong> to third<br />

order (zeroth order is zero)<br />

x µ = ∂xµ<br />

∂˜x ν ˜xν + 1 ∂x µ<br />

2 ∂˜x ν ∂˜x σ ˜xν ˜x σ + 1 ∂x µ<br />

6 ∂˜x ν ∂˜x σ ∂˜x ρ ˜xν ˜x σ ˜x ρ .<br />

We can use the first term to bring g to η. Show that after we have d<strong>on</strong>e this there<br />

are 1 ∂xµ<br />

d(d − 1) remaining degrees of freedom left for , where d is the dimensi<strong>on</strong>.<br />

2 ∂˜x ν<br />

Where do these remaining degrees of freedom corresp<strong>on</strong>d to?<br />

Show that we can bring ∂ σ g µν to zero using the sec<strong>on</strong>d term of the transformati<strong>on</strong>,<br />

by counting degrees of freedom.<br />

For arbitrary dimensi<strong>on</strong> there are not enough degrees of freedom to put ∂ σ ∂ ρ g µν to<br />

zero. Count the number of remaining degrees of freedom and show that it equals<br />

1<br />

12 d2 (d 2 −1). This is precisely the number of independent comp<strong>on</strong>ents of the Riemann<br />

tensor.<br />

Exercise 14. Solve equati<strong>on</strong> of moti<strong>on</strong> □X µ = 0 for the case of open string (take<br />

into account the open string boundary c<strong>on</strong>diti<strong>on</strong>s).<br />

where<br />

Exercise 15. C<strong>on</strong>sider soluti<strong>on</strong> of the closed string equati<strong>on</strong>s of moti<strong>on</strong><br />

X µ (σ, τ) = X µ L (τ + σ) + Xµ R<br />

(τ − σ)<br />

X µ R (τ − σ) = 1 2 xµ + pµ<br />

(τ − σ) +<br />

i<br />

4πT<br />

∑<br />

√<br />

4πT<br />

n≠0<br />

n≠0<br />

1<br />

n αµ ne −in(τ−σ)<br />

X µ L (τ + σ) = 1 2 xµ + pµ<br />

(τ + σ) +<br />

i ∑<br />

√ nᾱµ 1<br />

4πT<br />

ne −in(τ+σ)<br />

4πT<br />

Derive the Poiss<strong>on</strong> algebra of the variables (x µ , p µ , α µ n, ᾱ µ n) by using the fundamental<br />

Poiss<strong>on</strong> brackets of X µ (σ), P µ (σ) variables.


– 134 –<br />

Exercise 16. Prove that that (closed) string has an infinite set of integrals of<br />

moti<strong>on</strong>: for any functi<strong>on</strong> f the quantities<br />

are c<strong>on</strong>served.<br />

L f =<br />

∫ 2π<br />

0<br />

dσf(σ − )T −− , ¯Lf =<br />

∫ 2π<br />

0<br />

dσ f(σ + )T ++ ,<br />

Exercise 17. Obtain an expressi<strong>on</strong> for the Virasoro generators L m and ¯L m in<br />

terms of string oscillators.<br />

Exercise 18. Show that the operators D n = −ie inθ d dθ<br />

obey the Virasoro algebra.<br />

Exercise 19. C<strong>on</strong>sider an open string soluti<strong>on</strong> 0 ≤ σ ≤ π:<br />

X 0 = t = Lτ<br />

X 1 = L cos σ cos τ ,<br />

X 2 = L cos σ sin τ ,<br />

X i = 0, i = 3, . . . , d − 1 .<br />

• Show that this soluti<strong>on</strong> satisfies the Virasoro c<strong>on</strong>straints and open string boundary<br />

c<strong>on</strong>diti<strong>on</strong>s.<br />

• Compute the mass of string.<br />

• Compute the angular momentum J ≡ J 12 of string.<br />

• Show that J = α ′ M 2 , where α ′ = 1<br />

2πT<br />

is called a slope of the Regge trajectory.<br />

Exercise 20.<br />

Poincare group<br />

By using the Poiss<strong>on</strong> brackets between the generators of the<br />

{P µ , P ν } = 0<br />

{P µ , J ρσ } = η µσ P ρ − η µρ P σ<br />

{J µν , J ρσ } = η µρ J νσ + η νσ J µρ − η νρ J µσ − η µσ J νρ<br />

show that for a certain choice of a functi<strong>on</strong> f the following expressi<strong>on</strong><br />

J 2 = 1 ( )<br />

Jαβ J αβ + f(P 2 )P α J αλ P β J βλ , P 2 ≡ P µ P µ .<br />

2


– 135 –<br />

is an invariant of the Poincare group. Find the corresp<strong>on</strong>ding f.<br />

Exercise 21*.<br />

inequality holds<br />

Show that for any open classical string soluti<strong>on</strong> the following<br />

J ≡ √ J 2 ≤ α ′ M 2 ,<br />

(Hint. Make the computati<strong>on</strong> in the static gauge X 0 = t = τ and use the Schwarz<br />

inequality |( ¯f, g)| 2 ≤ ( ¯f, f)(ḡ, g) which holds for any two complex functi<strong>on</strong>s f and<br />

g.)<br />

Exercise 22. Show that the generators of the angular momentum commute with<br />

the Virasoso generators<br />

{L m , J µν } = 0 .<br />

Exercise 23. By using the first-order formalism and imposing the light-c<strong>on</strong>e<br />

gauge for the open string case<br />

• Solve the Virasoro c<strong>on</strong>straints,<br />

• Find the open string light-c<strong>on</strong>e Hamilt<strong>on</strong>ian.<br />

Exercise 24. Show that Virasoro c<strong>on</strong>straints T αβ = 0 which we have found in<br />

the c<strong>on</strong>formal gauge can be directly solved in the light-c<strong>on</strong>e gauge without using the<br />

first-order formalism. Show how the c<strong>on</strong>formal gauge Hamilt<strong>on</strong>ian turns into the<br />

light-c<strong>on</strong>e Hamilt<strong>on</strong>ian up<strong>on</strong> substituti<strong>on</strong> the light-c<strong>on</strong>e gauge c<strong>on</strong>diti<strong>on</strong>s and the<br />

Virasoro c<strong>on</strong>straints.<br />

Exercise 25. Rewrite the Poiss<strong>on</strong> algebra of the Poincaré generators in terms of<br />

light-c<strong>on</strong>e coordinates P ± , P i and J i± , J ij , J +− .<br />

Exercise 26. For the closed string case in the light-c<strong>on</strong>e gauge compute the<br />

Poiss<strong>on</strong> brackets between the zero mode variables p + , p − , p i , x − , x i . The full answer<br />

is given in the lecture notes.<br />

Exercise 27. Proof the fulfilment of the Poiss<strong>on</strong> algebra relati<strong>on</strong>s between the<br />

generators {J i+ , J j+ } and {J i+ , J j− } .<br />

Exercise 28. The Virasoro algebra relati<strong>on</strong> takes the form<br />

[L m , L n ] = (m − n)L m+n + A(m)δ m+n


– 136 –<br />

where A(m) is a functi<strong>on</strong> of m. The aim of this exercise is to find the c<strong>on</strong>straints <strong>on</strong><br />

A(m) that follow from the c<strong>on</strong>diti<strong>on</strong> that the relati<strong>on</strong>s above define the Lie algebra.<br />

• That does the antisymmetry requirement <strong>on</strong> a Lie algebra tells you about<br />

A(m)? What is A(0)?<br />

• C<strong>on</strong>sider the Jacobi identity for the generators L m , L n and L k with m+n+k =<br />

0. Show that<br />

(m − n)A(k) + (n − k)A(m) + (k − m)A(n) = 0.<br />

• Use the last equati<strong>on</strong> to show that A(m) = αm and A(m) = βm 3 , for c<strong>on</strong>stants<br />

α and β, yield c<strong>on</strong>sistent central extensi<strong>on</strong>s.<br />

• C<strong>on</strong>sider the last equati<strong>on</strong> with k = 1. Show that A(1) and A(2) determine all<br />

A(n)<br />

Exercise 29.<br />

• Use the Virasoro algebra to show that if a state is annihilated by L 1 and L 2<br />

then it is annihilated by all L n with n ≥ 1.<br />

• C<strong>on</strong>sider the Virasoro generators L 0 , L 1 and L −1 . Write out the relevant<br />

commutators. Do these operators form a subalgebra of the Virasoro algebra?<br />

Is there a central term here?<br />

Exercise 30. C<strong>on</strong>sider open string. The fundamental commutati<strong>on</strong> relati<strong>on</strong> is<br />

[X µ (σ, τ), P ν (σ ′ , τ)] = iη µν δ(σ − σ ′ ) , σ ∈ [0, π] .<br />

• Show that c<strong>on</strong>sistency with the oscillator expansi<strong>on</strong> implies that<br />

δ(σ − σ ′ ) = 1 ∞∑<br />

cos nσ cos nσ ′<br />

π<br />

n=−∞<br />

• Why the fundamental commutati<strong>on</strong> relati<strong>on</strong> compatible with open string boundary<br />

c<strong>on</strong>diti<strong>on</strong>s?<br />

• Prove this representati<strong>on</strong> for the δ-functi<strong>on</strong> by using the fact that any functi<strong>on</strong><br />

f(σ) with σ ∈ [0, π] and vanishing derivative at σ = 0, π can be expanded as<br />

f(σ) =<br />

∞∑<br />

A n cos nσ .<br />

n=0


– 137 –<br />

Exercise 31. Compute in the light-c<strong>on</strong>e gauge the commutator of the orbital<br />

momenta<br />

[l i− , l j− ] =?<br />

Exercise 32. Show that in the quantum theory the eigenvalues of the covariant<br />

number operator<br />

∞∑<br />

N =<br />

are always n<strong>on</strong>negative.<br />

n=1<br />

α µ −nα n,µ<br />

Exercise 33. Using the previous exercise show that for any fixed state all but<br />

a finite number of positively moded Virasoro operators automatically annihilate the<br />

state without imposing any c<strong>on</strong>diti<strong>on</strong>s. More precisely, show that any state |Φ〉 with<br />

the number eigenvalue N ≥ 0 automatically satisfies<br />

L n |Φ〉 = 0 for n > N.<br />

Exercise 34. Compute the open string propagator<br />

〈X(τ, σ)X(τ ′ , σ ′ )〉 = T ( X(τ, σ)X(τ ′ , σ ′ ) ) − : X(τ, σ)X(τ ′ , σ ′ ) : .<br />

Exercise 35. Show that the vertex operator of the open string<br />

V (k, τ) = e 1 √<br />

πT<br />

P ∞<br />

n=1<br />

kµα µ −n<br />

e n<br />

inτ<br />

e ikµ(xµ + pµ<br />

πT<br />

} {{ τ)<br />

} e − 1 P<br />

√ ∞<br />

πT n=1<br />

V − V 0 V +<br />

} {{ }<br />

kµα µ n<br />

n<br />

e−inτ<br />

} {{ }<br />

is the c<strong>on</strong>formal operator with the c<strong>on</strong>formal dimensi<strong>on</strong> ∆ = α ′ k 2 .<br />

Exercise 36. Compute the two-point correlati<strong>on</strong> functi<strong>on</strong> of the tachy<strong>on</strong> vertex<br />

operators<br />

〈0|V (k 2 , τ 2 )V (k 1 , τ 1 )|0〉<br />

Exercise 37. Compute the three-point correlati<strong>on</strong> functi<strong>on</strong> of the tachy<strong>on</strong> vertex<br />

operators<br />

〈0|V (k 3 , τ 3 )V (k 2 , τ 2 )V (k 1 , τ 1 )|0〉


– 138 –<br />

Exercise 38. Compute the three-point correlati<strong>on</strong> functi<strong>on</strong> of the tachy<strong>on</strong> vertex<br />

operators<br />

〈0|V (k 4 , τ 4 )V (k 3 , τ 3 )V (k 2 , τ 2 )V (k 1 , τ 1 )|0〉<br />

Exercise 40. Verify that ∂ n −X µ is a c<strong>on</strong>formal operator. Find the corresp<strong>on</strong>ding<br />

c<strong>on</strong>formal dimensi<strong>on</strong>. Find the singular terms of the OPE<br />

T −− (τ, σ) ∂ n −X µ (τ ′ , σ ′ ) .<br />

Exercise 41. Find the singular terms of the OPE<br />

T −− (τ, σ) V (k, τ ′ , σ ′ ) ,<br />

where V (k, τ ′ , σ ′ ) is the vertex operator of tachy<strong>on</strong>.<br />

Exercise 42. Find the singular terms of the OPE<br />

T −− (τ, σ) T −− (τ ′ , σ ′ ) .<br />

Exercise 43. Suppose that there are i = 1, . . . , n grassman (anticommuting)<br />

variables η i and ¯η i . Let us define the integrati<strong>on</strong> rules as<br />

∫<br />

∫<br />

dη = 0 , dηη = 1<br />

for any η i and ¯η i . Show that for any n × n matrix M the following formula is valid<br />

∫<br />

detM = dηd¯η e¯ηMη .<br />

Here ¯ηMη ≡ ¯η i M ij η j .<br />

Exercise 44. Show that the stress-tensor for the ghost fields implies the following<br />

expressi<strong>on</strong> for the ghost Virasoro generators of the closed string<br />

L gh<br />

m =<br />

¯L gh<br />

m =<br />

∞∑<br />

n=−∞<br />

∞∑<br />

n=−∞<br />

(m − n) : b m+n c −n :<br />

(m − n) : ¯b m+n¯c −n :


– 139 –<br />

Exercise 45. C<strong>on</strong>sider c<strong>on</strong>formal transformati<strong>on</strong>s z → z ′ = f(z). By definiti<strong>on</strong>,<br />

primary fields are the fields which transform as tensors under c<strong>on</strong>formal transformati<strong>on</strong>s<br />

( ∂z<br />

φ(z, ¯z) → φ ′ ′ ) h ( ∂¯z<br />

′ )¯hφ(z<br />

(z, ¯z) =<br />

′ (z), ¯z ′ (¯z))<br />

∂z ∂¯z<br />

Find how a primary field transforms under infinitezimal c<strong>on</strong>formal transformati<strong>on</strong>s<br />

φ → φ + δ ξ,¯ξφ, where z ′ = z + ξ(z).<br />

Exercise 46. Show that c<strong>on</strong>formal fields of weight h have the following mode<br />

expansi<strong>on</strong><br />

φ(z) =<br />

∑<br />

z −n−h φ n .<br />

n∈integer<br />

Exercise 47. In the radial quantizati<strong>on</strong> products of fields is defined by putting<br />

them in the radial order. Using the radial order prescripti<strong>on</strong> show that the c<strong>on</strong>formal<br />

transformati<strong>on</strong><br />

δ ξ φ(w) = [T ξ , φ(w)]<br />

can be written in the form<br />

∮<br />

δ ξ φ(w) =<br />

C w<br />

dz<br />

ξ(z)T (z)φ(w)<br />

2πi<br />

Here C w is a small c<strong>on</strong>tour in a complex plane around point z.<br />

Exercise 48.<br />

formula<br />

Using the previous exercise together with the Cauchy-Riemann<br />

∮<br />

C w<br />

dz f(z)<br />

2πi (z − w) = f (n−1) (w)<br />

n (n − 1)!<br />

show that any c<strong>on</strong>formal field must have the following R-ordered operator product<br />

with T (z):<br />

T (z)φ(w) =<br />

hφ(w)<br />

(z − w) + ∂φ(w) + regular terms<br />

2 (z − w)<br />

Exercise 49. Show that the following operator product of the stress tensor<br />

T (z)T (w) = c/2 2T (w) ∂T (w)<br />

+ +<br />

(z − w)<br />

4<br />

(z − w)<br />

2<br />

(z − w)<br />

+ regular terms<br />

corresp<strong>on</strong>ds to the following transformati<strong>on</strong> law<br />

δ ξ T (z) = c<br />

12 ∂3 ξ(z) + 2∂ξ(z)T (z) + ξ(z)∂T (z)


– 140 –<br />

as<br />

Here<br />

Exercise 50. Under finite transformati<strong>on</strong>s z → f(z) the stress tensor transforms<br />

T (z) → T ′ (z) = (∂f) 2 T (f(z)) + c<br />

12 D(f) z<br />

D(f) z = ∂f(z)∂3 f(z) − 3 2 (∂2 f(z)) 2<br />

(∂f) 2<br />

is the Schwarzian derivative. Show that if f(z) = az+b<br />

cz+d then D(f) z = 0.<br />

Exercise 51. C<strong>on</strong>sider a parallelogram <strong>on</strong> the complex plane determined by the<br />

following identificati<strong>on</strong><br />

z ≡ z + nλ 1 + mλ 2 ,<br />

n, m ∈ Z<br />

Show that a general transformati<strong>on</strong><br />

λ 1 → dλ 1 + cλ 2 , λ 2 → bλ 1 + aλ 2<br />

with the c<strong>on</strong>diti<strong>on</strong> ad − bc = 1 preserves the area of the parallelogram.<br />

Exercise 52. Show that the modular transformati<strong>on</strong>s<br />

T : τ → τ + 1 , S : τ → − 1 τ<br />

applied to the fundamental regi<strong>on</strong> of the modular group<br />

{<br />

M g=1 = − 1 2 ≤ Reτ ≤ 0, |τ|2 ≥ 1 ∪ 0 < Reτ < 1 }<br />

2 , |τ|2 > 1<br />

generate the whole upper-half plane.<br />

Exercise 53. Show that τ = i and τ = e 2πi<br />

3 are the fixed points of S and ST<br />

transformati<strong>on</strong>s respectively.<br />

Exercise 54. Verify that the acti<strong>on</strong><br />

S = − 1 ∫<br />

d 2 σ<br />

(∂ α X µ ∂ α X µ + 2i<br />

8π<br />

¯ψ<br />

)<br />

µ ρ α ∂ α ψ µ .<br />

is invariant under supersymmetry transformati<strong>on</strong>s w<br />

δ ɛ X µ = i¯ɛψ µ ,<br />

δ ɛ ψ µ = 1 2 ρα ∂ α X µ ɛ<br />

δ ɛ ¯ψµ = − 1 2¯ɛρα ∂ α X µ


– 141 –<br />

provided the parameter ɛ satisfies the following equati<strong>on</strong><br />

ρ β ρ α ∂ β ɛ = 0 .<br />

Exercise 55. If in the previous exercise the parameter ɛ is c<strong>on</strong>stant then we<br />

deal with global supersymmetry transformati<strong>on</strong>s. Derive the Noether current which<br />

corresp<strong>on</strong>ds to this global symmetry of the acti<strong>on</strong>.<br />

Exercise 56*. Express the spin c<strong>on</strong>necti<strong>on</strong> ω via vielbein e by using the c<strong>on</strong>diti<strong>on</strong><br />

of vanishing of the torsi<strong>on</strong><br />

T a αβ = D α e a β − D β e a α = 0 .<br />

Exercise 57. Derive the <strong>on</strong>-shell algebra of supersymmetry transformati<strong>on</strong>s<br />

[δ 1 , δ 2 ] =? .<br />

Exercise 58. Using the Noether method derive the currents corresp<strong>on</strong>ding to<br />

the Poincaré symmetry for the fermi<strong>on</strong>ic string. Express the corresp<strong>on</strong>ding Noether<br />

charges via oscillators for both Neveuw-Schwarz and Ram<strong>on</strong>d sectors.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!