Sie sind auf Seite 1von 38

Notes on Undergraduate String Theory

Moataz H. Emam

Department of Physics
SUNY College at Cortland
Cortland, New York 13045, USA

Abstract

These notes are intended to accompany my string theory for undergraduates course, first
taught in the spring semester of 2017. The material herein are based on Barton Zwiebachs
excellent textbook , in the hope that they would serve my students as easy reference as well
as put in perspective the big picture. The notes are perpetually a work in progress and I
will continuously update them. The reading assignments I gave to the class are listed in the
appendix. If you are teaching yourself string theory using Zwiebachs book, you might want
to follow these readings and use the notes as a supplement. The order of the material as well
as the content was chosen for students who come in not necessarily knowing special relativity,
advanced electrodynamics, or quantum mechanics. The only prerequisites to this class were
introductory physics and multi-variable calculus.

moataz.emam@cortland.edu. This document was last updated: May 9, 2017

A First Course in String Theory, 2nd Edition, 2009, Cambridge University Press.
1 Hamiltons principle of least action

For every physical process, there exist a function known as the Lagrangian L (t) such that the
so-called Action is defined by
Ztf
S= dtL (t), (1)
ti

where ti and tf are the starting (initial) time and ending (final) time of the process1 . The physical
process in question will follow the path defined by a stationary (also known as least or extremum)
action such that
S = 0 (2)

where is an operator, known as the variation, that follows the rules of derivatives; in other words
it satisfies the product rule: (f g) = f g + gf , the chain rule: [f (x)] = f 0 x and so on.
In the classical mechanics of point particles, the Lagrangian is defined by the difference between
the kinetic energies and potential energies of the process

L = K (qi ) U (qi ) (3)

making it dependent on the generalized coordinates qi (t) and generalized velocities qi (t).
These last are called generalized because they can be distances and linear velocities, or angles and
angular velocities respectively2 . For any given problem, there can be any number of them, hence
the index i = 1, , N which counts over the number of generalized coordinates N . In this case
applying the condition (2) leads to the Euler-lagrange equations for each qi :
 
L d L
=0 (4)
qi dt qi

The resulting equations after implementing (4) are second order differential equations in time
and are generally known as the equations of motion (or simply e.o.m) of the system under study.
They are identical to what one would have gotten if we just applied Newtons laws in the usual
way. It is important to note that we can define the so-called conjugate momentum pi as the
momentum associated with each qi , so for linear coordinates pi is the ordinary linear momentum
along the qth direction, and for angular coordinates pi becomes the angular momentum in the
1
In advanced physics it is sometimes traditional, as well as useful, to place the differential element before the
integrand rather than after.
2 dq d2 q
It is standard to denote time derivatives by an over-dot, i.e. q = dt
, and q = dt2
and so on.

2
plane of rotation defined by qi . The conjugate momentum for each generalized coordinate is easily
found from the Lagrangian by
L
pi = (5)
q
Finally, the total energy E of the system in question can be found by applying the so-called
Legendre transformation:
N
X
E= pi qi L (6)
i=1

In general, Hamiltons least action principle may be applied to a variety of situations. In addition
to particles, it is also valid for fields, so for example there is an action S for electromagnetic fields
where (2) leads directly to the Maxwell equations. It is also valid for relativistic phenomena as we
will see. It is a very powerful approach used in almost all branches of fundamental physics.

2 The classical string

An oscillating string is a continuous one-dimensional object, hence it is parameterized by two


parameters, the time t and the distance on the string itself x, whereas a point particle, being zero
dimensional, is only parameterized by time. One can then modify Hamiltons principle to apply to
Lagrangian functions that are dependent on two types of derivatives, one in time (symbolized by
a dot as before) and the other in space (derivatives with respect to x, designated by a prime 0 ). In
this case, one defines the so-called Lagrangian density L as follows

Zxf
y0

L (t) = dxL y, (7)
xi

where the multi-variable function y (t, x) replaces the generalized coordinates qi (t) of the point
particle, and xi and xf are the initial and final starting points of the parameter x; in other words
the endpoints of the string. The action is then

Ztf Zxf
y0 ,

S= dt dxL y, (8)
ti xi

In the case we are interested in, y is the transverse displacement of the string from its equilibrium
position. Assuming the string has a mass density 0 in units of mass per length, and a uniform

3
tension T0 , we find that the Lagrangian density, Lagrangian, and Action of an oscillating string are

1
0 y 2 T0 y 02

L =
2
Za
1
dx 0 y 2 T0 y 02

L =
2
0
Ztf Za
1
0 y 2 T0 y 02 ,

S = dtdx (9)
2
ti 0

where a is the length of the string under study. The variation of the action in (9) leads, perhaps
not too surprisingly, to the wave equation:

0
y 00 y = 0 (10)
T0

which is the equation of motion of the string. It can also be found by applying Newtons laws to a
differential element of length on the string. Following the standard theory of waves, it can be shown
p
that the velocity of waves on this string can be directly read from (10) and is just v0 = T0 /0 .
As is standard in mechanics, (10) can be solved for the function y (t, x) using appropriate boundary
conditions. There are two types of boundary conditions that are pertinent to our interests. These
are summarized next.

2.1 The Dirichlet String

This is what one would get if we constrain the end points of the string to specific static points in
space, such as tying the string between two nails. The so-called Dirichlet boundary conditions
that we get from this are simply
y (t, 0) = y (t, a) = 0, (11)

or equivalently
y
= 0. (12)
t x=0,a
Applying the Dirichlet boundary conditions to the wave equation, we find the solution that
represents standing waves on a string with fixed end-points; the so-called Dirichlet string solution:
 nx   
X nvt
y (t, x) = An sin sin . (13)
a a
n=1

4
2.2 The Neumann String

The second set of boundary conditions are those where the end points are allowed to move freely
in one (or more) directions that are transverse to the strings original flat shape. For example, a
string whose end-points are constrained to move vertically on a smooth vertical pole. In this case
we require that the string remains orthogonal to the vertical direction, or alternatively that the
tangent to the string at the end-points be orthogonal to the pole, ergo

y
=0 (14)
x x=0,a

This set of conditions leads the Neumann string solution:


 nx   
X nvt
y (t, x) = An cos sin . (15)
a a
n=1

2.3 For future reference

In this section we set the stage for future stuff. We begin by asking what is the momentum of
an oscillating string? This is easily shown to be
Za
p= dx0 y.
(16)
0

dp
It turns out that the momentum is only conserved (i.e. dt = 0) for Neumann strings. These
are allowed to oscillate as well as move through space carrying conserved momentum as one would
hope and expect. The Dirichlet string, on the other hand, being attached to something, does not
conserve momentum. In other words, one can say that momentum flows off of the Dirichlet string
at the end points into whatever is holding them in place. Classically, that would just be a nail
attached to the wall for example. In the case of the fundamental Dirichlet string of string theory,
the string is attached to what is known as D-branes, where the D stands for Dirichlet, as we will
eventually see.
Finally, just as we defined the Lagrangian density for the string, we also define a momentum
density. Recall that for a point particle the conjugate momentum is defined as the derivative of
the Lagrangian with respect to q (equation 5). In the string case, the conjugate momentum density
is defined as the derivative of the Lagrangian density with respect to y as follows

L
Pt = , (17)
y

5
where it is obvious that the total momentum (16) would just be
Za
p= dxP t . (18)
0

If the quantity P t is the momentum density conjugate to t, we also can, in the interests of a
symmetry between time and space, define a momentum density conjugate to x as follows

L
Px = . (19)
y 0

For the case of the classical string with mass density and tension, these are just

P t = 0 y

P x = T0 y 0 . (20)

As noted, this language assigns a symmetry between the time t and the distance x. They are
both being treated on equal footing and have similar constructions, which will be very useful later.
Based on this, the strings equation of motion (10) simply becomes

P t P x
+ =0 (21)
t x

and the boundary conditions become

P t end points = 0

Dirichlet

Neumann P x |end points = 0 (22)

This concludes our discussion of the classical, non-relativistic, non-quantum string.

3 The mathematics of spacetime

In 1905, Albert Einstein presented the postulates of the special theory relativity as follows

1. The postulate of relativity: The laws of nature have the same form in all inertial (non-
accelerating) reference frames.

2. The postulate of the constancy of the speed of light: The speed of light c in vacuum
is constant as measured by all inertial (non-accelerating) observers.

6
The first postulate was already well known to physicists since the time of Galileo; it is the
second one that was new. It implies that no matter how fast an inertial (non-accelerating) observer
is traveling, he or she will always observe the speed of light to be exactly c (in vacuo). This
certainly violates our classical experience and predicts that, at least at high velocities, new physical
phenomena will arise. Particularly, the way we measure distances and times with respect to various
observers will have to be redefined in order to preserve both postulates at once. For example, an
observer O measuring the displacement dx or the time dt of a specific physical event will necessarily
disagree with a different observer O0 , who will observe different values for the displacement dx0 and
the time dt0 , so dx 6= dx0 and dt 6= dt0 . This is true if one observer happens to be moving with a
constant speed v with respect to the other. Hence we hear about all kinds of interesting effects,
such as the contraction of meter sticks or the dilation of time measurements as seen by different
observers.
This sounds like a mess, however it was noted early on that although observers will disagree on
their dxs, dys, dzs, and dts, there is one quantity, derived from both postulates, that all observers
will agree on, this is
ds2 = c2 dt2 + dx2 + dy 2 + dz 2 . (23)

We state that all observers, no matter how fast they are moving with respect to each other
(as long as their speeds are constant, i.e. non-accelerating), will agree on the value of ds2 , usually
referred to (in modern language) as the Minkowski spacetime metric. Note that, the numerical
values of ds2 are particularly related to how fast the signal under study is crossing the distance
dx2 + dy 2 + dz 2 . There are three possibilities:

ds2 = 0, Null signal, moving at the speed of light.

ds2 < 0, Timelike signal, moving slower than light.

ds2 > 0, Spacelike signal, moving faster than light. (24)

The mathematician Hermann Minkowski was the one who made an observation, years after
Einstein published special relativity, that has effectively overturned our understanding of how we
view the space and time will live in. Minkowski noted that (23) looks like the Pythagorean theorem
in three dimensions
dl2 = dx2 + dy 2 + dz 2 (25)

with an extra term c2 dt2 . He proposed that (23) may be viewed as a new Pythagorean-like

7
theorem not in three spatial dimensions, but rather in four spacetime dimensions. This is the
origin of the saying that time is the fourth dimension. Equation (23) certainly treats time,
corrected by a factor of c, very similarly to the way the spatial pieces are treated. A major
difference between (23) and (25) is the minus sign. This must also be accounted for as we will
see below. To grasp what Minkowski did next, let us first go back to ordinary three-dimensional
vector spaces, and define, as is typical in intro textbooks, the position vector of a particle by the
Cartesian components
r = (x, y, z) (26)

The displacement vector is then


dr = (dx, dy, dz) (27)

as usual. Note that the scalar product of the displacement with itself leads to exactly the Pythagorean
(25):
dr dr = dx2 + dy 2 + dz 2 . (28)

An approach such as this would not work in four dimensional spacetime because of the minus
sign in (23). We need to define the spacetime dot product in such a way that it would lead to (23);
minus sign and all. One straightforward way to do this is by defining a complex version of t, known
as imaginary time as follows:
w = ict (29)

such that a spacetime position and displacement vectors would be3

r4 = (w, x, y, z)

dr4 = (dw, dx, dy, dz) (30)

and the dot product of the spacetime displacement with itself becomes

dr4 dr4 = dw2 + dx2 + dy 2 + dz 2

= c2 dt2 + dx2 + dy 2 + dz 2 , (31)

as needed. Although this would work, and in fact many early relativity books were written using
this language, a more modern language is much more frequently used. Instead of defining imaginary
3
In texts dealing with special relativity in four spacetime dimensions, vectors are called 4-vectors. We resist this
language and call them spacetime vectors instead. This is because we are generally interested in an arbitrary number
of spacetime dimensions.

8
time, we define the concept of the so-called dual vector, in addition to the usual vectors. So for
example, we define the spacetime position vector r4 and its dual r4 by

r4 = (ct, x, y, z)

r4 = (ct, x, y, z) , (32)

such that

dr4 = (cdt, dx, dy, dz)

dr4 = (cdt, dx, dy, dz) . (33)

Furthermore, we require that the dot product is always done between the vector and its dual,
so
dr4 dr4 = c2 dt2 + dx2 + dy 2 + dz 2 . (34)

Since both approaches lead to exactly the same thing (23) then they are both physically equiv-
alent. We will be using the modern vector/dual-vector approach.

4 The index notation and summation convention

We take all of the above further by defining an entirely new language for spacetime vectors. The
main advantage of this language is that it allows us, in the simplest possible way, to increase the
number of dimensions from 4 to as many as needed. This language is particularly economical in
that it greatly reduces the amount of writing one needs to express mathematical processes. It
is done by renaming the components of vectors; giving them numbers rather than names. For
example, the components of the position spacetime vector are renamed in terms of upper indices
(superscripts) as follows:

ct x0

x x1

y x2

z x3 (35)

9
whereas the components of the dual position vector are written as lower indices (subscripts):

ct x0

x x1

y x2

z x3 (36)

Clearly the components of the vector and its dual differ only by the minus sign in the zeroth
component; otherwise they are equal4 . The displacement vector and its dual are similarly redefined.
Using naming convention, the dot product of the displacement spacetime vector with its dual can
now be written as a summation over indices as follows
3
X
dr4 dr4 = dx dx
=0

= dx dx0 + dx1 dx1 + dx2 dx2 + dx3 dx3


0

= c2 dt2 + dx2 + dy 2 + dz 2 = ds2 (37)

Traditionally, Greek indices such as , , and so on, are used whenever time is included,
so they would run 0, 1, 2, d, while the Latin alphabet is used for space components only, so
i, j, k, a, b, c = 1, 2, d. Also traditionally, the letter d is reserved for the number of spatial
dimensions, as in d = 3, while the upper case D is the number of spacetime dimensions, so D = d+1.
Another simplifying convention is that one can drop the summation symbol in (37) entirely, so

ds2 = dx dx (38)

If there is no summation symbol how does one know that a summation exists? The rule is: if in
a given term you see two (exactly two, no less, no more) identical indices, one upper and one lower,
then a summation over that index exists. This is called the Einstein summation convention
and is heavily used in relativistic physics. Summations in this language are also sometimes referred
to as contractions, so one can say that (38) represents contracting dx with itself. Although we
have so far concentrated our attention on the position and displacement spacetime vectors only,
any spacetime vector and its dual can be defined by

V = V 0, V 1, V 2, V 3


V = (V0 , V1 , V2 , V3 ) (39)
4
This is only true in Cartesian coordinates.

10
where in Minkowskian (i.e. Cartesian with time) coordinates, they are related by

V0 = V 0 , V1 = V 1 , V2 = V 2 , V3 = V 3 (40)

such that the dot product between two vectors V and U goes like

V4 U4 = V U = V 0 U0 + V 1 U1 + V 2 U2 + V 3 U3

= V 0 U 0 + V 1 U 1 + V 2 U 2 + V 3 U 3 (41)

with the appropriate minus sign automatically included.

5 The Minkowski metric tensor and other issues

The dot product between two vectors (41) may be rewritten in terms of a double summation over
a two index quantity as follows
V U = V U (42)

where the components of this are defined by

00 = 1

11 = 22 = 33 = +1

01 = 21 = 23 = 23 = = 0, (43)

which can be appropriately generalized to D = d + 1 spacetime dimensions. Now, the expression


V V in (42) is clearly a double sum over and ; both of which run over 0, 1, 2, 3. Expanding
and using the definitions (43) we get
3 X
X 3
V U = V U
=0 =0

= 00 V 0 U 0 + 01 V 0 U 1 + 02 V 0 U 2 + + 11 V 1 U 1 +

= V 0 U 0 + V 1 U 1 + V 2 U 2 + V 3 U 3 (44)

identically with (41). The collective are components of what is known as the Minkowski
metric tensor. They are the components of the tensor in the same sense as V being components
of a vector. The Minkowski metric tensor has an important role to play in the mathematics of special
relativity. Aside from being the basis of the dot product as mentioned, it can be used to raise

11
or lower indices, which simply means that it provides a way of relating upper and lower index
vectors. For example it can be easily verified that

V = V (45)

while if we define the so-called inverse Minkowski metric as follows

00 = 1

11 = 22 = 33 = +1

01 = 21 = 23 = 23 = = 0, (46)

then we can also write


V = V (47)

One can easily show that


= (48)

where is the Kronecker delta defined as follows

00 = 11 = 22 = 33 = 1

10 = 23 = ( 6= ) = 0 (49)

The Kronecker delta is a commonly used quantity in physics and can simply be thought of as
the identity tensor, since if one contracts it with any vector or tensor quantity it just changes
the names of the indices but still returns exactly the same thing, for example

V = V

V = V

= (50)

and so on. The fact that contracting with gives as shown in (48) is what earned its
title as an inverse metric tensor, since an inverse is normally defined as that which when multiplied
with the original gives the identity, as in 1/x is the inverse of x and so on.

6 Lorentz invariance

If Einsteins first postulate is satisfied simultaneously with the second then there should exist a way
to calculate how different observers will view the various physical quantities in their own reference

12
frames. This is done via the Lorentz transformation tensor . There are many versions of
this . As an example consider an observer O0 traveling with a velocity v along the x direction
0
of another observer O. The moving observer O0 will view a vector quantity as V , while (the
stationary observer) O will see V . The two vectors are related to each other via the contraction

V0 = 0 V , (51)

where the non-vanishing components of 0 are given by

0 0
00 = 11 =
0 0
01 = 10 =
0 0
22 = 33 = 1, (52)

~ and is known as the Lorentz


where is the magnitude of the so-called rapidity vector ,
factor:

v
~ =
c
1
= p , where 2 = ~ ~ (53)
1 2
The Lorentz factor is an important quantity in physics; its subsequent appearance in many
relativistic equations is what guarantees that nothing can reach or exceed the speed of light since
p
this is the threshold at which the denominator 1 2 vanishes then becomes imaginary. It is
useful to keep the domains of and in mind as one inspects any relativistic equations containing
them; thus

= 01

= 1 . (54)

Now, going back to our observers, the stationary observer O will see

0
V = V 0 , (55)

with the inverse transformation tensor defined by

000 = 110 =

010 = 100 = +

220 = 330 = 1. (56)

13
0
One can show that is the inverse of 0 in the same sense that is the inverse of . In
other words, a relation similar to (48) can be defined

0
0 = . (57)

It is important to re-emphasize that the components of vectors (and tensors, although we


havent discussed this) change under Lorentz transformations, meaning that they will look different
to different inertial observers. However, the magnitudes of vectors (and tensors) are invariant under
the Lorentz transformations, since by definition they are scalars and the very definition of a scalar
is that it appears the same to everyone. Examples are x x , dx dx which is just the metric ds2 ,
any V V , as well as any ordinary scalar quantity such as the mass of a particle m and so on.
These so-called Lorentz invariants are important quantities in relativistic physics, and in many
cases provide invaluable insight into the meanings of things.

7 The proper time

Before proceeding, an important question to ask is, constrained to time measurements only, what
is the difference between what a moving observer sees and what a stationary observer measures?
Consider an event happening in the rest frame of observer O0 , in other words the frame attached
to O0 . The duration of the event, as measured by O0 , is generally known as the proper time of the
event d . Now imagine that O0 , carrying the event in question with him or her, is moving with a
velocity v away from another observer O, how long would O perceive the event to take? Certainly
not d but another, different, period of time dt. One way of showing how the proper time is related
to the time as measured by someone else is to write the metric with respect to O0 and compare
it with the metric with respect to O. Remember both observers will agree on the metric but will
disagree on its constituents. For O0 , the event is at rest with respect to his or her rest frame, so
their metric will just be
ds2 = c2 d 2 (58)

Note that this by itself provides a meaning to the metric, it is a quantity that is simply related
to the proper time of an event. Now, the observer O will see O0 and the event moving away from
him or her, hence they will measure both a duration and a distance change, so their metric will be

ds2 = c2 dt2 + dx2 (59)

14
But both observers will agree on ds2 , hence by equating (58) an (59) and noting that the speed
of the O0 reference frame with respect to O is just v = dx/dt, we find that

dt = d. (60)

8 The relativistic mechanics of a point particle

Just as in ordinary non-relativistic mechanics, our starting point is defining the position vector of
a point particle. We have already done this, it is

x = ct, xi

(61)

where xi are the spatial components x1 , x2 , . . . xd . We now define the spacetime velocity as the
derivative of (61) with respect to the proper time, which by using (60) can be written in terms of
any observers t:
dx dx
U = = = (c, v) (62)
d dt
Different observers will of course see different components of U , each observer with their own
value of . However, as noted earlier, they will all agree on the magnitude of U , as can be easily
checked by calculating
U U = c2 (63)

which is clearly independent of how fast the observer is moving, i.e. does not rely on . Next, we
define the spacetime momentum in the usual way

p = mU = m (c, v) (64)

where m is the mass of a moving particle. Clearly the spatial components of p are the relativistic
momentum, i.e.
p = mv (65)

The zeroth component has the surprising interpretation of being the total energy of the particle
divided by c, in other words
E
p0 = = mc, (66)
c
hence
E = mc2 . (67)

15
Note that the total energy of a particle in its own rest frame is found by setting = 1 which
gives the very familiar formula
E0 = mc2 (68)

Lets also not forget that we are discussing a free particle here, which means that there are no
external forces, and thus no potential energy. What the above analysis means is that a free particle
traveling in vacuum would have a total energy of mc2 , part of which is the rest energy of the
particle mc2 , while the remaining part can only be the particles relativistic kinetic energy. This
can be defined then by

K.E = E E0 = mc2 mc2 = ( 1) mc2 . (69)

A nice way of demonstrating that ( 1) mc2 is indeed kinetic energy is to see what it becomes
at the limit of slow speeds << 1. Expanding in powers of and keeping only the lowest order
terms gives
v2
1+ (70)
2c2
which when used in (69) gives
v2 v2
 
1
K.E = 1 + 2 1 mc2 = 2 mc2 = mv 2 (71)
2c 2c 2
as one would expect. The interpretation that energy is just the zeroth component of the spacetime
momentum is very interesting and hints at a relationship between time and energy which can further
be explored. The interested reader is referred to what is known as Noethers theorem for further
study. We now conclude this section by computing the Lorentz invariant scalar p p :
E2
p p = + p2 = m2 c2 (72)
c2
where p2 = p p = 2 m2 (v v). This leads to the very useful

E 2 = m2 c4 + p2 c2 (73)

Not surprisingly, if a particle is at rest, i.e. has vanishing momentum, its energy reduces to the
rest energy E = mc2 = E0 as expected. The interesting thing however is that the formula (73)
also seems to imply that maybe there could be a way to define a massless particle that carries a
non-mass based momentum p and has total energy that can be found by setting m = 0 in (73);
leading to E = pc. This is in fact quite possible, one of the consequences of quantum mechanics is
that massless particles can and do exist, and they carry both momentum and energy, for example
the photon.

16
9 Light cone coordinates

A particularly useful set of coordinates heavily used in string theory is the so-called light cone
coordinates. These are found by a linear combination between the x0 and x1 coordinates as shown
below, whilst leaving the remaining spatial directions the same:
1
x+ = x0 + x1

2
1
x = x0 x1 .

(74)
2
Hence the components of the spacetime position vector in this system are x+ , x , x2 , x3 , . . . xd .


Based on this, the metric becomes


2 2  2
ds2 = 2dx+ dx + dx2 + dx3 + + dxd (75)

and the components of the light cone metric tensor are

+ = + = 1

00 = 11 = 0

22 = 33 = = dd = 1, (76)

with the vanishing of all remaining components. Consequently, the components of any vector are
redefined
1
a+ = a0 + a1

2
1
a = a0 a1

(77)
2
such that the light-cone scalar product becomes

a4 b4 = a b = a b+ a+ b + a2 b2 + a3 b3 + (78)

where in this last formula the Greek indices count over (+, , 2, 3, . . .) rather than the usual
(0, 1, 2, 3, . . .). It can be readily shown that

a+ = a

a = a+ . (79)

Using this language, one can proceed to rewrite all of relativistic mechanics in terms of light
cone coordinates.

17
10 The action of a relativistic particle

The action for relativistic point particles has the following equivalent forms
Z
S = mc ds

Ztf
2 dt
= mc

ti

Zf r
dx dx
= mc d , (80)
d d
i

where is the Lorentz factor and is an arbitrary parameter of the world line that may or may not
R
be the proper time. Geometrically, the action is proportional to the length of the world line ds,
and requiring S = 0 is equivalent to requiring the world line to be as short as possible between
two spacetime points. This is important to note, since we will later guess the relativistic strings
action as being proportional to the area of the strings world sheet.
Requiring the vanishing of the variation of S leads to the expected free particle equation of
motion
dp
= 0. (81)
d
The action and the equations of motion are reparameterization invariant, so any parameter
leads to exactly the same physics. In other words, if one substitutes 0 ( ) in (80) we get
exactly (81) back. On the other hand, if is the proper time, then (81) leads to

d2 x
=0 (82)
d

as one would expect.

11 Relativistic strings and the Nambu-Goto action

As discussed above, the action of the relativistic point particle is proportional to the length, in
spacetime, of the particles world line:
Z
S = mc ds (83)

As such, the condition S = 0 is tantamount to calculating the shortest spacetime distance


between two events, i.e. we are minimizing the length s. The spacetime coordinates of the point

18
particle are then parameterized by a number which may or may not be the proper time of the
particles rest frame: x ( ).
Based on this experience, we define the relativistic strings action as the spacetime area of
its world sheet. The condition S = 0 is then geometrically analogous to finding the smallest
possible area of the world sheet between two pre-determined locations of the string. We define the
spacetime position vector as pointing to a point on the world sheet and parameterized by two
coordinates on the world sheet, one of of which we call and the other :

x X (, ) , = 0, d. (84)

Note the difference in notation, while x is any spacetime position vector, X (, ) is a position
vector pointing to the world-sheet. We will find that X (, ) behaves as a number D of scalar
fields living on the world-sheet. As such, X (, ) is sometimes referred to as the Bosonic world-
sheet fields; where the name implies that it defines Bosons not Fermions, as we will eventually
see.
The action can then be constructed (from the theory of minimal surfaces) as such:

Zf Z 1
r
T0 2
S= d d X X 0 X 2 X 02 . (85)
c 0
i

This is the Nambu-Goto action. It describes the dynamics of a free relativistic string. Over-
dots are derivatives with respect to and primes are derivatives with respect to . Although they
can be anything, the parameters and can be thought of as the proper time of the strings rest
frame and the proper length on the string. From that perspective, the number 1 is just the length
of the string. The constant T0 can be shown to be the strings tension, related to the strings rest
mass density 0 as follows:
T0 = 0 c2 . (86)

Finally, in (85) the following shorthand notations were also used:

X X X X X X
X 2 = , X 02 = , X X 0 = . (87)

The parameters and can be thought of as coordinates on the world-sheet with a metric ,
local to the world-sheet. This can be related to the spacetime metric. First, define 1 = and
2 = , hence
x = X ( ) (88)

19
where indices from the beginning of the Greek alphabet, such as and , count over 1 and 2 only.
The Minkowski metric can then be written as

X X
ds2 = dX dX = d d , (89)

where the chain rule was applied on (88). If we then define

X X
= (90)

such that
ds2 = d d (91)

then the quantity is a 22 matrix representing the metric tensor on the world sheet. The
components of can be read off directly from (90):

X 2 X X 0
[ ] = (92)
X X 0 X 02

This is known as the induced metric on the world sheet, while is the metric on spacetime,
sometimes referred to as the target space in which the world sheet is embedded, i.e. the world-
sheet metric is a metric within a metric. Comparing (92) with (85), it is clear that the later can
be re-written
Zf 1
Z
T0
S= d d (93)
c 0
i

where is the determinant of the matrix (92). Varying the action in either of its forms (85) or (93)
leads to the following equation of motion of the relativistic string

P P
+ =0 (94)

reminiscent of equations (21). The vectors P and P have a very similar definition as derivatives
of the Lagrangian density, except that in this case they lead to a bit more complicated forms:

L T0 h  i
P = = X X 0 X0 X 02 X
X c
L T0 h 0
 i
X 2 X 0 .
P = = X X X (95)
X 0 c

These will considerably simplify under a specific choice of reparameterization gauge, i.e. once
we choose what and really are.

20
12 The reparameterized equations of motion

From this point onwards, we switch our units system to the so-called natural units, where the
speed of light is chosen to be c = 1. Later on, we will also choose Plancks constant ~ to be
equal to the unity. This simplifies the equations by simply dropping the cs and ~s everywhere.
Furthermore, it is conventional, for historical reasons, to define the so-called Regge slope:

1
0 = ; (96)
2T0

essentially the inverse of the string tension.


The equations of motion derived in the previous section are particularly complicated. But
recall that we have not really defined the parameters and , we only know that they are arbitrary
coordinates on the world-sheet. We make the following reparameterization gauge choices. First,
we introduce a unit spacetime vector n defined to be as either timelike or null and orthogonal to
the plane defined by the string itself in spacetime. Then choose such that

= n X (97)

where is an arbitrary parameter. With the further choice that = t, i.e. ordinary time5 , (97)
guarantees that the string itself is spacelike, i.e. extended in the spacelike direction and propagating
in the timelike or null directions as one would want based on physical grounds. The parameter
can then be simply interpreted as a length parameter along the string itself, with 1 being the
total length of the string. For later convenience, we choose 1 = in the units we are using.These
choices simplify the equations of motion considerably, as well as introduces extra conditions on the
string world sheet. The consequences are as follows

n X = 0 (n p )
2
n p = n P , (98)

where is 2 for open strings and 1 for closed strings. The vector p is simply the overall spacetime
momentum of the string (i.e. analogous to equation 18)
Z

p = dP . (99)
0
5
This choice is known as the static gauge.

21
The momentum densities P and P now simplify to

1
P = X
20
1
P = X0 (100)
20

which, when substituted in (94) gives the much simpler equation of motion

X 00 = 0
X (101)

which we recognize immediately as the wave equation in spacetime. Finally we have the following
condition, which follows from the gauge choices (98), that must be satisfied for any solution of
(101):
 2
X X 0 = 0. (102)

13 The relativistic open string solution

Solving (101) with its associated conditions is a straightforward matter and gives, in terms of an
infinite Fourier series:

1
X (, ) = x0 + 20 p i 20
X
a in
an ein cos n.

n e (103)
n=1
n

The Neumann boundary conditions were also used, as this represents a freely propagating open
string. The vector x0 is the initial spacetime position vector of the strings center of mass (basically),
and p is the linear momentum of the strings center of mass. Hence the first and second terms
are essentially what one would expect from a point particle6 . The infinite Fourier expansion in the
last term represents the expected oscillatory modes of the string, with the complex constants an
and their conjugates a
n representing the Fourier amplitudes. The expression (103) is one way of

writing the relativistic strings solutions, another is found by defining:



0 = 20 p

n = an n

n = a
n n (104)
6
Analogous to the familiar Freshman physics formula: x (t) = x0 + vt.

22

where n 1. To guarantee that the solutions remain real, we also require n = (n ) . These
lead to the second form of the relativistic string solution:

1 in
X (, ) = x0 + 20 0 + i 20
X
n e cos n (105)
n=
n
n6=0

14 The relativistic string in light cone coordinates

It is straightforward to show that the choice


 
1 1
n = , , 0, 0, , 0 (106)
2 2
leads directly to the light cone coordinates defined earlier:
1
n X = X0 + X1 = X+

2
1
n p = p0 + p1 = p+ .

(107)
2
Applying this to (98) gives

X + = 0 p+
2 +
p+ = P (108)

The coordinate X + is then particularly simple and carries the dynamics of pure propagation
of the string, while it can be shown that the other independent parameter is not X , but is, in
fact, the constant vector x
0 . This will become particularly important when we get to the quantum

relativistic string. If we assign the upper case indices I, J, K to the spatial directions transverse to
the string (i.e. I, J, K = 2, 3, 4, . . . , d), then

X 1 I in
X I (, ) = X0I + 20 0I + i 20 e cos n (109)
n=
n n
n6=0

describes the transverse oscillatory dynamics of the string. The next step is to quantize the open
relativistic string solution in the light cone gauge, but before we do this, we must review some
quantum mechanics.

15 Non-relativistic quantum mechanics

Quantum mechanics was discovered in the early twentieth century. It is a complete re-write of how
we ask questions about the natural world; a paradigm shift paralleled only by the discovery of the

23
special theory of relativity. In the microscopic scale where quantum mechanics becomes prominent,
questions such as where is the particle, or how fast is it moving? no longer make much sense.
You cannot pinpoint the position or velocity of a quantum object, but you can calculate probabilities
of where it is and how fast it is moving. This discovery signaled the end of determinism in the
classical sense. Laplaces famous demon (see here) is now out of work! Or is it? While we can
no longer explore the evolution of a particles position or momentum, it turns out there is a single
quantity, one that has no classical analogue, whose history and evolution can still be explored in
deterministic detail. This is (r, t); the so-called wave function of quantum particles, also known
as the probability amplitude, and sometimes simply as the quantum state of the particle.
It is a complex-valued scalar function of space and time that has, by itself, no physical meaning
whatsoever. It turns out, however, that its magnitude squared gives the probability of finding the
particle in a given volume. So
Z
P = | (r, t)|2 dV (110)
V
is the probability of finding the particle in a box of volume V . Consequently, if one expands the
box to infinity, then
Z
P = | (r, t)|2 dV = 1 (111)
all space

in other words, the probability of finding the particle somewhere has to be 100%. Along with
the wave function, observable quantities, such as position and momentum, become operators in
quantum mechanics. The most important such operators are the position
r, momentum p
, and
(referred to as the Hamiltonian). In their simplest possible forms, the momentum and
energy H
Hamiltonian operators are, respectively:
*
= i~
p
2
= ~ 2 + V (r)
H (112)
2m

where V is potential energy. But these definitions are not universal, they are true only in the
non-relativistic case. However, the concept of turning observables into operators is universal and
is used to change classical theories, even relativistic ones, into quantum theories. Operators satisfy
important properties. Foremost among those is that they do not necessarily commute with each
and B
other, so given two operators A one can define the so-called commutator
h i
B
A, =A
B B
A (113)

24
which may or may not vanish. This means that the action of operators on the particles quantum
first then with B
state is order-sensitive. Operating with A will generally give different answers

from the other way around. The basic commutators of non-relativistic quantum mechanics are

[
xi , pj ] = i~ij

[
xi , x
j ] = [
pi , pj ] = 0, i, j = 1, 2, 3. (114)

Finally, we note that the action of these operators on the wave function yield numbers, gen-
erally known as the eigenvalues of the operator. In physical terms the eigenvalues represent the
observables themselves, so x
(r, t) = x (r, t) gives the x location of the particle, and px (r, t) =
px (r, t) gives the x component of its momentum, and so on. Of course, these quantities are
not quite measurable, however their probabilities are. So in order to find the probability that the
particle has, say, a specific momentum in the x direction, we calculate the so-called expectation
value of the momentum
Z
h
px i = (r, t) px (r, t) dV (115)
all space

and similarly for the other operators. Wave functions have some interesting properties that led to
deep insight of their nature. For example, one can show that (111) generalizes to
Z

P = m (r, t) n (r, t) dV = mn (116)
all space

where m and n are integral labels to two different states . The relations (116) are reminiscent
ei
of the properties of unit vectors in ordinary space, e.g. ej = ij , where
e1 = x
,
e2 = y
,
and
e3 = ei
z are the usual unit vectors in Cartesian coordinates. The formula ej = ij is
called the orthonormality relation of the unit vectors since it indicates that the unit vectors
are orthogonal to each other, and have a normal magnitude of 1. As such, equation(s) (116) are
the orthonormality relations of the wave functions. The states m are then taken to represent
unit vectors in an abstract infinite dimensional space known as Hilbert space. We can now
write them using the vector-like notation

m |m i = |mi (117)

where | i is known as a ket. If we define the complex conjugates of the wave function to be the
dual unit vectors:

m hm | = hm| (118)

25
where h | is called a bra, then the orthonormality relations (116) can be written as

hm | ni = mn (119)

where the bra(c)ket expression on the left hand side of (119) implies the integral (116); equivalent
to a scalar product of vectors in Hilbert space. General states can then be written as a linear
combination of unit vectors as follows
X
|i = cn |ni. (120)
n

In this language, known as the Dirac notation, operators and their expectation values are
re-written


A A |i
Z D E
(r, t) A
(r, t) dV
A . (121)
all space

The discussion so far has been focused on a representation of quantum mechanics known as the
Schr
odinger picture. In this picture, the wave function is time dependent and is assumed to
satisfy the Schr
odinger equation:

| (r, t)i = i~ | (r, t)i .


H (122)
t

Alternatively, there exist the so-called Heisenberg picture of quantum mechanics, where the
entire time dependance is carried not by the wave function, but by the operators themselves. It can
be shown that the time-dependent Heisenberg operators AH (t) are related to the time independent
Schrodinger operators AS via

AH (t) = eiHt AS eiHt (123)

where exponent operators can be evaluated using their infinite series expansion, i.e.

1 t2 H
eiHt = 1 + iHt H i t3 H
HH
+ (124)
2! 3!

The Heisenberg operators satisfy a differential equation, known as the Heisenberg equation,
that determines the time evolution of operators in the Heisenberg picture, just as the Schrodinger
equation determines the time evolution of the wave function in the Schrodinger picture thus:

dAH AH h i
i =i + AH , H . (125)
dt t

26
This indicates that an operator AH that is not explicitly dependent on time (i.e. its partial
derivative with respect to time vanishes) and commutes with the Hamiltonian is necessarily a
constant of motion, i.e. represents a conserved observable quantity.
Both of the Schr
odinger and Heisenberg pictures are mathematically equivalent to each other.
We choose, as is customary, to use the Heisenberg picture to quantize our relativistic theory of
strings. The reason is that it is more straightforward to define, or guess at the definition of,
operators that satisfy the Heisenberg equation, than it is to define a time-dependent quantum
state. Also note that the Schr
odinger equation in its given form works only non-relativistically,
while the Heisenberg equation is a general relation between operators and is valid in any quantum
theory, relativistic or not.
The recipe for quantization is somewhat simple: Start with a relativistic theory in
any system of coordinates you choose (we will choose light cone coordinates). Decide
on your independent variables, make them into operators. Find their commutation
relations, then finally construct a Hamiltonian and check that the Heisenberg equation
works for all operators. But before doing that to the relativistic string, lets first see what these
quantities look like in the simpler case of the relativistic point particle.

16 The quantum relativistic point particle in light cone gauge

Using the energy condition p p = m2 (where c = 1) as our reference, we choose the light cone
dynamical variables of the point particle to be the set

xI , xI0 , pI , p+

(126)

where the relations


p
x = x
0 +
m2
pI
xI = xI0 + 2 (127)
m
hold. The indices I count over 2, 3, 4, d in D = d + 1-dimensional spacetime. Imposing a hat
on each observable to make it an operator we construct the commutation relations similarly to the
non-relativistic case (with ~ = 1):
 I J
x
, p = i IJ
 +
x
0 , p = i. (128)

27
The Hamiltonian has the following equivalent representations
+
= p p = 1 pI pI + m2

H 2 2
(129)
m 2m

where pI pI is a summation over I. Using Heisenbergs equation, it can be shown that this is indeed
the correct form of the Hamiltonian, as it leads to relationships between the operators that are
exactly analogous to their classical counterparts.

17 The Dirac delta function

Before delving into the quantization of the free open string, let us define the so-called Dirac delta
function7 . Not really a function in the strictest sense of the word, the Dirac delta has the following
strange definition:
0 x 6= a
(x a) = (130)
x=a

where a is a constant that may or may not be zero. The importance of the Dirac delta stems from
requiring that
x>a
Z
dx (x a) = 1 (131)
x<a

as this allows us to define the integral


x>a
Z
dxf (x) (x a) = f (a), (132)
x<a

where f (x) is an ordinary function (smooth, differentiable etc, i.e. not another Dirac delta). In
other words, the Dirac delta, when multiplied by f (x) and integrated around x = a, picks a single
point, out of the infinity of points, on the function f , that is the point x = a and returns its
functional value f (a). This is reminiscent of the action of a Kronecker delta that tends to pick a
single component of a discrete number of components of a specific vector, i.e. V n nm = V m . Hence
the choice to call them both deltas. The reader can think of the Dirac delta as the continuum
limit to the Kronecker delta.
7
Another major contribution by Paul Dirac. There is one more, concerning the relativistic quantum mechanics of
Fermions that we will briefly discuss towards the end of this course.

28
18 The free relativistic quantum open string

In analogy with the dynamical variables of the point particle in light cone coordinates, we take the
basic dynamical operators of the relativistic string to be
 
X
I, x I +
0 ,P ,p (133)

with the commutation relations


h i
I () , P I 0 = i IJ 0

X
 +
x
0 , p = i (134)

leading to the following form for the Hamiltonian operator


Z "
0I X
0I
#
= 0 d P I P I + X
H (135)
(20 )2
0

with the expected


I
= 20 P I .
X (136)

Finally, we also find

I
nJ = m IJ m+n,0
 

m ,
h i
aIm , a
J
n = IJ mn
 I J
x
0 , p = i IJ . (137)

Note: The (dagger) symbol signifies a complex conjugate operator that is also Hermitian,
D E D E
. We wont have to worry
i.e. satisfies the quantum mechanical condition A = A
too much about that, I am just noting it for completeness.

18.0.1 The transverse Virasoro modes

First defined in the non-quantum relativistic theory by the combination

1X I
L
n =
I
nm m (138)
2
mZ

the so-called transverse Virasoro modes become incredibly useful in the quantum theory, when
they are turned into the Virasoro operators. While they can just be viewed as yet another set
of operators; they carry particular importance in the light cone gauge. The definition (138) is

29
based on the non-quantum relativistic theory, however in the quantum theory we know that the
operators do not commute (137), so we ask whether their order in (138) is correct. Of particular
importance is the operator L
0 as it will turn out to be proportional to the Hamiltonian, in that

sense the correct order of the operators in (138) is essential. Taking this question into consideration
we find that
H
=L0 1 (139)

where the so-called normal ordering constant 1 is added to guarantee the correct order of
. The calculation that leads to this also imposes the number of spacetime
oscillator operators in L 0

dimensions of the theory to be D = 26. This is essential as it is the only number that preserves
Lorentz invariance. Furthermore, the operators aI I
n and an are now re-interpreted as creation and

annihilation operators respectively. They act on the string quantum states to excite higher or lower
oscillatory modes, which carry higher or lower masses. Details to follow.

19 Constructing the quantum states of the open string

Now that we have a fully quantum mechanical theory of the relativistic open string, the next step
would be to construct all the possible quantum states of the string; i.e. the states that tell us what
kind of particle a specific string represents. However, we must first develop some terminology. In the
I
previous section we noted that the operators a In are relabeled creation and annihilation
n and a

operators. This is understood by analogy to the quantum mechanical harmonic oscillator whose
Hamiltonian function (the total energy in classical terms) is made into an operator as follows
2
= p + 1 m 2 x
H 2 . (140)
2m 2

with the usual commutation relation [


x, p] = i~. If we define the new operators

a
1
= (m x i
p)

a 2~m

N a
= a (141)

we can easily show that the following is true


h i
a
, a = 1
 
1
H = ~ N + . (142)
2

30
These operators have an extremely elegant interpretation. It can be shown that the eigenvalues
are always positive integers, which means that the action of the Hamiltonian operator on a
of N
given quantum oscillator returns discrete (quantized) energy states:
 
1
E = ~ n + , n = 0, 1, 2, . . . (143)
2

For example, the lower-most energy state allowed is not zero, but is rather E0 = 12 ~, arising from
3
n = 0. The next allowed state is when n = 1; i.e. E1 = 2 ~, and so on. No energy levels in
between these values are allowed! This is in direct contrast to the classical harmonic oscillator that
can oscillate at any energy (or amplitude) one wants it to. But this is quantum mechanics, where
everything is different. Note that an immediate consequence of these results is that a quantum
harmonic oscillator cannot just sit still, because this would correspond to an energy of zero, which
is not one of the allowed values. The quantum state with n = 0 is thus known as the ground state
of the quantum oscillator. Using the Dirac notation we can denote it by |0i. The state in the next
has the effect of raising the states
energy level is hence |1i, and arbitrarily: |ni. The operator a
to the next available energy, so

|0i = |1i
a

a
a |0i = |2i , (144)

hence the name creation operator. The operator a


has the reverse effect, it lowers the states to
the previous energy level

|1i = |0i
a

a
a |2i = |1i (145)

hence the name annihilation operator. Note that this means that the effect of the annihilation
operator on the ground state is to completely annihilate it, since there are no lower energy levels

|0i = 0.
a (146)

counts the number of rungs up the energy lader where a


In this language, the operator N
given state is in, thus
|ni = n |ni .
N (147)

31
For example

|1i = |1i
N
|2i = 2 |2i
N
|3i = 3 |3i ,
N (148)

also annihilates the ground state, thus N


and so on. Note that N |0i = 0. This counting property
is known as the number operator. Note that (148) can be rewritten as
is the reason N

a
N |0i = a
|0i
a
N a
|0i = 2
a a
|0i
a
N a
a
|0i = 3
a a
a
|0i . (149)

I
As it turns out, the oscillatory Fourier operators of the string, i.e. a In , relate to the
n and a

strings Hamiltonian in the same way the creation and annihilation operators of the quantum
oscillator relate to its Hamiltonian. We interpret them then as creation and annihilation operators
of the strings oscillatory modes, which the reader will recall are directly related to what particle
I
the string represents. So in a sense, a In are creation and annihilation operators of particles!
n and a

In this context, we also define the strings number operator



X
=
N aI
n In
n a (150)
n=1

that, upon action on a given string state (a.k.a particle) counts how many states above the strings
ground state it is. Furthermore, it can be shown, via the standard relativistic formula p p = m2
that the so-called mass operator of the string is

2 = 1 N
 
M 1 . (151)
0
2 acts upon a given string state, it returns the mass squared value
As one would expect, when M
of the particle; e.g.
2 |ni = m2 |ni = 1 (n 1) |ni
M (152)
0
for the particular quantum state |ni. Finally,
where the positive integer n is the eigenvalue of N
the following commutators are useful in calculations:
h i
N , a
I
n aI
= n n
h i
N , a
In = n aIn (153)

32
The quantum states of the string can now be specified. They are described by bras and kets
as usual. They are states of some arbitrary value of the strings linear momentum. So the ground
state is defined by
|0i = p+ , pI

(154)

where it is understood that it is a ground state in oscillations only, but the string is free to assume
any continuous value of its linear momenta p+ and pI . We can then build the states in analogy to
the quantum oscillator as follows

I
a n |0i = |1i

I
an aJ
n |0i = |2i , (155)

and so on. Also:

In |1i = |0i
a

In a
a Jn |2i = |1i (156)

and the number operator has a similar effect as before

a
N I I
n |0i = an |0i

a
I J aI J
N n a n |0i = 2 n a n |0i

a
N I
n aJ
n aK aI
n |0i = 3 J
n a K
n an |0i . (157)

The ground state is of course still annihilated by both the annihilation and number operators:

a |0i = 0
In |0i = N (158)

Finally, we note that the Hilbert space of the string has well-defined inner products of two
different ground states as follows:
D 0

0
E  0   0 
p+ , pI p+ , pI = p+ p+ pI pI (159)

which naturally leads to well-defined inner products of any two arbitrary states.
Using this language, the masses of the string states (the particles) can be almost trivially
calculated by operating on the states with the mass squared operator (151). The ground state
p , p , has n = 0, hence (152) gives m2 = 10 ! This is a particle known in the shady annals
+ I

of particle physics as a tachyon. It has complex mass, travels faster than the speed of light, and

33
backwards in time! The presence of this state signals an instability of the vacuum and almost
certain doom for string theory. Luckily, upon the introduction of supersymmetry (which we will
not do), the tachyonic ground state simply disappears. The next state has n = 1, hence m2 = 0.
These are massless particles. They can be represented by
25
cI aI
X
|1i = 1 |0i (160)
I=2

where cI are arbitrary constants. Veterans of quantum electrodynamics would recognize this as
exactly the state needed to describe a photon. This then is the first real life particle that appears
in string theory. Of course, this is a 26 dimensional photon, so it is not quite what we experience
1
from a beam of light. The states that follow are clearly massive m2 0 ; they are in fact an
infinite tower of successively more massive particles that string theory predicts but have not been
observed. The standard wisdom is that these are a lot more massive than our current technology is
able to detect. It is presumed that only the massless string states are responsible for the particles
of the standard model, acquiring mass via standard mechanisms in quantum field theory, but that,
of course, is another story.

20 Constructing the quantum states of the closed string

We have not, and we will not, discussed the details of how to build a quantum theory of a relativistic
closed string. The calculations however are analogous to the open string. The main difference is
that there are of course no boundary conditions; instead we have the periodicity condition:

X (, ) = X (, + 2) (161)

for all values of and . One finds that the oscillatory waves on the closed string are oriented,
i.e. allowed to travel clockwise and counterclockwise. This necessarily creates a double for each
n in addition to the usual n to represent
of the operators we have considered; particularly we get
left and right moving waves on the string. The full solution is then

r
0 X ein

n ein .
n ein +

X (, ) = x0 + 20 0 + i (162)
2 n
n6=0

To construct the quantum states we define the creation and annihilation operators as before

n = nan

n = na
n (163)

34
with similar expressions for the barred ones. If we switch to the usual light cone gauge and quantize
the operators, we find
h i h i
a I
Im , a n = a I
Im , a n = mn
IJ


L
= L
0 0

=
X X
=
N aI
n In ,
n a N I
na n a
In
n=1 n=1
= N
N
+ L
= L
H 2
0 0

2 = 2 
+ N

2 .
M N (164)
0

The closed string states are then defined by both sets of creation operators:
" 25 # " 25 #
Y Y  n,I Y Y   m,J
I J
|n, mi
= an a
m |0i . (165)
n=1 I=2 m=1 J=2

2 equation from (164), it is clear that the ground state is also tachyonic with
Using the M
M 2 = 4/0 . The first excited state however is a lot more interesting, it is constructed using both
creation operators aI I
1 and a

1 (since N = N ):

25
RIJ aI J
X
|1i = 1 a 1 |0i (166)
I,J=2

where RIJ is a matrix of arbitrary constants. Any square matrix can be decomposed into symmetric
and antisymmetric parts:

RIJ = SIJ + AIJ where SIJ = SJI and AIJ = AJI . (167)

Furthermore, any symmetric matrix, in this case SIJ , can be decomposed into the sum of a
0 + S 0 , where S 0 is traceless and S 0 is
traceless matrix and a multiple of the identity: SIJ = SIJ IJ IJ

a constant that in this case is equal to the trace of SIJ divided by (D 2). Hence

0
RIJ = SIJ + AIJ + S 0 IJ . (168)

In conclusion, the massless states of the closed string splits into three different particles:
25 25
aI J AIJ aI J 0 I I
X X
0
|1i = SIJ 1 a 1 |0i + 1 a 1 |0i + S a1 a
1 |0i . (169)
I,J=2 I,J=2

35
The first particle is what one would expect for a graviton state! This is a remarkable discovery.
String theory, in a sense, predicts gravity! The entire fame and notoriety of string theory started
with (169). The second particle represents the quantum particle state of the so-called Kalb-
Ramond antisymmetric field. Finally, the last quantum state in (169) is the particle of a scalar
field known as the dilaton.

21 Superstring Theory

We conclude the course with a quick look at open superstrings. The idea is to add a Fermionic
part to the Nambu-Goto action SN G such that the total superstring action becomes

S = SN G + SF . (170)

The complete action is invariant under supersymmetric variations (which we will not discuss).
What is needed is to construct a theory that would have states that respect Paulis exclusion
principle. Ever since Diracs seminal work on fermionic quantum theory it has been known that
the so-called Grassmann numbers play an important role in accomplishing just that. These are
numbers that satisfy the algebra:
{a, b} = ab + ba = 0, (171)

where the parenthesis {a, b} represent an anti-commutator, the opposite of the usual commutator.
So instead of commuting, like ordinary real numbers, Grassmanns numbers anti-commute. This
leads to the important property
a2 = 0. (172)

This is exactly the property needed to satisfy Paulis exclusion principle. Now, if Grassmann
numbers anti-commute, Grassmann operators may not; in other words
n o
, b 6= 0.
a (173)

Now lets get back to the superstring. In addition to the Bosonic fields X (, ), that may
not necessarily commute, we introduce two Fermionic world sheet fields 1 (, ) and 2 (, ) that
may not necessarily anti -commute. Based on Diracs work, the Fermionic part of the action is
Z Z
1
d 1I ( + ) 1I + 2I ( ) 2I
 
SF = d (174)
2
0

36
leading to the equations of motion

( + ) 1I = 0

( ) 2I = 0. (175)

The solutions split into two separate types of fields, known conventionally as sectors
X
The Neveu Schwarz Sector(NS) : I (, ) bIn ein( )
nZ+1/2
X
The Ramond Sector(R) : I (, ) dIn ein( ) . (176)
nZ

When quantized, the Fourier coefficients b and d become NS and R creation and annihilation
operators satisfying specific anti-commutator relations. The fact that b and d are Grassmannian
guarantees that two particles in the same quantum state cannot exist, in full realization of Paulis
principle; i.e. a state of the form
bI I
n bn |0i (177)

is annihilated by virtue of (172). Calculations very similar to what we have seen allow us to calculate
that the number of spacetime dimensions needed for superstring theory is reduced to D = 10. The
rest, as they say, is history.

37
Appendix: Reading assignments - Spring 2017

The section numbering refers to the second edition of Zwiebachs book. There are also some online
sources referenced.

1. Section 4.5. You may also find a lot of sources concerning Hamiltons principle of least Action
and Lagranges formula online. For example: Source 1, Source 2, Source 3. I also strongly
recommend Feynmans excellent discussion here. Finally, some solved problems are here for
your review and practice if you choose to do so (especially problem 4 which will be useful to
us later).

2. Sections 4.1, 4.2, 4.3, and 4.5.

3. Section 4.6. You may also wish to skim the on-line notes: Separation of variables, Solution
of partial differential equations with boundary conditions.

4. Sections 2.1, and 2.2. More detail on special relativity and the mathematics of spacetime
may be found in many online notes, for example here.

5. Sections 2.3, 2.4, 2.5, 5.1, 5.2, and 5.3.

6. Sections 6.1, 6.2, 6.3, 6.4, 6.5, 6.6, 6.7.

7. Sections 6.8, 6.9, 7.1, 7.2, 7.3, 7.4, 7.5, 9.1, 9.2, 9.3, 9.4.

8. Any introductory quantum mechanics notes, for example these.

9. Sections 9.5, 11.1, 11.2, 11.3.

10. Sections 12.1, 12.2, 12.3, 12.4.

11. Sections 12.5, 12.6, 12.8, 13.1, 13.2, 13.3, 14.1, 14.2, 14.4, and 14.5. In general you may just
skim over all of chapter 14.

38

Das könnte Ihnen auch gefallen