Preprint typeset in JHEP style - PAPER VERSION

STRING THEORY

Paul K. Townsend
Department of Applied Mathematics and Theoretical Physics
Centre for Mathematical Sciences, University of Cambridge
Wilberforce Road, Cambridge, CB3 0WA, UK.
Email: [email protected]

Abstract: Part III Course Notes. 24 Lectures.

Lent term 2018
Contents
1. What is String Theory? 1

2. The relativistic point particle 4

2.1 Gauge invariance 5
2.2 Hamiltonian formulation 7
2.2.1 Gauge invariance and first-class constraints 12
2.2.2 Gauge fixing 13
2.2.3 Continuous symmetries and Noether’s theorem 14
2.2.4 Canonical Quantization 16
2.2.5 Quantization for systems with first-class constraints 17

3. The Nambu-Goto string 18

3.1 Hamiltonian formulation 19
3.1.1 Alternative form of phase-space action 20
3.1.2 Gauge invariances 21
3.1.3 Symmetries of NG action 22
3.2 Open string boundary conditions 23
3.3 Monge gauge 24
3.3.1 Polyakov action 25
3.4 Conformal gauge 26
3.5 Solving the NG equations in conformal gauge 29
3.5.1 Conformal invariance of conformal gauge action 31
3.6 Fourier expansion: closed string 33
3.7 Fourier expansion: open string 35
3.8 The NG string in light-cone gauge 38
3.8.1 Closed string in light-cone gauge 40

4. Interlude: Light-cone gauge in field theory 40

4.1 Maxwell in light-cone gauge 41
4.2 Linearized Einstein in light-cone gauge 41
4.3 Massive fields 42

5. Quantum NG string 43
5.1 Light-cone gauge quantization: open string 43
5.1.1 Critical dimension 46
5.1.2 Quantum closed string 49
5.2 “Old covariant” quantization 50

–1–
 5.2.1 The Virasoro constraints 54
5.3 Conformal invariance and Vertex operators 58
5.3.1 Computation of anomalous dimension of V(k) 62
5.3.2 Looking ahead 64

6. Interlude: Path integrals and the point particle 64

6.1 Faddeev-Popov determinant 66
6.2 Fadeev-Popov ghosts 67
6.2.1 FP ghosts for systems with Lie algebra constraints 69

7. Path integrals and the NG string 71

7.0.1 Critical dimension again 74
7.1 Virasoro-Shapiro amplitude 77
7.1.1 The Virasoro amplitude 81
7.2 String theory at 1-loop: taming UV divergences 84
7.3 The dilaton and the string-loop expansion 86

8. Interlude: The spinning particle 89

8.1 Relativistic spin-1/2 90

9. Spinning strings 92
9.1 Conformal gauge and superconformal symmetry 93
9.2 Open spinning string: free ends 96
9.2.1 Ramond sector 97
9.2.2 Neveu-Schwarz sector 101

10. Superstrings 104

10.1 RNS Superstring 104
10.2 The Green-Schwarz superstring∗ 106
10.3 Closed superstrings∗ 107

11. M-Theory∗ 108

1. What is String Theory?

String theory is a theory of strings of a special kind. Consider a string of uniform ten-
sion T and uniform mass density ρ, stretched between two points. Small amplitude
oscillations of this string propagate along the string with speed
s
T
v= . (1.1)
ρ

–2–
If the string is tightened by pulling on the two ends, the tension goes up and the
mass-density decreases slightly, causing v to increase. However, it cannot increase
indefinitely because Special Relativity requires v ≤ c, which implies

T ≤ ρc2 . (1.2)

For everyday strings, such as violin strings, T  ρc2 ; these are non-relativistic.
String theory strings saturate the inequality, i.e. T = ρc2 ; they are therefore ultra-
relativistic. Cosmic strings, which arise as topological defects in scalar fields of
relevance to cosmology, are also ultra-relativistic, but they have a finite core: a cross
section of non-zero area. When probed at wavelengths less than the core size they
betray some internal structure. In contrast, string theory strings are assumed to have
a zero core size (classically, at least) and no internal structure; they are “elementary”
strings, and the oscillation modes are identified with elementary particles.
For an elementary ultra-relativistic string the only dimensionful constant of rel-
evance is the string tension T . In the quantum theory this serves to define a “string
length” r
~c
`s ∼ . (1.3)
T
Usually we will choose natural units for which ~ = c = 1 and sometimes we will write
1 √
T = (⇒ ` s ∼ α0 ) , (1.4)
2πα0
where α0 is called the “Regge slope parameter”. This terminology comes from the late
1960’s when it was found that hadron resonances could be predicted by extrapolating
straight lines of slope α0 in a plot of angular momentum J against m2 . At that time it
was supposed that `s ∼ 1 fm (sub-nuclear distances). The “dual model” theory that
was developed to explain such properties was interpreted as a string theory around
1970.
The application of string theory to hadron physics soon ran into difficulties.
There was a persistent prediction of a zero-mass spin-2 particle, and it was found
that the theory was intractable, or even inconsistent, unless Minkowski spacetime
was assumed to have dimension D = 26. This is now called the “critical dimension”.
It was also found that the high-energy scattering of string theory particles was par-
ticularly “soft”, in disagreement with experiments in which electrons were scattered
off nucleons. String theory also had to compete with the emerging theory of QCD,
which was ultimately much more successful.
By 1974 string theory was virtually abandoned, but that year it was resurrected
with a new interpretation. Now it was argued that string theory was a theory of
quantum gravity and that one should take (G is Newton’s gravitational constant)
r
~G
`s ∼ `P = (Planck length). (1.5)
c3

–3–
The massless spin-2 particle was now welcome; linearisation of Einstein’s gravita-
tional field equations yields a free field theory whose quanta are gravitons, and a
graviton is a massless particle of spin 2. The soft high-energy behaviour was also an
indication that string theory might resolve the problem of non-renormalizable UV
divergences that arise in attempts to quantize GR as an interacting spin-2 QFT.
The critical dimension of D = 26, later to be reduced to D = 10 in the context of
superstring theory, was now an opportunity to unify gravity with other forces that
could emerge on compactification to D = 4, along the lines suggested by Kaluza and
Klein in the 1920s.
In reality, the self-consistency of string theory as formulated in 1974 (and until
1995) requires `s  `P because otherwise the assumption of a Minkowski background
makes no sense; if `s ∼ `P then the gravitational back-reaction of the string is
not negligible. This restriction translates to a weak coupling requirement on string
theory. In fact, string theory as understood then (and in this course) is only defined
as an asymptotic perturbation series, with the first few terms providing a good
approximation at weak coupling. In other words, “string theory” is really a weak-
coupling expansion of some other theory in which strings might not be as fundamental
as they appear to be in perturbation theory.. More on this below.
Particles propagate along worldlines. To introduce interactions, one has to posit
point-like interaction vertices at which particles meet. One then gets a scattering
amplitude by summing over all possibilities consistent with given initial and final
states. This is how Feynman diagrams originated; fields were not necessary. This
particle viewpoint ended up in appendices to Feynman’s QED papers, because by
the time he wrote them his diagrams had been re-interpreted by Dyson as visual
aids to the organisation of QFT calculations. However, string theory did not emerge
as a generalisation of QFT. Instead, it is a generalisation of Feynman’s ideas about
particles. There is a string-theory analog of QFT called String Field Theory, but it
is much more complicated and still under construction; it is not a good entry point
to the subject.
Interactions of particles have to be pointlike because any “smearing” of the in-
teraction in a relativistic theory will violate causality (this is the locality requirement
on relativistic QFT). But point-like interactions lead to UV divergences; these can
be removed by renormalisation for some theories, at the cost of introducing a few
free parameters, but this doesn’t work for gravity. Strings can potentially avoid this
problem because strings propagate along worldsheets, and it is a simple matter to
draw smooth worldsheets that represent the scattering of strings. When the world-
sheet is viewed as a sequence of spacelike curves, which involves a choice of Lorentz
frame, the interaction looks like either (i) a splitting/joining of two ends, or (ii) the
merging/separation of two strings, but in either case the position of the “interaction
point” on the worldsheet depends on the choice of frame; it is not a Lorentz invariant
concept. There can be no “interaction point” on a smooth worldsheet.

–4–
 A string either has two ends (“open”) or it has no ends ( “closed”). Given a
theory of closed strings with interactions only of type (ii), open strings will never
appear. In contrast, given a theory of open strings, either type of interaction will
allow the appearance of a closed string. So all string theories must contain closed
strings. The graviton appears in the spectrum of every closed string, so all string
theories are (perturbative) quantum gravity theories.
How many string theories are there? Initially there was only what we now call
the “bosonic string” because there are no fermions in its spectrum; The critical
spacetime dimension is 26. Its ground state is tachyonic (particle of negative m2 )
which implies an instability of the Minkowski vacuum. It is now believed that there
is no stable vacuum state (it is similar in this respect to φ3 scalar QFT). These
problems can be resolved by introducing worldsheet fermions, in two ways: this gives
the Ramond (R) and the Neveu-Schwarz (NS) strings, and the critical dimension is
now 10. Ultimately, consistency requires a mixture of R and NS, resulting in the RNS
superstring, so called because the spectrum turns out to have (D = 10) spacetime
supersymmetry. An alternative formulation (the Green-Schwarz superstring) makes
this feature manifest.
In the 1980s it was found that there are five consistent superstring theories

• Type 1. Open and closed strings with N = 1 spacetime supersymmetry.

• Types IIA and IIB. Closed strings with N = 2 spacetime supersymmetry. We

shall be having a look at these two.

• Heterotic E and O. Closed strings with N = 1 spacetime supersymmetry.

All five have a low-energy effective description in terms of a D = 10 supergravity

theory.
String theory is not the only candidate for a theory of quantum gravity (two other
competing proposals go by the names “Asymptotic Safety” and “Loop Quantum
Gravity”). However, string theory has the property that consistency allows the string
of a string theory to couple only to itself1 . So if one proposes a string theory as the
theory of quantum gravity then one is necessarily proposing that this theory describes
everything else, e.g. standard model of particle physics, dark matter, dark energy,
and whatever else is discovered. This state of affairs led string theory to be called “A
Theory of Everything” (TOE). This was an unfortunate choice of terminology but
there is a serious point to it: ambition is not optional in string theory. On the other
hand, there are 5 distinct consistent superstring theories whereas one might expect
a TOE to be unique.
1
This is a statement about string perturbation theory, but what is currently understood about
M-theory (see below) indicates that some modified statement of this kind will still be true.

–5–
 In fact, the five superstring theories look different because they were all found as
weakly coupled theories. Around 1995 it was found that all five are actually weak-
coupling limits of a single theory for which the spacetime dimension is 11 rather than
10, and in the 11-dimensional vacuum there are membranes rather than strings, and
the effective description is provided by D = 11 supergravity. Remarkably, D = 11
is the highest dimension allowing a supergravity theory, and this theory is unique at
low energy. This unified theory acquired the name “M-theory”. It is more like a set
of interlocking ideas than a “theory” so the name is provisional. However, one thing
is clear: M-theory is not just a theory of strings but rather a “brane democracy”
because various types of “brane” (extended object, a string is a 1-brane) appear in
a symmetrical way.
Enough motivation. Now we start on particles.

2. The relativistic point particle

What is an elementary particle?

• (Maths) “A unitary irrep of the Poincaré group”. These are classified by mass
and spin.

• (Physics) “A particle without structure”. The classical action for such a particle
should depend only on the geometry of its worldline (plus possible variables
describing its spin).

Let’s pursue the physicist’s answer, in the context of a D-dimensional Minkowski

space-time. For zero spin the simplest geometrical action for a particle of mass m is
Z B Z B√ Z tB √
2 dxm
I = −mc dτ = −mc 2
−ds = −mc −ẋ2 dt , ẋm = , (2.1)
A A tA dt
where t is an arbitrary worldline parameter. In words, the action is the elapsed proper
time between an initial point A and a final point B on the particle’s worldline.

• The Minkowski coordinates are xm (m = 0, 1, . . . , D − 1) and the Minkowski

metric is η = diag.(−1, 1, . . . , 1). That is, we use a “mostly-plus” signature
convention and the ‘time’ coordinate x0 actually has dimensions of length (it
is really c, the speed of light, times some time coordinate).

• Latin indices (m, n, . . .) will be used for spacetime coordinates (but x will
become X for the string). When we later need to introduce coordinates for the
string worldsheet we will use greek indices (µ, ν, . . .).

• In many string theory texts, τ is used as a time coordinate for the string
worldsheet, and for the parameter of a particle worldline, but the use of τ will
be reserved here for proper time (and, much later on, for “Euclidean time”).

–6–
 We could include terms involving the extrinsic curvature K of the worldline
(inverse of the radius of curvature), which is essentially the D-acceleration, or yet
higher derivative terms, i.e.
Z √
I = −mc dt −ẋ2 1 + (`K)2 + . . . ,
 
(2.2)

where ` is a new length scale, which must be characteristic of some internal structure.
As long as K −1  ` this structure is invisible and we can neglect any extrinsic
curvature corrections. Or perhaps the particle is truly elementary, and ` = 0. In
either case, quantization should yield a Hilbert space carrying a unitary irrep of the
Poincaré group. For zero spin this means that the particle’s wavefunction Ψ should
satisfy the Klein-Gordon equation ( − m2 ) Ψ = 0. There are many ways to see that
this is true.

2.1 Gauge invariance

Because the worldline time t was an arbitrary parameter, we can change it to any
other parameter t∗ (t) (which should also be monotonic, so ṫ∗ 6= 0). Consider the
transformation
t → t∗ (t) x → x∗ (t∗ ) = x(t) . (2.3)
This definition of x∗ implies that it is a scalar function on the worldline. For an
infinitesimal transformation with

t∗ (t) = t − ξ(t) , (2.4)

where ξ(t) is an arbitrary (but infinitesimal) function, then2

x∗ (t − ξ) = x(t) ⇒ δξ x(t) ≡ x∗ (t) − x(t) = ξ ẋ . (2.5)

Let’s now compute the variation of the integrand induced by the variation δξ x of the
function x:
√ 1 d(δξ x) 1 ˙ 2 
δξ −ẋ2 = − √ ẋ · = −√ ξ ẋ + ξ ẋ · ẍ
−ẋ2 dt −ẋ2
√
√ d −ẋ2 d  √ 2
= ξ˙ −ẋ2 + ξ = ξ −ẋ . (2.6)
dt dt
This is the infinitesimal form of the transformation of a worldline scalar density; its
variation contributes only a time-boundary term to the variation of I. This time-
boundary term is cancelled by a shift of the integration limits. Let’s verify this.
s  s 
Z t∗B ∗ 2
Z tb −ξ(tB ) 2
dx∗

∗ dx ∗
I[x] → −mc dt − = −mc dt − . (2.7)
t∗A dt∗ tA −ξ(tA ) dt∗
2
Use ξ ẋ∗ = ξ ẋ + O(ξ 2 ).

–7–
Now we relabel the integration variable t∗ → t to get
Z tb −ξ(tB ) q Z tb −ξ(tB ) n√ √ o
∗ 2
I[x] → −mc dt − (ẋ ) = −mc dt −ẋ2 + δξ −ẋ2
tA −ξ(tA ) tA −ξ(tA )
Z tb −ξ(tB ) √ Z tB √
= −mc dt −ẋ2 − mc δξ −ẋ2 . (2.8)
tA −ξ(tA ) tA

We are keeping only the terms linear in ξ and discarding anything of higher order
(that’s what we mean by “infinitesimal”). Now we expand the first term to first
order in ξ, which gives us I plus a time-boundary term, and we use (2.6) to rewrite
the last integral as another time-boundary term:
 h √ iB  h √ iB
I[x] → I[x] + mc ξ −ẋ 2 − mc ξ −ẋ 2 = I[x] . (2.9)
A A

The time-boundary terms have cancelled.

Gauge invariance is not a symmetry. Instead it implies a redundancy in the
description. We can remove the redundancy by imposing a gauge-fixing condition.
For example, in standard Minkowski spacetime coordinates. (x0 , ~x) we may choose
the “temporal gauge”
x0 (t) = c t . (2.10)
Since δξ x0 = c ξ when x0 = c t, insisting on this gauge choice implies ξ = 0; i.e. no
gauge transformation is compatible with the gauge choice, so the gauge is fixed. In
this gauge
Z p Z  
2 2 2 2 1 2 2 2


I = −mc dt 1 − v /c = dt −mc + mv 1 + O v /c , (2.11)
2
where v = |~x˙ |. The potential energy is therefore the rest mass energy mc2 , which we
can subtract because it is constant. We can then take the c → ∞ limit to get the
non-relativistic particle action
Z
1
IN R = m dt |d~x/dt|2 . (2.12)
2
From now on we set c = 1.

2.2 Hamiltonian formulation

√
If we start from the gauge-invariant action with L = −m −ẋ2 , then
∂L mẋ
p= =√ ⇒ p2 + m 2 ≡ 0 . (2.13)
∂ ẋ −ẋ2
So not all components of p are independent, which means that we cannot solve for
ẋ in terms of p. Another problem is that
mẋ2 √
H = ẋ · p − L = √ + m −ẋ2 ≡ 0 , (2.14)
−ẋ2

–8–
so the canonical Hamiltonian is zero.
What do we do? Around 1950 Dirac developed methods to deal with such cases.
We call the mass-shell condition p2 + m2 = 0 a “primary” constraint because it is
a direct consequence of the definition of conjugate momenta. Sometimes there are
“secondary” constraints but we will never encounter them. According to Dirac we
should relax the constraint to allow all components of p to be independent but then
impose it with a Lagrange multiplier. Since the canonical Hamiltonian H is zero we
end up with the phase-space action
Z  
1 2 2


I[x, p; e] = dt ẋ · p − e p + m , (2.15)
2
where e(t) is the Lagrange multiplier.
We can easily check this result by eliminating the variables p and e:

• Use the p equation of motion p = e−1 ẋ to get the new action

Z
1
dt e−1 ẋ2 − em2 .

I[x; e] = (2.16)
2
At this point it looks as though we have ‘1-dim. scalar fields” coupled to 1-
dim. “gravity”, with “cosmological constant” m2 ; in this interpretation e is
the square root of the 1-dim. metric, i.e. the “einbein”.
√
• Now eliminate e from (2.16) using the e equation of motion3 me = −ẋ2 , to
R √
get the standard point particle action I = −m dt −ẋ2 .

Elimination lemma. When is it legitimate to solve an equation of motion and sub-

stitute the result back into the action to get a new action? Let the action I[ψ, φ]
depend on two sets of variables ψ and φ, such that the equation δI/δφ = 0 can be
solved algebraically for the variables φ as functions of the variables ψ, i.e. φ = φ(ψ).
In this case 
δI 
≡ 0. (2.17)
δφ φ=φ(ψ)
The remaining equations of motion for ψ are then equivalent to those obtained by
ˆ = I[ψ, φ(ψ)], i.e. that obtained by back-substitution.
variation of the new action I[ψ]
This follows from the chain rule and (2.17):

δ Iˆ
  
δI  δφ(ψ) δI  δI 
= + = . (2.18)
δψ δψ φ=φ(ψ) δψ δφ φ=φ(ψ) δψ φ=φ(ψ)

Moral: If you use some equations of motion to eliminate a set of variables {A} then
you can substitute the result into the action, to get a new action for the remaining
3
√
The equation is actually me = ± −ẋ2 , where the sign corresponds to the sign of the energy
(relativity always allows both signs). Here we are choosing the energy to be positive.

–9–
variables, provided that the equations used are those found by varying the original
action with respect to the set of variables {A}. However, if you solve for {A} by
using the equations of motion of {B} then back-substitution into the action is not
legitimate unless {B} = {A}, even though it is legitimate to make this substitution
into the equations of motion.

The action (2.15) is still gauge invariant. The gauge transformations are now4

d
δξ x = ξ ẋ , δξ p = ξ ṗ , δξ e = (eξ) , (2.19)
dt

which shows that x and p are worldline scalars but e is is scalar density (e ≡ det e in
1D). We shall call the above variations the Diff 1 variations because they are induced
by a diffeomorphism (1D general coordinate transformation, or reparametrization)
of the worldline.
The action (2.15) is also invariant under the much simpler gauge transformations

δα x = α(t)p , δα p = 0 , δα e = α̇ . (2.20)

Let’s call this the “canonical” gauge transformation (for reasons that will become
clear). In fact,
1
t
δα I = α p2 − m2 tB , (2.21)
2 A

which is zero if α(tA ) = α(tB ) = 0.

The Diff 1 and canonical gauge transformations are equivalent (at least if we
ignore time-boundary terms) because they differ by a “trivial” gauge transformation
(see Q.I.1):

• Trivial gauge invariances. Consider I[ψ, φ] again and transformations

δI δI
δf ψ = f , δf φ = −f , (2.22)
δφ δψ

for arbitrary function f (we assume that φ and ψ satisfy b.c.s for which the
functional derivatives are defined). This gives δf I = 0, so the action is gauge in-
variant. As the gauge transformations are zero “on-shell” (i.e. using equations
of motion) they have no physical effect. Any two sets of gauge transformations
that differ by a trivial gauge transformation have equivalent physical implica-
tions.
4
Recall that the variation δt = −ξ(t) produces a time-boundary term; we’ll ignore these terms
now, or assume that they are zero because of endpoint conditions on the gauge transformation
parameters.

– 10 –
If we fix the gauge invariance by choosing the temporal gauge x0 (t) = t then we have

ẋm pm = ~x˙ · p~ − p0 , (2.23)

so in this gauge the canonical Hamiltonian is

p
H = p0 = ± |~p|2 + m2 , (2.24)

where we have used the constraint to solve for p0 . The sign ambiguity is typical for
a relativistic particle.
The canonical Hamiltonian depends on the choice of gauge. Another possible
gauge choice is light-cone gauge. Choose phase-space coordinates
1
x± = √ x1 ± x0 , x = x2 , . . . , xD−1



2
1
p± = √ (p1 ± p0 ) , p = (p2 , . . . , pD−1 ) . (2.25)
2
Then

ẋm pm = ẋ+ p+ + ẋ− p− + ẋ · p , p2 ≡ η mn pm pn = 2p+ p− + |p|2 . (2.26)

The latter equation follows from the fact that the non-zero components of the
Minkowski metric in light-cone coordinates are

η +− = η −+ = 1 = η−+ = η+− , η IJ = ηIJ = δIJ , I, J = 1, . . . , D − 2. (2.27)

It also follows from this that

p+ = p− , p− = p+ . (2.28)

The light-cone gauge (which can be viewed as an infinite Lorentz boost of the tem-
poral gauge) is
x+ (t) = t . (2.29)
Since δα x+ = αp+ = αp− the gauge is fixed provided that p− 6= 0. In this gauge

ẋm pm = ẋ · p + ẋ− p− + p+ , (2.30)

so the canonical Hamiltonian is now

|p|2 + m2
H = −p+ = , (2.31)
2p−

where we have used the mass-shell constraint to solve for p+ .

– 11 –
 • Poisson brackets. For mechanical model with action5
Z
I[q, p] = dt q̇ I pI − H(q, p)
 
(2.32)

the Poisson bracket of any two functions (f, g) on phase space is

"←−− ←−− #
∂ ∂ ∂ ∂
{f, g}P B = f − g (2.33)
∂q I ∂pI ∂pI ∂q I

where the backwards arrow over a derivative indicates that it acts to the left6 .
In particular,
 I
q , pJ P B = δJI .

(2.34)

• More generally, we start from a symplectic manifold, a phase-space with coor-

dinates z A and a symplectic (closed, invertible) 2-form Ω = 21 ΩAB dz A ∧ dz B .
Locally, since dΩ = 0,

Ω = dω , ω = dz A fA (z) , (2.35)

and the action in local coordinates is

Z
I = dt ż A fA (z) − H(z) .
 
(2.36)

The PB of functions (f, g) is defined as

∂f ∂g
{f, g}P B = ΩAB , (2.37)
∂z A ∂z B
where ΩAB is the inverse of ΩAB . The PB is an antisymmetric bilinear product,
from its definition. Also, for any three functions (f, g, h),

dΩ = 0 ⇔ {{f, g}P B , h}P B + cyclic permutations ≡ 0 . (2.38)

In other words, the PB satisfies the Jacobi identity, and is therefore a Lie
bracket, as a consequence of the closure of the symplectic 2-form.

• Darboux theorem. This states that there exist local coordinates7 such that

Ω = dpI ∧ dq I ⇒ ω = pI dq I + d() . (2.39)

This leads to the action (2.32) and the definition (2.33) of the PB.
5
This action is not invariant under time reparametrizations unless H ≡ 0 but it can be converted
to one that is; see Q.I.2 for an illustration of this point.
6
This way of writing the PB has advantages when it is generalised to phase spaces with anti-
commuting coordinates.
7
“Local” means at a point and in some neighbourhood of that point.

– 12 –
 • Canonical transformations. Any function Q on phase-space is the generator of
an infinitesimal change of phase-space coordinates, which implies an infinites-
imal variation of any function f of these coordinates; for parameter , this
variation is
δ f = {f, Q}P B  . (2.40)
Suppose that we have Darboux coordinates (q I , pI ). Then
∂Q ∂Q
δ q I =  δ pI = − . (2.41)
∂pI ∂q I
Notice that
     
I

∂Q ∂Q I
δ dpI ∧ dq =  dpI ∧ d −d ∧ dq = 0. (2.42)
∂pI ∂q I
The last equality follows from the symmetry of mixed partial derivatives.
In other words, the transformation generated by Q preserves the form of the
symplectic 2-form, equivalently the Poisson bracket. Such transformations
are called symplectic diffeomorphisms (Maths) or canonical transformations
(Phys.)
In differential geometry a symplectic diffeomorphism is equivalent to a choice
of vector field ξ such that Lξ Ω = 0, where Lξ denotes the Lie derivative with
respect to ξ. This condition (combined with dΩ = 0) implies that ξ is a
Hamiltonian vector field, which means that it takes the form ξ = ΩAB ∂A Q∂B for
some function Q, which is the function generating the corresponding canonical
transformation.

2.2.1 Gauge invariance and first-class constraints

Consider the action
Z
I = dt q̇ I pI − λi ϕi (q, p) ,

I = 1, . . . , N ; i = 1, . . . , n < N . (2.43)

The Lagrange multipliers λi impose the phase-space constraints ϕi = 0. Let us

suppose that
{ϕi , ϕj }P B = fij k ϕk (2.44)
for some phase-space structure functions fij k = −fji k . In this case we say that the
constraints are “first-class”. Otherwise the constraints are either “second-class” or a
mixture of first and second class, but here we consider only the simple case in which
all constraints are first-class8 . The special feature of first-class constraints is that
they generate gauge invariances.
8
Strictly speaking, the adjective “first class” applies to sets of constraints, not individual con-
straints, unless there is only one.

– 13 –
Lemma. The infinitesimal transformations of (2.41), but with constant parameter
 promoted to a function (t), are such that9
d
δ q̇ I pI = Q


˙ + () . (2.45)
dt
Here’s the proof:
     
d ∂Q I ∂Q d ∂Q ∂Q I ∂Q
LHS =  pI − q̇ =  pI −  ṗI + q̇
dt ∂pI ∂q I dt ∂pI ∂pI ∂q I
    
d ∂Q d ∂Q
= pI − Q̇ =  pI − Q + Q ˙ . (2.46)
dt ∂pI dt ∂pI
This becomes a total time derivative for ˙ = 0 because the transformation is then
canonical.

Applying this for Q = i ϕi we get

d
δ q̇ I pI = ˙i ϕi + () ,


(2.47)
dt
and we also have

δ λi ϕi = δ λi ϕi + λi j {ϕi , ϕj }P B = δ λk + λi j fij k ϕk ,



(2.48)

where we use (2.44) in the second equality. Putting these result together, we have
Z  
k k i j k

d
δ I = dt ˙ − δ λ − λ  fij ϕk + () . (2.49)
dt
As the Lagrange multipliers are not functions of canonical variables, their transfor-
mations can be chosen independently. If we choose

δ λk = ˙k + i λj fij k , (2.50)

then δ I is a surface term, which is zero if we impose the b.c.s i (tA ) = i (tB ) = 0.
The point particle is a very simple (abelian) example. The one constraint is
1 2
p + m2 ,


ϕ= (2.51)
2
and it is trivially first-class. It generates the canonical gauge transformations:
1  2
α x, p + m2 P B = αp ,

δα x =
2
1 
δα p = α p, p2 + m2 P B = 0 ,

(2.52)
2
9
There is also a ∂Q/∂t term if Q has an explicit t-dependence, but this possibility is not relevant
when t is just the integration parameter of a time-reparametrization invariant action.

– 14 –
and if we apply the formula (2.50) to get the gauge transformation of the einbein,
we find that δα e = α̇.
The general model (2.43) also includes the string, as we shall see later. This is
still a rather simple case because the structure functions are constants, which means
that the constraint functions ϕi span a (non-abelian) Lie algebra. In such cases
the transformation (2.50) is a Yang-Mills gauge transformation for a 1D YM gauge
potential.

2.2.2 Gauge fixing

We can fix the gauge generated by a set of n first-class constraints by imposing n
gauge-fixing conditions
χi (q, p) = 0 i = 1, . . . , n . (2.53)
The gauge transformation of these gauge-fixing functions is

δ χi = χi , ϕj P B j ,

(2.54)

so if we want δ χi = 0 to imply j = 0 for all j (which is exactly what we do want in

order to fix the gauge completely) then we must choose the functions χi such that

det χi , ϕj P B 6= 0 .

(2.55)

This is a useful test for any proposed gauge fixing condition.

In addition to requiring that the gauge-fixing conditions χi = 0 actually do fix
the gauge, it should also be possible to make a gauge transformation to ensure that
χi = 0 if this is not already the case. In particular, if χi = f i for infinitesimal
functions f i , and χ̂i = χi + δ χi , then χ̂i = f i + δ χi and we should be able to find
parameters i such that χ̂i = 0. This requires us to solve the equation

χi , ϕj j = −f i

PB
(2.56)

for i , but a solution exists for arbitrary f i iff the matrix {χi , ϕj }P B has non-zero
determinant.

Corollary. Whenever {χi , ϕj }P B has zero determinant, two problems arise. One
is that the gauge fixing conditions don’t completely fix the gauge, and the other
is that you can’t always arrange for the gauge fixing conditions to be satisfied by
making a gauge transformation. This is a very general point. Consider the Lorenz
gauge ∂ · A = 0 in electrodynamics (yes, that’s Ludwig Lorenz, not Henrik Lorentz
of the Lorentz transformation). A gauge transformation A → A + dα of the gauge
condition gives α = 0, which does not imply that α = 0; the gauge has not been
fixed completely. It is also true, and for the same reason, that you can’t always
make a gauge transformation to get to the Lorenz gauge if ∂ · A is not zero, even

– 15 –
if it is arbitrarily close to zero: the reason is that the operator  is not invertible
because there are non-zero solutions of the wave equation that cannot be eliminated
by imposing the b.c.s permissible for hyperbolic partial differential operators. The
Coulomb gauge ∇ · A = 0 does not have this problem because ∇2 is invertible for
appropriate b.c.s (but it breaks manifest Lorentz invariance).
A similar problem arises if we try to fix the gauge invariance of the action (2.43)
by imposing conditions on the Lagrange multipliers. More on this later.

2.2.3 Continuous symmetries and Noether’s theorem

In addition to its gauge invariance, the point particle action is invariant (in a
Minkowski background) under the Poincaré transformations, i.e, spacetime trans-
lations xm → xm + am and Lorentz “rotations”

xm → Λ m n xn , ΛT ηΛ = η ,
n
pm → pn Λ−1 m = Λm n pn .


(2.57)

Because these are continuous transformations, we may consider transformations close

to the identity. For Lorentz transformations we write

Λm n = δnm + ω m n + O(ω 2 ) , ηmp ω p n ≡ ωmn = −ωnm , (2.58)

where ω m n is assumed “small”. Similarly, we may suppose am to be “small”. The

resulting “infinitesimal” transformations are

xm → xm + δxm , δxm = am + ω m n xn ,
pm → pm + δpm , δpm = ωm n pn . (2.59)

For constant parameters, this infinitesimal transformation leaves the point particle
action unchanged.
Noether’s theorem says that any continuous symmetry is associated with a cor-
responding constant of the motion, i.e. a “conserved charge” There is an easy way
to find these Noether charges. Let I[φ] be an action functional invariant under an
infinitesimal transformation δ φ for constant parameter . Then, when  is promoted
to an arbitrary function of t, it must be possible to write the variation of I in the
form Z
δ I = dt ˙ Q . (2.60)

The phase space function Q is just what the computation of δ I produces10 .

10
If any terms with higher derivatives of  appear at an intermediate point in this computation
then we just integrate by parts; any resulting boundary terms must cancel because δ I = 0 when
˙ = 0, by hypothesis.

– 16 –
 • Noether’s Theorem. Q is a constant of motion. Proof: choose (t) to be
zero at the endpoints of integration. In this case, integration by parts gives us
Z
δ I = − dt  Q̇ . (2.61)

The LHS must vanish if the equations of motion hold because these equations
are found by extremizing the action for arbitrary variations of the functions on
which I depends. But the RHS is zero for all possible (t) iff Q̇(t) = 0 for all t
within the integration limits (which we can choose arbitrarily). It follows that
the equations of motion imply Q̇ = 0, i.e. that Q is a constant of motion.

This proves Noether’s theorem11 : a continuous symmetry implies a conserved charge

(constant of motion). The symmetry has to be continuous for us to be able to
consider its infinitesimal form.
Also, given Q we can recover the symmetry transformation from the formula

δ f = {f, Q}P B (2.62)

for any phase-space function f . There may be conserved charges for which the RHS
of this formula is zero. These are “topological charges”, which do not generate
symmetries; they are not Noether charges.

To apply this proof of Noether’s theorem to Poincaré invariance of the point

particle action, we allow the parameters am and ω m n of (2.59) to be time-dependent.
A calculation then shows that
Z  
m 1 n m
δI = dt ȧ Pm + ω̇ m J n , (2.63)
2
where
Pm = pm , J m n = xm p n − xn p m , (2.64)
which are therefore the Poincaré charges. Notice that they are gauge-invariant; this
is obvious for Pm , and for J m n we have

δα J m n = α (pm pn − pn pm ) = 0 . (2.65)

The gauge-invariance of Noether charges could have been anticipated from the fact
that gauge-fixing cannot break symmetries., which we will now prove for gauge fixing
conditions of the type just considered.

Gauge-fixing and symmetries. If we have fixed a gauge invariance by imposing gauge-

fixing conditions χi = 0, then what happens if our gauge choice does not respect a
symmetry with Noether charge Q, i.e. what happens if {Q, χi }P B is non-zero.
11
See arXiv:1605.07128 for details, and a history of this proof

– 17 –
 The answer is that the symmetry is not broken. The reason is that there is an
intrinsic ambiguity in the symmetry transformation generated by Q whenever there
are gauge invariances. We may take the symmetry transformation to be

δ f = {f, Q}P B  + {f, ϕj }P B αj () . (2.66)

That is, a symmetry transformation with parameter  combined with a gauge trans-
formtion for which the parameters αi are fixed, in a way to be determined, in terms
of . Because gauge transformations have no physical effect, such a transformation
is as good as the one generated by Q alone. The parameters αi () are determined by
requiring that the modified symmetry transformation respect the gauge conditions
χi = 0, i.e.
0 = χi , Q P B  + χi , ϕj P B αj () .
 
(2.67)
As long as {χi , ϕj }P B has non-zero determinant, we can solve this equation for all
αi in terms of .

Moral: gauge-fixing never breaks symmetries, because it just removes redundancies

(see Q.I.3 for an example). If a symmetry is broken by some gauge choice then there
is something wrong with the gauge choice!

2.2.4 Canonical Quantization

We will use Dirac’s quantization prescription
i I
q I , pJ
 
PB
→− q̂ , p̂J , (2.68)
~
which gives us the canonical commutation relations for the operators q̂ I and p̂I that
replace the classical phase-space coordinates:
 I
q̂ , p̂J = i~ δJI .

(2.69)

Let’s apply this to the point particle in temporal gauge. In this case the canonical
commutation relations are [x̂i , p̂j ] = i~ δji for i, j, = 1, . . . , D − 1. We can realise these
relations on eigenfunctions Ψ(t, ~x) of ~x ˆ by setting pˆ~ = −i~∇. ~ The Schroedinger
equation is
∂Ψ √
ĤΨ = i~ , Ĥ = ± −~2 ∇2 + m2 . (2.70)
∂t
Acting on this with Ĥ we get Ĥ 2 = i~∂t Ψ = −∂t2 Ψ, and hence
 2
−∂t + ∇2 − (m/~)2 Ψ(t, ~x) = 0 .

(2.71)

This is the Klein-Gordon equation for a scalar field Ψ with mass parameter m/~ (the
mass parameter of the field equation is the particle mass divided by ~). The final
result is Lorentz invariant even though this was not evident at each step.

– 18 –
2.2.5 Quantization for systems with first-class constraints
An alternative procedure that maintains Lorentz covariance at each step is provided
by Dirac’s method for quantization of systems with first-class constraints. We’ll use
the point particle to illustrate the idea.

• Step 1. We start from the manifestly Lorentz invariant phase space action and
quantise canonically as if there were no constraint. This gives us the canonical
commutation relations
[x̂m , p̂n ] = i~δnm . (2.72)
We can realise this on eigenfunctions Ψ(x) of x̂m by setting p̂m = −i~∂m .

• Step 2. Because of the gauge invariance there are unphysical states in the
Hilbert space. We need to remove these with a constraint. The mass-shell
constraint encodes the full dynamics of the particle, so we now impose this in
the quantum theory as the physical state condition

p̂2 + m2 |Ψi = 0 .


(2.73)

This is equivalent to the Klein-Gordon equation

D − (m/~)2 Ψ(x) = 0 ,
 
Ψ(x) = hx|Ψi . (2.74)

where D = η mn ∂m ∂n is the wave operator in D-dimensions.

More generally, for the general model with first-class constraints, we impose the
physical state conditions

ϕ̂i |Ψi = 0 , i = 1, . . . , n . (2.75)

The consistency of these conditions requires that

[ϕ̂i , ϕ̂j ] |Ψi = 0 ∀ i, j . (2.76)

This would be guaranteed if we could apply the PB-to-commutator prescription of

(2.68) to arbitrary phase-space functions, because this would give

[ϕ̂i , ϕ̂j ] = i~fij k ϕ̂k , (?) (2.77)

and the RHS annihilates physical states. However, because of operator ordering
ambiguities there is no guarantee that (2.77) will be true when the functions ϕi are
non-linear. We can use some of the ambiguity to redefine what we mean by ϕ̂i ,
but this may not be sufficient. There could be a quantum anomaly. The string will
provide an example of this.
From now on we set ~ = 1.

– 19 –
3. The Nambu-Goto string
The string analog of a particle’s worldline is its “worldsheet”: the 2-dimensional
surface in spacetime that the string sweeps out in the course of its time evolution.
Strings can be open, with two ends, or closed, with no ends. We shall start by
considering a closed string. This means that the parameter σ specifying position on
the string is subject to a periodic identification. The choice of period has no physical
significance12 we will choose it to be 2π; i.e. (∼ means “is identified with”)

σ ∼ σ + 2π . (3.1)

The worldsheet of a closed string is topologically a cylinder, parametrised by σ and

some arbitrary time parameter t. We can consider these together as σ µ (µ = 0, 1),
i.e.
σ µ = (t, σ). (3.2)
The map from the worldsheet to Minkowski space-time is specified by worldsheet
fields X m (t, σ). Using this map we can pull back the Minkowski metric on space-
time to the worldsheet to get the induced worldsheet metric

gµν = ∂µ X m ∂ν X n ηmn . (3.3)

The natural string analog of the point particle action proportional to the proper
length of the worldline (i.e. the elapsed proper time) is the Nambu-Goto action,
which is proportional to the area of the worldsheet in the induced metric, i.e.
Z I p
IN G [X] = −T dt dσ − det g , (3.4)

where the constant T is the string tension, and

X(t, σ + 2π) = X(t, σ) . (3.5)

We are still free to change the coordinates locally, and the action doesn’t change if
we suppose that the X m are worldsheet scalars. In this case, a change of coordinates
σ µ → σ µ − ζ µ for infinitesimal worldsheet vector field ξ induces the Diff2 gauge
transformation
δξ X m = ξ µ ∂ µ X m . (3.6)
This yields p   p 
µ
δξ − det g = ∂µ ξ − det g , (3.7)

12
It would fix units for length if σ were supposed to have dimensions of length. For this reason
we assume here, as we may, that the parameters σ and t are both dimensionless; in this case, and
for c = 1, we have [X] = L and [P ] = M .

– 20 –
confirming that the variation of the action is at most a boundary term (with potential
contributions only at initial and final times since there is no space boundary).
Varying IN G with respect to X we get the NG equation of motion13
p 
∂µ − det g g µν ∂ν X = 0 . (3.8)

This is just the massless wave equation for a set of 2D “scalar fields” X m but the
metric on the “2D spacetime” (the worldsheet) depends on the scalar fields, so it
looks hopelessly non-linear!

3.1 Hamiltonian formulation

Denoting derivatives with respect to t by an overdot and derivatives with respect to
σ by a prime, we have
Ẋ 2 Ẋ · X 0
 
gµν = . (3.9)
Ẋ · X 0 X 02
This shows that the NG action can be written as
Z I q
IN G = dt LN G , LN G = −T dσ (Ẋ · X 0 )2 − Ẋ 2 X 02 . (3.10)

The worldsheet momentum density Pm (t, σ) canonically conjugate to the wordsheet

fields X m is therefore
δL T h  i
Pm = =√ Ẋm X 02 − Xm
0
Ẋ · X 0 . (3.11)
δ Ẋ m − det g
This implies the following identities [Exercise: check this]
2
P 2 + (T X 0 ) ≡ 0 , X 0m Pm ≡ 0 . (3.12)

In addition, the canonical Hamiltonian is

I
H = dσ Ẋ m Pm − L ≡ 0 . (3.13)

As for the particle, we should take the Hamiltonian to be a sum of Lagrange

multipliers times the constraints, so we should expect the phase-space form of the
action to be
Z I  
m 1 h 2 0 2
i
0m
I[X, P ; e, u] = dt dσ Ẋ Pm − e P + (T X ) − u X Pm , (3.14)
2
where e(t, σ) and u(t, σ) are Lagrange multipliers14 . To check this, we eliminate P
by using its equation of motion:

P = e−1 Dt X , Dt X ≡ Ẋ − uX 0 . (3.15)
13
Check, as exercise, or wait until we derive it later in a simpler way.
14
Analogous to the “lapse” and “shift” functions in the Hamiltonian formulation of GR.

– 21 –
We are assuming here that e is nowhere zero (as for the particle). Back substitution
takes us to the action
Z I
1 n
2
o
I[X; e, u] = dt dσ e−1 (Dt X)2 − e (T X 0 ) . (3.16)
2
Varying u in this new action we find that
Ẋ · X 0 det g
u= ⇒ (Dt X)2 = . (3.17)
X 02 (X 0 )2
Here we assume that X 02 is non-zero (but we pass over this point). Eliminating u
we arrive at the action
Z I  
1 −1 det g 0 2
I[X, e] = dt dσ e − e (T X ) . (3.18)
2 X 02
Varying this action with respect to e we find that
√
− det g
e= , (3.19)
T (X 0 )2
and back-substitution returns us to the Nambu-Goto action in its original form.

3.1.1 Alternative form of phase-space action

Notice that the phase-space constraints are equivalent to
1 2
H± = 0 , H± ≡ (P ± T X 0 ) , (3.20)
4T
so we may rewrite the action as
Z I n o
−
m +
I = dt dσ Ẋ Pm − λ H− − λ H+ , λ± = T e ± u . (3.21)

Notice that the Lagrange multipliers λ± are dimensionless if the worldsheet coordi-
nates are taken to be dimensionless.

3.1.2 Gauge invariances

From the Hamiltonian form of the NG string action (3.14) we read off the canonical
Poisson bracket relations15

{X m (σ), Pn (σ 0 )}P B = δnm δ(σ − σ 0 ) , (3.22)

where, for any two functionals F [X, P ] and G[X, P ],

I "←−−−−− ←−−−−− #
δ δ δ δ
{F, G}P B = dσ F − G.
δX m (σ) δPm (σ) δPm (σ) δX m (σ)
15
By means of the Fourier series representations for X and P to be given later, one can verify
that this is equivalent to the definition (2.33).

– 22 –
This implies, in particular, that {(X m )0 (σ), Pn (σ 0 )}P B = δnm δ 0 (σ − σ 0 ).
Using (3.22), one may now compute the PBs of the constraint functions. One
finds that

{H+ (σ), H+ (σ 0 )}P B = [H+ (σ) + H+ (σ 0 )] δ 0 (σ − σ 0 ) ,

{H− (σ), H− (σ 0 )}P B = − [H− (σ) + H− (σ 0 )] δ 0 (σ − σ 0 ) ,
{H+ (σ), H− (σ 0 )}P B = 0 . (3.23)

This shows that

• The constraints are “first-class”, with constant structure functions, which are
therefore the structure constants of a Lie algebra

• This Lie algebra is a direct sum of two isomorphic algebras (−H− obeys the
same algebra as H+ ). In fact, it is the algebra

Diff 1 ⊕ Diff 1 . (3.24)

We will verify this later. Notice that this is a proper subalgebra of Diff 2 .
Only the Diff 1 ⊕ Diff 1 subalgebra has physical significance because all other
gauge transformations of Diff 2 are “trivial” in the sense explained earlier for
the particle.

The gauge transformation of any function F on phase space is

 I 
− +


δξ F = F, dσ ξ H− + ξ H+ , (3.25)
PB

where the parameters ξ ± (t, σ) are arbitrary functions. This gives

1 − 1 +
δX = ξ (P − T X 0 ) + ξ (P + T X 0 ) ,
2T 2T
1h i0 1  0
δP = − ξ − (P − T X 0 ) + ξ + (P + T X 0 ) . (3.26)
2 2
Notice that
δξ− (P + T X 0 ) = 0 , δξ+ (P − T X 0 ) = 0 , (3.27)

and hence δξ∓ H± = 0, as expected from the fact that the algebra is a direct sum
(H+ has zero PB with H− ).
To verify invariance of the action one needs
  d
δξ Ẋ m Pm = ξ˙+ H+ + ξ˙− H− + () , (3.28)
dt

– 23 –
and, using (3.23),
I
δξ± H± (σ) = dσ 0 ξ ± (σ 0 ) {H± (σ), H± (σ 0 )}P B
I
= ± dσ 0 ξ ± (σ 0 ) [H± (σ) + H± (σ 0 )] δ 0 (σ − σ 0 )
0 0
= ± ξ ± (σ) H± (σ) ± ξ ± (σ)H± (σ) .
 
(3.29)

Using these results, one finds that the action is invariant provided that

0
0
0
0
δλ− = ξ˙− + λ− ξ − − ξ − λ− , δλ+ = ξ˙+ − λ+ ξ + + ξ + λ+ . (3.30)

We see that λ± is a gauge potential for the ξ ± -transformation, with each being inert
under the gauge transformation associated with the other, as expected from the direct
sum structure of the gauge algebra. The sign differences in these two transformations
are a consequence of the fact that H+ and −H− have the same PB algebra.

3.1.3 Symmetries of NG action

The closed NG action has manifest Poincaré invariance, with Noether charges
I I
Pm = dσPm , Jmn = 2 dσX[m Pn] . (3.31)

These are constants of the motion. [Exercise: verify that the NG equations of motion
imply that Ṗm = 0 and J˙mn = 0.]

N.B. We use the following notation

1 1
T[mn] = (Tmn − Tnm ) , T(mn) = (Tmn + Tnm ) . (3.32)
2 2
In other words, we use square brackets for antisymmetrisation and round brackets for
symmetrisation, in both cases with “unit strength” (which means, for tensors of any
rank, that A[m1 ...mn ] = Am1 ...mn if A is totally antisymmetric, and S(m1 ...mn ) = Sm1 ...mn
if S is totally symmetric).

The closed NG string is also invariant under worldsheet parity: σ → −σ (mod

2π). The worldsheet fields (X, P ) are parity even, which means that X 0 is parity odd
and hence (P + T X 0 ) and (P − T X 0 ) are exchanged by parity. This implies that H±
are exchanged by parity.

3.2 Open string boundary conditions

An open string has two ends. We shall choose the parameter length to be π, so the
action in Hamiltonian form is
Z Z π  
m 1  2 0 2
 0
I = dt dσ Ẋ Pm − e P + (T X ) − uX · P . (3.33)
0 2

– 24 –
What are the possible boundary conditions at the ends of the string?

Principle: the action should be stationary when the equations of motion are sat-
isfied. In other words, when we vary the action to get the equations of motion,
the boundary terms arising from integration by parts must be zero; otherwise the
functional derivative of the action is not defined.

Applying this principle to the above action, we see that boundary terms can arise
only when we vary X 0 and integrate by parts to get the derivative (with respect to σ)
off the δX variation (we can ignore any boundary terms in time). These boundary
terms are [Exercise: check this].
Z
σ=π
δI|on−shell = − dt T 2 eX 0 + uP · δX σ=0 ,


(3.34)

where, “on-shell” is shorthand for “using the equations of motion”.

It would make no physical sense to fix X 0 at the endpoints, and if X 0 is free
then so is Ẋ 0 and hence P 0 when we use the equations of motion, so the boundary
term with the factor of δX 0 will be zero only if we impose the conditions

0 
u|ends = 0 , X0  = 0. (3.35)
ends

Given that e 6= 0 we conclude that

 
~ 0 ~
X · δX 

= 0. (3.36)
ends

~
What this means is that at each end we may choose cartesian space coordinates X
such that for any given component, call it X∗ , we have

either X∗0 = 0 (Neumann b.c.s) ,

or δX∗ = 0 ⇒ X∗ = X̄∗ (constant) (Dirichlet b.c.s) (3.37)

There are many possibilities. The only one that does not break Lorentz invari-
ance is free-end boundary conditions

X 0 |ends = 0 . (3.38)

This implies that (X 0 )2 is zero at the ends of the string. The open string mass-shell
constraint then implies that P 2 is zero at the endpoints, and since
  
P |ends = e−1 Ẋ − uX 0  = e−1 Ẋ  , (3.39)
 
ends ends

we deduce that Ẋ 2 is zero at the ends of the string; i.e. the string endpoints move
at the speed of light.

– 25 –
3.3 Monge gauge
A natural analogue of the temporal gauge for the particle is a gauge in which we set
not only X 0 = t, to fix the time-reparametrization invariance, but also X 1 = σ, to
fix the reparametrization invariance of the string16 . This is often called the “static
gauge” but this is not a good name because there is no restriction to static configu-
rations. A better name (standard in Mathematical Biology) is “Monge gauge”, after
the French geometer who used it in the study of surfaces. So, the Monge gauge for
the NG string is17
X 0 (t, σ) = t X 1 (t, σ) = σ . (3.40)
In this gauge the action (3.14) becomes
Z I n
I = dt dσ Ẋ I PI + P0 − u P1 + X 0I PI


1  2 2 2 2 0 2


− e −P0 + P1 + |P| + T 1 + |X | , (3.41)
2

where I = 2, . . . , D − 2, and X is the (D − 2)-vector with components X I (and

similarly for P). We may solve the constraints for P1 , and P02 . Choosing the sign of
P0 corresponding to positive energy, we arrive at the action
Z I n p o
I = dt dσ Ẋ I PI − T 1 + |X0 |2 + T −2 [|P|2 + (X0 · P)2 ] . (3.42)

The expression for the Hamiltonian in Monge gauge simplifies if P is momentarily

18
zero ; we then have
I p
H = T dσ 1 + |X0 |2 (P = 0) . (3.43)

The integral equals the proper length L of the string. To see this, we observe that
the induced worldsheet metric in Monge gauge is
 2
0

2 2 2
ds ind = −dt + dσ + Ẋdt + X dσ 
 
   
2
= − 1 − |Ẋ|2 dt2 + 2Ẋ · X0 dσdt + 1 + |X0 | dσ 2 , (3.44)

and hence I p I p
L= ds |ind (t = const.) = dσ 1 + |X0 |2 .
2 (3.45)

16
We could choose any linear combination of the space components of X to equal σ but locally
we can always orient the axes such that this combination equals X 1 .
17
This fixes the units of length if (t, σ) are dimensionless; for this reason, it is convenient to
assume here that they have dimensions of length and to leave unspecified the periodicity in σ.
18
We assume here a moment at which the entire string is simultaneously at rest; such a moment
exists only for very special string configurations, but these suffice for the argument.

– 26 –
 Also, when P = 0 the equations of motion in Monge gauge imply that Ẋ = 0,
so the string is momentarily at rest. The energy of such a string is H = T L, and
hence the (potential) energy per unit length, or energy density, of the string is

E =T, (3.46)

as expected for an ultra-relativistic string.

An ultra-relativistic string cannot support tangential momentum. Given X 0 = t,
the constraint X 0 · P = 0 becomes

~ 0 · P~ = 0 ,
X (3.47)

which tells us that the (space) momentum density at any point on the string is
orthogonal to the tangent to the string at that point; the momentum density has no
tangential component. This has various consequences. One is that there can be no
longitudinal waves on the string (i.e. sound waves). Only transverse fluctuations are
physical.
Another consequence is that a plane circular loop of NG string cannot be sup-
ported against collapse by rotation in the plane (which can be done if T < E).
This does not mean that a plane circular loop of string cannot be supported against
collapse by rotation in other planes; we’ll see an example later.

3.3.1 Polyakov action

Recall that elimination of P , using P = e−1 Dt X, gives the action
Z
1 n
2
o  
I= d2 σ e−1 (Dt X)2 − e (T X 0 ) Dt = Ẋ − uX 0 (3.48)
2
This is quadratic in derivatives of X. In fact,
Z  
T 2 µν µν 1 −1 u
I=− d σ γ̃ ∂µ X · ∂ν X , γ̃ = . (3.49)
2 Te u T 2 e2 − u2

Notice that det γ̃ ≡ −1, which allows us to write

p
γ̃ µν = − det γ γ µν , (3.50)

where γ µν is the inverse of an independent and unconstrained worldsheet metric. This

gives us the “Polyakov” action for the NG string19
Z
T p
Ipoly [X; γ] = − d2 σ − det γ γ µν ∂µ X · ∂ν X , (3.51)
2
19
The Polyakov action was actually introduced by Brink, DiVecchia and Howe, and by Deser and
Zumino. Polyakov used it in the context of a path-integral quantization of the NG string, which
we will consider later.

– 27 –
The construction is such that only the conformal class of γµν has been defined. In
fact,
−λ+ λ− 12 [λ+ − λ− ]
 2
u − T 2 e2 u
  
2 2
γµν = Ω =Ω (3.52)
u 1 1
2
[λ+ − λ− ] 1
for some arbitrary conformal factor Ω. Equivalently,

ds2 (γ) = Ω2 dσ + λ+ dt dσ − λ− dt .



(3.53)

The conformal factor Ω cancels from the Polyakov action because this action is
“Weyl invariant”; i.e. invariant under an arbitrary local rescaling of the metric. This
is easily seen since
p p
γ µν → Ω−2 γ µν , − det γ → Ω2 − det γ , (3.54)

so the factors of Ω cancel from γ̃ µν .

We have found the Polyakov action by elimination of P from the phase-space
action. Recall that we established equivalence with the NG action by subsequent
elimination of the Lagrange multipliers. This step can now be done more easily by
eliminating the Polyakov metric directly. Variation with respect to γ µν gives us the
equation  
−2 −2 1 µν
gµν = Ω γµν Ω = γ gµν , (3.55)
2
and hence p
γµν = Ω2 gµν ⇒ γ̃ µν = − det g g µν . (3.56)
In words, the equation of motion for γµν sets it equal to the induced metric gµν up
to a conformal factor, which cancels from γ̃ µν . Back substitution now gives us
Z Z
T p p
Ipoly → − d σ − det g g gµν = −T d2 σ − det g ,
2 µν
(3.57)
2
which is the NG action.
The Polyakov action also simplifies the derivation of the NG equations of motion.
Varying it with respect to X, we find the equation of motion

∂µ (γ̃ µν ∂ν X m ) = 0 . (3.58)

On using (3.56), we recover the NG equation of motion (3.8).

3.4 Conformal gauge

On a 2D manifold any metric is conformally flat, which means that we may choose
local coordinates20 for which the metric is a conformal factor times a standard flat
metric. For a Lorentzian metric on the worldsheet of a string, this means that local
20
Recall that “local” means at a point and in some neighbourhood of that point

– 28 –
coordinates exist for which the Polyakov metric γµν is some conformal factor Ω2 times
the standard 2D Minkowski metric ηµν = diag.(−1, 1), i.e.

γµν = Ω2 ηµν ⇔ ds2 (γ) = Ω2 (dσ + dt)(dσ − dt) . (3.59)

From (3.53) we see that this is equivalent to imposing the following gauge-fixing
conditions on the Lagrange multipliers:
1
λ+ = λ− = 1 ⇔ e= & u = 0. (3.60)
T
This is the conformal gauge.
A feature of the conformal gauge choice is that it does not completely fix the
gauge; there is a residual gauge invariance. One way to see this is to use the conformal
gauge in the expressions of (3.30) for the gauge transformations of λ± . This gives us

0
δλ− = ξ˙− + ξ − = 2 ∂+ ξ −

0
δλ+ = ξ˙+ − ξ + = 2∂− ξ + , (3.61)

where ∂± are partial derivatives with respect to the light-cone worldsheet coordi-
nates21
1
σ± = t ± σ ⇔ ∂± = (∂t ± ∂σ ) . (3.62)
2
We see that δλ± = 0 if

∂± ξ ∓ = 0 ξ ± = ξ ± (σ ± ) .
 
⇒ (3.63)

In other words the conformal gauge conditions are preserved by those gauge trans-
formations for which ξ + is a function only of σ + and ξ − is a function only of σ −
Using P = T Ẋ, the residual gauge transformation of X can be read off from
(3.26):
δξ X m = ξ − ∂− X m + ξ + ∂+ X m . (3.64)
By interpreting ξ ± as the lightcone components of a wordsheet vector field ξ = ξ µ ∂µ ,
we can rewrite this residual transformation as

δξ X m = ξ µ ∂ µ X m . (3.65)

Using this, we can compute the residual transformation of the induced metric:

δξ gµν = 2∂(µ X · ∂ν) ξ λ ∂λ X

= ξ λ ∂λ gµν + 2∂(µ ξ λ gν)λ ≡ [Lξ g]µν . (3.66)

21
Notice that we are defining worldsheet lightcone coordinates with different conventions to those
used for spacetime lightcone coordinates.

– 29 –
This is the Lie derivative of the induced metric with respect to the vector field
ξ, otherwise known as the infinitesimal Diff2 transformation generated by ξ. The
transformation of X is also its Lie derivative with respect to ξ (in its action on a
worldsheet scalar).
However, the induced metric in conformal gauge is conformal to the 2D Minkowski
metric. This is because it is conformal to the Polyakov metric (by the Polyakov met-
ric equations, which are also the Hamiltonian constraints) and the Polyakov metric
is conformal to the 2D Minkowski metric (by definition of conformal gauge). In other
words, gµν = Ω2 ηµν in conformal gauge and we can choose to set Ω2 = 1 as long as
we allow this to change under a residual gauge transformation. This means that the
worldsheet vector field ξ must satisfy
[Lξ η]µν = χ ηµν , (3.67)
where the function χ is a small change in the conformal factor caused by the residual
gauge transformation. This is the conformal Killing equation, in this case for the
standard 2D Minkowski metric. As this metric is independent of the worldsheet
coordinates, the conformal Killing equations simplifies to
∂µ ξν + ∂ν ξµ = χ ηµν . (3.68)
By choosing worldsheet lightcone coordinates, and using the fact that η++ = η−− = 0,
one finds that the lightcone components ξ ± of the worksheet vector field ξ = ξ µ ∂µ
are restricted precisely by (3.63).
This provides us with an interpretation for the parameters ξ ± (σ ± ) of a residual
gauge transformation. They are solutions of the conformal Killing equation for 2D
Minkowski spacetime, and hence associated with infinitesimal conformal transfor-
mations of coordinates for this spacetime. It is a special feature of 2D Minkowski
spacetime that there are an infinite number of conformal Killing vector fields. For
example, in 4D, there are precisely 15 independent ones, and their commutator alge-
bra is the algebra of SO(4, 2). More generally one finds that they span the algebra
SO(D, 2). For D = 2 this is
SO(2, 2) = Sl(2; R) ⊕ Sl(2; R) , (3.69)
but each Sl(2; R) factor is actually a subalgebra of an infinite-dimensional Diff1 alge-
bra. So the algebra of residual gauge transformations of the NG string in conformal
gauge is
Diff 1 ⊕ Diff 1 . (3.70)
This is exactly the same as the algebra of canonical gauge transformations prior to
gauge fixing. All that the gauge fixing has done is to restrict the parameters22 .
22
In this respect, the conformal gauge is analogous to the Lorenz gauge ∂ · A in electrodynamics;
the gauge condition is Lorentz invariant but there is a residual U (1) gauge invariance A → A + dα
with the parameter restricted by α = 0.

– 30 –
3.5 Solving the NG equations in conformal gauge
In conformal gauge, the Polyakov action simplifies to
Z
T
I=− d2 σ η µν ∂µ X · ∂ν X . (3.71)
2
The field equation for X is now linear:

2 = −∂t2 + ∂σ2 .


2 X = 0 (3.72)

Equivalently,
∂+ ∂− X = 0 ⇒ X = XL (σ + ) + XR (σ − ) . (3.73)
In other words, the general solution of the NG equation of motion in conformal gauge
is a linear superposition of a left-moving wave profile (given by XL ) with a right-
moving wave profile (given by XR ). These wave profiles move at the speed of light,
which is characteristic of an ultra-relativistic string.
However, there is more than this to solving the NG equations. To linearize the
equations we had to impose gµν = Ω2 ηµν for some conformal factor Ω2 , and this tells
us that
Ẋ · X 0 = 0 , −Ẋ 2 = (X 0 )2 = Ω2 ⇒ Ẋ 2 + (X 0 )2 = 0 . (3.74)
These are just the Hamiltonian constraints H± = 0, simplified by using the conformal-
gauge relation P = T Ẋ. So, to solve the NG equations we must ensure that our
solution X = XL + XR also satisfies (Ẋ ± X 0 )2 = 0 (or verify that the induced metric
is conformally flat). Using the lightcone worldsheet coordinates σ ± , we can rewrite
these conditions as
(∂± X)2 = 0 . (3.75)
Notice that 2 X 0 = 0 is solved by X 0 = t. There are more general solutions
but we can use the residual gauge invariance to arrange for X 0 = t without loss of
generality. Let’s check that X 0 = t fixes the residual gauge invariance. The residual
gauge transformation of X 0 is

δξ X 0 = ξ + (σ + )∂+ X 0 + ξ − (σ − )∂− X 0 . (3.76)

Setting X 0 = t and insisting that δξ X 0 = 0 (to maintain the gauge choice) we find
that
0 = ξ + (σ + ) + ξ − (σ − ) ⇒ ± ξ ± = ξ¯ (constant). (3.77)
Thus, the choice X 0 = t has now fixed all of the residual gauge invariance except for
the transformation
δξ̄ X = ξ¯ (∂+ − ∂− ) X = ξX
¯ 0, (3.78)
which is the transformation induced by a constant shift of σ; this is an invariance
of the closed string (it makes no difference which point on the string corresponds

– 31 –
to σ = 0) but the b.c.s for an open string require ξ¯ = 0. So we have just enough
residual gauge invariance (and slightly more for a closed string) to set X 0 = t. If we
do so then the constraints (3.75) become
 2
 ~  = 1 .
2 ∂± X (3.79)
See Q.II.1 for a geometrical interpretation of this result.
Let’s apply these ideas to the closed string configuration in a 5D Minkowski
spacetime with X 0 = t and
1 in(σ−t) 1 im(σ+t)
Z ≡ X 1 + iX 2 = e , W ≡ X 3 + iX 4 = e . (3.80)
2n 2m
This configuration solves the 2D wave equation and is periodic in σ with period 2π.
If the induced metric is conformal to 2D Minkowski then it will also solve the full
NG equations. A calculation gives

2
ds2 ind = − dX 0 + |dZ|2 + |dW |2


1 1
= −dt2 + (dσ − dt)2 + (dσ + dt)2
4 4
1 2 2


= −dt + dσ . (3.81)
2
In other words,
1
gµν = ηµν . (3.82)
2
This is a conformal factor (1/2) times the Minkowski metric, so the given configura-
tion is a solution of the NG equations23 .
Let’s compute the proper length of the string. Setting t = t0 in the induced
worldsheet metric (for some constant t0 ) we see that d`2 = 12 dσ 2 , so
1
I √
L= √ dσ = 2 π . (3.83)
2
It is rather surprising that this should be constant, i.e. independent of t0 . This can
only happen if the motion of the string supports it against collapse due to its tension.
To check this, we may compute the total energy, which is
I I
H = dσP = T dσ Ẋ 0 = 2πT .
0
(3.84)

We see that √ √ 
H= 2TL = TL + 2 − 1 TL. (3.85)
The first term is the potential energy of the string. The second term is therefore
kinetic energy. The string is supported against collapse by rotation in the Z and
W planes. This solution has the special property of being stationary; the string is
motionless in a particular rotating frame, supported against collapse by its rotation
in the Z and W planes (neither of which coincide with the plane of the string).
23
It is not necessary to verify (3.79), but this is a good check.

– 32 –
3.5.1 Conformal invariance of conformal gauge action
Let’s now verify the residual gauge invariance of the conformal gauge action for the
closed string. We can write this action as24
Z I
I[X] = 2T dt dσ ∂+ X · ∂− X . (3.86)

The variation of this action that results from the transformation δX = ξ + ∂+ X is

Z I
δξ+ I = 2T dt dσ ∂+ ξ + ∂+ X · ∂− X + ∂+ X · ∂− ξ + ∂+ X .



(3.87)

Integrating by parts in the first term25 we find that

Z I
δξ+ I = 2T dt dσ ∂− ξ + (∂+ X)2 . (3.88)

We thus confirm that the action is invariant if ∂− ξ + = 0.

But is this a gauge invariance rather than a symmetry? For it to be a symmetry
there would have to be corresponding Noether charges. Let’s attempt to compute
them. As σ ∼ σ + 2π, we also have σ ± ∼ σ ± + 2π, so ξ ± (σ ± ) must be periodic
functions of their arguments, which we can express as a Fourier series:
±
X
ξ ± (σ ± ) = einσ ξn± . (3.89)
n∈Z

We have verified invariance of the action for constant coefficients ξn± . Now we allow
these coefficients to depend on t:
X ± 1 X inσ± ˙±
ξ ± (σ + , t) = einσ ξn+ (t) ⇒ ∂∓ ξ ± = e ξn (t) . (3.90)
2
n∈Z n∈Z

Now we have
Z I
±
X
δξ± I = dt ξ˙n± Q±
n , Q±
n =T dσ einσ (∂± X)2 , (3.91)
n∈Z

and it looks as though we have Noether charges Q±n (n ∈ Z). Let’s verify that the
±
Qn are constants of motion. We have
I h i I h i
± inσ + ±
Q̇n = T dσ ∂t e 2
(∂± X) = 2T dσ ∂∓ einσ (∂± X)2
I I
inσ ± ±
= 2T dσ e ∂∓ (∂± X) = −T dσ einσ ∂± X · 2 X
2

= 0 given 2 X = 0 . (3.92)
24
(∂X)2 = −4 ∂+ X · ∂− X for our definition of worldsheet lightcone coordinates.
25
As usual, we ignore time-boundary terms. There is no space boundary for a closed string.

– 33 –
 The Q± n are Noether charges associated to symmetries of the action (3.86) but in
the process of arriving at this action we have lost the constraints (∂± X)2 = 0. This
cannot happen if the gauge-fixing procedure is legitimate, so there must be something
wrong with the conformal gauge! What we have overlooked is that the conformal
gauge condition can be made only locally on the worldsheet. On the cylindrical
worldsheet of a closed string propagating from some initial time, we could arrange
(for example) to set λ± = 1 everywhere immediately after an initial time t0 , but then
we could not set λ± = 1 at the initial time. In this case, the action would still depend
on λ± (t0 ) and variation with respect to λ± (t0 ) would lead to the initial conditions
(∂± X)2 |t=t0 = 0 . (3.93)
However, the identity
1 0
∂t (∂± X)2 ≡ − ∂± X · 2 X ± (∂± X)2
  
(3.94)
2
ensures that (∂± X)2 remains zero if it is initially zero, as a consequence of the equa-
tion of motion 2 X = 0 . In other words, the equation of motion plus constraints as
initial conditions is equivalent to the full NG equations (which must be the outcome
of any legitimate resolution of the problem). And once we include the constraints,
the Noether charges Q± n are all zero, which means that the residual gauge invari-
ance of the NG action in conformal gauge is indeed a gauge invariance, and not a
symmetry.
Another statement of the NG constraints in the context of the conformal gauge
action is that the stress tensor for X is zero. Let’s check this. The stress tensor is
 
1 λ


Θµν = T ∂µ X · ∂ν X − ηµν ∂ X · ∂λ X . (3.95)
2
This tensor is traceless because X is massless (as a set of 2D fields) and this translates
to Θ+− = 0; to check this you need to use the fact that η+− = 1/2 for our worldsheet
lightcone conventions. The non-zero components are
Θ±± = T (∂± X)2 , (3.96)
so the constraints are equivalent to the vanishing of Θµν . This implies, in particular,
that the total energy of the 2D fields X m is zero (which is possible because X 0 enters
the action with the “wrong” sign). However, this does not mean that the energy of
the string is zero. The string energy is
I I
Estring = dσ P = T dσ Ẋ 0 ,
0
(3.97)

where we have used the conformal gauge relation P = T Ẋ. As we have shown, the
residual gauge invariance in the conformal gauge can be fixed by setting X 0 = t, but
this implicitly fixes a unit of length if (t, σ) are assumed dimensionless (such that
Estring = 2πT in these units). It is usually more convenient to suppose (t, σ) to have
units of length and then allow the solutions to determine the periodicity in σ.

– 34 –
3.6 Fourier expansion: closed string
The worldsheet fields of the closed string are periodic functions of σ with (by con-
vention) period 2π, so we can express them as Fourier series. It is convenient to
express (X, P ) as Fourier series by starting with the combinations P ± T X 0 (because
the gauge transformations act separately on P + T X 0 and P − T X 0 ), so we write
r
0 T X ikσ
P − TX = e αk (t) (α−k = αk∗ )
π
k∈Z
r
T X
P + T X0 = e−ikσ α̃k (t) (α̃−k = α̃k∗ ) . (3.98)
π
k∈Z

Recall that worldsheet parity σ → −σ exchanges P + T X 0 with P − T X 0 . Because

of the relative minus sign in the exponents of the Fourier series, this means that
worldsheet parity exchanges αk with α̃k :

αk ↔ α̃k . (3.99)

We can integrate either of equations (3.98) to determine the total D-momentum

in terms of Fourier modes because X 0 integrates to zero for a closed string; this gives
us I √
4πT α0 p
p = P dσ = √ ⇒ α0 = α̃0 = √ . (3.100)
4πT α̃0 4πT
By adding the Fourier series expressions for P ± T X 0 we now get
r
p(t) T X ikσ
P (t, σ) = + e [αk (t) + α̃−k (t)] . (3.101)
2π 4π k6=0

By subtracting we get
1 X ikσ
X0 = −√ e [αk (t) − α̃−k (t)] , (3.102)
4πT k6=0

which we may integrate to get the Fourier series expansion for X:

1 X i ikσ
X(t, σ) = x(t) + √ e [αk (t) − α̃−k (t)] . (3.103)
4πT k6=0 k

The integration constant (actually a function of t) can be interpreted as the position

of the centre of mass of the string; we should expect it to behave like a free particle.
Using the Fourier series expansions of (X, P ) we now find that
I ∞
X i
d
m m
dσ Ẋ Pm = ẋ pm + α̇k · α−k + α̃˙ k · α̃−k + () (3.104)
k=1
k dt

– 35 –
Exercise: check this [Hint. Cross terms that mix α with α̃ are all in the total time
derivative term, and the k < 0 terms in the resulting sum double the k > 0 terms].
Next we Fourier expand the constraint functions H± :
1 X inσ 1 X −inσ
H− = e Ln , H+ = e L̃n . (3.105)
2π 2π
n∈Z n∈Z

Inverting to get the Fourier coefficients in terms of the functions H± , we get

I I
−inσ 1 2
Ln = dσ e H− = dσ e−inσ (P − T X 0 )
4T
I I
1 2
inσ
L̃n = dσ e H+ = dσ einσ (P + T X 0 ) . (3.106)
4T
Inserting the Fourier expansions of (P ± T X 0 ) given in (3.98), we find that (Exercise:
check this)
1X 1X
Ln = αk · αn−k , L̃n = α̃k · α̃n−k . (3.107)
2 2
k∈Z k∈Z
We may similarly expand the Lagrange multipliers as Fourier series but it should
be clear in advance that there will be one Fourier mode of λ− for each Ln (let’s call
this λ−n ) and one Fourier mode of λ+ for each L̃n (let’s call this λ̃−n ). We may now
write down the closed string action in terms of Fourier modes. It is
 
Z  ∞
Xi   

X
I = dt ẋm pm + α̇k · α−k + α̃˙ k · α̃−k − λ−n Ln + λ̃−n L̃n . (3.108)

k=1
k 
n∈Z

This action is manifestly Poincaré invariant. The Noether charges are

I
Pm = dσPm = pm ,
I
mn
J = 2 dσX [m P n] = 2x[m pn] + S mn , (3.109)

where the spin part of the Lorentz charge is (Exercise: check this)
∞
mn
X i  [m n] [m n]

S = −2 α−k αk + α̃−k α̃k . (3.110)
k=1
k

• Lemma. For a Lagrangian of the form

i
α̇α∗ − H(α, α∗ )
L= (3.111)
c
for constant c, the PB of the canonical variables takes the form

{α, α∗ }P B = −ic . (3.112)

p
To see this set α = |c|/2 [q + i sign(c)p] to get L = q̇p − H, for which we know
that {q, p}P B = 1. This implies the above PB for α and α∗ .

– 36 –
Using this lemma we may read off from the action that the non-zero Poisson brackets
of canonical variables are {xm , pn }P B = δnm and
n
αkm , α−k = −ikη mn , n
α̃km , α̃−k = −ikη mn .
 
PB PB
(3.113)

Using these PBs, and the Fourier series expressions for (X, P ), we may compute the
PB of X(σ) with P (σ 0 ). [Exercise: check that the result agrees with (3.22).]
We may also use the PBs (3.113) to compute the PBs of the constraint functions
(Ln , L̃n ). The non-zero PBs are (Exercise: check this)
n o
{Lk , Lj }P B = −i (k − j) Lk+j , L̃k , L̃j = −i (k − j) L̃k+j . (3.114)
PB

We may draw a number of conclusions from this result:

• The constraints are first class, so the Ln and L̃n generate gauge transformations,
for each n ∈ Z.

• The structure functions of the algebra of first-class constraints are constants.

This means that the (Ln , L̃n ) span an infinite dimensional Lie algebra.

• The Lie algebra of the gauge group is a direct sum of two copies of the same
algebra, sometimes called the Witt algebra.

The Witt algebra is also the algebra of diffeomorphisms of the circle. Suppose we
have a circle parameterized by θ ∼ θ + 2π (we could take θ to be σ + or σ − ). The
algebra Diff 1 of diffeomorphisms is spanned by the vector fields on the circle, and
since these are periodic we may take as a basis set the vector fields {Vn ; n ∈ Z},
where
d
Vn = einθ . (3.115)
dθ
The commutator of two basis vector fields is

[Vk , Vj ] = −i (k − j) Vk+j . (3.116)

Corollary: the algebra of the gauge group is Diff 1 ⊕ Diff 1 , as claimed earlier.

3.7 Fourier expansion: open string

The open string has two ends. We will choose the ends to be at σ = 0 and σ = π,
so the parameter length of the string is π (this is just a convention). We shall con-
sider the case of free-end (Neumann) boundary conditions, as these preserve Lorentz
invariance.
If the ends of the string are free, we must require X 0 to be zero at the ends, i.e.
at σ = 0 and σ = π. We shall proceed in a way that will allow us to take over results
from the closed string; we shall use a “doubling trick”:

– 37 –
 • First, we extend the definition of (X, P ) from the interval [0, π] to the interval
[0, 2π] in such a way that (X, P ) are periodic on this doubled interval. This
will allow us to use the closed string Fourier series expressions.

• Next, we impose a condition that relates (X, P ) in the interval [π, 2π] to (X, P )
in the interval [0, π]; this will ensure that any additional degrees of freedom
that we have introduced by doubling the interval are removed. Notice that if
σ ∈ [0, π] then −σ ∼ −σ + 2π ∈ [π, 2π], so we need to relate the worldsheet
fields at σ to their values at −σ. The condition that does this should be
consistent with periodicity in the doubled interval, but it should also imply the
free-end b.c.s at σ = 0, π.
The solution to these requirements is to impose the condition

(P + T X 0 ) (σ) = (P − T X 0 ) (−σ) . (3.117)

This is consistent with periodicity, and setting σ = 0 it implies that X 0 (0) = 0. It

also implies that X 0 (π) = 0 because −π ∼ π by periodicity. In terms of Fourier
modes, the condition (3.117) becomes

α̃k = αk (k ∈ Z). (3.118)

Using this in (3.98) we have

r
0 T X ∓ikσ
P ± TX = e αk (α−k = αk∗ ) . (3.119)
π
k∈Z

Equivalently,
r
T X i X
P = cos(kσ) αk , X0 = −√ sin(kσ) αk . (3.120)
π πT k∈Z
k∈Z

Integrating to get X, and defining p(t) by

p
α0 = √ , (3.121)
πT
we have
1 Xi
X(t, σ) = x(t) + √ cos(kσ) αk ,
πT k6=0 k
r
p(t) T X
P (t, σ) = + cos(kσ) αk . (3.122)
π π k6=0

Notice that p is again the total momentum since

Z π
dσ P (t, σ) = p , (3.123)
0

– 38 –
but its relation to α0 differs from that of the closed string.
Using the Fourier series expansions for (X, P ) we find that
Z π ∞
m m
X i d
Ẋ Pm dσ = ẋ pm + α̇k · α−k + () . (3.124)
0 k=1
k dt
Because of (3.117) we also have H+ (σ) = H− (−σ), so we should impose a similar
relation on the Lagrange multipliers
λ+ (σ) = λ− (−σ) (⇒ u|ends = 0 & e0 |ends = 0) . (3.125)
Then
Z π Z π Z π
− −
+
dσ λ− (−σ)H− (−σ)


dσ λ H− + λ H+ = dσ λ (σ)H− (σ) +
0
Z0 π Z0 0+2π
= dσ λ− (σ)H− (σ) + dσ λ− (σ)H− (σ)
−π+2π
I0
= dσ λ− H− , (3.126)

and we can now use the Fourier series expansions of the closed string.
The final result for the open string action in Fourier modes is
 
Z  Xi X 
I = dt ẋm pm + α̇k · α−k − λ−n Ln . (3.127)
 k k=1

n∈Z

The difference with the closed string is that we have one set of oscillator variables
instead of two. We can now read off the non-zero PB relations
{xm , pn }P B = δnm ,
 m n
αk , α−k P B = −ikη mn , (3.128)
and we can use this to show that
{Lj , αkm }P B = ikαk+j . (3.129)
This means that the gauge variation of αk is
X X
δξ αkm = ξ−n {αkm , Ln }P B = −ik m
ξ−n αn+k , (3.130)
n∈Z n∈Z

where ξn are √
parameters. To compute the gauge transformation of (x, p) we use the
relation p = πT α0 and the fact that
1
L0 = α02 + . . . , Ln = α0 · αn + . . . , (3.131)
2
where the dots indicate terms that do not involve α0 , to compute
1 X
δξ x m = √ ξ−k αkm , δpm = 0 . (3.132)
πT k∈Z
Finally, one may verify that the action is invariant if
X
δξ λj = ξ˙j + i (2k − j) ξk λj−k . (3.133)
k∈Z

– 39 –
3.8 The NG string in light-cone gauge
We shall start with the open string (with free-end b.c.s). We shall impose the gauge
conditions
X + (t, σ) = x+ (t) , P− (t, σ) = p− (t)/π . (3.134)
where x( t) and p− (t) are freely variable functions of t. It customary to also set
x+ (t) = t, as for the particle, but it is simpler not to do this. This means that we will
not be fixing the gauge completely since we will still be free to make σ-independent
reparametrizations of the worldsheet time t.
The above gauge-fixing conditions are equivalent to
+
(P ± T X 0 ) = p− (t)/π ⇔ αk+ = 0 ∀k 6= 0 . (3.135)

In other words, we impose a light-cone gauge condition only on the oscillator variables
of the string, not on the zero modes (centre of mass variables). Let’s check that
the gauge has been otherwise fixed. We can investigate this using the criterion
summarised by the formula (2.55); we compute
+ +
= −ikα0+ δjk (using gauge condition)

Lj , α−k PB
= −ikαj−k
p−
= −ik √ δjk . (3.136)
πT
This is invertible if we exclude j = 0 and k = 0, so we have fixed all but the gauge
transformation generated by L0 . Now we have, since αk+ = 0,
∞ ∞
X i X i
α̇k · α−k = α̇k · α−k , (3.137)
k=1
k k=1
k

where the (D − 2)-vectors αk are the transverse oscillator variables.

We also have,
∞
1 2 X 1
p2 + 2πT N ,


L0 = α0 + α−k · αk = (3.138)
2 k=1
2πT

where N is the level number:

∞
X
N= α−k · αk . (3.139)
k=1

For n 6= 0,
1X + −
1X
Ln = αk αn−k + αk− αn−k
+
+ αk · αn−k
2 2
k∈Z k∈Z
1 X
= α0+ αn− + αk · αn−k (using gauge condition). (3.140)
2
k∈Z

– 40 –
 √
We can solve this for αn− ; using p = πT α0 , we get
√
πT X
αn− =− αk · αn−k (n 6= 0). (3.141)
2p−
k∈Z

As we have solved the constraints Ln = 0 for n 6= 0, only the L0 = 0 constraint

will be imposed by a Lagrange multiplier in the gauge-fixed action, which is
Z ( ∞
)
X i 1
dt ẋm pm + α̇k · α−k − e0 p2 + M 2


I= , (3.142)
k=1
k 2

where e0 = λ0 /(πT ) and

 
0 1
2
M = 2πT N = N/α α0 ≡ . (3.143)
2πT

Notice that the action does not involve αn− (for n 6= 0) but the Lorentz charges do.
Recall that the spin part of the Lorentz charge J mn is S mn = −2 ∞ i [m n]
P
k=1 k α−k αk . Its
non-zero components of S mn in light-cone gauge are (I, J, = 1, . . . D − 2)
∞ ∞
X i [I J] X i − I
S IJ = −2 S −I = − I
αk− .


α−k αk , α−k αk − α−k (3.144)
k=1
k k=1
k

The canonical PB relations that we read off from the action (3.142) are

{xm , pn }P B = δnm , αkI , α−k

J
= −ik δ IJ .

PB
(3.145)

These may be used to compute the PBs of the Lorentz generators; since J = L + S
where {L, S}P B = 0, the PB relations among the components of S alone must be the
same as those of J . The PBs of S IJ are those of the Lie algebra of the transverse
rotation group SO(D − 2), and their PBs with S −K are those expected from the fact
that S −K is a (D − 2) vector. Finally, Lorentz invariance requires that

S −I , S −J

PB
= 0. (3.146)

This can be confirmed by making use of the PB relations

√ √
πT πT I
αk− , α`− P B −
  − I
=i (k − `) αk+` , αk , α` P B = −i ` αk+` . (3.147)
p− p−

This has to work because gauge fixing cannot break symmetries; it can only obscure
them.

– 41 –
3.8.1 Closed string in light-cone gauge
Now we fix the gauge invariances associated with Ln and L̃n for n 6= 0 by setting

αk+ = 0 & α̃k+ = 0 ∀k 6= 0 . (3.148)

This leaves unfixed the gauge invariances generated by L0 and L̃0 , which are now
∞
1 2 X p2
L0 = α0 + α−k · αk = +N,
2 k=1
8πT
∞
1 2 X p2
L̃0 = α̃0 + α̃−k · α̃k = + Ñ . (3.149)
2 k=1
8πT

Here we have used the closed string relation (3.100) between p and α0 = α̃0 . By
adding and subtracting the two constraints L0 = 0 and L̃0 = 0 we get the two
equivalent constraints
 
p2 + 4πT N + Ñ = 0 & Ñ − N = 0 , (3.150)

which will be imposed by the Lagrange multipliers e0 = 2(λ0 + λ̃0 ) and u0 = 4πT (λ −
λ̃0 ) in the gauge-fixed action. The remaining constraints Ln = 0 and L̃n = 0 for n 6= 0
we solve for αk− and α̃k− , as for the open string. The closed string action in light-cone
gauge is therefore
Z ( ∞
)
X i   1 
˙ k · α̃−k − e0 p2 + M 2 − u0 N − Ñ

I = dt ẋm pm +


α̇k · α−k + α̃ ,
k=1
k 2
(3.151)
where
 
M 2 = 4πT N + Ñ
 
= 8πT N using Ñ = N constraint . (3.152)

We now have two first-class constraints, a mass-shell constraint and a level-matching

constraint26 .

4. Interlude: Light-cone gauge in field theory

We shall consider Maxwell’s equation, for the vector potential Am , and the linearised
Einstein equations for a symmetric tensor potential hmn , which may be interpreted
as the perturbation of the space-time metric about a Minkowski vacuum metric.
26
The constraint function (N − Ñ ) generates the residual gauge transformation δβ αk (t) =
−ikβ(t)αk (t) and δβ α̃k (t) = ikβ(t)α̃k (t) which originates from invariance under t-dependent shifts
of σ.

– 42 –
 We choose light-cone coordinates (x+ , x− , xI ) (I = 1, . . . , D − 2). Recall that
for the particle we assumed that p− 6= 0, which is equivalent to the assumption that
the differential operator ∂− is invertible. We shall make the same assumption in the
application to field theory.

4.1 Maxwell in light-cone gauge

Maxwell’s equations are

D Am − ∂m (∂ · A) = 0 , m = 0, 1, . . . , D − 1 . (4.1)

They are invariant under the gauge transformation Am → Am + ∂m α. The light-cone

gauge is
A− = 0 . (4.2)
To see that the gauge is fixed we set to zero a gauge variation of the gauge-fixing
condition: 0 = δα A− = ∂− α. This implies that α = 0 if the differential operator ∂−
is invertible.
In this gauge the m = − Maxwell equation is ∂− (∂ · A) = 0, which implies that
∂ · A = 0 since we assume invertibility of ∂− . And since 0 = ∂ · A = ∂− A+ + ∂I AI ,
we can solve for A+ :
A+ = −∂−−1 (∂I AI ) . (4.3)
This leaves AI as the only independent variables. The m = + equation is D A+ = 0,
but this is a consequence of the m = I equation, which is

D AI = 0 , I = 1, . . . , D − 2 . (4.4)

So this is what Maxwell’s equations look like in light-cone gauge: wave equations for
D − 2 independent polarisations.

4.2 Linearized Einstein in light-cone gauge

The Einstein field equations of GR are equations for a spacetime metric gmn . Given
that Minkowski spacetime is a solution, we can write gmn = ηmn + hmn and view hmn
as a symmetric tensor field on a Minkowski background; this makes sense if hmn is
“small”. To first order in this metric perturbation, the (source-free) Einstein field
equations become

D hmn − 2∂(m hn) + ∂m ∂n h = 0 , hm ≡ ∂ n hnm , h = η mn hmn . (4.5)

This equation is invariant under the gauge transformation (Exercise: verify this)

hmn → hmn + 2∂(m ξn) . (4.6)

The light-cone gauge choice is

h−n = 0 (n = −, +, I) ⇒ h− = 0 & h = hJJ . (4.7)

– 43 –
 In the light-cone gauge the “m = −” equation is −∂− hn + ∂− ∂n hJJ = 0, which
can be solved for hn :
hn = ∂n hJJ . (4.8)
But since we already know that h− = 0, this tells us that hJJ = 0, and hence that
hn = 0 and h = 0. At this point we see that the equations reduce to D hmn = 0,
but
h+ = 0 ⇒ h++ = −∂−−1 (∂I hI+ ) ,
hI = 0 ⇒ h+I = −∂−−1 (∂J hJI ) , (4.9)
so the only independent components of hmn are hIJ , and this has zero trace. We
conclude that the linearised Einstein equations in light-cone gauge are
D hIJ = 0 & hII = 0 . (4.10)
The number of polarisation states of the graviton in D dimensions is therefore
1 1
(D − 2)(D − 1) − 1 = D(D − 3) . (4.11)
2 2
For example, for D = 4 there are two polarisation states27 , and the graviton is a
massless particle of spin 2.
It is a general feature of massless particles that their independent polarisation
states form a representation of the transverse rotation group SO(D − 2).

4.3 Massive fields

Notice that for both Maxwell and linearised Einstein, the independent fields in
light-cone gauge form an irreducible representation of the transverse rotation group
SO(D − 2). In the former case we get a (D − 2)-vector. In the latter case we get a
traceless symmetric tensor whose components span a space of dimension D(D −3)/2.
Adding a mass usually breaks the gauge invariance that we used to fix the light-
cone gauge, but the light-cone decomposition is still useful. Consider the Proca
equations
(D − m2 )Am = 0 , ∂ m Am = 0 . (4.12)
The second of these equations is called the subsidiary condition. In light-cone coordi-
nates we have A = dxm Am = dx− A− + dx+ A+ + dxI AI , and the subsidiary condition
is
∂− A+ + ∂+ A− + ∂I AI = 0 . (4.13)
We can’t now set A− = 0 because there isn’t a gauge invariance to justify it, but we
can still suppose that ∂− is invertible, in which case the subsidiary condition can be
solved for A+ . This leaves us with the D − 1 equations
D − m2 AI = 0 & D − m2 A− = 0 .



(4.14)
27
Massless particles of any non-zero spin have two independent polarization states in D = 4.

– 44 –
There are now (D − 1) independent polarisation. This is the (D − 1)-vector repre-
sentation of the (full) rotation group SO(D − 1).. For D = 4 we have a 3-vector
and hence (since 3 = 2s + 1 for s = 1) the Proca equations are field equations for a
massive particle of spin 1.
It is a general feature of massive particles that their polarisation states form a
representation of the rotation group.

5. Quantum NG string
Now we pass to the quantum theory. This is simplest in light-cone gauge because
this gauge choice removes all unphysical components of the oscillator variables prior
to quantization, but it obscures Lorentz invariance. Then we consider how the same
results might be found in the conformal gauge, where Lorentz invariance is still
manifest.

5.1 Light-cone gauge quantization: open string

The canonical PB relations of the open string in Fourier modes are (3.145). These
become the canonical commutation relations

[x̂m , p̂n ] = iδnm ,

 I J 
α̂k , α̂−k = k δ IJ , (5.1)

where the hats now indicate operators, and the hermiticity of the operators (X̂, P̂ )
requires that
α̂−k = α̂†k . (5.2)
A state of the string of definite momentum is the tensor product of a momentum
eigenstate |pi with a state in the oscillator Fock space, built upon the Fock vacuum
state |0i annihilated by all annihilation operators:

α̂k |0i = 0 ∀ k ∈ Z+ . (5.3)

We get other states in the Fock space by acting on the oscillator vacuum with the
creation operators α̂−k any number of times, and for any k > 0. This gives us a
basis for the entire infinite-dimensional space.
Next, we need to replace the level number N by a level number operator N̂ ,
but there is an operator ordering ambiguity; different orderings lead to operators N̂
that differ by a constant. We shall choose to call N̂ the particular operator that
annihilates the oscillator vacuum; i.e.
∞
X
N̂ = α̂−k · α̂k ⇒ N̂ |0i = 0 . (5.4)
k=1

So the oscillator vacuum has level number zero. This removes the ambiguity in the
definition of N̂ but it does not remove the ambiguity in passing from the classical to

– 45 –
the quantum theory; whenever we see N in the classical theory we may still replace
it by N̂ plus a constant in the quantum theory.
Notice now that h i
N̂ , α̂−k = k α̂−k . (5.5)

This tell us that acting on a state with any component of α̂−k raises the level number
by k, and this tells that N̂ is diagonal in the Fock state basis constructed in the
way described above, and that the possible level numbers (eigenvalues of N̂ ) are
N = 0, 1, 2, . . . , ∞. We can therefore organise the states according to their level
number. There is only one state in the Fock space with N = 0, the oscillator
I
vacuum. At N = 1 we have the (D − 2) states α̂−1 |0i. At N = 2 we have the states

I I J
α̂−2 |0i , α̂−1 α̂−1 |0i . (5.6)

At N = 3 we have the states

I I J I J K
α̂−3 |0i , α̂−2 α̂−1 |0i , α̂−1 α̂−1 α̂−1 |0i , (5.7)

and so on.
A generic state of the string at level N in a momentum eigenstate |pi of D-
momentum p takes the form
|pi ⊗ |ΨN i , (5.8)

where |ΨN i is some state in the oscillator Fock space with level number N . The
mass-shell constraint for such a state implies that p2 = −M 2 , where

M 2 = 2πT (N − a) . (5.9)

The constant a is introduced to take care of the operator ordering ambiguity in

passing from the classical to the quantum theory.
Now we proceed to analyse the string spectrum level by level. We shall use the
standard notation
2πT = 1/α0 (5.10)

• N = 0. There is one state, and hence a scalar, with α0 M 2 = −a.

• N = 1. There are now (D − 2) states, α̂−1 |0i with α0 M 2 = (1 − a). The only
way that these states could be part of a Lorentz-invariant theory is if they de-
scribe the polarization states of a massless vector (a massive vector has (D − 1)
polarisation states). It is important to appreciate that Lorentz invariance of
the quantum theory in light-cone gauge is not guaranteed because the Lorentz
invariance of the classical action in light-cone gauge is not manifest (i.e. not lin-
early realised; the Lorentz transformations are non-linear in light-cone gauge).

– 46 –
 In fact, we now see that a necessary condition for Lorentz invariance of the
quantum theory is that we choose
a = 1. (5.11)
This ensures that the N = 1 state is massless, but it also tells us that the
N = 0 scalar is a tachyon; i.e. “particle” with spacelike D-momentum. This
looks bad; we’ll return to it later.

• N = 2. Given a = 1 the N = 2 states are massive, with α0 M 2 = 1. The states

are those of (5.6), which are in the symmetric 2nd-rank tensor plus vector
representation of SO(D − 2). These states form a symmetric traceless tensor
of SO(D − 1) and hence describe a massive spin-2 particle28 .
We now know that the open string ground state is a scalar tachyon, and its first
excited state is a massless vector, a “photon”. All higher level states are massive, and
so should be in SO(D − 2) representations that can be combined to form SO(D − 1)
representations (i.e. representations of the rotation group). We have seen that this
is true for N = 2 (in this case, exceptionally, the SO(D − 1) representation is
irreducible) and it can be shown to be true for all N ≥ 2.
The parameter −a has a physical meaning as the total zero-point energy of the
oscillator modes of the string. To see why, let’s suppose that we had chosen the
standard Weyl ordering of operators for a harmonic oscillator. This tells us to take
N → N̂weyl with
∞ ∞
1X 1 X m
α̂−k , α̂nk ηmn

N̂weyl = (α̂−k · α̂k + α̂k · α̂−k ) = N̂ +
2 k=1 2 k=1
∞
(D − 2) X
= N̂ + k. (5.12)
2 k=1

The extra constant term is the sum over the zero-point contributions of the oscilla-
tors29 . The eigenvalues of N̂weyl are N − a where (as before) N is the eigenvalue of
N̂ , and
∞
(D − 2) X
−a= k. (5.13)
2 k=1
The sum on the RHS diverges, so it doesn’t appear to make sense to use Weyl
ordering, but Euler was of the opinion that ∞
P
k=1 k = −1/12. That’s because he was
working on properties of the infinite series.
∞
X
ζ(s) = k −s . (5.14)
k=1
28
A symmetric traceless tensor field of rank n is usually said to describe a particle of “spin n”
even though “spin” is not sufficient to label states in space times of dimension D > 4.
29
There are (D − 2) oscillators of angular frequency |k| associated to the pair (αk , α†k )).

– 47 –
When this sum converges, it defines a function (now known as the Riemann zeta
function) that can be analytically continued to the entire complex s-plane except at
s = 1, where it has a simple pole. In particular
1
ζ(−1) = − . (5.15)
12
If we use this to justify Euler’s intuition, then we conclude that

(D − 2)
a= . (5.16)
24
Hence, since we know that Lorentz invariance requires a = 1, we also conclude that
D = 26. Remarkably, it is indeed true that Lorentz invariance requires D = 26.

5.1.1 Critical dimension

Because Lorentz invariance is not manifest (i.e. linearly realized) in the light-cone
gauge, there is no gauarantee that it will be preserved in the passage from the
classical to the quantum theory. To check Lorentz invariance we have to compute
the commutators of the quantum Lorentz charges Jˆmn . Since Jˆ = L̂ + Ŝ and it is
easy to see that [L̂, Ŝ] = 0, we can focus on the spin Ŝ mn . As SO(D − 2) invariance
is manifest in light-cone gauge, it cannot be broken; indeed, we saw that the states
are all tensors of SO(D − 2). For the same reason, the commutation relations of Ŝ IJ
with itself and with Ŝ −I must be those expected from the corresponding PB relations,
which merely tell us that Ŝ IJ is an SO(D − 2) tensor and that Ŝ −I is an SO(D − 2)
vectors; recall that S +I is zero in light-cone gauge. The only commutation relation
that is required by Lorentz invariance30 but not guaranteed to hold in light-cone
gauge is h i
Ŝ −I , Ŝ −J = 0 (?) (5.17)

If the {, }P B → −i[, ] rule were to apply here then Lorentz invariance of the quantum
string would follow, but it does not obviously apply in light-cone gauge because
in this gauge Ŝ −I is cubic in transverse oscillator operators. The product of two
such operators is 6th-order in transverse oscillators, so the commutator is 4th-order,
in principle, but as there is no quartic term classically there must be an operator
ordering that leads to its cancellation in the quantum theory. However, any re-
ordering to achieve this could generate a term quadratic in transverse operators,
which would be inconsistent with Lorentz invariance.
From the classical expression (3.147) for αk− we have

πT X
α̂n− = − α̂j · α̂n−j , (5.18)
2p−
j∈Z

30
As can be verified by computing the same commutator for L; see Q.II.3.

– 48 –
which is unambiguous since there is no operator ordering ambiguity. Similarly, from
the classical expression for S −I in (3.145) we may guess that
∞
X i − I
−I I
α̂k− .


Ŝ =− α̂−k α̂k − α̂−k (5.19)
k=1
k

There is an ordering ambiguity here because

√
 −  πT
α̂k , α̂` = α̂k+` , (5.20)
p−

but the order of operators must be as given in order that Ŝ −I |0i = 0.

The commutation relation (5.20) is exactly as we should expect from the corre-
sponding PB relation. The only other commutation relation that we need to compute
the commutator (5.17)is [α̂k− , α̂`− ]. From the PB relations of (3.147) we might expect
√
 − − πT −
α̂k , α̂` = − (k − `) α̂k+` (?). (5.21)
p−
However, this commutator must be examined carefully; it must equal some expression
quadratic in the transverse oscillator operators but if a re-ordering is needed to
get agreement with the above guess then a constant term (times 1/p2− ) could be
generated. In fact, one finds that
√   
 − − πT − 2πT (D − 2) (D − 2) 3
α̂k , α̂` = − (k − `) α̂k+` + 2 a− k+ k δk+` ,
p− p− 24 24
(5.22)
where 
1 n=0
δn = (5.23)
0 n 6= 0 .
We see that there is indeed
√ an −additional non-classical term when k + ` = 0.
Using the fact that πT α0 = p+ , and the mass-shell constraint in the form

−2p+ p− = |p|2 + 2πT (N − a) , (5.24)

we may rewrite the ` = −k commutation relation of (5.22) as

 − −  kπT |p|2
    
2πT (D − 2) (D − 2) 3
α̂k , α̂−k = 2 + 2(N − a) + 2 a− k+ k
p− πT p− 24 24
(5.25)
Noticing that the terms involving the constant a cancel, we can further rewrite this
commutation relation as
 
 − −  kπT 2

2πT D − 2
k3 − k .


α̂k , α̂−k = 2 |p| + 2πT N + 2 (5.26)
p− p− 24
This commutator is the essential one to verify, and we shall now do so.

– 49 –
 • Proof of (5.26). It suffices to consider k > 0 since the k < 0 case then follows
by hermitian conjugation. In this case α̂k− |0i = 0, so that

h0| α̂k− , α̂−k

−
|0i = ||α̂k− |0i||2 .
 
(5.27)

Observe also, again for k > 0, that

√ k−1
!
− πT 1X
α̂−k |0i = − α0 · α̂−k + α̂−j · α̂j−k |0i . (5.28)
p− 2 j=1

This means that

" k−1 X
k−1
#
πT 1 X
||α̂k− |0i||2 = 2 k|α0 |2 + h0| (α̂−j · α̂j−k ) (α̂−j · α̂j−k ) |0i
p− 4 j=1 `=1
k−1
k|p|2 πT X
= 2 + 2 (D − 2) j(k − j) , (5.29)
p− 2p− j=1

√
where we have used πT α0 = p. Now using the identities
k−1 k−1
X 1 X 1
j ≡ k(k − 1) , j 2 ≡ k(k − 1)(2k − 1) , (5.30)
j=1
2 j=1
6

and recalling (5.27), we deduce that

 − −  k|p|2 πT (D − 2) 3

h0| α̂k , α̂−k |0i = 2 + 2 k −k . (5.31)
p− p− 24

This is consistent with (5.26), since N = 0 in the ground state, and it gives
the stated non-classical addition to the commutation relation (5.21) that is the
naive expectation based on the classical PB result. When combined with the
classical result, this quantum compuation therefore implies (5.26).

The result (5.22) is just a first step but it shows how the spacetime dimension
D and the parameter a can enter into the final result, which is
∞    
h
−I −J
i 4πT X (D − 2) 1 (D − 2) [I J]
Ŝ , Ŝ = 2 −2 k+ 2a − α̂−k α̂k . (5.32)
p− k=1 12 k 12

This is zero for D > 3 iff

a = 1 & D = 26 . (5.33)
We therefore confirm that Lorentz invariance requires a = 1, but we now see that it
also requires D = 26; this is the critical dimension of the NG string.

– 50 –
5.1.2 Quantum closed string
Finally, we consider light-cone gauge quantization of the closed string. There are
now two sets of oscillator operators, with commutation relations
h I J i
ˆ k , α̃
ˆ −k .
 I J 
α̂k , α̂−k = kδ IJ = α̃ (5.34)

The oscillator vacuum is now

|0i = |0iR ⊗ |0iL , ˆ k |0iL = 0

α̂k |0iR = 0 & α̃ ∀k > 0 . (5.35)
ˆ such that they annihilate the oscillator
We define the level number operators N̂ and Ñ
vacuum. Again, we can choose a basis for the Fock space built on the oscillator
vacuum for which these operators are diagonal, with eigenvalues N and Ñ . In the
space of states of definite p and definite (N, Ñ ) the mass-shell and level matching
constraints are
h  i  
2
p + 4πT (N − aR ) + Ñ − aL = 0 & (N − aR ) = Ñ − aL . (5.36)

Since we want |0i to be a physical state we must choose aL = aR = a, and then we

have
p2 + 8πT (N − a) = 0 & Ñ = N . (5.37)
This means that we can organise the states according to the level N , with M 2 =
8πT (N − a). We must do this respecting the level-matching condition Ñ = N . Let’s
look at the first few levels

• N = 0. There is one state, and hence a scalar, with α0 M 2 = −4a.

• N = 1. There are now (D − 2) × (D − 2) states

I
α̂−1 ˆ J−1 |0iL
|0iR ⊗ α̃ (5.38)

We can split these into irreducible representations by taking the combinations

I
[hIJ (p) + δIJ φ(p) + bIJ (p)] α̂−1 ˆ J−1 |0iL ,
|0iR ⊗ α̃ (5.39)

where hIJ is symmetric traceless tensor, bIJ an antisymmetric tensor and φ a

scalar. The only way that these could be part of a Lorentz-invariant theory is if
hIJ are the physical components of a massless spin-2 field because massive spin-
2 would require a symmetric traceless tensor of the rotation group SO(D − 1).
Then bIJ must be the physical components of a massless antisymmetric tensor
field, and φ a massless scalar (the dilaton).
Since we require M 2 = 0 we must choose a = 1 again31 . This means that the
ground state is again a tachyon.
31
This is usually given as a = 2 but that’s due to a different definition of a for the closed string.

– 51 –
 • N = 2. Since a = 1, the N = 2 states are massive, with α0 M 2 = 4. Recalling
that the open string states at level 2 combined into a symmetric traceless tensor
of SO(D − 1), we see that the level-2 states of the closed string will combine
into those SO(D − 1) representations found in the product of two symmetric
traceless SO(D −1) tensors. This includes a 4th-order totally symmetric trace-
less tensor describing a massive particle of spin-4; there will be several lower
spins too.

The most remarkable fact about these results is that the closed string spectrum
contains a massless spin-2 particle, suggesting that a closed string theory will be a
theory of quantum gravity. As for the open string, one finds that Lorentz invariance
is preserved only if D = 26 (the calculation needed to prove this is a repeat of the
open string calculation because the spin operator is a sum of a contribution from
“left” oscillators and a contribution from “right” oscillators).
As for the open string, the closed string in its oscillator ground state is a tachyon.
Formally, this is a “particle” of spacelike D-momentum but it is really an indication
that the Minkowski ground state is unstable. The tachyon is absent in superstring
theory, for which the critical dimension is D = 10, and there are various ways to
compactify dimensions so as to arrive at more-or-less realistic models of gravity
coupled to matter.

5.2 “Old covariant” quantization

Dirac’s method of dealing with first-class constraints would appear to allow us to
quantize the string in a way that preserves manifest Lorentz invariance. Let’s consider
the open string with free-end b.c.s. Recall that the action in terms of Fourier modes
is
 
Z  ∞
Xi X  1X
I = dt ẋm pm + α̇k · α−k − λ−n Ln , Ln = αk · αn−k , (5.40)

k=1
k  2
n∈Z k∈Z

and that {Lk , L` }P B = −i (k − `) Lk+` . Applying the {, } → −i[, ] rule to the PBs of
the canonical variables, we get the canonical commutation relations

[x̂m , p̂n ] = iδnm ,

 m n 
α̂k , α̂−k = kη mn . (5.41)

Now we define the oscillator vacuum |0i by

α̂km |0i = 0 ∀k > 0 (m = 0, 1, . . . , D − 1). (5.42)

m
The Fock space is built on |0i by the action of the creation operators α̂−k , but this
gives a space with many unphysical states since we now have D oscillators for each k,
whereas we know (from light-cone gauge quantization) that D −2 suffice to construct
the physical states.

– 52 –
 Can we remove unphysical states by imposing the physical state conditions

L̂n |physi = 0 ∀n ? (5.43)

If this were possible then we would have achieved a Lorentz covariant quantization
of the massless spin-1 particle at level 1 without the need for unphysical polarisation
states, but this is not possible: the Lorentz-invariant Lorenz gauge p · A = 0 reduces
their number from D to D − 1 but no Lorentz-invariant condition will reduce it to
D − 2. So we are going to run into a problem!
Notice that we do not encounter an operator ordering ambiguity when passing
from the classical phase-space function Ln to the corresponding operator L̂n except
when n = 0, so the operator L̂n is unambiguous for n 6= 0 and it is easy to see that

L̂n |0i = 0 , n > 0. (5.44)

However, it is also easy to see [exercise: check these statements] that

1X
L̂−1 |0i ≡ α̂k · α̂−1−k |0i = α̂0 · α̂−1 |0i
2 k
 
1X 1 2
L̂−2 |0i ≡ α̂k · α̂−2−k |0i = α̂0 · α̂−2 + α̂−1 |0i , (5.45)
2 k 2

so it looks as though not even |0i is physical. In fact, there are no states in the
Fock space satisfying (5.43) because the algebra of the operators L̂n has a quantum
anomaly, which is such that the set of operators {L̂n ; n ∈ Z} is not “first-class”. That
is what we shall now prove.
Since the L̂n are quadratic in oscillator variables, the product of two of them
is quartic but the commutator [L̂m , L̂n ] is again quadratic. That is what we expect
from the PB, which is proportional to L̂m+n , but to get this from the expression that
results from computing the commutator we may need to re-order operators, and that
would produce a constant term. So, we are bound to find that
h i
L̂m , L̂n = (m − n) L̂m+n + Amn (5.46)

for some constants Amn . We can compute the commutator using the fact that
h i
L̂m , α̂k = −k α̂k+m . (5.47)

This can be verified directly but it also follows from the corresponding PB result
because no ordering ambiguity is possible either on the LHS or the RHS. Using this,
we find that
h i 1 X h i h i
L̂m , L̂n = L̂m , α̂k · α̂n−k + α̂k · L̂m , α̂n−k
2 k
1X 1X
=− k α̂k+m · α̂n−k − (n − k) α̂k · α̂n+m−k . (5.48)
2 k 2 k

– 53 –
As long as n + m 6= 0 this expression is not affected by any change in the order of
operators, so it must equal what one gets from an application of the {, }P B → −i[, ]
rule. In other words, Amn = 0 unless m + n = 0. We can check this by using the
fact that α̂−k = α̂k† , so that

α̂k |0i = 0 ⇔ h0|α̂−k = 0 . (5.49)

From this we see that for m + n 6= 0,

h0|α̂k+m · α̂n−k |0i = 0 = h0|α̂k · α̂n+m−k |0i (m + n 6= 0) , (5.50)

and hence that h0|[L̂m , L̂n ]|0i = 0 unless m + n = 0. This tells us that

Amn = A(m)δm+n . (5.51)

We now focus on the m + n = 0 case, for which

h i
L̂m , L̂−m = 2mL̂0 + A(m) ⇒ A(−m) = −A(m) . (5.52)

Because of an operator ordering ambiguity, the operator L̂0 is only defined up to the
addition of a constant, so what we find for A(m) will obviously depend on how we
define L̂0 . We shall define it as
∞
1 X
L̂0 = α̂02 + N̂(cov) , N̂(cov) ≡ α̂−k · α̂k , (5.53)
2 k=1

The operator N̂(cov ) is a Lorentz covariant analog of the level number operator of
the light-cone gauge. As in that case, it annihilates the oscillator vacuum and has
eigenvalues that are non-negative integers; these are the level numbers, which we
again call N

From now on we drop the hats on operators

We could now return to (5.48), set n = −m, and then complete the computation
to find A(m). This can be done, but it has to be done with great care to avoid
illegitimate manipulations of infinite sums (see Q.III.1). Here we take an indirect
route. First we use the Jacobi identity32

[Lk , [Lm , Ln ]] + cyclic permutations ≡ 0 , (5.54)

32
This is a consequence of the associativity of the product of operators; i.e. we use the fact
that (Lk Lm )Ln = Lk (Lm Ln ). The antisymmetric Lie product of two operators defined by their
commutator is not associative (the Jacobi identity tells us that) but what is relevant here is the
product used to define the commutator, not the Lie product defined by the commutator.

– 54 –
to deduce that [Exercise]

[(m − n)A(k) + (n − k)A(m) + (k − m)A(n)] δm+n+k = 0 . (5.55)

Now set k = 1 and m = −n − 1 (so that m + n + k = 0) to deduce that

(n + 2)A(n) − (2n + 1)A(1)

A(n + 1) = n ≥ 2. (5.56)
n−1
This is a recursion relation that determines A(n) for n ≥ 3 in terms of A(1) and A(2),
so there are two independent solutions of the recursion relation. You may verify that
A(m) = m and A(m) = m3 are solutions, so now we have

[Lm , L−m ] = 2mL0 + c1 m + c2 m3 , (5.57)

for some constants c1 and c2 . Observing that (m > 0)

h0| [Lm , L−m ] |0i = h0|Lm L−m |0i = ||L−m |0i||2 , (5.58)

and that
1 p2
h0|L0 |0i = α02 = , (5.59)
2 2πT
we deduce that  2
2 p
||L−m |0i|| − m = c1 m + c2 m3 . (5.60)
πT
We can now get two equations for the two unknown constants (c1 , c2 ) by evalu-
ating ||L−m |0i||2 for m = 1 and m = 2. Using (5.45) we find that

1 pm pn p2
||L−1 |0i||2 = h0|p · α1 p · α−1 |0i = h0|α1m α−1
n
|0i = , (5.61)
πT πT πT
and that
  
2 1 2 1 2
||L−2 |0i|| = h0| α0 · α2 + α1 α0 · α−2 + α−1 |0i
2 2
1 1
= h0|p · α2 p · α−2 |0i + h0|α12 α−1
2
|0i
πT 4
2p2 D
= + , (5.62)
πT 2
from which we see that
p2 2p2 D
||L−1 |0i||2 − = 0, ||L−2 |0i||2 − = , (5.63)
πT πT 2
and hence that
D D
c1 + c2 = 0 , c1 + 4c2 = ⇒ c2 = −c1 = . (5.64)
4 12

– 55 –
Inserting this result into (5.57), we have
D
m3 − m ,


[Lm , L−m ] = 2mL0 + (5.65)
12
and hence that
D
m3 − m δm+n ,


[Lm , Ln ] = (m − n) Lm+n + (5.66)
12
where 
1 k=0
δk = (5.67)
6 0.
0 k=
This is an example of the Virasoro algebra. In general, the Virasoro algebra
takes the form
c
m3 − m δm+n ,


[Lm , Ln ] = (m − n) Lm+n + (5.68)
12
where c is the central charge. In the current context we get this algebra with c = D.

5.2.1 The Virasoro constraints

We have just seen that without breaking manifest Lorentz covariance, it is not possi-
ble to impose the physical state conditions that we would need to impose to eliminate
all unphysical degrees of freedom. In view of our light-cone gauge results, this should
not be too much of a surprise. We saw that the level-one states of the open string
are the polarisation states of a massless spin-1 particle, but a manifestly covariant
description of a massless spin-1 particle necessarily involves unphysical degrees of
freedom.
This argument suggests that we should aim to impose weaker conditions, leaving
unphysical states that will have to be dealt with in some other way. The strongest
constraints that we can consistently impose are the Virasoro constraints:

(Ln − aδn ) |Ψi = 0 ∀n ≥ 0 . (5.69)

Here we allow for the fact that the operator L0 appearing here could differ by a
constant from the one defined in (5.53). We shall refer to states satisfying these
conditions as “Virasoro allowed”. Notice that for any two Virasoro allowed states
|Ψi and |Ψ0 i,
hΨ0 | (Ln − aδn ) |Ψi = 0 ∀n , (5.70)
because it follows from α−k = αk† that

L−n = L†n . (5.71)

Notice too that the oscillator vacuum |0i is Virasoro allowed, because of (5.44), with
p2 = 2πT a; i.e. α0 M 2 = −a, just as we found in light-cone gauge. We may construct

– 56 –
a basis for the oscillator Fock space by the successive actions of creation operators
α−k on |0i. As for the light-cone case, we can organise the basis states according
to their level number N , which is now the eigenvalue of the covariant level number
operator. In a basis for which the D-momentum operator is also diagonal, with
D-vector eigenvalues p, the mass-shell condition for a state |ΨN i at level N is

p2 + M 2 = 0 , α0 M 2 = N − a . (5.72)

Let’s consider the first few levels. The only N = 0 state in the oscillator Fock space
is |0i; this is a scalar with α0 M 2 = −a.
The general N = 1 state is
m
Am (p) α−1 |0i , p2 = 2πT (a − 1) . (5.73)

This is obviously annihilated by Ln for n > 1 and by L1 if

0 = L1 (A · α−1 )|0i = A · α0 |0i ⇒ p · A = 0. (5.74)

So the level-1 state (5.73) is Virasoro allowed iff p · A = 0. Its norm-squared is

kA · α−1 |0ik2 = Am An h0|α1m α−1

n
|0i = η mn Am An ≡ A2 . (5.75)

This could be positive, negative or zero according to the value of the constant a:

• a > 1. In this case M 2 < 0, so p is spacelike. In a frame where p = (0; p, 0)

the constraint p · A = 0 is equivalent to A1 = 0, so the general allowed level-1
state in this frame is
0
|Ψ1 i = A0 α−1 |0i + A · α−1 |0i , |||Ψ1 i||2 = −A20 + |A|2 . (5.76)

The state with A = 0 has negative norm; it is a “ghost”. This implies a

violation of unitarity (non-conservation of probability); we should not allow
a > 1.

• a < 1. In this case M 2 > 0, so p is timelike. In a frame where p = (p, ~0), the
constraint p · A = 0 implies that A0 = 0, so the general allowed level-1 state in
this frame is
~·α
|Ψ1 i = A ~ −1 |0i , |||Ψ1 i||2 = |A|
~ 2. (5.77)
Now, all non-zero allowed states have positive norm. There are (D − 1) inde-
pendent such states, exactly the number required for a massive spin-1 particle.
There is nothing unphysical about this but it doesn’t agree with the light-cone
gauge result. Covariant quantization with a < 1 breaks the conformal gauge
invariance of the classical NG string, resulting in an additional level-1 state that
was absent classically. This leads to difficulties in constructing an interacting
theory33 .
33
Attempts to do this go by the name of “non-critical string theory”.

– 57 –
 • a = 1. In this case M 2 = 0, so p is null, which agrees with the light-cone gauge
result. In a frame where p = (1; 1, 0) the constraint p · A = 0 implies that
A0 = A1 ; equivalently A− = 0. The general allowed level-1 state is then
+


|Ψ1 i = A+ α−1 + A · α−1 |0i . (5.78)

Now we get

k|Ψ1 ik2 = h0| A+ α1+ + A · α1 +

+ A · α−1 |0i = |A|2



A+ α−1 (5.79)
+
There are no ghosts, but α−1 |0i is a state of zero norm; i.e. a null state. In
fact, it is orthogonal to all allowed states, including itself, as the equation above
shows.
Although the number of independent allowed level-1 states is (D − 1) we may
identify any two states that differ by the addition of some multiple of the null
+
state α−1 |0i; in other words we consider the equivalence class of allowed states
defined by the equivalence relation
+
A · α−1 |0i ∼ A · α−1 |0i + A+ α−1 |0i (5.80)

for any A+ . The dimension of the space of these equivalence classes is (D − 2)

+
because the basis state α−1 |0i is now equivalent to the zero state. If physical
states are defined in this way we get agreement with the light-cone gauge.
What we are finding here is essentially the Gupta-Bleuler quantization of elec-
trodynamics (in D space-time dimensions).

Let’s now look at the level-2 states assuming that the level-1 states are massless;
i.e. assuming a = 1. These states have α0 M 2 = 1 and the general such state is
m n m


|Ψ2 i = Amn α−1 α−1 + Bm α−2 |0i . (5.81)

This is trivially annihilated by Lk for k > 2. However, L1 |Ψ2 i = 0 imposes the

condition
Bn = −α0m Amn , (5.82)
and L2 |Ψ2 i = 0 imposes the further condition

η mn Amn = −2α0 · B . (5.83)

This means that only the traceless part of Amn is algebraically independent, so the
dimension of the Virasoro-allowed level-2 space is
 
1 1
D(D + 1) − 1 = D(D − 1) − 1 + D (5.84)
2 2

– 58 –
The dimension is D larger than the physical level-2 space that we found from light-
cone gauge quantization (these formed the irreducible traceless symmetric rank-2
tensor of the SO(D − 1) rotation group). However, equivalence with the light-cone
gauge results is still possible if there are sufficient null states, and no ghosts.
To analyse this we need to consider the norm-squared of |Ψ2 i, which is

|||Ψ2 i||2 = 2Amn Amn + 2B 2 . (5.85)

Then we need to consider the implications for this norm of (5.82) and (5.83). We will
not carry out a complete analysis; the final result is that there are no ghosts only if
D ≤ 26 and then there are sufficient null states for equivalence with the light-cone
gauge results iff D = 26.
It is simple to see that there are ghosts if D > 26, and null states if D = 26.
Consider the special case of (5.81) for which

Amn = η mn + k1 α0m α0n , B m = k2 α0m . (5.86)

This gives us the Lorentz scalar state

α−1 + k1 (α0 · α−1 )2 + k2 α0 · α−2 |0i .

 2 
(5.87)

The conditions (5.82) and (5.83) determine the constants (k1 , k2 ) to be

D+4 D−1
k1 = , k2 = , (5.88)
10 5
and then one finds that the norm-squared is (see Q.III.2)
2
− (D − 1) (D − 26) . (5.89)
25
This is negative for D > 26, so the state being considered is a ghost. To avoid ghosts
we require D ≤ 26.
For D < 26 the scalar state (5.87) has positive norm, and this implies that
there is a physical scalar field at level 2, in disagreement with the result of light-
cone gauge quantization. If fact, for D < 26 one finds not only an additional scalar
but also an additional (massive) vector; these account for the increase by D in the
dimension of the space of level-1 physical states in comparison to the light-cone
gauge. Quantization has broken the gauge invariance of the classical theory, thereby
introducing additional degrees of freedom that have no classical analog.
Finally, if D = 26 the scalar state (5.87) is null; it is also orthogonal to all other
allowed states (as a further calculation shows). The same is true of the additional
(D − 1)-vector state, so if we define physical states as equivalence classes (in the way
described for level 1) then we recover the level-2 results of the light-cone gauge.
It can be shown that provided a = 1 and D = 26, the results of the light-cone
gauge are recovered in this way at all higher levels. This is the “no-ghost theorem”
of string theory.

– 59 –
5.3 Conformal invariance and Vertex operators
The Virasoro constraints have the following interpretation: a physical state is a high-
est weight state for an irreducible representation of the Virasoro algebra of conformal
weight a. As we have seen, Lorentz invariance of the quantum open string requires
a = 1.
We shall focus on the closed string. In this case we have two sets of Virasoro
conditions, so the physical states are highest weight states with weights (1, 1). This
statement can be transferred from states to operators. We shall seek those operators
that create a given state in the string spectrum by their action on the following string
state representing a string at rest in its ground state:

|p = 0i ⊗ |0i . (5.90)

These “vertex operators” should have conformal dimension (1, 1), but what does this
mean?
Let us first consider this question at the classical level, in the context of the
conformal gauge action
Z
I = T dσ + dσ − ∂− X · ∂+ X . (5.91)

This action is invariant under conformal isometries of the worldsheet Minkowski

metric, for which the algebra is Diff1 ⊕ Diff 1 . The first Diff1 acts on functions of
σ − and the second acts on functions of σ + , but more generally these “functions”
are one-component tensors that transform in the following way: if A± is a function
of σ ± and we consider new coordinates σ̌ ± then the new “function” Ǎ± of the new
coordinates is  ± h±
± dσ
Ǎ± (σ̌ ) = ±
A± (σ ± ) , (5.92)
dσ̌
for some number h± called the conformal dimension (or weight). More precisely,
we should say that A− has conformal dimensions (h− , 0) and A+ has conformal
dimensions (0, h+ ). Their product A− A+ has conformal dimensions (h− , h+ ).
The infinitesimal form of (5.92), found by setting σ̌ ± = σ ± − ξ ± (σ ± ), is

δξ± A± (σ ± ) = ξ ± ∂± A± + h± (∂± ξ ± )A± . (5.93)

The h± = 1 case is special because in this case

δξ± A± = ∂± ξ ± A± ,


(5.94)

and hence, since ∂± A∓ = 0, the product A+ A− of conformal dimension (1, 1) varies

into a total derivative. This means that we can add A+ A− as a perturbation to
the conformal gauge Lagrangian without destroying the conformal invariance of the
action; this is an important property that we will exploit later.

– 60 –
 The transformation law (5.93) can be re-expressed as PB relations between Diff1
generators and the Fourier coefficients of A± (see Q.III.4). For A− , for example

{Lm , An }P B = −i [m(h − 1) − n] Am+n , (h ≡ h− ) (5.95)

The corresponding PB relation of L̃m with the Fourier coefficients Ãn of A+ is the
same but with h̃ ≡ h+ replacing h. As a check, recall that αk are the Fourier
components of T ∂− X (which is what P − T X 0 becomes upon using the conformal
gauge relation P = T Ẋ). and also that

{Lm , αn }P B = inαn+m , (5.96)

which confirms that h = 1 in this case. As a further check, recall that

{Lm , Ln }P B = −i(m − n)Lm+n , (5.97)

which tells us that h = 2 for Θ−− because {Ln } are the Fourier coefficients of
Θ−− = T (∂− X)2 . This is expected because the conformal dimension of a product is,
classically, the sum of the conformal dimensions of the factors of that product.
So far, we have discussed the classical meaning of conformal dimension for a
class of worldsheet tensors. The quantum version is similar; for example, an operator
Â− (σ− ) with Fourier coefficients Ân is a “conformal primary” operator of conformal
dimension h = h (and  (σ ) with Fourier coefficients à ˆ is a “conformal primary”
− + + n
operator of conformal dimension h+ = h̃) if
h i
L̂m , Ân = [m(h − 1) − n] Âm+n ,
h i h i
L̃m , Ãn = m(h̃ − 1) − n Ãm+n . (5.98)

For example, ∂− X̂ is a conformal primary operator of conformal dimension h = 1

because h i
L̂m , α̂m = −nα̂n+m . (5.99)

However, the operator Θ̂−− is no longer a conformal primary operator because of

the central charge in the Virasoro algebra (its quantum conformal dimension is not
defined).
What about X? For the closed string in the conformal gauge, the equations of
motion have the solution
 p 
x(t) = x(0) + t, αn (t) = e−int αn (0) , α̃n (t) = e−int α̃n (0) . (5.100)
4πT
Using these relations in (3.103), we find that

X(t, σ) = XR (σ − ) + XL (σ + ) , (5.101)

– 61 –
where
pσ −
 
− 1 i X 1 −inσ−
XR (σ ) = x(0) + +√ e αn (0) ,
2 4πT 4πT n6=0 n
pσ +
 
+ 1 i X 1 −inσ+
XL (σ ) = x(0) + +√ e α̃n (0) . (5.102)
2 4πT 4πT n6=0 n

However, although X is periodic in σ, the functions XR (σ − ) and XL (σ + ) are not

periodic (because of the term linear in σ ± ). For this reason, the functions XR and
XL are not conformal tensors on the cylindrical worldsheet of the closed string; they
do not have the right transformation properties. However, this difficulty disappears
if we consider eik·X , for some Lorentz vector k:
− +)
eik·X = eik·XR (σ ) eik·XL (σ (5.103)

Both factors are periodic and hence well-defined conformal tensors on the cylindrical
worldsheet; both of conformal dimension zero, so eik·X has conformal dimensions
(0, 0). However, this is a classical result.
In the quantum theory we first need to define the operator that replaces eik·X
by choosing an ordering of the operator products in eik·X̂ . We first define (now
omitting hats on operators)

V (k) = (V< )V0 (V> ) , Ṽ (k) = (Ṽ< )V0 (Ṽ> ) , (5.104)

where
−
!
1 X e−inσ
V< = exp − √ k · αn (0) ,
4πT n<0 n
!
−inσ −
1 X e
V> = exp − √ k · αn (0) , (5.105)
4πT n>0 n

and similarly for Ṽ< and Ṽ> with σ + replacing σ − and α̃(0) replacing α(0). We can
now define an operator V(k) replacing the classical eik·X by

V(k) = (Ṽ< )V (k)(Ṽ> ) ≡ (V< )Ṽ (k)(V> ) . (5.106)

k 2
• Proposition: V has conformal dimensions h± = 8πT . The proof will be
sketched shortly. It will be seen there that this is a purely quantum effect
since the classical result is h± = 0.

• Corollary: V(k) has conformal dimensions (1, 1) if k 2 = 8πT , which is pre-

cisely the value of k 2 for the closed string tachyon. This suggests that V(k)|p =
0i ⊗ |0i could be related to the closed string tachyon state. We shall now in-
vestigate this.

– 62 –
 The motivation behind the definition of V(k) is that all the annihilation operators
are in V> and Ṽ> , so that
V> |0i = |0i , Ṽ> |0i = |0i . (5.107)
Moreover (exercise)
k2
 
p̂
ik·[x̂(0)+ 4πT ] i t ik·x(0)
V0 |p = 0i = e =e 8πT
e |p = 0i = eiω(k)t |ki , (5.108)
where
k2
ω(k) = . (5.109)
8πT
We see that
V(k)|p = 0i ⊗ |0i = eiω(k)t |ki ⊗ (V< )|0iR ⊗ (Ṽ< )|0iL . (5.110)
This is not yet the tachyon state when k 2 = 8πT because of the operator V< , but if we
replace t by (1 − i)t for arbitrarily small positive number  then V< → 1 as t → −∞.
In other words, with this prescription the tachyon state |ki ⊗ |0i represents a tachyon
in the far past, which then propagates into the far future, so that the worldsheet is
a cyclinder that is infinite in both directions.
By taking t → (1 − i)t we have implicitly supposed that all worldsheet fields
are restrictions of functions of a complex variable t to the real axis in the complex t-
plane, and we have just infinitesimally rotated the line on which we initially restricted
these fields. We could rotate by −π/2 instead of −. This is equivalent to setting
t = −iτ and then taking τ (the “imaginary time”) to be real; this is analogous to
“Wick-rotation” in QFT. If we do this then
σ − = −iw , σ + = −iw̄ ; w = τ − iσ , (5.111)
We have now replaced the real coordinates on the cylindrical closed string worldsheet
by a complex coordinate w, subject to the identification
w ∼ w + 2πi , (5.112)
and the Minkowski worldsheet metric −dt2 + dσ 2 ≡ −dσ − dσ + has become the Eu-
clidean worldsheet metric dτ 2 + dσ 2 ≡ dwdw̄. The conformal invariance of our
conformal gauge string action will allow us to conformally map the cyclinder to the
complex plane with complex coordinate
z = ew = eτ eiσ . (5.113)
Notice that circles in the z-plane represent the closed string at a fixed “imaginary
time” τ (it is actually real after the Wick rotation). Also the “far past” becomes a
neighbourhood of the origin z = 0. We can now write V< as
( )
−n
1 X z
V< = exp − √ k · αn (0) → 1 as z → 0 , (5.114)
4πT n<0 n

– 63 –
and similarly for Ṽ< . At z = 0 we have V< = 1 and Ṽ< = 1, so

V(k)|z=0 |p = 0i ⊗ |0i = eiω(k)t |ki ⊗ |0i . (5.115)

This is a phase times the closed string tachyon state provided that k 2 = 8πT , but
(by the above proposition) this is precisely the condition for V(k) to have conformal
dimensions (1, 1).
The operator V(k) for k 2 = 8πT is called the tachyon vertex operator. Now
consider the operator (m, n are Lorentz indices here)

∂− X m ∂i X n hmn (k) , hmn (k) = cmn V(k) . (5.116)

for constants cmn . As ∂− X m ∂+ X n cmn has conformal dimensions (1, 1) (which is

also its classical conformal dimension) we should expect to have to choose k such
that V(k) has conformal dimensions (0, 0). According to the proposition above, this
requires k 2 = 0. An analysis similar to the one above for the tachyon shows that this
operator acting on the state representing a string at rest in its ground state yields a
level one state of the closed string provided that one takes z → 0 and imposes the
conditions
k m hmn = 0 , (5.117)
which also follow by requiring that the level 1 state satisfy the Virasoro conditions
(this is the closed string analog of the k m Am = 0 constraint that we we derived from
the Virasoro conditions imposed on a level-1 state of the open string). Notice that
in this level-1 case all operators in the level 1 vertex operator (which include the
graviton, dilaton and notoph vertex operators) have their classical dimension.
More generally, every state of the string corresponds to the action, at z = 0,
of a vertex operator of conformal dimensions (1, 1) on the string ground state, and
there are no other operators of conformal dimensions (1, 1). The classical conformal
dimensions of vertex operators for states at level N is (N, N ), so it is only at level-1
that the classical dimension matches the quantum dimension.

5.3.1 Computation of anomalous dimension of V(k)

• Starting from the expressions of (5.105), show that
" #
1 X inσ
−i∂− V< = (V< ) √ e k · αn ,
4πT n<0
" #
1 X inσ
−i∂V> = √ e k · αn V> (5.118)
4πT n>0

and  
1 k · α0 k · α0
−i∂− V0 = V0 √ +√ V0 . (5.119)
2 4πT 4πT

– 64 –
 Thereby deduce that
   
1 X X
−i∂− V (k) = √ V<  einσ k · αn V0 + V0  einσ k · αn V> (5.120)
2 4πT n∈Z n∈Z

• Now show that

!
1 X
[Lm , V> ] = √ einσ k · αm+n V>
4πT n>0
k · αm
[Lm , V0 ] = √ V0 . (5.121)
4πT

These are exactly the results expected from the corresponding PB relations, so
they are essentially classical. Because of the need to re-order operators, there
is a deviation from the classical result in the commutator
!  2 
1 X k
[Lm , V< ] = √ (V< ) inσ
e k · αm+n + (m − 1) e−imσ V< (5.122)
4πT n<0
8πT

The first term is classical and the last term is quantum.

Combining these results one finds that
 
 2 
1 X k
[Lm , V (k)] = √ (V< )  inσ
e k · αm+n  V0 V> + (m − 1) e−imσ V
4πT 8πT
n∈Z
(5.123)
We now use !
X X
inσ −imσ inσ
e k · αm+n = e e k · αn (5.124)
n∈Z n>0

to get
   
2
 
1 X k
[Lm , V (k)] = e−imσ  √ V<  einσ k · αn  V0 V> + (m − 1) V
4πT 8πT
n∈Z
    
 1 X X
= e−imσ √ V<  einσ k · αn V0 + V0  einσ k · αn  V>
 2 4πT
n∈Z n∈Z
 2  
k
m V
8πT
  2 
−imσ k
= −ie ∂− + im V. (5.125)
8πT

– 65 –
 • From the above result, and the observation that the Fourier coefficients of V (k)
are I
Vn = dσe−inσ V , (5.126)

one deduces that

k2
[Lm , Vn ] = [m(h − 1) − n] Vm+n , h= (5.127)
8πT

A similar calculation can be done for Ṽ (k), so we conclude that the conformal
dimensions of V(k) are as stated in the proposition.

5.3.2 Looking ahead

The worldsheet of a free string has been turned into a complex surface, with the
topology of a cylinder, by “Wick-rotation”, and this cylinder has been mapped to
the complex z-plane. The origin of this plane represents an incoming string, in some
quantum state. A basis of these states is provided by the level-N states of the free
string, and a choice of any one of these basis states corresponds to “insertion” of
the corresponding vertex operator at the origin of the z-plane. There is another
insertion at z = ∞ corresponding to the outgoing particle, so the worldsheet is
really the Riemann sphere with two marked points at which vertex operators are
inserted, but we could insert more vertex operators at other points. In this way, we
can compute amplitudes for scattering of string states from a computation of the
ground state expectation value of a product of vertex operators at any point on the
sphere. A computation of this type with 4 tachyon vertex operators (which is the
simplest non-trivial case) yields the Virasoro-Shapiro amplitude for the scattering of
two tachyons, but we are going to derive this later by a path-integral method.
For open strings, the Riemann-sphere Euclidean worldsheet is replaced by the
unit disc, with the boundary being the union of the worldlines of the two ends of
the string. The incoming and outgoing strings now correspond to insertions of open
string vertex operators on this boundary. The simplest non-trivial amplitude is found
from an insertion on the disc boundary of four open-string tachyon vertex operators,
and this gives the Veneziano amplitude.
We shall not pursue this operator approach here. Instead we turn to path inte-
grals, with the aim of arriving at the Virasoro-Shapiro amplitude in this way. But
first we need to see how the path-integral formulation of QM applies to the relativistic
particle.

6. Interlude: Path integrals and the point particle

Let A(X) be the quantum-mechanical amplitude for a particle to go from the origin
of Minkowski coordinates to some other point in Minkowski space-time with cartesian

– 66 –
coordinates X. As shown by Feynman, A(X) has a path-integral representation. In
the case of a relativistic particle of mass m, with phase-space action I[x, p; e] we have
Z Z
A(X) = [de] [dxdp] eiI[x,p;e] , x(0) = 0 , x(1) = X . (6.1)

Here we are parametrising the path such that it takes unit parameter time to get
from the space-time origin to the space-time point with coordinates X. The integrals
have still to be defined, but we proceed formally for the moment.
An immediate problem is the gauge invariance of the action, which implies that
we are integrating over too many variables. We can solve this problem by gauge
fixing, but we don’t want to break manifest Lorentz covariance. For the string we
chose the conformal gauge; the particle analog is
e(t) = 1 (?) (6.2)
However (as for the conformal gauge) one cannot make this choice everywhere on the
R1
worldline. This can be seen from the fact that 0 edt is gauge invariant; in fact, it’s
proportional to the elapsed proper time, which could be any positive number whereas
R1
0
dt = 1. We could choose to set e(t) = 1 everywhere except near t = 0, in which
case the mass-shell constraint would become an initial condition (as discussed for the
string in conformal gauge). Alternatively, we can set
e(t) = s (6.3)
for a “variable constant” s. The action then becomes
Z 1
s 1
Z
m
dt p2 + m2 .


I → Is [x, p] = dt ẋ pm − (6.4)
0 2 0
Variation of this action with respect to the constant s yields the integrated constraint
R
dt (p2 + m2 ) = 0, but the integrand is a constant of the motion so the equations of
motion will imply that p2 + m2 = 0 everywhere on the worldline. We now have the
simpler path integral34
Z ∞ Z
A(X) = ds [dxdp] eiIs [x,p] . (6.5)
0

This is not quite right (although close enough for some purposes) because we have
not done the gauge fixing properly; we’ll return to this point later.
We have to integrate over functions x(t) and p(t). Let us first write these func-
tions as
x(t) = Xt + q(t) , q(0) = q(1) = 0
Z 1
p(t) = P + π(t) , dt π(t) = 0 , (6.6)
0
34
We have assumed that e > 0 (by assuming that s > 0). We do this because the worldline time-
reversal invariance of the action is an additional discrete diffeomorphism, which we fix by requiring
e > 0.

– 67 –
so q satisfies simpler boundary conditions and P is the average D-momentum. We
now find that Z ∞ Z
m s 2 2
A(X) = ds dDP eiX Pm −i 2 (P +m ) F (s) (6.7)
0
where Z  Z 1  
m 1 2
F (s) = [dqdπ] exp i dt q̇ πm − sπ . (6.8)
0 2
The functions q and π can be expressed as Fourier series:
∞ ∞
X xk X
q(t) = sin(kπt) , π(t) = pk cos(kπt) . (6.9)
k=1
kπ k=1

This gives us
∞ Z Z  h 
Y
D D i s 2i
F (s) = const. d xk d pk exp xk · p k − p
k=1
2 2 k
∞ Z Z   
Y
D D i 1 2
= const. d uk d vk exp uk · vk − v (6.10)
k=1
2 2 k
√ √
where we have set xk = s uk and pk = vk / s to arrive at the second equality. This
shows that F (s) is actually independent of s. It is just an infinite constant, which
we can absorb into the definition of the normalisation of A(X), which is now
Z Z ∞
s
iX m Pm 2 2
A(X) = const. d P eD
ds e−i 2 (P +m ) . (6.11)
0

We are now going to view the s integral as a contour integral in the complex s-plane.
We will rotate the contour to the negative imaginary axis and then set s = −is̃ for
real s̃. This gives us
Z Z ∞
s
iX m Pm 2 2
D
A(X) = const. d P e ds̃ e− 2 (P +m )
0
Z iXP
e
= const. dDP 2 , (6.12)
P + m2
which is the Feynman propagator.

6.1 Faddeev-Popov determinant

The problem with the formula (6.1) is that, because of gauge-invariance, we are
integrating over too many functions. We circumvented this problem by imposing the
gauge condition e = s, effectively by inserting a delta functional δ[e − s] into the
R
[de] functional integral. But is this procedure correct?
We can express any function e(t) as a gauge transform of e = s:

e(t) = es (t) ≡ s + α̇(t) , α(1) = α(0) (6.13)

– 68 –
where the parameter α(t) is a map from the worldline to the gauge group (abelian
and 1-dimensional in this case). The restriction on α(t) arises because we insist that
R1
s = 0 dte(t). The problem with the formula (6.1) is that we are implicitly including
a functional integral over α(t) that gauge invariance would allow us to omit if it were
explicit. We can make it explicit by writing
Z Z ∞ Z Z ∞ Z
[de] = ds [de0 [α]] = ds [dα] ∆F P , (6.14)
0 0

where ∆F P is the Jacobian for the change of variables from e0 [α](t) to α(t):
   
δe0 [α](t) δ α̇(t)
∆F P = det = det = det [δ 0 (t − t0 )] (6.15)
δα(t0 ) δα(t0 )
This functional Jacobian is called the Faddeev-Popov determinant. Provided that
the functional integrand is gauge invariant, the [dα] functional integral will produce
only an infinite constant that can be absorbed into the undetermined normalization
of the path integral for the amplitude A[X], which now takes the corrected form
Z ∞ Z
A(x) = ds ∆F P [dxdp] eiIs [x,p] . (6.16)
0

This replaces (6.5), from which the ∆F P factor was missing. That’s why (6.5) was
“not quite right” but in this simple case the ∆F P factor is independent of any of the
other variables that we have to integrate over, so it can only change the normalization
of A(X) which anyway depends on the detailed definitions of the path integral35 .
However, it is essential to take into account the FP determinant for generic gauge
theories; in particular, it is very important to the path-integral quantization of the
NG string, which we will get to soon.

6.2 Fadeev-Popov ghosts

Let (bi , ci ) (i = 1, . . . , n) be n pairs of anticommuting variables. This means that

bi , cj = 0 ,
  i j
{bi , bj } = 0 , c , c = 0 ∀ i, j = 1, . . . , n , (6.17)

where {, } means anticommutator: {A, B} = AB + BA. Any function of anticom-

muting variables has a terminating Taylor expansion because no one anticommuting
variable can appear twice. Consider the n = 1 case

f (b, c) = f0 + bf1 + cf−1 + bcf˜0 , (6.18)

where (f0 , f±1 , f˜0 ) are independent of both b and c. Then

∂ ∂ ∂
f = f1 + cf˜0 ⇒ f = f˜0 . (6.19)
∂b ∂c ∂b
35
It is for this reason that the FP determinant can be ignored, for most purposes, in the path-
integral approach to QED.

– 69 –
Essentially, a derivative with respect to b removes the part of f that is independent of
b and then strips b off what is left. However, we should move b to the left of anything
else before stripping it off; this is equivalent to the definition of the derivative as a
“left derivative”. Using this definition we have
∂ ∂ ∂
f = f−1 − bf˜0 ⇒ f = −f˜0 . (6.20)
∂c ∂b ∂c
There is minus sign relative to (6.19) because we had to move c to the left of b. This
result shows that  
∂ ∂
, = 0. (6.21)
∂b ∂c
That is, partial derivatives with respect to anticommuting variables anti-commute. In
particular, since a function of anticommuting variables is necessarily linear in any
one of them,  2  2
∂ ∂
= 0, = 0. (6.22)
∂b ∂c
We can also integrate over anticommuting variables. The (Berezin) integral over
an anticommuting variable is defined to be the same as the partial derivative with
respect to it. Consider, for example, the Gaussian integral
Z   
n n bi M i j cj ∂ ∂ ∂ ∂ bi M i j cj
d bd c e = ··· · · · e . (6.23)
∂bn ∂b1 ∂cn ∂c1
If we expand the integrand in powers of bM c the expansion terminates at the nth
term, which is also the only term that contributes to the integral because it is the
only one to contain all bi and all ci . Because of the anti-commutativity of the partial
derivatives, we then find that
Z
i 1
dn c dn b ebi M j cj ∝ εi1 ···in M i1 j1 · · · M in jn εj1 ···jn = det M (6.24)
n!
We can use a functional variant of this result to write the FP determinant as a
Gaussian integral over anticommuting “worldline fields” b(t) and c(t):
Z Z Z 
0 0 0 0 0
det [δ(t − t )∂t ] = [db dc] exp dt dt b(t ) [δ (t − t )] c(t)
Z  Z 
= [db dc] exp − dt bċ . (6.25)

The anticommuting worldline fields are known collectively as the FP ghosts, although
it is useful to distinguish between them by calling c the ghost and b the anti-ghost.

N.B. There is no relation between the FP ghosts and the ghosts that
appear in the NG string spectrum for D > 26. The same word is being
used for two entirely different things!

– 70 –
 Using (6.25) in the expression (6.16) we arrive at the result
Z ∞ Z Z
A(X) = ds [dxdp] [db dc] eiIqu , (6.26)
0

where the “quantum” action is

Z
s 2
Iqu = dt {ẋm pm + ibċ − Hqu } , p + m2 .


Hqu = (6.27)
2

We now have a mechanical system with an extended phase-space with additional,

anticommuting, coordinates (b, c). The factor of i multiplying bċ is needed by the
convention that a product of two “real” anticommuting variables is “imaginary”. Of
course, an anticommuting number cannot really be real; it is “real” if we declare it
to be unchanged by complex conjugation.

6.2.1 FP ghosts for systems with Lie algebra constraints

Now we consider how these ideas apply to the more general mechanical system with
action Z
I[q, p; λ] = dt q̇ I pI − λi ϕi , {ϕi , ϕj }P B = fij k ϕk .

(6.28)

For simplicity, we shall assume that the structure functions fij k are constants, in
which case they are the structure constants of a Lie algebra, and hence subject to
the constraint (following from the Lie algebra Jacobi identity)

f[ij p fk]p q = 0 . (6.29)

We shall suppose that we have chosen to gauge fix (perhaps only partially) by
imposing the condition
λi = λ̄i (constants). (6.30)
The FP operator is then found by varying the gauge transformation of the gauge-
fixing function (λi − λ̄i ) with respect to the gauge parameter, so

δ λi (t0 ) 
 
∆F P = det , δ λi = ˙i − j λ̄k fkj i . (6.31)
δj (t) λi =λ̄i

This gives
Z
0
δji ∂t k i
[dbdc] eiIF P [b,c] ,


∆F P = det − λ̄ fkj δ(t − t ) ∝ (6.32)

where the FP action is

Z
dt ibi ċi − cj λ̄k fkj i .
  
IF P [b, c] = (6.33)

– 71 –
This must be added to the original action to get the “quantum action”
Z
Iqu = dt q̇ I pi + ibi ċi − Hqu , Hqu = λ̄k Φk ,

(6.34)

where
(gh) (gh)
Φk = φk + φk , φk = icj fjk i bi . (6.35)
The reason for writing the “quantum Hamiltonian” this way will become clear shortly.
We now have an action for a mechanical system with an extended phase space
(actually a superspace) for which some coordinates are anticommuting. Leaving aside
the Hamiltonian, the action can be constructed from the following closed 2-form on
this space:
Ω = d pm dxm + ibi dci = dpm ∧ dxm + idbi ∧ dci .


(6.36)
This is an “orthosymplectic” 2-form because the anticommutativity of the FP ghosts
means that36
dbi ∧ dci = dci ∧ dbi . (6.37)
This leads to a Poisson bracket for the new canonical variables bi and ci that is
symmetric rather than antisymmetric:

bi , cj P B = cj , bi P B = −iδij .
 
(6.38)

The only other non-zero PB is the usual {q I , pJ }P B = δJI . Using these PB relations
one finds that n o
(gh) (gh) (gh)
φi , φj = fij k φk , (6.39)
PB
and hence that
{Φi , Φj }P B = fij k Φk . (6.40)
Thus, the new functions Φi , defined on the extended phase space, satisfy the same
PB relations as the original constraint functions ϕi .
Recall that the gauge invariance is completely fixed when the equation δ λi = 0
has no non-zero solutions for . In this case, the operator appearing in the FP ghost
action will be invertible, and the bi equation will imply that ci = 0, and vice-versa;
in other words, the FP ghosts can be trivially eliminated from the action. This is
precisely why they are not needed in light-cone gauge. They are needed only when
the gauge-fixing condition does not completely fix the gauge, leaving some residual
gauge invariance and hence unphysical degrees of freedom; these are “cancelled out”
(in a way made precise by the BRST formalism) by the FP ghosts. So let us now
suppose that δ λi = 0 allows non-zero solutions for the parameters i . These residual
gauge parameters satisfy
˙i = k λ̄j fjk i . (6.41)
36
The usual minus sign coming from the antisymmetry of the wedge product of 1-forms is cancelled
by the minus sign coming from changing the order of b and c.

– 72 –
The residual gauge transformation of the “quantum” Hamiltonian is

δ Hqu = λ̄j Φj , k Φk P B = k λ̄j fjk i = ˙k Φk ,


(6.42)

where the equality uses (6.41). The variation of −Hqu therefore cancels the usual
˙k Φk contribution from the rest of the Lagrangian.
The residual gauge transformations of the (q, p) variables are found as before,
and a similar calculation yields the residual gauge transformation of the FP ghost
variables, e.g.
n o
(gh)
δ ci = ci , k Φk P B = k ci , ϕk = k cj fjk i .

(6.43)
PB

Compare this with the transformation of λi (it is the same if one omits the ˙i term).
Similarly,
δ bi = bi , k Φk P B = k fik j bj .

(6.44)
Notice that, in general, bi and ci transform differently (as will be the case for the
string).

7. Path integrals and the NG string

We now aim to use the conformal gauge in a path-integral approach to quantisation
of the closed NG string. Recall that the conformal gauge for the NG string is λ± = 1,
and that in this gauge the gauge variation of λ± is

δξ λ± = 2∂∓ ξ ± . (7.1)

The FP determinant is therefore

2∂+ [δ(t − t0 )δ(σ − σ 0 )]

 
0
∆F P = det . (7.2)
0 2∂− [δ(t − t0 )δ(σ − σ 0 )]

Following the steps spelled out for the particle we arrive at the following FP-ghost
contribution to the action:
Z I n o
IF P = 2i dt dσ b ∂+ c + b̃ ∂− c̃ . (7.3)

Adding the FP action to the usual phase-space action, we get the “quantum” action
for the closed NG string in conformal gauge
Z I n o
Iqu [X, P ; b, c; b̃, c̃] = dt dσ Ẋ m Pm + ibċ + ib̃c̃˙ − Hqu ,
P2 T  
Hqu = + (X 0 )2 − i bc0 − b̃c̃0 . (7.4)
2T 2

– 73 –
From this action we can read off the PB relations; in particular
{b(σ), c(σ 0 )}P B = −iδ(σ − σ 0 ) = {b̃(σ), c̃(σ 0 )}P B . (7.5)
As explained previously for the generic mechanical model with (Lie algebra) first-class
constraints, this action is automatically invariant under the residual gauge invariance
that survives the conformal gauge; in other words, it is conformal invariant, with the
ghosts (c, c̃) transforming in the same way as the parameters under a composition
of two residual gauge transformations; this tells us that (c, c̃) will transform as the
light-cone components of a vector field.
δξ c = ξ − ∂− c − ∂− ξ − c , δξ c̃ = ξ + ∂+ c̃ − ∂+ ξ + c̃ .



(7.6)
Invariance of the action then requires (b, b̃) to transform as the light-cone components
of a (symmetric traceless) “quadratic differential ”:
δξ b = ξ − ∂− b + 2 ∂− ξ − b , δξ b̃ = ξ + ∂+ b̃ + 2 ∂+ ξ + b̃ .



(7.7)
The Noether charges are the Fourier coefficients of the two non-zero light-cone com-
ponents of the FP-ghost stress tensor
I I
(gh) −imσ (gh) (gh)
Lm = dσ e Θ−− , (gh)
L̃m = dσ eimσ Θ++ . (7.8)

These stress tensor components are

(gh)
Θ−− = i [2b ∂− c − c ∂− b] = −i [2b c0 + b0 c] ,
h i h i
(gh) 0 0
Θ++ = i 2b̃ ∂+ c̃ − c̃ ∂+ b̃ = i 2b̃ c̃ + b̃ c̃ , (7.9)

where the FP ghost equations of motion have been used for the second equalities.
We now pass to the Fourier-mode form of the action. In addition to the Fourier
series for P ± T X 0 , we will need the Fourier series expansions
X 1 X ikσ
c= eikσ ck , b= e bk ,
2π
k∈Z k∈Z
X 1 X
c̃ = e−ikσ c̃k , b̃ = e−ikσ b̃k . (7.10)
2π
k∈Z k∈Z

This gives us
X X
L(gh)
m = (m + k) bm−k ck , L̃(gh)
m = (m + k) b̃m−k c̃k . (7.11)
k∈Z k∈Z

For m = 0 we have
∞
X
(gh)
L0 = N(gh) ≡ k (b−k ck + c−k bk ) ,
k=1
X∞
(gh)
L̃0 = Ñ(gh) ≡ k(b̃−k c̃k + c̃−k b̃k ) . (7.12)
k=1

– 74 –
As we shall see, N(gh) and Ñ(gh) will have the interpretation as (covariant) ghost level
numbers.
The “quantum” action in terms of Fourier modes is
 
Z ∞
 X i X   
Iqu = dt ẋm pm + α̇k · α−k + α̃˙ k · α̃−k + i b−n ċn + b̃−n c̃˙ n − Hqu ,


 k k=1

n∈Z

Hqu = L0 + L̃0 (7.13)

where
1
L0 = L0 + N(gh) = α02 + Nqu , Nqu = N + N(gh) ,
2
1
L̃0 = L̃0 + N(gh) = α02 + Ñqu , Ñqu = Ñ + Ñ(gh) . (7.14)
2
We can now read off the PBs of the Fourier modes. For the new, anticommuting,
variables we have37
n o
{cn , b−n }P B = −i , c̃n , b̃−n = −i , (n ∈ Z). (7.15)
PB

Notice that n = 0 is included, although the (anti)ghost zero modes (b0 , c0 ) and (b̃0 , c̃0 )
do not appear in the Hamiltonian. Using these PB relations, one finds that
 (gh)  (gh)
Lm , cn P B = i (2m + n) cn+m , Lm , bn P B = −i (m − n) bm+n . (7.16)

This leads to
(gh)
L(gh) (gh)

m , Ln PB
= −i (m − n) Lm+n , (7.17)
(gh)
and similarly for L̃m .
The total “quantum” conformal charges are

Lm = Lm + L(gh)
m , L̃m = L̃m + L̃(gh)
m . (7.18)

These are the Fourier modes of the non-zero components of the energy-momentum
stress tensor of the quantum action (7.4), and their algebra is that of Witt ⊕ Witt:

{Lm , Ln }P B = −i (m − n) Lm+n
{Lm , L̃n }P B = 0
{L̃m , L̃n }P B = −i (m − n) L̃m+n . (7.19)

The PBs of the Lm with the Fourier modes of the various fields determine the trans-
formations of these fields under the residual conformal invariance of the conformal
gauge.
37
These PB relations are equivalent to (7.5).

– 75 –
 (h)
Recall that A(h) (σ − ) has conformal dimension (h, 0) if its Fourier coefficients An
satisfy
(h)
Lm , A(h)

n PB
= −i [m (h − 1) − n] An+m , (7.20)
and similarly for A(h̃) (σ + ) of conformal dimension (0, h̃). For instance, the PB rela-
tions

{Lm , αk }P B = ikαk+m ,
{Lm , ck }P B = i (2m + k) ck+m ,
{Lm , bk }P B = −i (m − k) bk+m . (7.21)

tell us that ∂− X has conformal dimension (1, 0) (as we knew). Let’s use the following
notation for this:
[∂− X] = (1, 0) , [∂+ X] = (0, 1) , (7.22)
which are the conformal dimensions of light-cone components a worldsheet one-form.
Compare this with
[c] = (−1, 0) , [c̃] = (0, −1) , (7.23)
which are the conformal dimensions of the light-cone components of a worldsheet
vector. Finally, h i
[b] = (2, 0) , b̃ = (0, 2) . (7.24)
These are the conformal dimensions of the light-cone components of a (symmetric
traceless) quadratic differential.
After elimination of P from the action (7.4), we get the generalisation of the
conformal gauge action (3.86):
Z n  o
2
Iqu = d σ 2T ∂+ X · ∂− X + i b ∂+ c + b̃ ∂− c̃ . (7.25)

From the above results for conformal dimensions, and taking into account that acting
with ∂− raises the conformal dimension by (1, 0) and acting with ∂+ raises it by (0, 1),
we see that all terms in the Lagrangian have conformal dimension (1, 1), as required
for conformal invariance of the action
This is all classical, but it carries over to the quantum theory with the same
changes that we encountered previously: central charge in the Virasoro algebra and
anomalous dimensions for vertex operators.

7.0.1 Critical dimension again

For the anticommuting FP ghosts, with Fourier modes satisfying the symmetric PB
relations (7.5), we adapt the usual {, }P B → −i[, ] rule to {, }P B → −i{, }. This
gives us the canonical anticommutation relations (we omit the hats on operators)
n o
{cn , b−n } = 1 , c̃n , b̃−n = 1 n ∈ Z. (7.26)

– 76 –
We define the (anti)ghost oscillator vacuum to be the state

|0igh = |0igh gh
R ⊗ |0iL , (7.27)

such that
cn |0igh
R = 0, c̃n |0igh
L = 0, n > 0, (7.28)
and
bn |0igh
R = 0, b̃n |0igh
L , n ≥ 0. (7.29)
Notice the asymmetry for n = 0. This is because {c0 , b0 } = 1, and we choose b0 to
be the annihilation operator.
We can act on |0igh with the (anti)ghost creation operators, (c−n , c̃−n ) for n ≥ 0
and (b−n , b̃−n ) for n > 0, to get states in an (anti)ghost Fock space; The full oscillator
vacuum is now the tensor product state38

|0) = |0i ⊗ |0igh , (7.30)

where |0i is the oscillator vacuum of the “old covariant” quantization procedure; i.e.
|0i = |0iR ⊗ |0iL .
(gh)
For m 6= 0 there is no ordering ambiguity in the operator versions of Lm and
(gh)
L̃m of (7.11). For m = 0 we take the ordering to be as given in (7.12). We then
have
gh gh
L(gh)
m |0iR = 0 , L̃gh
m |0iL = 0 , m ≥ 0. (7.31)
(gh)
We know that the classical Lm satisfy a Witt algebra with respect to PBs, so the
(gh)
same logic that we used previously for the Lm tells us that the quantum Lm satisfy
a Virasoro algebra with some central charge. In other words,
  c
 (gh) (gh)  (gh) gh
m3 − m δm+n


Lm , Ln = (m − n) Lm+n − aδm+n + (7.32)
12
(gh)
for some constants a and cgh . A similar statement applies to L̃m so we focus now
(gh)
on the Lm operators.
The constants a and cgh can be found as follows. First we observe that
h i
(gh) (gh) (gh)
kL−m |0igh k2 = gh h0|L(gh)
m L−m |0i gh
= gh
h0| L(gh)
m , L−m |0i
gh

cgh
m3 − m ,


= −2ma + (7.33)
12
where the last equality follows from (7.39) and the fact that Lgh
0 |0i
gh
= 0. Choosing
m = 1 and then m = 2 we deduce that
(gh) (gh) cgh
kL−1 |0igh k2 = −2a , kL−2 |0igh k2 = −4a + . (7.34)
2
38
We will use the notation | ) to indicate a state in the space obtained by taking the tensor
product of the Fock space built on the oscillator vacuum |0i with the Fock space built on the ghost
oscillator vacuum |0igh .

– 77 –
We can now compute the LHSs directly using
(gh)
X
L−1 |0i = (k − 1)b−1−k ck |0igh = − (b−1 c0 + 2b0 c−1 ) |0i , (7.35)
k∈Z
(gh)
X
L−2 |0i = (k − 2)b−2−k ck |0igh = − (2b−2 c0 + 3b−1 c−1 + 4b0 c−2 ) |0i .
k∈Z

For example39
 (gh) 2
 
L−1 |0i = h0| (c0 b1 + 2c1 b0 ) (b−1 c0 + 2b0 c−1 ) |0i
= −2h0| (c0 b0 b1 c−1 + b0 c0 c1 b−1 ) |0i (using b20 = c20 = 0)
= −2h0| (c0 b0 {b1 , c−1 } + {c1 , b−1 } b0 c0 ) |0i (using b1 |0i = c1 |0i = 0)
= −2h0| {c0 , b0 } |0i = −2 , (7.36)

from which we conclude that a = 1. Similarly,

 (gh) 2
 
L−2 |0i = h0| (2c0 b2 + 3c1 b1 + 4c2 b0 ) (2b−2 c0 + 3b−1 c−1 + 4b0 c−2 ) |0i
= −8h0| (c0 b0 b2 c−2 + b0 c0 c2 b−2 ) |0i − 9h0|c1 b−1 b1 c−1 |0i
= −8h0| (c0 b0 {b2 , c−2 } + b0 c0 {c2 , b−2 }) |0i − 9h0| {c1 , b−1 } {b1 , c−1 } |0i
= −8h0|({c0 , b0 } |0i − 9 = −17 , (7.37)

from which we conclude that

cgh
−4 + = −17 ⇒ cgh = −26 . (7.38)
2
Now we know that
  26
 (gh) (gh)  (gh)
m3 − m δm+n .


Lm , Ln = (m − n) Lm+n − δm+n − (7.39)
12
But we also know that
D
m3 − m δm+n .


[Lm , Ln ] = (m − n) Lm+n + (7.40)
12
(gh)
Combining these results, we find that the operators Lm = Lm + Lm have the
commutation relations
(D − 26)
m3 − m .


[Lm , Ln ] = (m − n) (Lm+n − δm+n ) + (7.41)
12
The central charge of the Virasoro algebra is now D − 26. Since the conformal
symmetry of the conformal gauge action is really a gauge invariance “in disguise”, as
39
Here we assume that c0 and b0 are hermitian since the classical anticommuting variables are
“real”. This in spite of the fact that these operators satisfy c20 = b20 = 0. Can an operator that
squares to zero be hermitian? Yes, for a particular choice of inner product on the state space.

– 78 –
discussed previously, we should expect to have to impose physical state constraints
[Lm −δm ]|phys) = 0, and this is only possible when the central charge of the Virasoro
algebra spanned by the Lm is zero, and this requires D = 26. In this spacetime
dimension, and only in this dimension, the FP ghosts must cancel the conformal
anomaly40 .

7.1 Virasoro-Shapiro amplitude

Now we shall compute the amplitude for the scattering of N tachyons using the path
integral. It is not very physical to consider the “scattering of tachyons” but this is
the simplest amplitude to compute, and it illustrates a number of important features
of string theory.
We work in the semi-classical approximation in which the worldsheet is assumed
to be a Riemann sphere with N punctures, at which we insert a tachyon vertex
operator. This approximation allows two simplifications. One is that the FP ghosts
contribute only to the overall normalisation, so we now omit them. The other is that
there are no modular parameters to sum over because the Riemann sphere is unique
up to conformal equivalence.
Let’s suppose that the ith tachyon has D-momentum pi , and that its vertex
operator is inserted at the point z = zi on the Riemann sphere. In the path-integral
version of the computation, we have to insert eip·X(zi ) into the path integral and then
integrate over zi . For N such insertions we get the amplitude
Z N Z
Y
−IE
A(p1 , . . . , pN ) = [dX] e d2 zi eipi ·Xi , (7.42)
i=1 RS

where Xi = X(zi , z̄i ). This amplitude is a function of the D-momenta of all the
tachyons. Amplitudes involving any of the particles in the string spectrum can be
calculated in a similar way but the calculation is more complicated because the vertex
operators are more complicated.
We can rewrite the tachyon scattering amplitude as
N Z Z
Y PN
A(p1 , . . . , pN ) = 2
d zi [dX] e−IE +i j=1 pj ·Xj
. (7.43)
i=1 RS

Now we observe that

N Z ( " N # )
X
¯ − i X
− IE + i pj · Xj = −T d2 z ∂X · ∂X δ 2 (z − zj )pj · X . (7.44)
j=1 RS T j=1

40
In fact, the physical state condition, in this Lorentz covariant formulation with FP ghosts, is
the stronger condition that Q̂BRST |physi = 0, where Q̂BRST is the operator version of the Noether
charge associated to a kind of supersymmetry that interchanges the original variables with the FP
ghosts. Consistency requires that Q̂2BRST = 0, and this turns out to be true only if D = 26.

– 79 –
 R
where the delta function is defined such that41 d2 z δ 2 (z − zi )f (z) = f (zi ). Integrat-
ing by parts, we can replace ∂X · ∂X ¯ = −X · ∂∂X
¯ since ∂ is acting on functions
defined on the Riemann sphere, which has no boundary. This gives us
N Z ( " N
#)
X
¯ i X
−IE + i p j · Xj = T d2 z X · ∂∂X + δ 2 (z − zj )pj . (7.45)
j=1 RS T j=1

The idea now is to complete the square in X but to do this we need to invert
¯ = 1 ∇2 and there is a problem with this because ∇2 has a zero eigenvalue on the
∂∂ 4
sphere. The eigenfunction is the constant function, i.e. X(z) = X0 , so we should
write Z Z Z
[dX] = dD X0 [dX]0 , (7.46)

where [dX]0 is an integral over all functions except to the constant function42 . Iso-
lating the X0 -dependence we now have
Z 
i( j pj )·X0
P
D
A(p1 , . . . , pN ) = d X0 e Â(p1 , . . . , pN )
N
!
X
∝ δ pj Â(p1 , . . . , pN ) , (7.47)
j=1

where the path integral for  excludes the integration over the constant function.
The delta-function prefactor imposes conservation of the total D-momentum.
¯ the inverse is the 2D Green function, which we define
We can now invert ∂∂;
such that43
¯ 1
∂∂G(z − zi ) = δ 2 (z − zi ) ⇒ G(z) = ln |z| . (7.48)
π
Setting
N
i X
X(z) = Y (z) − G(z − zi )pi , (7.49)
2T i=1

we have44 [dX]0 = [dY ]0 , and [Exercise]

N Z
X
¯ 1 XX
−IE + i pj · Xj = T d2 z Y · ∂∂Y + G(zi − zj )pi · pj . (7.50)
j=1 RS 4T i j

41
Notice that d2 z = 2dxdy ⇒ 2δ 2 (z) = δ(x)δ(y).
42
More precisely, orthogonal to the constant function but we pass over the specification of the
norm needed to define orthogonality.
43
The formula for G can be verified by using the formulas ∂z¯ −1 = 2πδ 2 (z) and the fact that
2
2δ (z) = δ(x)δ(y) for z = x + iy.
44
A shift in the integration variable has no effect because we integrate over all values of the
(non-constant) functions X.

– 80 –
The terms in the double sum are infinite when i = j, but also independent of the mo-
menta, so these terms can be omitted; they can only affect the overall normalisation.
The Gaussian [dY ]0 path integral also contributes only to the overall normalisation.
Using the specific form of the Green function, we are then left with
N Z
Y Y p i · pj
Â(p1 , . . . , pN ) ∝ d2 zi |zj − zk |αjk , αij = . (7.51)
i=1 RS j<k
2πT

We ignored the FP ghosts in arriving at this formula on the grounds that inte-
gration over them contributes only to the overall normalisation. Let’s reassess this
claim, focusing on the (b, c) ghosts; the problem is that the action for (b, c) will be
¯ = 0 or ∂b
zero for any solution of either ∂c ¯ = 0 that is defined everywhere on the RS,
so the Berezin integral over it will give a zero amplitude. As it happens, there are no
¯ = 0 on the RS but there are three independent glob-
globally defined solutions of ∂b
¯ = 0, corresponding to the conformal Killing vector fields
ally defined solutions of ∂c
that generate the Sl(2; C) group of conformal isometries of the RS. [See Q.IV.1]. In-
tegrating over these functions gives a zero amplitude, apparently, but actually there
is a compensating infinity in our result (7.51). This amplitude is Sl(2; C)-invariant,
which means that there is an implicit integration over the group Sl(2; C) resulting
in a factor equal to the volume of Sl(2; C), but this volume is infinite.
Let’s verify the Sl(2; C) invariance of  as given in (7.51). The Sl(2; C) trans-
formation of z is
az + b
z → z0 = , a, b, c, d ∈ C , ad − bc = 1 . (7.52)
cz + d
Using this one finds that
zi − zj d2z
zi0 − zj0 = , d2z 0 = , (7.53)
(czi + d) (czj + d) |cz + d|4
and hence that
N
" N
#" N #
−4− 0j αij
Y Y Y Y Y P
d2 zi0 |zj0 − zk0 |αjk = d2 zi |zj − zk |αjk |czi + d| , (7.54)
i=1 j<k i=1 j<k i=1

where
X N
X
0
αij = αij − αii (i = 1, . . . , N )
j j=1
N
!
1 X p2i
= pi · pj −
2πT j=1
2πT
p2i
=− (by momentum conservation). (7.55)
2πT

– 81 –
We see from this that the amplitude is Sl(2; C) invariant only if

p2i
−4 + =0 ⇔ p2i = 8πT . (7.56)
2πT
This is the mass-shell condition for the tachyonic ground state of the string!
The Sl(2; C) invariance of the amplitude (7.51) means that there are three too
many integrals. We could fix this problem by the gauge choice

fi ≡ zi − ui = 0 i = 1, 2, 3. (7.57)

Inserting δ 2 (f1 )δ 2 (f2 )δ 2 (f3 ) into the integrand of (7.51) removes three of the integrals,
but this would not be the correct thing to do. The problem is that the Sl(2; C)
invariance of the amplitude implies that it is proportional to the volume of Sl(2; C),
which is infinite because the group is non-compact, but this factor does not appear
explicitly in the expression (7.51) for the amplitude.
We know how to solve this problem. When we fix the positions of the first
three points by insertion of delta functions, we must also include a Fadeev-Popov
determinant. The infinitesimal form of the transformation (7.52) is45

δz = α0 + α1 z + α2 z 2 . (7.58)

For fi = zi − ui we have δfi = δzi and hence

 1 z1 z12 
 
 
∂(δfi )  
det =  1 z2 z22  = (z1 − z2 )(z2 − z3 )(z3 − z1 ) . (7.59)
∂αj  1 z z 2 
3 3

Because the variables are complex (we are inserting three 2D delta functions), the
FP determinant is the modulus squared of this, so

∆F P = |z1 − z2 |2 |z2 − z3 |2 |z3 − z1 |2 . (7.60)

Following the earlier argument for gauge fixing the particle action, the insertion of
the delta functions with the FP determinant allows us to factor out the (infinite)
volume Ω of Sl(2; C); dividing by this volume we then get
N Z
Y Y
−1
Ω Â(p1 , . . . , pN ) ∝ d2 zi δ 2 (f1 )δ 2 (f2 )δ 2 (f3 )∆F P |zj − zk |αjk . (7.61)
i=1 j<k

This can be checked as follows. Multiply both sides by |(u1 − u2 )(u2 − u3 )(u3 − u1 )|−2
and integrate over (u1 , u2 , u3 ). On the RHS the u integrals can be done using the
45
This shows that (∂, z∂, z 2 ∂) are the globally defined conformal Killing vector fields.

– 82 –
delta functions, the ∆F P factor is then cancelled and we recover the expression (7.51).
On the LHS the integral cancels the factor of Ω−1 because, formally,

d2 u1 d2 u2 d2 u3
Z
Ω= . (7.62)
|(u1 − u2 )(u2 − u3 )2 (u3 − u1 )|2

This integral is infinite but the integrand is the Sl(2; C) invariant measure on the
Sl(2; C) group manifold, parametrised by three complex coordinates on which Sl(2; C)
acts by the fractional linear transformation (7.52).
We may now do the (z1 , z2 , z3 ) integrals of (7.61) to get

Ω−1 Â(p1 , . . . , pN ) ∝ |u1 − u2 |2+α12 |u2 − u3 |2+α23 |u3 − u1 |2+α13 (7.63)

YN Z YN Y
2
× d zi |u1 − zi |α1i |u2 − zi |α2i |u3 − zi |α3i |zi − zj |αjk .
i=4 i=4 4≤j<k

This can be simplified enormously by the choice

u3 = 1 , u2 = 0 , u1 → ∞ . (7.64)

In this limit we get a factor of

P
|u1 |4−α11 + i α1i
= 1, (7.65)

where the equality follows upon using both the mass-shell condition and momentum
conservation. The remaining terms give the Virasoro-Shapiro amplitude
N Z
Y N
Y Y
2
ÂV S (p1 , . . . , pN ) ∝ d zi |zi |α2i |zi − 1|α3i |zj − zk |αjk . (7.66)
i=4 i=4 3<j<k

The result for N = 3 is a constant, which can be interpreted as a coupling constant.

For N = 4 we have the Virasoro amplitude
Z
Â(p1 , p2 , p3 , p4 ) = d2 z |z|α24 |z − 1|α34 . (7.67)

7.1.1 The Virasoro amplitude

Consider the elastic scattering of two identical particles of mass m. In the rest frame,
the incoming particles have D-momenta

p1 = (E, p~) , p2 = (E, −~p) . (7.68)

The outgoing particles have D-momenta

−p3 = (E, p~ 0 ) , −p4 = (E, −~p 0 ) . (7.69)

– 83 –
Notice that p1 + p2 + p3 + p4 = 0, as required by D-momentum conservation. In
addition, since p2i = −m2 for i = 1, 2, 3, 4, we have

|~p|2 = |~p 0 |2 = E 2 − m2 . (7.70)

The scattering angle θs is defined by

p~ · p~ 0
cos θs = 2 . (7.71)
E − m2
We may trade the frame-dependent variables (E, cos θs ) for the Lorentz scalar Man-
delstam variables

sM = − (p1 + p2 )2 = 4E 2
tM = − (p1 + p3 )2 = −2(E 2 − m2 ) (1 − cos θs )
uM = − (p1 + p4 )2 = −2(E 2 − m2 ) (1 + cos θs ) (7.72)

These are not all independent because

sM + tM + uM = 4m2 . (7.73)

The variable sM is the square of the centre of mass energy. For fixed sM , the variable
tM determines the scattering angle.
In the context of closed string tachyon scattering, for which m2 = −8πT , it is
convenient to use the rescaled Mandelstam variables46
1
(s, t, u) = (sM , tM , uM ) . (7.74)
8πT
This gives us
1 1
s=− (p1 + p2 )2 = −2 − α12
8πT 2
1 1
t=− (p1 + p3 )2 = −2 − α13
8πT 2
1 1
u=− (p1 + p4 )2 = −2 − α14 (7.75)
8πT 2
Using this and momentum conservation yields

α34 = α12 = −4 − 2s ,
α24 = α13 = −4 − 2t ,
α23 = α14 = −4 − 2u , (7.76)

and
s + t + u = −4 . (7.77)
46
Equivalently, we can choose units for which 8πT = 1.

– 84 –
We can also write s and t as
E2  s
s= , t = −2 1 + (1 − cos θs ) . (7.78)
2πT 4
Using (7.76) we can rewrite the Virasoro amplitude of (7.67) as
Z
 = d2 z |z|2α |z − 1|2β , α = −2 − t , β = −2 − s . (7.79)

Next, using the identity

Γ(α + 1)Γ(β + 1)Γ(−α − β − 1)

Z
1
d2 z |z|2α |z − 1|2β ≡ , (7.80)
π Γ(α + β + 2)Γ(−α)Γ(−β)

we arrive at the Virasoro amplitude for the scattering of two closed string tachyons:

Γ(−1 − t)Γ(−1 − s)Γ(−1 − u)

A(s, t) ∝ (u = −4 − s − t). (7.81)
Γ(u + 2)Γ(s + 2)Γ(t + 2)

Here, Γ(z) is Euler’s Gamma function, with the properties

zΓ(z) = Γ(z + 1) , Γ(n) = (n − 1)! for n ∈ Z+ . (7.82)

It also has an analytic continuation to a meromorphic function on the complex z-

plane with no zeros and simple poles at z = −n for n ≥ 0, with residues (−1)n /n!.
For fixed t the Virasoro amplitude A becomes a function of s with simple poles
at
s = −1, 0, 1, 2, . . . (7.83)
These poles correspond to resonances, i.e. to other particles in the spectrum (stable
particles, in fact, because the poles are on the real axis in the complex s-plane). The
position of the pole on the real axis gives the mass-squared of the particle in units
of 8πT . The pole at s = −1 is the tachyon itself; in other words, the tachyon can
be considered as a bound state of two other tachyons. The pole at s = 0 implies the
existence of a massless particle, or particles. The residue of this pole is

Γ(−1 − t)Γ(3 + t)
− = t2 − 4 . (7.84)
Γ(−2 − t)Γ(t + 2)

This is a quadratic function of t and hence of cos θs , which implies that there must
be a massless particle of spin 2 (but none of higher spin). The residue of the pole
at s = n is a polynomial in t of order 2(n + 1), so that 2(n + 1) is the maximum
spin of particles in the spectrum with mass-squared n × (8πT ). In a plot of Jmax
against s, such particles appear at integer values of Jmax on a straight line with slope
α0 /2 and intercept 2 (value of Jmax at s = 0). This is the leading Regge trajectory.
All other particles in the spectrum appear on parallel “daughter” trajectories in the

– 85 –
(J, s) plane (e.g. the massless spin-zero particle in the spectrum is the first one on
the trajectory with zero intercept. In fact, the entire string spectrum can be found
in this way!
If we had computed the amplitude for scattering gravitons instead of tachyons
then we would have found the tachyon as a resonance. This shows that it is not
consistent to simply omit the tachyon from the spectrum.
Another feature of the Virasoro amplitude is its s ↔ t symmetry. Poles in A
as a function of s at fixed t therefore reappear as poles in A as a function of t at
fixed s. These correspond to the exchange of a particle. In particular, a massless
spin-2 particle is exchanged, and arguments imply that such a particle must be the
quantum associated to the gravitational force, so a theory of interacting closed strings
is a theory of quantum gravity.
The Virasoro amplitude for closed strings was preceded by the Veneziano ampli-
tude for open strings47

Γ(−1 − s)Γ(−1 − t)
A(s, t) = , (7.85)
Γ(−2 − s − t)

where now
1 1
s=− (p1 + p2 )2 , t=− (p1 + p3 )2 . (7.86)
2πT 2πT
This amplitude also has poles at s = −1, 0, 1, 2, . . ., but the maximum spin for s = n
is now Jmax = n + 1, and the leading Regge trajectory has slope α0 and intercept
1 (this is the constant a that equals the zero point energy in the light-cone gauge
quantization of the open string).

7.2 String theory at 1-loop: taming UV divergences

We will now take a brief look at what happens at one string-loop. In this case
amplitudes are found from the path integral by considering vertex operator insertions
at points on flat torus, which we can define by a doubly periodic identification in the
complex z-plane. Without loss of generality (because we are free to rescale z and
change its sign) we can choose

z ∼ z + 1, z ∼z+τ Im τ > 0 . (7.87)

Any further analytic transformation that preserves the first of these identifications
will leave τ unchanged48 . However, not all values of τ in the upper half plane define
47
Veneziano did not compute it from string theory (which did not then exist) but just proposed
it on the basis of its properties.
48
By a non-analytic coordinate transformation we can put the equivalence relations into the form
z ∼ z + n + im for integers (n, m), but the metric is then conformal to |dz + µdz̄|2 , and µ now
parametrises the conformally inequivalent metrics on the torus with standard identifications.

– 86 –
inequivalent tori because the identifications are obviously unchanged by the transla-
tion
T : τ → τ + 1. (7.88)
They are also unchanged by the inversion
1
S: τ →− . (7.89)
τ
To see this multiply the equivalence relations of (7.87) by −1/τ and then rescale
z → −τ z to get49
1
z∼z− , z ∼ z − 1. (7.90)
τ
Composition of the S and T maps generates elements of the group P Sl(2; Z) which
acts on τ by a fractional linear transformation
 
aτ + b ab
τ→ , ∈ Sl(2; Z) (7.91)
cτ + d cd

Two matrices of Sl(2; Z) that differ by a sign have the same action on τ , so P Sl(2; Z) ∼
=
Sl(2; Z)/ {±1}. Inequivalent tori are parametrised by complex numbers τ lying in a
fundamental domain of P Sl(2; Z) in the complex τ -plane.
In the path integral representation of the one string-loop amplitudes, we have to
sum over all inequivalent tori (Euclidean worldsheets of genus 1). This leads to an
integral over τ : Z
A∝ d2 τ . . . (7.92)
F

where F is any fundamental domain of P Sl(2; Z); it is convenient to choose it to be

the one in which we may take Im τ → ∞ on the imaginary axis. In this limit the
torus becomes long and thin and it starts to look like a one-loop Feynman diagram
(with vertices at various points if we had vertex operators at points on the torus). In
this infra-red limit we can interpret Im τ as the modular parameter s of a particle
worldline. In the particle case we would have
Z ∞
A∝ ds . . . (7.93)
0

and we typically get divergent results from the part of the integral where s → 0.
These are ultra-violet divergences. They are absent in string theory because the
domain of integration F does not include the origin of the τ -plane.

Moral There are no UV divergences in string theory. We have discussed only one
string-loop but the result is general. This can be understood in other ways. For
49
The relation z ∼ z + 1 is equivalent to z ∼ z − 1.

– 87 –
example, the force due to exchange of the massless states of the closed string in-
cludes the force of gravity because it includes a spin-2 field. Whereas this leads to
unacceptable UV behaviour in GR, the UV divergences are cut off in string theory
at the string length scale √
`s ∼ α0 (7.94)
because at this scale the exchange of the massive string states becomes as important
as the graviton exchange.

7.3 The dilaton and the string-loop expansion

The way that the dilaton field Φ(X) couples to the string in its Polyakov formulation
is through the scalar curvature of the independent worldsheet metric γµν . In two
dimensions the Riemann curvature tensor is entirely determined by its double trace,
the Ricci scalar R(γ), but this allows us to add to the Euclidean NG action the term
√
Z
1
IΦ = d2 z Φ(X) γ R(γ) . (7.95)
4π
Here are some features of this term:

• If Φ = φ0 , a constant then
√
Z
1
IΦ = φ0 χ , χ= d2 z γ R(γ) . (7.96)
4π
The integral χ is a topological invariant of the worldsheet, called the Euler
number. For a compact orientable Riemann surface without boundary (which
we’ll abbreviate to “Riemann surface” in what follows) the Euler number is
related to the genus g (the number of doughnut-type “holes”) by the formula

χ = 2 (1 − g) . (7.97)

• In conformal gauge, we can write the line element for the (Euclidean signature)
metric γ as ds2 (γ) = 2eσ dzdz̄, i.e. a conformal factor eσ (an arbitrary function
√
of z and z̄) times the Euclidean metric. We then find that γ R(γ) = 2∇2 σ
and hence, after integrating by parts,
Z
1 ¯ n ∂n ∂m Φ .
IΦ = d2 z σ ∂X m ∂X (7.98)
2π
This dependence on σ shows that IΦ is not conformal invariant, unless Φ is
constant. This is allowed because IΦ , being independent of the string tension
T , comes with an additional factor of α0 relative to the NG action (we have
to consider the lack of conformal invariance of IΦ at the same time that we
consider possible conformal anomalies).

– 88 –
 These properties suggest that we write

Φ = φ0 + φ(X) , (7.99)

where φ(X) is zero in the vacuum; i.e. the constant φ0 is the “vacuum expectation
value” of Φ(X). Then there will appear a factor in the path integral of the form

g−1
e−φ0 χ = gs2 , gs ≡ eφ0 . (7.100)

For g = 0 this tells us that the Riemann sphere contribution to scattering amplitudes
is weighted by a factor of 1/gs2 . If we use these amplitudes to construct an effective
field theory action S from which we could read off the amplitudes directly (by looking
at the various interaction terms) then this action will come with a factor of 1/gs2 (we
can then absorb all other dimensionless factors into a redefinition of gs , i.e. of φ0 ).
If we focus on the amplitudes for scattering of massless particles then we find that
the effective action for the massless fields {hmn , bmn , φ} is
Z  
1 D
p −2φ 1 2 2 0
S= (D−2)
d x − det g e 2Λ + R(g) − H + 4 (∂φ) + O(α ) , (7.101)
gs2 `s 3

where g is the spacetime metric (gmn = ηmn + hmn ), H = db is the 3-form field
strength for the 2-form potential b, and
(D − 26)
Λ= , (7.102)
3α0
although the perturbative expansion in powers of α0 is justifiable only if Λ = 0, i.e.
if D = 26. Some other features of this space-time action are

• The exact result for S will involve a series of all order in α0 since the coupling
of the background fields to the string introduces interactions into the 2D QFT
defined by the string worldsheet action I.

• Coupling the string only to background fields associated to the massless parti-
cles yields a renormalizable 2D QFT. This is equivalent to the statement that
it is consistent to truncate the full equations of motion of the background fields
associated to all particles in the string spectrum by setting to zero the back-
ground fields of the massive particles. In other words, the massless fields do
not act as sources for the massive fields. In contrast any massive field will act
as a source for more massive fields, implying that they must all be included
once one is included. This statement is equivalent to the non-renormalizability
of the 2D QFT with higher-derivative interactions.

• The integrand involves a factor of e−2φ . This is because the action must be
such that φ0 ≡ ln gs and φ(X) must appear only through the combination
Φ = φ0 + φ(X).

– 89 –
 In the effective spacetime action, gs2 plays the role of ~. This suggests that we
have been considering so far only the leading term in a semi-classical expansion. This
is true because we have still to consider Riemann surfaces with genus g > 0, and
a string amplitude at genus g is weighted, according to to (7.100), by a factor of
(gs2 )g−1 , i.e. a factor of (gs2 )g relative to the zero-loop amplitude. This confirms that
the string-loop expansion is a semi-classical expansion in powers of gs2 . Taking into
account all string loops gives us a double expansion of the effective field theory50
Z ∞ ∞
1 D
p X
2g 2(g−1)φ
X
S= (D−2)
d x − det g gs e Lg , Lg = `2l (l)
s Lg . (7.103)
2
gs `s g=0 l=0

In effect, the expansion in powers of `s comes from first-quantisation of the string, and
the expansion in powers of gs comes from second-quantisation. How can we quantise
twice? Is there not a single ~? The situation is actually not so different from that of
the point particle. When we first-quantise we get a Klein-Gordon equation but with
a mass m/~; we then relabel this as m so that it becomes the mass parameter of the
classical field equation, and then we quantise again. For the string, first quantization
would have led to α0 ~ as the expansion parameter if we had not set ~ = 1; if we
relabel this as α0 then ~ appears only in the combination gs2 ~.
To lowest order in α0 we have what looks like GR coupled to an antisymmetric
tensor and a scalar. The D-dimensional Newton gravitational constant GD is

GD ∝ gs2 `(D−2)
s . (7.104)

Consistency of the string-loop expansion (in powers of gs2 ) relies on this formula.
Particles in the string spectrum have masses proportional to 1/`s , independent of gs ,
so their contribution to the gravitational potential in D dimensions is proportional
to gs2 , and hence zero at zero string coupling. This means that the strings of free
(gs = 0) string theory do not back-react on the space-time metric; the metric is
changed by the presence of strings only within perturbation theory. If this had not
been the case it would not have been consistent to start (as we did) by considering
a string in Minkowski spacetime.
Why is gs called the string coupling constant? Consider a g string-loop vacuum
to vacuum diagram with the appearance of a chain of g tori connected by long
“throats”, and think of it as a “fattened” Feyman diagram in which a chain of g
loops connected by lines; where each line meets a loop we have a 3-point vertex. As
50 (l)
There is a lot of freedom in the form of the Lagrangians Lg beyond leading order. Recall that
the construction of S involves a prior determination of scattering amplitudes of the level-1 fields,
which we then arrange to replicate from a local spacetime Lagrangian. Since the amplitudes are
all “on-shell” they actually determine only the field equations for the background fields, and then
only up to field redefinitions. Even with all this freedom it is not obvious why it should be possible
to replicate the string theory scattering amplitudes in this way, although this has been checked to
low orders in the expansion and there are general arguments that purport to prove it.

– 90 –
there are (g − 1) lines that link the loops, and each of the two ends of each line ends
at a vertex, we have a total of 2(g − 1) vertices. If we associate a coupling constant
to each vertex, call it gs , we see that this particular diagram comes with a factor of
(gs2 )g−1 , which agrees with our earlier result.
Is there a g-loop Riemann surface with the appearance just postulated. Yes,
there is. For g > 0 there is no longer a unique flat metric, as we have already seen
for g = 1. For g ≥ 2 there is a 3(g − 1)-parameter family of conformally inequivalent
flat metrics; these parameters are called “moduli” (see Q.IV.2). This number can be
understood intuitively from the “chain of tori” diagram if we associate one parameter
with each propagator. For g loops we have, in addition to the (g − 1) links, (g − 2)
“interior” loops with 2-propagators each, and two “end of chain” loops with one
propagator each. The total number of propagators is therefore

(g − 1) + 2(g − 2) + 2 = 3(g − 1) . (7.105)

This is also, and not coincidentally, the dimension of the space of quadratic differen-
tials on a Riemann surface of genus g ≥ 2 (see Q.IV.2).

8. Interlude: The spinning particle

A point particle with non-zero spin can be accommodated by including additional

anticommuting coordinates. Consider the non-relativistic particle action

|~p|2
Z  
˙ i ~ ~˙
I = dt ~x · p~ + ψ · ψ − , (8.1)
2 2m

~ are anti-commuting and “real”. The action is invariant

where the 3-vector variables ψ
under space rotations; the infinitesimal transformations are

~ × ~x ,
δω ~x = ω ~ × p~ ,
δω p~ = ω ~=ω
δω ψ ~.
~ ×ψ (8.2)

The Noether charge is the angular momentum J~ = L

~ + S,
~ where51

~ = ~x × p~ ,
L ~×ψ
~ = −iψ
S ~. (8.3)

The Poisson brackets of the anticommuting variables are

ψa, ψb = −iδ ab .

PB
(8.4)

51 ~~ is non-zero because ψ
~ is anticommuting.
ψ×ψ

– 91 –
 • Check: For consistency with our earlier conventions for FP ghosts, the La-
grangian L = iψ ∗ ψ̇ for complex anticommuting ψ gives the Poisson bracket
√
relations {ψ ∗ , ψ}P B = −i. Now set ψ = (ψ1 + iψ2 )/ 2 to get

1
i

ψ1 = √ ψ + ψ̄ , ψ2 = − √ ψ − ψ̄ . (8.5)
2 2
This gives us the following Lagrangian and corresponding PB relations
2
iX
L= ψi ψ̇i , → {ψi , ψj }P B = −iδij . (8.6)
2 i=1

~˙ term in (8.1).
~·ψ
This explains why the factor of 1/2 appears in the ψ

Upon quantization we get canonical anticommutation relations that can be re-

alised by Pauli matrices:
n o r
~ a
ψ̂ a , ψ̂ b = δ ab ⇒ ψ̂ a = σ . (8.7)
2
The spin vector operator is then

i ~ ~
Ŝ a = − εabc ψ̂ b ψ̂ c = −i εabc σb σc = σa (8.8)
2 4 2
and hence  2 3  
−2  a 1
~ Ŝ  = = 2s + 1 for s = (8.9)
4 2
which shows that the action (8.1) describes a spin- 12 particle.

8.1 Relativistic spin-1/2

The relativistic generalization, in D spacetime dimensions, is simplest for a massless
particle52 . The spin can be accommodated by including an anticommuting D-vector
coordinate ψ m . The action is
Z  
m i m n 1 2
I = dt ẋ pm + ηmn ψ ψ̇ − e p − iχ ψ · p . (8.10)
2 2

We read off from this action the Poisson brackets

{xm , pn }P B = δnm , {ψ m , ψ n }P B = −iη mn . (8.11)

52
The massive case can be found by starting with the massless particle action (8.10) in (D + 1)
dimensions and reducing to D dimensions with PD = m.

– 92 –
Using this we find that

{ψ · p, ψ · p}P B = −ip2 , p2 , ψ · p

PB
= 0, (8.12)

which shows that the constraints are first-class and hence that they both generate
gauge invariances. These are

δxm = αpm + iψ m , δψ m = −pm , (8.13)

where  is an infinitesimal anticommuting parameter. The action is invariant if the

Lagrange multipliers transform as

δe = α̇ , δχ = ˙ . (8.14)

To be precise, one finds that

1 2 t
δI = α p + i ψ · p tB , (8.15)
2 A

which is zero if α and  are zero at t = tA and t = tB .

The simplest way to see that this action does indeed describe a massless spin- 12
relativistic particle is to quantize à la Dirac. Recall that we first quantize as if there
were no constraints, which gives us the (anti)commutation relations

[xm , pn ] = iδm
n
, {ψ m , ψ n } = η mn . (8.16)

These can be realised on a spinor wave function Ψ(x) by

1
pm = −i∂m , ψ m = √ Γm , (8.17)
2
where Γm are the 2[D]/2 × 2[D]/2 spacetime Dirac matrices. Next, we impose the con-
straints as physical state conditions. We have only to impose the condition resulting
from the constraint ψ · p = 0 because this implies the mass-shell condition; the result
is the massless Dirac equation

Γm ∂m Ψ(x) = 0 . (8.18)

If D = 5, 6, 7 mod 8 then Ψ must be complex, i.e. a “Dirac spinor”, but if

D = 2, 3, 4, 8, 9 mod 8 then we may choose it to be real53 . For any D we can define

ΓD+1 = Γ0 Γ1 · · · ΓD , (8.19)
53
This can be interpreted as a “gauging” of the time-reversal invariance of the particle action. In
general, time reversal is represented in QM by an anti-unitary operator K, with the property that
K 2 = ±1. When K 2 = 1, as is the case for the spinning particle when D = 2, 3, 4, 8, 9 mod 8, we
can impose the condition KΨ = ±Ψ as a new physical state condition. This is a reality condition
because K involves taking the complex conjugate. When K 2 = −1, as is the case for the spinning
particle when D = 5, 6, 7 mod 8, it is not consistent to impose a reality condition so Ψ is necessarily
complex; this is a simple illustration of Kramer’s degeneracy in QM.

– 93 –
but this matrix is ±1 for odd D; for even D it anticommutes with each of the D
Dirac matrices and its square is either +1 or −1. If Γ2D+1 = 1 we can define a chiral
(anti-chiral) spinor as an eigenspinor of ΓD+1 . For D = 2 mod 8, but not otherwise,
a real spinor can also be chiral. This fact is of importance for superstring theory.

9. Spinning strings

We can find a ‘spinning string” by following the example of the spinning particle.
We need to take “square roots” of the constraint functions H± , and to do this we
need to introduce anticommuting worldsheet fields ψ± . For the closed string, this
gives us an action of the form
Z I 
i 
I = dt dσ Ẋ m Pm + ψ+ ψ̇+ + ψ− ψ̇−
2
−λ H− − λ H+ − iχ− Q− − iχ+ Q+ .
− +

(9.1)

where χ± are Lagrange multipliers. The constraint functions are

1 2 i 1
H± = (P ± T X 0 ) ± ψ± · ψ±
0
, Q± = √ (P ± T X 0 ) · ψ± . (9.2)
4T 2 2 T

This action reduces to the Hamiltonian form of the NG action when all anti-commuting
variables are omitted. This means that all classical solutions of the NG string are
solutions of the spinning string, but the new anti-commuting variables make a sig-
nificant difference to the quantum theory.
We can read off the non-zero PB relations from the above action. In particular,

m n
(σ 0 ) P B = −iη mn δ(σ − σ 0 ) .

ψ± (σ), ψ± (9.3)

A calculation using the PBs of the canonical variables shows that the non-zero PBs
of the constraint functions are

{Q± (σ), Q± (σ 0 )}P B = −i H± (σ) δ(σ − σ 0 )

 
1
{Q± (σ), H± (σ )}P B = ± Q± (σ) + Q± (σ ) δ 0 (σ − σ 0 )
0 0
2
{H± (σ), H± (σ )}P B = ± [H± (σ) + H± (σ 0 )] δ 0 (σ − σ 0 ) .
0
(9.4)

Let’s verify the first of these:

• First, you should verify that

m n
(P ± T X 0 ) (σ), (P ± T X 0 ) (σ 0 ) P B = ±2T η mn δ 0 (σ − σ 0 ) .

(9.5)

– 94 –
 Using this and the canonical PB relation (9.3), we find that

0 1
{Q± (σ), Q± (σ )}P B = −i (P ± T X 0 ) (σ) · (P ± T X 0 ) (σ 0 )δ(σ − σ 0 )
4T

i 0 0 0
± ψ± (σ) · ψ± (σ )δ (σ − σ ) (9.6)
2
2
Now we use the fact that ψ± ≡ 0 (because of the anticommutativity of ψ± ) to
deduce that

0 = ∂σ0 [ψ± (σ) · ψ± (σ)δ(σ − σ 0 )] = ∂σ0 [ψ± (σ) · ψ± (σ 0 )δ(σ − σ 0 )]

0
= ψ± (σ) · ψ± (σ 0 )δ(σ − σ 0 ) − ψ± (σ) · ψ± (σ 0 )δ 0 (σ − σ 0 ) , (9.7)

and hence that

i
± ψ± (σ) · ψ± (σ 0 )δ 0 (σ − σ 0 ) = ψ± (σ) · ψ±
0
(σ 0 )δ(σ − σ 0 ) . (9.8)
2
Using this in (9.6) we deduce that {Q± (σ), Q± (σ 0 )}P B = −i H± (σ) δ(σ − σ 0 ).

From (9.4) we see that the constraints are all first-class, so they generate gauge
invariances of the canonical variables via their PBs with ξ ± H± +i± Q± for parameters
ξ ± (t, σ) and anticommuting parameters ± (t, σ). The gauge transformations are
1 − 1 + i
ξ (P − T X 0 ) + ξ (P + T X 0 ) + √ + ψ+ + − ψ−


δX =
2T 2T 2 T
√
1 −
h i 0 1 + i T +
0 0
0

0
 ψ+ − − ψ−

δP = − ξ (P − T X ) + ξ (P + T X ) +
2 2 2
1 ± 0 1 ±
0 1 0 ±
δψ± = ± ξ ψ± ± ξ ψ± − √ (P ± T X )  . (9.9)
2 2 2 T
One then finds that the action is invariant provided the Lagrange multipliers are
assigned the gauge transformations

0
0 1 ±
0 ±
0
δξ λ± = ξ˙± ∓ λ± ξ ± ± ξ ± λ± , δξ χ± = ∓ ξ χ ± ξ ± χ± ,
2
i ± 0 1 ±
0 ±
δ λ± = √ χ± ± , δ χ± = ˙ ± ∓ λ±


 ± λ  . (9.10)
T 2
9.1 Conformal gauge and superconformal symmetry
The conformal gauge for the spinning string is

λ± = 1 , χ± = 0 . (9.11)

This leaves residual gauge invariances with parameters restricted by

ξ ± = ξ ± (σ ± ) , ± = ± (σ ± ) . (9.12)

– 95 –
 Using the conformal gauge conditions in (9.1) yields
Z I  
m 1  2 0 2

I = dt dσ Ẋ Pm + i (ψ+ · ∂− ψ+ + ψ− · ∂+ ψ− ) − P + (T X ) . (9.13)
2T

After elimination of P , by its equation of motion P = T Ẋ, this becomes

Z I  
i
I = 2T dt dσ ∂+ X · ∂− X + (ψ+ · ∂− ψ+ + ψ− · ∂+ ψ− ) . (9.14)
2T

The residual gauge invariance is a symmetry of this action, with transformations54

i
δX = ξ − ∂− + ξ + ∂+ X + √ − ψ− + + ψ+ ,



2 T
1 √
δψ± = ξ ± ∂± ψ± + ∂± ξ ± ψ± − T ∂± X ± .


(9.15)
2
The Noether charges are precisely H± and Q± .

• Let’s check invariance of the integral of ψ+ · ∂− ψ+ with respect to the ξ-

transformation of ψ+ . Since δ(ψ+ · ∂− ψ+ ) = 2δξ ψ+ · ∂− ψ+ + ∂− (· · ·) and we can
ignore total derivatives, we have
 
1 1
ψ+ · ∂− ψ+ = ξ + ∂+ ψ+ · ∂− ψ+ + ∂+ ξ + ψ+ · ∂− ψ+



δξ
2 2
1
= − ∂+ ξ + ψ+ · ∂− ψ+ − ξ + ψ+ · ∂− ∂+ ψ+


2
1 1
= − ∂+ ξ + ψ+ · ∂− ψ+ + ξ + ∂− ψ+ · ∂+ ψ+


2 2
= 0, (9.16)

where the penultimate line comes from integrating by parts and using ∂− ξ + = 0
to deduce (ignoring total derivatives) that
 
+ 1 1 +
−ξ ψ+ · ∂− ∂+ ψ+ = + ξ ∂− ψ+ · ∂+ ψ+
2 2
1 1
= − ξ + ∂+ ψ+ · ∂− ψ+ + ξ + ∂− ψ+ · ∂+ ψ+ , (9.17)
2 2
and the last line comes from the similar deduction (again using ∂− ξ + = 0 and
ignoring total derivatives) that

1 1
− ∂+ ξ + ψ+ · ∂− ψ+ = − ξ + ∂− ψ+ · ∂+ ψ+ .


(9.18)
2 2
54
To check the agreement with (9.9) you must use the fact that the conformal gauge action has
a trivial gauge invariance with transformation δψ± = f ± ∂∓ ψ± for any functions f ± .

– 96 –
 Each component of the Lorentz D-vector ψ± is also a worldsheet lightcone com-
ponent of a real 2D spinor. To demonstrate this, we will work backwards from the
2D Dirac action:

• 2D Dirac action. In 2D the Dirac matrices satisfying {γµ , γν } = 2ηµν can be

chosen to be real. For example,

γ 0 = iσ2 , γ 1 = σ1 (⇒ γ 0 γ 1 = σ3 ). (9.19)

This means that that it is consistent to suppose that a 2D Dirac spinor Ψ is

real55 . The Dirac Lagrangian for a real anticommuting spinor Ψ is

i
LDirac = ΨT γ0 γ µ ∂µ Ψ . (9.20)
2

Notice that the matrices γ 0 γµ are symmetric; the Lagrangian would be a total
derivative if Ψ were a commuting spinor! We include the factor of i because this
is needed for “reality” if we adopt the usual convention for taking the complex
conjugate of products of anticommuting variables.
We may choose a light-cone basis for the Dirac matrices

γ± = γ0 ± γ1 ⇒ γ µ ∂µ = γ + ∂+ + γ − ∂− . (9.21)

Noticing that
 
± 1 ∂− 0
γ0 γ + ∂+ + γ − ∂− =


γ0 γ = (1 ∓ σ3 ) ⇒ , (9.22)
2 0 ∂+

we have
 
Ψ+
Ψ= ⇒ LDirac = i (Ψ+ ∂− Ψ+ + Ψ− ∂+ Ψ− ) . (9.23)
Ψ−

The action for the closed spinning string in conformal gauge is therefore the
action for a free massless 2D field theory for D real scalar fields and D real spinor
fields. This action has a (1, 1) superconformal symmetry generated by the Fourier
modes of the Noether charges Q± and H± . Of course, the string theory constraints
are equivalent to Q± = 0 and H± = 0, implying that the superconformal invariance
is actually a gauge invariance of the spinning string theory
55
A spinor that is real in a real basis for the Dirac matrices becomes a Majorana spinor in a
general basis.

– 97 –
9.2 Open spinning string: free ends
The phase-space action for the open spinning string, of parameter length π, is
Z Z π 
m i 
I = dt dσ Ẋ Pm + ψ+ · ψ̇+ + ψ− · ψ̇−
0 2
−λ− H− − λ+ H+ − iχ− Q− − iχ+ Q+ .

(9.24)

We must choose the boundary conditions at the string endpoints such that the action
is stationary at solutions of the equations of motion, which means that the boundary
term arising from a general variation must vanish. The relevant terms in the action
giving rise to new boundary terms are
Z Z π n
i √  o
λ− ψ − · ψ −
0 0
− λ+ ψ+ · ψ+ + T χ− ψ− − χ+ ψ+ · X 0 . (9.25)

Irelevant = dt
0 2

and the new boundary terms are

i
Z h √ − +

iπ
δI|on−shell = dt T e (ψ− · δψ− − ψ+ · δψ+ ) + T χ ψ− − χ ψ+ · δX .
2 0
(9.26)
To simplify things, we’ll assume that the previous open string NG b.c.s still apply,
so that we now need to find b.c.s for the anticommuting variables such that the new
boundary terms are zero. We shall restrict attention to the spinning string with free
ends, in which case the δX term is zero only if the D-vector coefficient is zero. The
other term must be zero too so we arrive at the conclusion that

χ− ψ−
m
− χ+ ψ+
m


ends
= 0, (ψ− · δψ− − ψ+ · δψ+ )ends = 0 , (9.27)

The least restrictive solution to these requirements is to impose, at each end sepa-
rately, the boundary conditions
m
ψ+ m
|end = ± ψ− |end & χ+ |end = ± χ− |end . (9.28)

The sign at any given end is not significant because we are free to redefine ψ± → ±ψ± ,
so we may choose
ψ+ |σ=0 = ψ− |σ=0 . (9.29)
However, the relative sign is significant so we now have two cases to consider:

• Ramond sector. Same sign boundary conditions: ψ+ |σ=π = ψ− |σ=π . In this

case the Fourier series expansions of ψ± are

1 X ∓ikσ
ψ± (t, σ) = √ e dk (t) , (9.30)
2π k∈Z

– 98 –
 where the dk are anticommuting coefficient functions of t. This obviously satis-
fies the boundary condition (9.29) at σ = 0, and the same boundary condition
is satisfied at σ = π because e−ikπ = eikπ for integer k.
There is a similar expansion for the parameters ± . In particular, in the con-
formal gauge, where ± is a function only of σ ± , we have
±
X
± (σ ± ) = ±
0 + eikσ k (9.31)
k6=0

The constants ±
0 are the parameters for a 2D (1, 1) supersymmetry (one left-
handed and one right-handed spinor charge).

• Neveu-Schwarz sector. Opposite sign boundary conditions: ψ+ |σ=π = −ψ− |σ=π .

In this case
1 X ∓irσ
ψ± (t, σ) = √ e br (t) . (9.32)
2π 1
r∈Z+ 2

The NS boundary condition is satisfied because e−irπ = −eirπ for r ∈ Z + 12 .

It might now seem that we have two different types of spinning string. However,
consistency ultimately requires that we include both the Ramond and the Neveu-
Schwarz strings as two “sectors” of a single RNS string. We shall now examine these
two sectors separately, and obtain the light-cone gauge action for both.

9.2.1 Ramond sector

For the Fourier series expansion (9.30) we have
Z π ∞
i n o iX
˙ i ˙
X
dσ ψ− · ψ̇− + ψ+ · ψ+ = d−k · dk = d0 · d0 + i d−k d˙k . (9.33)
2 0 2 2 k=1
k∈Z

The full Ramond string action in Fourier space therefore takes the form
Z n ∞  
i X 1
IR = dt ẋ pm + d0 · d˙0 + i
m
α−k · α̇k + d−n · d˙n
2 k
X ok=1
− (λ−n Ln + iχ−n Fn ) , (9.34)
n∈Z

where Ln and Fn are Fourier coefficients of the periodic functions H(σ) and Q(σ)
constructed from H± and Q± by the “doubling trick” that we used previously56 :
1X X
Ln = [αn−k · αk + k dk−n · dk ] , Fn = αk · dn−k . (9.35)
2
k∈Z k∈Z

56
For example, Q(σ) = Q− (σ) for 0 ≤ σ ≤ π and Q(σ) = Q+ (σ) for π ≤ σ ≤ 2π.

– 99 –
The canonical PB relations are

αkm , α−k
n
= −ik η mn , dm n
= −iη mn .
 
PB −k , dk PB
(9.36)

Using these relations, one can show that

{Lm , Ln }P B = −i (m − n) Lm+n ,
m 
{Lm , Fn }P B = −i − n Fm+n ,
2
{Fm , Fn }P B = −2iLm+n . (9.37)

Notice that this implies that the Fn are (classically) the Fourier components of an
operator57 of conformal dimension 3/2.
As for the NG string, the constraints generate gauge invariances via Poisson
brackets, and the combination
X
(ξ−n Ln + i−n Fn ) , (9.38)
n∈Z

generates the gauge transformations

X
δαn = −in (ξn α0 + in d0 ) − in (ξk αn−k + ik dn−k ) ,
k6=n
X
δdn = n α0 − inξn d0 + (k αn−k − inξk dn−k ) . (9.39)
k6=n

We can fix all but the zero-mode gauge invariance with parameter α0 by the gauge-
fixing conditions
αk+ = 0 (k 6= 0) , d+
k = 0 (∀k) . (9.40)
A gauge transformation of these conditions yields, assuming that α0+ 6= 0, the equa-
tions ξk = 0 for k 6= 0 and k = 0 for all k. We may then solve the equations Ln = 0
(n 6= 0) and Fn = 0 (all n) to get expressions for αn− (n 6= 0) and d− n (all n). These
expressions are needed to get the Lorentz generators in terms of transverse canonical
variables, but they are not needed for the action, which is
Z ( ∞   )
i X i
IR = dt ẋm pm + d0 · ḋ0 + α−k · α̇k + id−k · ḋk − λ0 L0 , (9.41)
2 k=1
k
√
where, now, recalling that α0 = p/ πT for an open string,
∞
p2 X
L0 = + (α−k · αk + k d−k · dk ) . (9.42)
2πT k=1

57
Light-cone component of a worlsheet supersymmetry current.

– 100 –
From this we see that the mass-shell constraint is p2 + M2 = 0 with

M2 = 2πT NR , NR = Nb + Nf , (9.43)

where
∞
X ∞
X
Nb = α−n · αn , Nf = k d−k · dk . (9.44)
n=1 k=1

In the quantum theory, the transverse oscillator variables have the canonical
(anti)commutation relations
 I J 
αk , α−k = kδ IJ ,
 I
d−k , dJk = δ IJ

k ≥ 0. (9.45)

We define the oscillator vacuum in the usual way:

αkm |0i = 0 & dm

k |0i = 0 , k > 0. (9.46)

We also choose the operator ordering in the operator versions of Nb and Nf as given
in (9.66), so that both annihilate the oscillator vacuum. Other states in the Fock
space are found by acting with the creation operators {α−k , d−k ; k > 0}. Acting with
α−k raises the eigenvalue of Nb by k and acting with d−k raises the eigenvalue of Nf
by k, so the eigenvalues of the level number N are non-negative integers and we can
organise the spectrum according to this level number.
As for the NG string in light-cone gauge, Lorentz invariance of the quantum the-
ory is not guaranteed; this has to be checked by a computation of the commutation
relations of the operators representing the Lorentz generators. Whereas this calcu-
lation yields the result that D = 26 for the NG string, it yields the result D = 10
for the Ramond string, so we now restrict to D = 10.
Recall that requiring Lorentz invariance of the quantum NG string in light-cone
gauge also fixed the intercept parameter a. For the Ramond string we find a = 0.
This is to be expected from its interpretation as the oscillator zero-point energy
because every contribution from a Bose oscillator is now cancelled by a contribution
from a Fermi oscillator. This means that the mass-squared at level NR = Nf + Nb
is (2πT )NR . In particular, the NR = 0 states are massless; it is “states” and not
“state” because the oscillator vacuum is degenerate; the operators d0 commute with
NR but satisfy the anti-commutation relations

1
dI0 , dJ0 = δ IJ dI0 → √ γ I ,

⇒ (9.47)
2

where γ I are the 16×16 Dirac matrices, which we may choose to be real. The 16×16
matrices
1
S0IJ = γ [I γ J] (9.48)
2

– 101 –
obey the commutation relations of the Lie algebra of Spin(8), which is the same as
the Lie algebra of SO(8). A special feature of Spin(8) is that its real 16-component
spinor representation is reducible. Observe that the matrix
γ9 = γ 1 γ 2 · · · γ 8 (9.49)
has the properties
γ92 = 1 γ9 , γ I = 0 .

(9.50)
The latter property implies that γ9 has zero trace, so we can choose a basis for which
 
I8 0
γ9 = . (9.51)
0 −I8
This basis is consistent with reality of the matrices γ I , so a real 16-component spinor
of Spin(8) is the sum of two 8-dimensional representations: the eigenspinors of γ9
with eigenvalues ±1; equivalently a chiral and an anti-chiral spinor. In math-speak
they are the 8s (spinor) and 8c (conjugate spinor) of Spin(8).
The restriction to D = 10 is confirmed when we quantise the Ramond string
covariantly. In the path integral approach, we will now have new FP ghosts, usually
denoted by (β, γ), arising from the conformal gauge fixing of the gauge symmetries
generated by Q± . These FP ghosts are commuting because the gauge parameters
are anticommuting, and this flips the sign of their contribution to the central charge.
From Q.III.6 we know that an anticommuting (b, c) system for which [b] = J and
[c] = 1 − J (or vice versa since the formula is invariant under J → 1 − J) has central
charge
c = −2(6J 2 − 6J + 1) (⇒ c = −26 for J = 2). (9.52)
In the standard (b, c) case we have J = 2 (because the Lm are Fourier components of a
worldsheet field (2D stress tensor) whose classical conformal dimension is 2) and this
gives c = −26. In the (β, γ) case we have J = 3/2 because Fn are Fourier components
of a worldsheet field (2D supercurrent) whose classical conformal dimension is 3/2.
Allowing for the change of sign due to the opposite “statistics” we deduce that the
(β, γ) ghost system has central charge
c = +2(6J 2 − 6J + 1)|J= 3 = 11 . (9.53)
2

The total ghost central charge has to cancel against the non-ghost contribution. We
know that the worldsheet scalar fields contribute c = D. As we are discussing the
open string, for which ψ± share the same Fourier coefficients, we need only consider
ψ+ , say. Its action is essentially of the (b, c) form but with J = 12 (since J = 1 − J)
and also with 1/2 the number of variables since both b and c are replaced by ψ+ .
This gives c = 1/2, and hence c = D/2 for D-vector ψ+ . Therefore, the central
charge cancels when
1
D + D − 26 + 11 = 0 ⇒ D = 10 . (9.54)
2

– 102 –
To summarise: the quantum Ramond string is Lorentz invariant in light-cone gauge
only if D = 10, and the NR = 0 state is then massless and degenerate. These massless
states transform as the 8s ⊕ 8c representation of the transverse rotation group. All
NR > 0 states are massive.

9.2.2 Neveu-Schwarz sector

For the Fourier series expansion (9.32) we have
∞
i π n
Z o i X X
dσ ψ− · ψ̇− + ψ+ · ψ+ = b−r · ḃr = i b−r ḃr , (9.55)
2 0 2 1 1
r∈Z+ 2 r= 2

The full NS string action in Fourier space is

Z n ∞ ∞
m
X i X
IN S = dt ẋ pm + α−k · α̇k + ib−r · ḃr
k=1
k 1
r=
X X o 2
− λ−n Ln − i χ−r Gr , (9.56)
n∈Z r∈Z+ 21

where
1X 1 X
Ln = αn−k · αk + r bn−r · br (n ∈ Z)
2 2 1
k∈Z r∈Z+ 2
 
X 1
Gr = αk · br−k r ∈Z+ . (9.57)
2
k∈Z

The canonical PB relations are

 m n
αk , α−k P B = −ik η mn , bm n
= −iη mn .

−r , br PB
(9.58)
The PB algebra of the constraint functions is
{Lm , Ln }P B = −i (m − n) Lm+n ,
m 
{Lm , Gr }P B = −i − r Gm+r ,
2
{Gr , Gs }P B = −2iLr+s . (9.59)
The gauge transformations generated by the linear combination of constraint
functions X X
ξ−n Ln + i−r Gr , (9.60)
n∈Z r∈Z+ 21
are
X X
δαn = −inξ0 αn − in ξm αn−m + n r bn−r
m6=n r
X X
δbr = α0 r + αr−s s − ir ξm br−m . (9.61)
s6=r m

– 103 –
Invariance under these gauge transformations may be fixed by an extension of the
light-cone gauge condition, as for the Ramond string:

αn+ = 0 (n 6= 0) & b+
r = 0 (∀r) . (9.62)

Let’s check this: a gauge variation of these conditions yields

0 = −inα0+ ξn , (n 6= 0) & 0 = α0+ r . (9.63)

This tells us that the gauge invariance is completely fixed except for the ξ0 transfor-
mation (assuming that α0+ is non-zero). Having fixed the gauge in this way, we may
now solve Ln = 0 for αn− (n 6= 0) and Gr = 0 for b− r . The resulting expressions for
these variables are needed only for the Lorentz generators, not for the gauge-fixed
action, which is
 
Z ∞ ∞

m
X i X 
I = dt ẋ pm + α−k · α̇k + i b−r · ḃr − λ0 L0 , (9.64)
 kk=1 1

r= 2

√
where now, recalling that α0 = p/ πT for open strings,
∞ ∞
p2 X X
L0 = + α−n · αn + r b−r · br . (9.65)
2πT n=1 1
r= 2

From this we see that the mass-shell constraint is p2 + M2 = 0, with

∞
X ∞
X
2
M = 2πT (Nb + Nf ) ; Nb = α−n · αn , Nf = rb−r · br . (9.66)
n=1 r= 21

In the quantum theory, the oscillator variables have the canonical (anti)-commutation
relations

αkI , α−k
J
= kδ IJ ,
 
k = 1, 2, . . .
 I J 1 3
b−r , br = δ IJ , r = , ,... (9.67)
2 2
and the oscillator vacuum |0i is defined by

αk |0i , k > 0, br |0i = 0 , r > 0. (9.68)

The operators Nb and Nf both annihilate the oscillator vacuum for the ordering as
given in (9.66), so in a basis where they are diagonal,

M 2 = 2πT (NN S − a) , N = Nb + Nf , (9.69)

– 104 –
where the constant a is now introduced to allow for the ambiguity due to operator
ordering. We should no longer expect a bose-fermi cancellation of the zero-point
energies because of the different “moding” of the bose and fermi oscillators. In
contrast to the Ramond string, the oscillator ground state is non-degenerate, so it
corresponds to a scalar particle, which will be a tachyon if a > 0.
Let’s now look at the first excited states. The operator Nb has integer eigenvalues
0, 1, 2, . . . but the operator Nf has eigenvalues 12 , 32 , . . ., so 2NRS is a non-negative
integer, and the first excited states have NN S = 12 . These states are

bI− 1 |0i , I = 1, . . . , D − 2 . (9.70)

2

By the same argument that we used at level-1 of the NG string, these must be the
polarization states of a massless vector if the quantum theory is to preserve Lorentz
invariance. This means that we must now choose
1
a= . (9.71)
2
On the other hand, the zero point energy −a is formally given by
"∞ ∞  #
1 X X 1
−a = (D − 2) (n + 1) − n+ . (9.72)
2 n=0 n=0
2

The relative sign is due to the opposite sign contribution of bosonic and fermionic
oscillators. We can perform the sum using the fact that the generalized ζ-function,
∞
X
ζ(s, q) = (n + q)−s , (9.73)
n=0

has a unique value at s = −1:

1
6q 2 − 6q + 1 .


ζ(−1, q) = − (9.74)
12
This gives  
1 1 1 (D − 2)
−a = (D − 2) − − =− . (9.75)
2 12 24 16
Using this formula, we see that a = 12 implies that D = 10. This value is confirmed
by a computation of the Lorentz commutators. The theory is Lorentz invariant only
if a = 21 and D = 10.

To summarize: Lorentz invariance of the NS string requires D = 10 and then we

have a scalar tachyon with α0 M 2 = − 12 at level NN S = 0 and a massless vector at
level NN S = 1/2 with physical polarizations in the 8v (vector) representatiuon of the
transverse Spin(8) rotation group. All NN S ≥ 1 states are massive.

– 105 –
10. Superstrings
Open string interactions always allow closed string states to appear as virtual par-
ticles in loops of the string loop expansion, so the quantum theory is inconsistent
unless we include closed strings. For closed strings, the choice between R and NS
boundary conditions becomes the choice between worldsheet “fermions” are periodic
or anti-periodic in σ; in the context of a path-integral formulation of the quantum
theory, for which we use the conformal gauge action (in which ψ± are the compo-
nents of a worldsheet spinor) this choice amounts to a choice of spin structure on the
Riemann surface worldsheet. Consistency of the quantum theory forces a sum over
spin structures and this forces us to combine the R and NS strings.
The NS sector of the spinning string still has a tachyon. At the free-string level
there would be nothing to stop us from simply discarding the tachyon state; we could
have already done that for the NG string. However, we cannot expect arbitrary trun-
cations of the spectrum to be consistent with interactions; anything we throw out
will usually reappear in loops in the quantum theory, making the truncation incon-
sistent. The only way to guarantee consistency of some truncation that removes the
tachyon is by means of a symmetry. If there is a symmetry that excludes the tachyon,
then its exclusion will be consistent if we can introduce interactions consistent with
the symmetry. Nothing of that kind is available for the NG string, or for the NS-
sector of the spinning string, but once we combine the NS sector with the R sector,
a possibility presents itself: spacetime supersymmetry.

10.1 RNS Superstring

Let’s take a closer look at the low-lying states of the combined R and NS spinning
strings, organised according to the value of α0 M 2 (recall that for integer α0 M 2 both
R and NS states contribute, with NN S = NR + 21 ):

α0 M 2 = − 12 : n |0iN S o 1
α0 M 2 = 0 : I
|0iR ; b− 1 |0iN S (8s ⊕ 8c ) ⊕ 8v
n 2 o
α0 M 2 = 1
2
: I
α−1 |0iN S , bI− 1 bJ− 1 |0iN S 8v ⊕ 28
2 2
α0 M 2 = 1 :
 I
n α−1 |0iR , dI−1 |0iR o (8v ⊕ 8v ) ⊗ (8s ⊕ 8c )
I J I I J K
α−1 b− 1 |0iN S , b− 3 |0iN S , b− 1 b− 1 b− 1 |0iN S (8v ⊗ 8v ) ⊕ 8v ⊕ 56v
2 2 2 2 2
(10.1)
A useful lemma for Spin(8) tensor products is

• 8 × 8 lemma. Let i = v, s, c label the three 8-dimensional representations of

Spin(8). Then

8i ⊗ 8i = 1 ⊕ 28 ⊕ 35i
8i ⊗ 8j = 8k ⊕ 56k (i, j, k cyclic) . (10.2)

– 106 –
 The 35v is a 2nd-rank symmetric traceless tensor, the 28 is a 2nd-rank antsym-
metric tensor, and the 56v is a 3rd-rank antisymmetric tensor. The 35s (35c )
is an (anti-)self-dual 4th-rank antisymmetric tensor. The 56s is a chiral vector-
spinor and the 56c is an anti-chiral vector-spinor.

Now we define a notion of “G-parity” and declare that all physical states must
be G-parity even. First we define worldsheet bosons (commuting worldsheet fields)
to have even G-parity and worldsheet fermions (anticommuting worldsheet fields) to
have odd G-parity. So far G-parity is determined by Grassmann parity, but the way
this works for d0 is that if the chiral component of |0iR (the 8s ) is assigned even
G-parity then the anti-chiral component (the 8c ) must be assigned odd G-parity;
we’ll call the chiral component |0iR+ and the anti-chiral component |0iR− . The NS
oscillator vacuum is assigned odd G-parity (this removes the tachyon). The G-parity
assignments of the oscillator vacua are therefore

G − even : |0iR+
(10.3)
G − odd : |0iR− & |0iN S

The requirement of even G-parity is called the Gliozzi-Scherk-Olive (GSO) projec-

tion, and the combined R and NS spinning strings with this projection is called the
RNS superstring.
To see why it’s called a “superstring” we look at the low-mass even G-parity
states
n o
α0 M 2 = 0 : |0iR+ ; bI− 1 |0iN S 8s ⊕ 8v
2
α0 M 2 = 1 : n
 I

α−1 |0iR+ , dI−1 |0iR− o (8v ⊗ 8s ) ⊕ (8v ⊗ 8c )
I J I I J K
α−1 b− 1 |0iN S , b− 3 |0iN S , b− 1 b− 1 b− 1 |0iN S (8v ⊗ 8v ) ⊕ 8v ⊕ 56v
2 2 2 2 2
(10.4)
We have 8+8 massless states. The 8v are the states of a D = 10 photon and the 8s are
the states of a D = 10 Majorana (read that as “real”) chiral spinor. These make up
the fields of the Maxwell supermultiplet of D = 10, i.e. spacetime, supersymmetry.
At the next level we have massive states.
 
Spacetime fermions (8v ⊗ 8s ) ⊕ (8v ⊗ 8c ) = 128
(10.5)
Spacetime bosons (8v ⊗ 8v ) ⊕ 8v ⊕ 56v = 44 ⊕ 128

where the last column gives the Spin(9) representations. This is the physical light-
cone content of a D = 10 massive spin-2 supermultiplet. In fact, at every level the
spectrum is a supermultiplet of D = 10 spacetime supersymmetry. There is a simple
way to see why, found by Green and Schwarz.

– 107 –
10.2 The Green-Schwarz superstring∗
There is a way to reformulate the RNS superstring that makes manifest its spacetime
supersymmetry. It relies on the triality property of the Spin(8) algebra. The repre-
sentation theory for Spin(8) is unchanged by a permutation of its three 8-dimensional
representations. Consider the Ramond string in light-cone gauge: if we make the re-
placement
 I
→ {θkα ; α = 1, . . . , 8} ,

dk ; I = 1, . . . , 8 (10.6)
where the θkα are an 8c of anticommuting variables then we have only renamed and
relabeled the anticommuting light-cone gauge variables, giving them a new spacetime
interpretation (as an anti-chiral spinor of the transverse rotation group). Quantizing
this (light-cone gauge) Green-Schwarz superstring leads to exactly the same results
that we found previously for the Ramond string but with a permutation of the Spin(8)
representations
8v → 8s 8s → 8c , 8c → 8v . (10.7)
The ground state of the GS string has zero mass, like the Ramond string, but the
Spin(8) representations at this level are

8v ⊕ 8s . (10.8)

This coincides with the ground states of the RNS superstring. In other words, the
result of imposing the GSO projection on the RNS string is reproduced by imposing
Ramond boundary conditions on the GS superstring, discarding its NS sector. This
obviously ensures bose-fermi matching at all levels58 . To show that one gets super-
multiplets at all levels requires more work, which we are not going to do but let’s
look at the first excited states:

I α
α−1 |0iGS , θ−1 |0iGS . (10.9)

This gives us states with α0 M 2 = 1 in the Spin(8) representations

(8v ⊕ 8s ) ⊗ (8v ⊕ 8c ) = [(1 ⊕ 8v ⊕ 28) ⊕ (35v ⊕ 56v )] ⊕ [(8s ⊕ 56c ) ⊕ (8c ⊕ 56s )]
= [44 ⊕ 84] ⊕ 128 , (10.10)

where the representations in the second line are those of Spin(9). Those in the square
bracket are the bosons: the 44 is a symmetric traceless tensor, which describes
a massive spin-2 particle, and the 84 is a third-order antisymmetric tensor. All
fermions are in the 128, which describes a massive spin-3/2 particle; all together we
have the massive spin-2 multiplet of N = 1 D=10 supersymmetry.
58
The tensor product of 8c ⊕ 8v with any spin(8) tensor T is (8c ⊗ T ) ⊕ (8v ⊗ T ), which exhibits
a bose-fermi matching, and the same is true if T is a tensor-spinor.

– 108 –
10.3 Closed superstrings∗
As for the NG string, the physical states at each level of the closed superstring are
just tensor products of two copies of the physical states of the open superstring at
that level. In other words, the spin(8) representation content of the massless states
will be the tensor product of two copies of the spin(8) representations of the massless
states of the open string. For the latter we had to choose between the two possibilities
of (??); the choice didn’t matter there but now we have a relative choice to make
because, for the closed GS superstring we have two sets of oscillators, one for the
left-movers and one for the right-movers, and this means that we get two distinct
closed superstring theories according to the relative choice of spin(8) chirality for the
anticommuting variables θ±
α α̇
• IIA. Opposite spin(8) chirality: (θ− , θ+ ). We use α̇ = 1, . . . , 8 for the anti-
chiral 8c spinor.
α α
• IIB. Same spin(8) chirality: (θ− , θ+ ).

As Hamlet put it so eloquently: IIB or not IIB, that is the question.

Using this lemma we find the following results, which we organise according to
their RNS origin:

• IIA string. The Spin(8) representation content of massless states is



 1 ⊕ 28 ⊕ 35v NS − NS

8v ⊕ 56v R−R

(8v ⊕ 8c ) ⊗ (8v ⊕ 8s ) = (10.11)
 8s ⊕ 56s
 R − NS

 8 ⊕ 56
c c NS − R

We get a spinor ground state for the Ramond open string, so the fermions
of the closed superstring come from its R-NS and NS-R sectors. Notice that
these give spinorial spin(8) states of opposite chirality. The states in the R-R
sector are bi-spinors, which are equivalent to antisymmetric tensors; for the IIA
superstring we get a vector AI and a third-order antisymmetric tensor AIJK .

• IIB string.The Spin(8) representation content of massless states is



 1 ⊕ 28 ⊕ 35v NS − NS

1 ⊕ 28 ⊕ 35c R − R

(8v ⊕ 8c ) ⊗ (8v ⊕ 8c ) = (10.12)

 8s ⊕ 56s R − NS

 8 ⊕ 56
s s NS − R

The spinorial states from the R-NS sector now have the same chirality as those
from the NS-R sector. The R-R states are now a scalar A, a 2nd-order anti-
symmetric tensor AIJ and a 4th-order self-dual anti-symmetric tensor AIJKL ;

– 109 –
 the self-duality means that
1
AIJKL = IJKLM N P Q AM N P Q (10.13)
4!
where  is the alternating spin(8) invariant tensor.

In both IIA and IIB cases we get the same NS-NS massless states, which are also
the same as those of the closed NG string; as we saw in that case, the 35v can be
interpreted as the physical polarisation states of a massless spin-2 particle.
All remaining massless bosonic states come from the R-R sector. For example,
for the IIA superstring, the full set of tensorial spin(8) representations, coming from
the combined NS-NS and R-R sectors, is

(1 ⊕ 8v ⊕ 35i ) ⊕ (288v ⊕ 56v ) = 44 ⊕ 84 , (10.14)

where the second equality gives the spin(9) representations. They are the same that
we found at the first massive level of the open superstring. In that context the
spin(8) representations had to combine into spin(9) representations for consistency
with Lorentz invariance. That’s not the case here because we are now dealing with the
massless particles in the IIA superstring spectrum; in the massless particle context
the spin(9) representations are what would be required for Lorentz invariance in
eleven spacetime dimensions, i.e. D=11. In fact, this is also true for the massless
fermions of the IIA superstring.

11. M-Theory∗
Here are a few facts about the fermions: each of the 8s ⊕ 56s and 8c ⊕ 56c states are
the physical polarisation states of a massless D=10 spin-3/2 particle, either chiral
or anti-chiral. Consistency of the interactions of massless spin-3/2 particles requires
supersymmetry, so their presence in the massless spectrum of the closed spinning
string is a simple way of seeing why the GSO projection is necessary for consistency59 .
It follows that the effective D=10 spacetime action for the massless states of either
the IIA or the IIB superstring is an N = 2 D = 10 supergravity theory. There are
two of them, according to whether the two spin-3/2 fields have the same (IIB) or
opposite (IIA) chirality.
The maximal spacetime dimension for which a supergravity theory exists is
D=11, and dimensional reduction of the unique D=11 supergravity theory to D=10
yields the IIA supergravity theory, which is the effective low-energy theory for the
massless states of the IIA superstring. For a long time this was seen as just a coinci-
dence, since superstring theory appeared to require D=10. Another “coincidence” is
59
There is no analogous argument for the open spinning string but quantum consistency of any
open string theory requires the inclusion of closed strings.

– 110 –
that the Green-Schwarz construction of a Lorentz covariant and manifestly spacetime
supersymmetric action for the IIA and IIB superstrings, also applies to membranes
in D=11 (but not to any other p-branes, for any p > 0, for any D > 11), and its
dimensional reduction yields the IIA GS superstring action.
In string theory one can compute, in principle, the amplitude for scattering of any
particles in the string spectrum to arbitrary order in a string-loop expansion, with
each term being UV finite. However, this expansion is a divergent one; we cannot
sum the series, even in principle. This is also typically the case in QFT but the
perturbation expansions of QFT are usually derived from an action, and some QFT’s
can be defined non-perturbatively, e.g. as a continuum limit of a lattice version.
String theory is different because all amplitudes are found “on-shell”, and a spacetime
action constructed order by order from these amplitudes has no more information in
it than the computed amplitudes. String theory just gives us a perturbation series; it
does not tell us what it is that is being perturbed. The completed non-perturbative
theory could be something completely different. The fact that D=10 is the critical
dimension of superstring shows only that D ≥ 10 because some dimensions could be
invisible in perturbation theory.
Fortunately, the constraints due to maximal supersymmetry are so strong that
the effective spacetime field theory for the massless particles of the superstring con-
tains a lot of information about non-perturbative string theory, sufficient to show
that the five distinct superstring theories60 are unified by some 11-dimensional the-
ory, known as M-theory, and that this theory includes D=11 supergravity. Unfortu-
nately, we don’t really know what this theory is, so this is a good place to finish this
course on String Theory.

60
In addition to the IIA and IIB closed superstring theories we have the Type I open superstring
(this is the string theory that results from inclusion of the additional features needed to get quantum
consistency of interacting open superstrings) and two heterotic superstring theories (for which the
worldsheet action has only (1, 0) D=2 supersymmetry).

– 111 –