eScholarship provides open access, scholarly publishing

services to the University of California and delivers a dynamic

research platform to scholars worldwide.
Lawrence Berkeley National Laboratory
Peer Reviewed
Title:
M-theory through the looking glass: Tachyon condensation in the E8 heterotic string
Author:
Horava, Petr
Publication Date:
07-11-2008
Publication Info:
Lawrence Berkeley National Laboratory
Permalink:
http://www.escholarship.org/uc/item/5j30g398
Abstract:
We study the spacetime decay to nothing in string theory and M-theory. First we recall a
nonsupersymmetric version of heterotic M-theory, in which bubbles of nothing -- connecting the
two E_8 boundaries by a throat -- are expected to be nucleated. We argue that the fate of
this system should be addressed at weak string coupling, where the nonperturbative instanton
instability is expected to turn into a perturbative tachyonic one. We identify the unique string
theory that could describe this process: The heterotic model with one E_8 gauge group and
a singlet tachyon. We then use worldsheet methods to study the tachyon condensation in the
NSR formulation of this model, and show that it induces a worldsheet super-Higgs effect. The
main theme of our analysis is the possibility of making meaningful alternative gauge choices for
worldsheet supersymmetry, in place of the conventional superconformal gauge. We show in a
version of unitary gauge how the worldsheet gravitino assimilates the goldstino and becomes
dynamical. This picture clarifies recent results of Hellerman and Swanson. We also present
analogs of R_\xi gauges, and note the importance of logarithmic CFT in the context of tachyon
condensation.
M-Theory Through the Looking Glass:
Tachyon Condensation in the E
8
Heterotic String*
Petr Horava and Cynthia A. Keeler
Berkeley Center for Theoretical Physics and Department of Physics
University of California, Berkeley, CA, 94720-7300
and
Theoretical Physics Group, Lawrence Berkeley National Laboratory
Berkeley, CA 94720-8162, USA
September 2007
*This work was supported in part by the Director, Office of Science, Office of High Energy Physics, of
the U.S. Department of Energy under Contract No. DE-AC02-05CH11231.
DISCLAIMER
This document was prepared as an account of work sponsored by the United States Government. While this document is believed to contain correct
information, neither the United States Government nor any agency thereof, nor The Regents of the University of California, nor any of their employees,
makes any warranty, express or implied, or assumes any legal responsibility for the accuracy, completeness, or usefulness of any information, apparatus,
product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product,
process, or service by its trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or
favoring by the United States Government or any agency thereof, or The Regents of the University of California. The views and opinions of authors
expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof or The Regents of the University of
California.
a
r
X
i
v
:
0
7
0
9
.
3
2
9
6
v
1


[
h
e
p
-
t
h
]


2
0

S
e
p

2
0
0
7
Preprint typeset in JHEP style - HYPER VERSION
M-Theory Through the Looking Glass:
Tachyon Condensation in the E
8
Heterotic String
Petr Horava and Cynthia A. Keeler
Berkeley Center for Theoretical Physics and Department of Physics
University of California, Berkeley, CA, 94720-7300
and
Theoretical Physics Group, Lawrence Berkeley National Laboratory
Berkeley, CA 94720-8162, USA
Abstract: We study the spacetime decay to nothing in string theory and M-theory. First
we recall a nonsupersymmetric version of heterotic M-theory, in which bubbles of nothing
connecting the two E
8
boundaries by a throat are expected to be nucleated. We argue that
the fate of this system should be addressed at weak string coupling, where the nonperturba-
tive instanton instability is expected to turn into a perturbative tachyonic one. We identify
the unique string theory that could describe this process: The heterotic model with one E
8
gauge group and a singlet tachyon. We then use worldsheet methods to study the tachyon
condensation in the NSR formulation of this model, and show that it induces a worldsheet
super-Higgs eect. The main theme of our analysis is the possibility of making meaningful
alternative gauge choices for worldsheet supersymmetry, in place of the conventional supercon-
formal gauge. We show in a version of unitary gauge how the worldsheet gravitino assimilates
the goldstino and becomes dynamical. This picture claries recent results of Hellerman and
Swanson. We also present analogs of R

gauges, and note the importance of logarithmic CFT

in the context of tachyon condensation.
Contents
1. Introduction 2
2. Spacetime Decay to Nothing in Heterotic M-Theory 4
2.1 The E
8
E
8
Heterotic M-Theory 4
2.2 The Other Side of the E
8
Wall 5
2.3 Heterotic String Description at Weak Coupling 6
3. The Forgotten E
8
Heterotic String 6
3.1 The Free Fermion Language 7
3.2 The Language of Free Bosons 9
4. Tachyon Condensation in the E
8
Heterotic String 10
4.1 The General Philosophy 10
4.2 The Action 10
4.2.1 Linear dilaton 11
4.2.2 Superpotential and the tachyon prole 11
4.3 The Lone Fermion as a Goldstino 12
5. Tachyon Condensation and the Worldsheet Super-Higgs Mechanism 13
5.1 Superconformal Gauge 14
5.2 Alternatives to Superconformal Gauge 16
5.3 Liberating the Worldsheet Gravitino in an Alternative Gauge 17
5.4 R

Gauges 21
6. The Condensed Phase: Exploring the Worldsheet Theory at = 23
7. Conclusions 24
7.1 Overview 24
7.2 Further Analysis of the E
8
System 25
7.3 Towards Non-Equilibrium String Theory 25
A. Supersymmetry Conventions 26
A.1 Flat worldsheet 26
A.2 Local worldsheet supersymmetry 27
A.3 Lightcone coordinates 28
1
B. Evaluation of the Determinants 29
B.1 J

31
B.2

J 32
1. Introduction
The motivation for this paper is to further the studies of time-dependent backgrounds in string
theory. In particular, we concentrate on the problem of closed-string tachyon condensation,
and its hypothetical relation to the spacetime decay to nothing.
Open-string tachyon condensation is now relatively well-understood (see, e.g., [1, 2] for
reviews), as a description of D-brane decay into the vacuum (or to lower-dimensional stable
defects). On the other hand, the problem of the bulk closed-string tachyon condensation
appears related to a much more dramatic instability in which the spacetime itself decays, or
at least undergoes some other extensive change indicating that the system is far from equi-
librium. In the spacetime supergravity approximation, this phenomenon has been linked to
nonperturbative instabilities due to the nucleation of bubbles of nothing [3]. One of the rst
examples studied in the string and M-theory literature was the nonsupersymmetric version of
heterotic M-theory [4], in which the two E
8
boundaries of eleven-dimensional spacetime carry
opposite relative orientation and consequently break complementary sets of sixteen super-
charges. At large separation between the boundaries, this system has an instanton solution
that nucleates bubbles of nothing. In eleven dimensions, the nucleated bubbles are smooth
throats connecting the two boundaries; the nothing phase is thus the phase on the other
side of the spacetime boundary.
In addition to this eect, the boundaries are attracted to each other by a Casimir force
which drives the system to weak string coupling, suggesting some weakly coupled heterotic
string description in ten dimensions. In the regime of weak string coupling, we expect the
originally nonperturbative instability of the heterotic M-theory background to turn into a
perturbative tachyonic one.
We claim that there is a unique viable candidate for describing this system at weak
string coupling: The tachyonic heterotic string with one copy of E
8
gauge symmetry, and
a singlet tachyon. In this paper, we study in detail the worldsheet theory of this model
in the NSR formalism with local worldsheet (0, 1) supersymmetry when the tachyon
develops a condensate that grows exponentially along a lightcone direction X
+
. There is a
close similarity between this background and the class of backgrounds studied recently by
Hellerman and Swanson [58].
1
The main novelty of our approach is the use of alternative
gauge choices for worldsheet supersymmetry, replacing the traditional superconformal gauge.
1
Similar spacetime decay has also been seen in solutions of noncritical string theory in 1+1 dimensions [9],
and noncritical M-theory in 2 + 1 dimensions [10].
2
We show that the worldsheet dynamics of spacetime tachyon condensation involves a super-
Higgs mechanism, and its picture simplies considerably in our alternative gauge.
Our main results were briey reported in [11]; in the present paper, we elaborate on the
conjectured connection to spacetime decay in heterotic M-theory, and provide more details
of the worldsheet theory of tachyon condensation, including the analysis of the super-Higgs
mechanism and its compatibility with conformal invariance.
Section 2 reviews the nonsupersymmetric version of heterotic M-theory, as a simple con-
guration that exhibits the spacetime decay to nothing. We argue that the dynamics of
this instability should be studied at weak string coupling, and advocate the role of the tachy-
onic E
8
heterotic model as a unique candidate for this weakly coupled description of the
decay. In Section 3, we review some of the worldsheet structure of the tachyonic E
8
heterotic
string. In particular, we point out that the E
8
current algebra of the nonsupersymmetric
(left-moving) worldsheet sector is realized at level two and central charge c
L
= 31/2; this is
further supplemented by a single real fermion of c
L
= 1/2.
Sections 4 and 5 represent the core of the paper, and are in principle independent of
the motivation presented in Section 2. In Section 4, we specify the worldsheet theory in the
NSR formulation, before and after the tachyon condensate is turned on. The condensate is
exponentially growing along a spacetime null direction X
+
. Conformal invariance then also
requires a linear dilaton along X

if we are in ten spacetime dimensions. We point out that

when the tachyon condensate develops, transforms as a candidate goldstino, suggesting a
super-Higgs mechanism in worldsheet supergravity.
Section 5 presents a detailed analysis of the worldsheet super-Higgs mechanism. Tradi-
tionally, worldsheet supersymmetry is xed by working in superconformal gauge, in which
the worldsheet gravitino is set to zero. We discuss the model briey in superconformal gauge
in Section 5.1, mainly to point out that tachyon condensation leads to logarithmic CFT.
Since the gravitino is expected to take on a more important role as a result of the super-
Higgs eect, in Section 5.2 and 5.3 we present a gauge choice alternative to superconformal.
This alternative gauge choice is inspired by the unitary gauge known from the conventional
Higgs mechanism in Yang-Mills theories. We show in this gauge how the worldsheet grav-
itino becomes a dynamical propagating eld, contributing c
L
= 11 units of central charge.
Additionally, we analyze the Faddeev-Popov determinant of this gauge choice, and show that
instead of the conventional right-moving superghosts , of superconformal gauge, we get
left-moving superghosts

, of spin 1/2. In addition, we show how the proper treatment
of the path-integral measure in this gauge induces a shift in the linear dilaton. This shift is
precisely what is needed for the vanishing of the central charge when the ghosts are included.
Thus, this string background is described in our gauge by a worldsheet conformal (but not
superconformal) eld theory. Section 6 points out some interesting features of the worldsheet
theory in the late X
+
region, deeply in the condensed phase.
In Appendix A we list all of our needed worldsheet supergravity conventions. Appendix B
presents a detailed evaluation of the determinants relevant for the body of the paper.
3
2. Spacetime Decay to Nothing in Heterotic M-Theory
The anomaly cancelation mechanism that permits the existence of spacetime boundaries in M-
theory works locally near each boundary component. The conventional realization, describing
the strongly coupled limit of the E
8
E
8
heterotic string [12, 13], assumes two boundary
components, separated by xed distance R
11
along the eleventh dimension y, each breaking
the same sixteen supercharges and leaving the sixteen supersymmetries of the heterotic string.
In [4], a nonsupersymmetric variant of heterotic M-theory was constructed, simply by
ipping the orientation of one of the boundaries. This ipped boundary breaks the comple-
mentary set of sixteen supercharges, leaving no unbroken supersymmetry. The motivation
behind this construction was to nd in M-theory a natural analog of D-brane anti-D-brane
systems whose study turned out to be so illuminating in superstring theories. Dp-branes
dier from Dp-branes only in their orientation. In analogy with Dp-Dp systems, we refer to
the nonsupersymmetric version of heterotic M-theory as E
8
E
8
to reect this similarity.
2
2.1 The E
8
E
8
Heterotic M-Theory
This model, proposed as an M-theory analog of brane-antibrane systems in [4], exhibits two
basic instabilities. First, the Casimir eect produces an attractive force between the two
boundaries, driving the theory towards weak coupling. The strength of this force per unit
boundary area is given by (see [4] for details):
T =
1
(R
11
)
11
5
2
14
_

0
dt t
9/2

2
(0[it), (2.1)
where R
11
is the distance between the two branes along the eleventh dimension y.
Secondly, as was rst pointed out in [4], at large separations the theory has a nonpertur-
bative instability. This instanton is given by the Euclidean Schwarzschild solution
ds
2
=
_
1
_
4R
11
r
_
8
_
dy
2
+
dr
2
1
_
4R
11
r
_
8
+ r
2
d
2

9
(2.2)
under the Z
2
orbifold action y y. Here r and the coordinates in the S
9
are the other ten
dimensions. This instanton is schematically depicted in Figure 1(b).
The probability to nucleate a single bubble of nothing of this form is, per unit boundary
area per unit time, of order
exp
_

4(2R
11
)
8
3
4
G
10
_
, (2.3)
where G
10
is the ten-dimensional eective Newton constant. As the boundaries are forced
closer together by the Casimir force, the instanton becomes less and less suppressed. Eventu-
ally, there should be a crossover into a regime where the instability is visible in perturbation
theory, as a string-theory tachyon.
2
Actually, this heterotic M-theory conguration is an even closer analog of a more complicated unstable
string theory system: A stack of D-branes together with an orientifold plane, plus anti-D-branes with an
anti-orientifold plane, such that each of the two collections is separately neutral. These collections are only
attracted to each other quantum mechanically, due to the one-loop Casimir eect.
4
R
(a) (b)
11
E
8
E
8
Figure 1: (a) A schematic picture of the E
8
E
8
heterotic M-theory. The two boundaries are
separated by distance R
11
, carry opposite orientations, and support one copy of E
8
gauge symmetry
each. (b) A schematic picture of the instanton responsible for the decay of spacetime to nothing.
The instanton is a smooth throat connecting the two boundaries. Thus, the bubble of nothing is in
fact a bubble of the hypothetical phase on the other side of the E
8
boundary.
2.2 The Other Side of the E
8
Wall
The strong-coupling picture of the instanton catalyzing the decay of spacetime to nothing
suggests an interesting interpretation of this process. The instanton has only one boundary,
interpolating smoothly between the two E
8
walls. Thus, the bubble of nothing that is being
nucleated represents the bubble of a hypothetical phase on the other side of the boundary of
eleven-dimensional spacetime in heterotic M-theory. In the supergravity approximation, this
phase truly represents nothing, with no apparent spacetime interpretation. The boundary
conditions at the E
8
boundary in the supergravity approximation to heterotic M-theory are
reective, and the boundary thus represents a perfect mirror. However, it is possible that
more rened methods, beyond supergravity, may reveal a subtle world on the other side of
the mirror. This world could correspond to a topological phase of the theory, with very few
degrees of freedom (all of which are invisible in the supergravity approximation).
At rst glance, it may seem that our limited understanding of M-theory would restrict
our ability to improve on the semiclassical picture of spacetime decay at strong coupling.
However, attempting to solve this problem at strong coupling could be asking the wrong
question, and a change of perspective might be in order. Indeed, the theory itself suggests
a less gloomy resolution: the problem should be properly addressed at weak string coupling,
to which the system is driven by the attractive Casimir force. Thus, in the rest of the paper,
our intention is to develop worldsheet methods that lead to new insight into the hypothetical
phase behind the mirror, in the regime of the weak string coupling.
5
2.3 Heterotic String Description at Weak Coupling
We conjecture that when the Casimir force has driven the E
8
boundaries into the weak
coupling regime, the perturbative string description of this system is given by the little-
studied tachyonic heterotic string model with one copy of E
8
gauge symmetry [14].
3
The
existence of a unique tachyonic E
8
heterotic string theory in ten spacetime dimensions has
always been rather puzzling. We suspect that its role in describing the weakly coupled stages
of the spacetime decay in heterotic M-theory is the raison detre of this previously mysterious
model.
We intend to review the structure of this nonsupersymmetric heterotic string model in
sucient detail in Section 3. Anticipating its properties, we list some preliminary evidence
for this conjecture here:
The E
8
current algebra is realized at level two. This is consistent with the anticipated
Higgs mechanism E
8
E
8
E
8
, analogous to that observed in brane-antibrane systems
where U(N) U(N) is rst higgsed to the diagonal U(N) subgroup. (This analogy is
discussed in more detail in [4].)
The nonperturbative decay to nothing instanton instability is expected to become
at weak string coupling a perturbative instability, described by a tachyon which is a
singlet under the gauge symmetry. The tachyon of the E
8
heterotic string is just such
a singlet.
The spectrum of massless fermions is nonchiral, with each chirality of adjoint fermions
present. This is again qualitatively the same behavior as in brane-antibrane systems.
The nonsupersymmetric E
8
E
8
version of heterotic M-theory can be constructed as
a Z
2
orbifold of the standard supersymmetric E
8
E
8
heterotic M-theory vacuum.
Similarly, the E
8
heterotic string is related to the supersymmetric E
8
E
8
heterotic
string by a simple Z
2
orbifold procedure.
The problem of tachyon condensation in the E
8
heterotic string theory is interesting in its
own right and can be studied independently of any possible relation to instabilities in heterotic
M-theory. Thus, our analysis in the remainder of the paper is independent of this conjectured
relation to spacetime decay in M-theory. As we shall see, our detailed investigation of the
tachyon condensation in the heterotic string at weak coupling provides further corroborating
evidence in support of this conjecture.
3. The Forgotten E
8
Heterotic String
Classical Poincare symmetry in ten dimenions restricts the number of consistent heterotic
string theories to nine, of which six are tachyonic. These tachyonic models form a natural
3
Another candidate perturbative description was suggested in [15].
6
hierarchy, terminating with the E
8
model. We devote this section to a review of some of the
salient aspects of the nearly forgotten heterotic E
8
theory. Most of these features have been
known for quite some time but are scattered in the literature [14, 1620].
3.1 The Free Fermion Language
The tachyonic E
8
string was rst discovered in the free-fermion description of the nonsuper-
symmetric left-movers [14]. The starting point of this construction is the same for all heterotic
models in ten dimensions (including the better-known supersymmetric models): 32 real left-
moving fermions
A
, A = 1, . . . 32, and ten right-moving superpartners

of X

, described
(in conformal gauge; see Appendix A for our conventions) by the free-eld action
S
fermi
=
i
2

_
d
2

A
+

A
+
+

_
. (3.1)
The only dierence between the various models is in the assignment of spin structures to
various groups of fermions, and the consequent GSO projection. It is convenient to label
various periodicity sectors by a 33-component vector whose entries take values in Z
2
= ,
4
U = (, . . .
. .
32
[). (3.2)
The rst 32 entries indicate the (anti)periodicity of the A-th fermion
A
, and the 33rd entry
describes the (anti)periodicity of the right-moving superpartners

of X

.
A specic model is selected by listing all the periodicities that contribute to the sum over
spin structures. Modular invariance requires that the allowed periodicities U are given as
linear combinations of n linearly independent basis vectors U
i
U =
n

i=1

i
U
i
(3.3)
with Z
2
-valued coecients
i
. Modular invariance also requires that in any given periodicity
sector, the number of periodic fermions is an integer multiple of eight. All six tachyonic
heterotic theories can be described using the following set of basis vectors:
U
1
= ([),
U
2
= (+ + + + + + + + + + + + + + + +[),
U
3
= (+ + + + + + + ++ + + + + + + +[),
U
4
= (+ + + ++ + + ++ + + ++ + + +[),
U
5
= (+ ++ ++ ++ ++ ++ ++ ++ +[),
U
6
= (++++++++++++++++[).
4
Here + and correspond to the NS sector and the R sector, respectively. This choice is consistent
with the grading on the operator product algebra of the corresponding operators. Hence, the sector labeled
by + (or ) corresponds to an antiperiodic (or periodic) fermion on the cylinder.
7
The theory which has only U
1
as a basis vector has 32 tachyons. Adding U
2
reduces the
number of tachyons to sixteen. This process can be continued until the allowed periodicities
are spanned by all six vectors U
i
; here only one bosonic tachyon is present. This most intricate
of the tachyonic theories is the single-tachyon E
8
model we wish to discuss.
5
Thus, in our case there are 2
6
= 64 dierent periodicity sectors. Note that there is a
perfect permutation symmetry among the left-moving fermions, but this symmetry is lifted
by the coupling to the spin structure of the right-moving fermions

. There is precisely
one left-moving fermion whose spin structure is always locked with the spin structure of the
supersymmetric sector

. Since this fermion

32
+
plays a special role, we shall denote it by

+
and refer to it as the lone fermion for brevity.
The tachyon in this theory is a singlet, and comes from the (+. . . +[+) sector in which all
the fermions are Neveu-Schwarz. The left-moving vacuum is excited by the lowest oscillation
mode b
1/2
of the lone fermion
+
; the right-moving vacuum is in the ground state. The
vertex operator for the tachyon (in picture 0) is thus
1 = (F + p

) exp(ip

). (3.4)
The spectrum also contains 248 massless vector bosons. Their spacetime Yang-Mills
group structure is rather obscure in the free fermion language, but they do form the adjoint
of E
8
. There is one family of adjoint massless fermions for each chirality; string loop eects
are likely to combine these into one massive eld. More information about the spectrum at
higher levels can be extracted from the one-loop partition function calculated below, in (3.8).
The 31 free fermions
A
, A = 1 . . . 31 realize a level-two E
8
current algebra. The E
8
current algebra at level k has central charge
c
E
8
,k
=
k dimE
8
k + h
=
248k
k + 30
. (3.5)
Here h = 30 is the dual Coxeter number of E
8
. At level k = 2, this corresponds to the
central charge of 31/2, which agrees with the central charge of the 31 free fermions
A
which
comprise it. It is now convenient to switch to a more compact, mixed representation of the
left-moving sector of the worldsheet CFT. In this representation, the left-movers are succinctly
described by the lone fermion
+
(
+
) together with the algebra of E
8
currents J
I
(
+
); here
I is the adjoint index of E
8
. The spin structure of
+
is locked with the spin structure of the
right-moving superpartners

of X

.
At level two, the E
8
current algebra has three integrable representations [1), [248) and
[3875), where n denotes the representation whose highest weight is in n of E
8
. The conformal
weights of the highest weight states in [1), [248) and [3875) are 0, 15/16 and 3/2, respectively.
5
Adding other generators, for example U0 = (. . . |+) which would relax the lock between the spin
structures of the left and right moving fermions, will produce other heterotic models. In the case of U0 added
to any portion of the basis U1, . . . U6, the familiar SO(16) SO(16) model is produced. Adding generators
which have only eight periodic fermions will produce the remaining supersymmetric theories, as well as extra
copies of the tachyonic ones [14].
8
In the spectrum of physical states, the NS and R sectors [) of
+
and

are interwined
with the representations of (E
8
)
2
, which leads to the following sectors of the physical Hilbert
space,
[+) ([1) [3875)) ,
[) [248).
One of the advantages of this representation is that the states charged under the E
8
symmetry
are now generated by the modes of the currents J
I
, making the E
8
symmetry manifest and
the need for its realization via 31 free fermions with a complicated GSO projection obsolete.
3.2 The Language of Free Bosons
The bosonization of this model is nontrivial. In order to rewrite a free fermion model in
terms of bosonic elds, one typically associates a bosonic eld with a pair of fermions. This
is, however, impossible in the E
8
heterotic model: No two fermions carry the same spin
structure in all sectors, due to the intricate interlacing of the spin structures reviewed above,
and a twist of the conventional bosonization is needed.
As a result, the model cannot be constructed as a straight lattice compactication; how-
ever, it can be constructed as a Z
2
orbifold of one [16]. In fact, the starting point can be the
supersymmetric E
8
E
8
heterotic string in the bosonic language.
The E
8
E
8
lattice of left-moving scalars in the supersymmetric E
8
E
8
string has a
nontrivial outer automorphism 1 that simply exchanges the two E
8
factors. One can use it
to dene a Z
2
orbifold action on the CFT via
= 1 exp (iF
s
) , (3.6)
where F
s
is the spacetime fermion number. Note that exp(iF
s
) can be conveniently realized
as a 2 rotation in spacetime, say in the X
1
, X
2
plane. The orbifold breaks supersymmetry
completely, and yields the tachyonic E
8
heterotic string.
This surprisingly simple orbifold relation between the supersymmetric E
8
E
8
theory and
the tachyonic E
8
model is possible because of some unique properties of the fusion algebra and
the characters of E
8
. The fusion rules for the three integrable highest-weight representations
of (E
8
)
2
are isomorphic to those of a free CFT of a single real fermion
+
, with [248) playing
the role of the spin eld, and [3875) that of the fermion. This is related to the fact that the
c = 1/2 CFT of
+
can be represented as a coset
(E
8
)
1
(E
8
)
1
(E
8
)
2
. (3.7)
This explains why there is such a simple relation between the supersymmetric E
8
E
8
heterotic
string and the tachyonic E
8
model: They can be viewed as two dierent ways of combining
a sigle free-fermion theory (3.7) with the level-two E
8
current algebra.
9
In the bosonic form, the construction of the tachyonic E
8
heterotic string model is quite
reminiscent of the CHL string backgrounds [21]. In those models, a single copy of E
8
symmetry
at level two is also obtained by a similar orbifold, but the vacua are spacetime supersymmet-
ric [22]. It is conceivable that such supersymmetric CHL vacua in lower dimensions could
represent endpoints for decay of the E
8
model when the tachyon prole is allowed an extra
dependence on spatial dimensions, as in [5, 6].
The one-loop partition function of the heterotic E
8
string theory can be most conveniently
calculated in lightcone gauge, by combining the bosonic picture for the left-movers with the
Green-Schwarz representation of the right-movers. The one-loop amplitude is given by
Z =
1
2
_
d
2

(Im)
2
1
(Im)
4
[()[
24
_
16
E
8
(2)

4
10
(0, )

4
10
(0, )
+
E
8
(/2)

4
01
(0, )

4
01
(0, )
+
E
8
(( + 1)/2))

4
00
(0, )

4
00
(0, )
_
. (3.8)
For the remainder of the paper, we use the representation of the worldsheet CFT in terms of
the lone fermion and the level-two E
8
current algebra, represented by 31 free fermions.
4. Tachyon Condensation in the E
8
Heterotic String
4.1 The General Philosophy
We wish to understand closed-string tachyon condensation as a dynamical spacetime process.
Hence, we are looking for a time-dependent classical solution of string theory, which would
describe the condensation as it interpolates between the perturbatively unstable conguration
at early times and the endpoint of the condensation at late times. Classical solutions of string
theory correspond to worldsheet conformal eld theories; thus, in order to describe the con-
densation as an on-shell process, we intend to maintain exact quantum conformal invariance
on the worldsheet. In particular, in this paper we are not interested in describing tachyon
condensation in terms of an abstract RG ow between two dierent CFTs. In addition, we
limit our attention to classical solutions, and leave the question of string loop corrections for
future work.
4.2 The Action
Before any gauge is selected, the E
8
heterotic string theory with the tachyon condensate
tuned to zero is described in the NSR formalism by the covariant worldsheet action
S
0
=
1
4

_
d
2
e
_

(h
mn

m
X

n
X

+ i

i
m

n
X

)
+ i
A

A
F
A
F
A
_
. (4.1)
where, as usual, h
mn
=
ab
e
m
a
e
n
b
, e = det(e
m
a
). We choose not to integrate out the aux-
iliary elds F
A
from the action at this stage, thus maintaining its o-shell invariance under
10
local supersymmetry, whose transformation rules on elds are given by (A.8-A.10). We have
collected other useful formulae and our choices of conventions in Appendix A.
4.2.1 Linear dilaton
In order to obtain a description of tachyon condensation in terms of an exactly solvable CFT,
we will consider the tachyon condensate that evolves along a null direction. Thus, our tachyon
condensate will depend on a eld, say X
+
, which has trivial OPE with itself. In order for
such a condensate to maintain conformal invariance, we also need to turn on a linear dilaton
background,
(X) = V

, (4.2)
for some constant vector V

. If we wish to maintain the critical dimension equal to ten,

the linear dilaton must be null, V V = 0. Hence, we can adjust our choice of spacetime
coordinates X

, X
i
such that V is only nonzero in the X

direction. Later on, when we turn

on the tachyon prole, the linear dilaton will depend on the light-cone direction X
+
instead.
6
In the presence of the linear dilaton, the covariant action of the heterotic model is S =
S
0
+ S
V
, with S
0
given in (4.1) and
S
V
=
1
4
_
d
2
e V

_
X

R(h) + i
m

m
D
n

_
. (4.3)
Recall that we are in Minkowski signature both on the worldsheet and in spacetime; this
accounts for the negative sign in front of S
V
. In the case of the null dilaton, V V = 0, both
S
0
and S
V
are separately Weyl invariant; the proof of this fact for S
V
requires the use of
the equations of motion that follow from varying X
+
in the full action. In addition, o-shell
supersymmetry of S
0
+S
V
also requires a modication of the supersymmetry transformation
rules in the supersymmetric matter sector, which now become
X

= i

=
m

m
X

m
D
m
. (4.4)
The remaining supersymmetry transformations (A.9) and (A.10) remain unmodied.
The rst term in (4.3) produces the standard V -dependent term in the energy-momentum
tensor, while the second term yields the well-known improvement term in the supercurrent
of the linear dilaton theory. The second term also contributes a gravitino dependent term to
the energy-momentum tensor, as we will show below.
4.2.2 Superpotential and the tachyon prole
At the classical level, the tachyon couples to the string as a worldsheet superpotential. Clas-
sically, its coupling constant would be dimensionful. Additionally, the superpotential would
6
The need for a nonzero dilaton gradient at weak string coupling is somewhat reminiscent of a similar
phenomenon at strong coupling: the perturbative Casimir force in the E8 E8 heterotic M-theory. In both
cases, the decay-to-nothing instability plays out on top of a nontrivial spacetime dependence of the string
coupling. Given the absence of a reliable interpolation between the weakly and strongly coupled regimes, it is
not possible to determine whether these two phenomena are directly related.
11
be neither Weyl nor super-Weyl invariant: It would depend on the Liouville mode as well
as its superpartner
+
.
We are only interested in adding superpotentials that are, in conformal gauge, exactly
marginal deformations of the original theory. The leading-order condition for marginality
requires the tachyon condensate T (X) to be a dimension (1/2, 1/2) operator, and the quantum
superpotential takes the following form,
S
W
=

_
d
2

_
F T (X) i

T (X)
_
; (4.5)
is a dimensionless coupling.
With T (X) exp(k

) for some constant k

, the condition for T (X) to be of dimension

(1/2, 1/2) gives
k
2
+ 2V k =
2

. (4.6)
If we wish to maintain quantum conformal invariance at higher orders in conformal perturba-
tion theory in , the prole of the tachyon must be null, so that S
W
stays marginal. Together
with (4.6), this leads to
T (X) = exp(k
+
X
+
), (4.7)
V

k
+
=
1
2

. (4.8)
Since our k
+
is positive, so that the tachyon condensate grows with growing X
+
, this means
that V

is negative, and the theory is weakly coupled at late X

.
From now on, we will only be interested in the specic form of the superpotential that
follows from (4.7) and (4.8),
S
W
=

_
d
2

_
F ik
+

_
exp(k
+
X
+
). (4.9)
Interestingly, the check of supersymmetry invariance of (4.9) requires the use of (4.8) together
with the V -dependent supersymmetry transformations (4.4).
4.3 The Lone Fermion as a Goldstino
Under supersymmetry, the lone fermion
+
transforms in an interesting way,

+
= F
+
. (4.10)
F is an auxiliary eld that can be eliminated from the theory by solving its algebraic equation
of motion. In the absence of the tachyon condensate, F is zero, leading to the standard (yet
slightly imprecise) statement that
+
is a singlet under supersymmetry. In our case, when
the tachyon condensate is turned on, F develops a nonzero vacuum expectation value, and

+
no longer transforms trivially under supersymmetry. In fact, the nonlinear behavior of
+
under supersymmetry in the presence of a nonzero condensate of F is typical of the goldstino.
12
Traditionally, the goldstino eld
+
is normalized such that its leading order transfor-
mation under supersymmetry is just
+
= + . . ., where . . . indicates eld-dependent
corrections. In our case, choosing

+
=

+
F
(4.11)
gives the proper normalization for a goldstino under supersymmetry. Classically,
+
trans-
forms as

+
=
+
i
+
(
m
D
m
). (4.12)
This is the standard nonlinear realization of supersymmetry on the goldstino in the Volkov-
Akulov sense. This realization of supersymmetry has also played a central role in the
Berkovits-Vafa construction [2327]. This construction has been directly linked by Heller-
man and Swanson to the outcome of tachyon condensation, at least in the case of Type 0
theory.
5. Tachyon Condensation and the Worldsheet Super-Higgs Mechanism
Now that we have precisely dened the worldsheet action in covariant form, we will show how
alternative gauge choices for worldsheet supersymmetry can elucidate the dynamics of the
system, and in particular, make the worldsheet super Higgs mechanism manifest.
Our alternative gauge choices will have one thing in common with superconformal gauge:
For xing the bosonic part of the worldsheet gauge symmetry, we always pick the conventional
conformal gauge, by setting (locally) e
m
a
=
m
a
. This is logically consistent with the fact
that we turn on the tachyon condensate as an exactly marginal deformation, maintaining
worldsheet conformal invariance throughout. In conformal gauge (and in worldsheet lightcone
coordinates (

,
+
)), the full worldsheet action becomes
S =
1

_
d
2

+
X
i

X
i
+
i
2

+
X
+

i
2

+
i
2

A
+

A
+

2
exp(2k
+
X
+
) i
++

X
i
+
i
2

++

+
i
2

++

X
+
+ i

++

+ ik
+

exp(k
+
X
+
)
_
. (5.1)
In this action, we have integrated out the auxiliaries F
A
using their algebraic equations of
motion: All F
A
s are zero with the exception of the superpartner F of the lone fermion, which
develops a nonzero condensate:
F = 2 exp(k
+
X
+
). (5.2)
The supersymmetry algebra now closes only on-shell, with the use of the
+
equation of
motion. In the rest of the paper, F always refers to the composite operator in terms of X
+
as given by (5.2). We will use products of powers of F with other elds several times below.
Because the OPE of F with any eld other than X

is trivial, these objects are quantum

mechanically well-dened so long as X

does not appear in them.

13
We also present the energy-momentum tensor and supercurrent, again in conformal gauge
and worldsheet lightcone coordinates:
T
++
=
1

_
2i
++

+
X

2i

_
3
2

++

+
1
2
(
+

++
)

_
+
+
X

+
X

+
i
2

+
X

_
, (5.3)
T

=
1

+
i
2

_
, (5.4)
G

=
2

X
i

1
2

1
2

X
+

_
. (5.5)
In the classical action (5.1), the condensate (5.2) induces a bosonic potential term

2
exp(2k
+
X
+
). As shown in (5.3), this potential term does not contribute to the world-
sheet vacuum energy, since it is an operator of anomalous dimension (1, 1) and hence its
integration over the worldsheet does not require any dependence on the worldsheet metric.
Since this potential term does not contribute to the energy-momentum tensor, it will not
contribute to the BRST charge either.
On the other hand, this bosonic potential does contribute to the equation of motion for
X

, which can be written locally as

X
+
= 0,

= 2
2
k
+
exp(2k
+
X
+
) +
i
2

(
++

) ik
2
+

exp(k
+
X
+
). (5.6)
These equations imply that the generic incoming physical excitations of the string are eec-
tively shielded by the tachyon condensate from traveling too deeply into the bubble of nothing.
Thus, fundamental string excitations are pushed away to innity in the X

direction by the
walls of the bubble of nothing. A similar phenomenon has been observed numerous times
in the previous studies of closed-string tachyon condensation, see e.g. [5, 6, 2830].
5.1 Superconformal Gauge
In the conventional treatment of strings with worldsheet supersymmetry, superconformal
gauge is virtually always selected. In this gauge, the worldsheet gravitino is simply set to
zero:

++
= 0. (5.7)
In our background, however, we expect the gravitino to take on a prominent role as a result
of the super-Higgs mechanism. For that reason, we will explore alternative gauge choices,
friendlier to this more important role expected of the gravitino.
Before we introduce alternative gauge choices, however, we address some aspects of the
theory in the conventional superconformal gauge. This exercise will reveal at least one in-
triguing feature of the model: The emergence of logarithmic CFT in the context of tachyon
condensation.
14
Superconformal gauge leaves residual superconformal symmetry which should be realized
on all elds by the action of the supercurrent G

. Consider in particular the lone fermion

+
. Before the tachyon condensate is turned on, the operator product of G

with
+
is
nonsingular, in accord with the fact that
+
transforms trivially under on-shell supersym-
metry. As we have seen in Section 4, when the auxiliary eld F develops a nonzero vacuum
expectation value in the process of tachyon condensation,
+
transforms under supersymme-
try nontrivially, as a candidate goldstino. This raises an interesting question: How can this
nontrivial transformation be reproduced by the action of G

on
+
, if, as we have seen in
(5.5), the supercurrent G

is unmodied by ?
The resolution of this puzzle must come from nontrivial OPEs that develop at nite
between the originally leftmoving eld
+
and the originally rightmoving elds

. Here and
in the following, it will be useful to introduce a rescaled version of the elds

/F,

= F
+

. (5.8)
We will encounter this particular rescaled version of

again below, in another gauge. In

terms of

, the supercurrent (5.5) simplies to

G

=
2

X
i

1
2

_
. (5.9)
The supersymmetry variations of elds are reproduced as follows. Consider for example the
supermultiplet
i

and X
i
. In superconformal gauge, these are free elds, satisfying standard
OPEs such as

)
j
(

ij

, (5.10)
which imply
G

)
i

)
2

X
i
(

, (5.11)
correctly reproducing the supersymmetry variation
i

= 2

X
i
.
Similarly, the last term in (5.9) will reproduce the supersymmetry transformation
+
=
F if the following OPE holds,
2

)
+
(

)
F(

)/F(

. (5.12)
This required OPE can be checked by an explicit calculation: Starting with the free-eld
OPEs

)
2

, (5.13)
we get for the rescaled elds (5.8)

)
2

F(

)/F(

n=0
1
n!
(
+

+
)
n
_

n
+
F(

)
_
/F(

). (5.14)
15
Note that since X
+
satises locally the free equation of motion
+

X
+
= 0, the coecient
of the 1/(

) term in this OPE is only a function of

+
and
+
. The OPE between

and
+
can then be determined from the the equation of motion for
+
:

+
() = k
+

exp(k
+
X
+
) =
k
+
2

. (5.15)
Combining these last two equations, we get

+
(

) =
k
+
2

) =

k
+
F(

)/F(

. (5.16)
Integrating the result with respect to

, we nally obtain

+
(

) =

k
+
_

n=0
1
n!
(
+

+
)
n
_

n
+
F(

)
_
/F(

)
_
log(

). (5.17)
One can then easily check that this OPE implies (5.12) when (4.8) is invoked. This in turn
leads to
G

)
+
(

)
F(

, (5.18)
and the supersymmetry transformation of
+
is correctly reproduced quantum mechanically,
even in the presence of the tachyon condensate.
As is apparent from the form of (5.17), our theory exhibits in superconformal gauge
OPEs with a logarithmic behavior. This establishes an unexpected connection between
models of tachyon condensation and the branch of 2D CFT known as logarithmic CFT (or
LCFT; see, e.g., [31, 32] for reviews). The subject of LCFT has been vigorously studied in
recent years, with applications to a wide range of physical problems, in particular to systems
with disorder. The logarithmic behavior of OPEs is compatible with conformal symmetry, but
not with unitarity. Hence, it can only emerge in string backgrounds in Minkowski spacetime
signature, in which the time dependence plays an important role (such as our problem of
tachyon condensation). We expect that the concepts and techniques developed in LCFTs
could be fruitful for understanding time-dependent backgrounds in string theory. In the
present work, we will not explore this connection further.
5.2 Alternatives to Superconformal Gauge
In the presence of a super-Higgs mechanism, another natural choice of gauge suggests itself.
In this gauge, one anticipates the assimilation of the Goldstone mode by the gauge eld,
by simply gauging away the Goldstone mode altogether. In the case of the bosonic Higgs
mechanism, this gauge is often referred to as unitary gauge as it makes unitarity of the
theory manifest.
Following this strategy, we will rst try eliminating the goldstino as a dynamical eld,
for example by simply choosing

+
= 0 (5.19)
16
as our gauge xing condition for local supersymmetry. Of course, this supplements the confor-
mal gauge choice that we have made for worldsheet dieomorphisms. This gauge choice gives
rise to non-propagating bosonic superghosts, together with the usual propagating fermionic
b, c system of central charge c = 26. We refer to this gauge as unitary gauge only due to
its ancestry in the similar gauges that proved useful in the Higgs mechanism of Yang-Mills
gauge theories. In fact, as we shall see, the proper implementation of this unitary gauge
will lead to propagating superghosts, and therefore no manifest unitarity of the theory.
In addition to the conventional fermionic b, c ghosts from conformal gauge, this gauge
choice would lead to non-propagating bosonic ghosts, which might be integrated out alge-
braically. More importantly, this gauge choice still leaves complicated interaction terms, such
as

++

, (5.20)
in the gauge-xed version of (5.1). Moreover, if the algebraic superghosts are integrated out,
the equation of motion arising from variation of the action under
+
,
k
+

exp(k
+
X
+
) = 0, (5.21)
needs to be imposed on physical states by hand. This could be accomplished by simply setting

= 0. This leads to another constraint, imposing the equation of motion obtained from
the variation of
+

in the original action as a constraint on physical states.

Instead of resolving such diculties, we will restart our analysis with
+

= 0 as the gauge
condition. As we will see below, this condition makes the gauge-xing procedure transparent.
5.3 Liberating the Worldsheet Gravitino in an Alternative Gauge
We will now explicitly consider the gauge that begins simply by setting

= 0. (5.22)
If
+

is eliminated from the action, the remaining

++
,

system can be rewritten as a

purely left-moving rst-order system of conformal weights (3/2, 0) and (1/2, 0).
7
This can
be seen as follows. Consider rst the terms in the action that are bilinear in these two elds,
1
2

++

X
+
+

++

. (5.23)
The presence of

here suggests
++
and

ought to become purely left-moving. Addition-

ally, these elds show up only in the energy momentum tensor T
++
in this gauge. However,
their conformal weights do not reect their left-moving nature. In order to obtain elds whose
conformal weights are nonzero only in the left-moving sector, we can rescale

++
= F
++
,

F
. (5.24)
7
This system is still coupled to the transverse elds
i
, X
i
but the strength of the interaction goes to zero
with growing F.
17
This rescaling leads to an additional benet: The bilinear terms (5.23) in the classical action
now assemble into the canonical kinetic term of a rst-order system of spin 3/2 and 1/2,
iV

_
d
2

++

, (5.25)
with central charge c
L
= 11.
The rst-order system (5.25) describes the worldsheet gravitino sector of the theory.
In superconformal gauge in the absence of the tachyon condensate, the gravitino was non-
dynamical and led to a constraint. Here, instead, the gravitino has been liberated as a result
of the super-Higgs mechanism: together with its conjugate

, at late times it appears to have

formed a left-moving free-eld massless system, of central charge c = 11. In this modied
unitary gauge, the gravitino has been literally set free: It has become a free propagating
massless eld!
Our gauge choice (5.22) reduces the classical action signicantly, to
S =
1

_
d
2

+
X
i

X
i
+
i
2

+
X
+

+
i
2

A
+

A
+

2
exp(2k
+
X
+
)
i
++

X
i
+
i
2

++

X
+
+ i

++

_
. (5.26)
Note that unlike in the case of superconformal gauge, the gauge xing condition (5.22) leaves
no residual unxed supersymmetries, at least at nite . Thus, in this gauge, the theory will
be conformal, but not superconformal. In Section 6, we shall return to the issue of residual
supersymmetry in this gauge, in the late-time limit of .
The action (5.26) will be corrected by one-loop eects. The rst such correction is due to
the Faddeev-Popov (super)determinant
FP
. As we will now see, the inherent X
+
dependence
in our gauge xing condition renders
FP
dependent on X
+
as well.
Our full gauge xing condition consists of the bosonic conformal gauge, e
m
a
=
m
a
,
as well as the fermionic condition (5.22). Note rst that the corresponding Faddeev-Popov
superdeterminant factorizes into the ratio of two bosonic determinants,
8

FP
= J
bc
/J

. (5.27)
Here J
bc
arises from the conformal gauge condition, and produces the standard set of fermionic
b, c ghosts with central charge c = 26. On the other hand, J

, which comes from the

change of variables between the gauge-xing condition
+

and the innitesimal supersymme-

try parameter , turns out to be more complicated. It will be useful to rewrite the variation
of
+

as

= 2

X
+
+ 4

=
4

F
D

(F). (5.28)
8
This is true because the operator whose superdeterminant is being calculated has a block-triangular form,
as the action of dieomorphisms on the fermionic gauge-xing condition
+

vanishes when (5.22) is satised.

18
Here, we are only xing worldsheet dieomorphisms and not Weyl transformations, in order
to allow a systematic check of the vanishing of the Weyl anomaly. We are thus in a gauge
where h
mn
= e
2

h
mn
. The derivative D

preserves information about the spin of the eld it

acts on. We can now see the denominator of (5.27) should be dened as
J

= det
_
1
F
D

F
_
. (5.29)
As in the case of J
bc
, we wish to represent this determinant as a path integral over a conjugate
pair of bosonic elds , : the bosonic ghosts associated with the xing of local supersym-
metry. Note, however, that the variation of our xing condition is dependent on X
+
. This
fact leads to interesting additional subtleties, which we discuss in detail in Appendix B.
As a result, the Jacobian turns out to be given by a set of ghosts

, of central charge
c = 1 whose path integral measure is independent of X
+
, plus an extra term that depends
explicitly on X
+
. As shown in Eqn. (B.17) of Appendix B, we nd
1
J

= exp
_

_
d
2

k
2
+

+
X
+

X
+
__
T

T exp
_
iS
e
e
_
, (5.30)
where S
e
e
is the ghost action for a free left-moving bosonic ghost system of spin 1/2.
The Faddeev-Popov determinant thus renormalizes the spacetime metric. If this were the
whole story, we could re-diagonalize the spacetime metric to the canonical light-cone form,
by redening the spacetime light-cone coordinates
Y
+
= X
+
,
Y

= X

k
2
+
X
+
. (5.31)
In these new coordinates, the linear dilaton would acquire a shift:
= V

,
= V

Y

V

k
2
+
X
+
,
= V

Y

+
k
+
2
Y
+
. (5.32)
The eective change in the central charge from this shift in the linear dilaton would be
c
dil
= 6

V
2
= 24

V
+
= 6, (5.33)
where we have again used (4.6).
However, understanding the Faddeev-Popov determinant is not the whole story. In addi-
tion, the X
+
-dependent rescaling of
++
and

as in (5.24) also produces a subtle Jacobian

J. As shown in Appendix B, this Jacobian can be expressed in Minkowski signature as

J = exp
_

i
4

_
d
2
e
_

k
2
+

h
mn

m
X
+

n
X
+
+

k
+
X
+
R
_
_
. (5.34)
19
In this sense, the original
++
,

system is equivalent to the canonical

++
,

system
when this additional renormalization of the linear dilaton term and the the spacetime metric
are taken into account. In lightcone coordinates, the rst factor in (5.34) becomes
exp
_
i

_
d
2

k
2
+

+
X
+

X
+
_
. (5.35)
This contribution to the renormalization of the metric precisely cancels the contribution
obtained from the Faddeev-Popov determinant (5.30). Thus, the combined eect of

J and
J

on the X

elds is just a simple shift of the linear dilaton in the original X

variables,
as implied by the second term in (5.34). The contribution to the central charge due to this
dilaton shift is c = 12, as c
dil
= 6

V
2
= 12.
The full action, still in conformal gauge, is thus given by:
S =
1

_
d
2

+
X
i

X
i
+
i
2

+
X
+

+
i
2

A
+

A
+

2
exp(2k
+
X
+
)
i

++
F

i

X
i
+ i

++

_
+ S
bc
+ S
e
e
. (5.36)
The information about the linear dilaton prole is absent in this action, but the energy-
momentum tensor of the theory reveals that the dilaton is given by
V

= (k
+
, V

0). (5.37)
In this form, the action has only canonical kinetic terms, plus the potential term and an
interaction term which will vanish at late times (since it goes as 1/F). Specically, we note
that the elds
++
,

have now become purely left-moving. This matches their conformal

weights, which are (3/2, 0) and (1/2, 0) respectively. In addition, they contribute only to
the left moving energy momentum tensor; their contribution results in 11 units of central
charge for c
L
, in the late X
+
limit when they decouple from the transverse degrees of freedom
X
i
,
i
.
Now, let us summarize the central charge contributions of each eld present in this gauge,
at late times. For comparison, we also present the central charge breakdown for the free theory
in superconformal gauge.
Superconformal gauge The alternative gauge
Field c
L
c
R
X
+
, X

, X
i
10 10
linear dilaton 0 0
bc ghosts 26 26

i
0 4

+
,

0 1
, ghosts 0 11

+
1/2 0
(E
8
)
2
31/2 0
Field c
L
c
R
X
+
, X

, X
i
10 10
linear dilaton 12 12
bc ghosts 26 26

i
0 4
,

11 0

, ghosts 1 0

+
1/2 0
(E
8
)
2
31/2 0
(5.38)
20
Our new gauge choice has resulted in 12 units of central charge in the right-moving sector,
from

and the system, eectively moving to become 12 units of central charge on

the left. These left-moving central charge units come from ,

, and the new

ghosts.
We also see that the shifted linear dilaton precisely compensates for this relocation of central
charge, resulting in the theory in the alternative gauge still being exactly conformal at the
quantum level.
Interestingly, the equation of motion that follows from varying
+

in the original action

allows us to make contact with the original unitary gauge. Classically, this equation of motion
is

+
++

+ 2k
+

+
exp(k
+
X
+
) = 0. (5.39)
This constraint can be interpreted and solved in a particularly natural way: Imagine solving
the X

and
++
,

sectors rst. Then one can simply use the constraint to express
+
in
terms of those other elds. Thus, the alternative gauge still allows the gravitino to assimilate
the goldstino in the process of becoming a propagating eld. This is how the worldsheet
super-Higgs mechanism is implemented, in a way compatible with conformal invariance.
We should note that this classical constraint (5.39) could undergo a one-loop correction
analogous to the one-loop shift in the dilaton. We might expect a term

++
, from varying
a one-loop supercurrent term
++

in the full quantum action. Such a correction would

not change the fact that
+
is determined in terms of the oscillators of other elds, it would
simply change the precise details of such a rewriting.
5.4 R

Gauges
The history of understanding the Higgs mechanism in Yang-Mills theories was closely linked
with the existence of a very useful family of gauge choices, known as R

gauges. R

gauges
interpolate as one varies a control parameter between unitary gauge and one of the more
traditional gauges (such as Lorentz or Coulomb gauge).
In string theory, one could similarly consider families of gauge xing conditions for world-
sheet supersymmetry which interpolate between the traditional superconformal gauge and our
alternative gauge. We wish to maintain conformal invariance of the theory in the new gauge,
and will use conformal gauge to x the bosonic part of the worldsheet gauge symmetries. Be-
cause they carry disparate conformal weights (1, 1/2) and (0, 1/2) respectively, we cannot
simply add the two gauge xing conditions
++
and
+

with just a relative constant. In

order to nd a mixed gauge-xing condition compatible with conformal invariance, we need
a conversion factor that makes up for this dierence in conformal weights. One could for
example set
(

X
+
)
2

++
+ F
2

= 0, (5.40)
with a real constant (of conformal dimension 0). The added advantage of such a mixed gauge
is that it interpolates between superconformal and alternative gauge not only as one changes
, but also at any xed as X
+
changes: At early lightcone time X
+
, the superconformal
21
gauge xing condition dominates, while at late X
+
, the alternative gauge takes over, due to
the relative factors of F between the two terms in (5.40).
Even though (5.40) is compatible with conformal invariance, it is highly nonlinear, and
therefore impractical as a useful gauge xing condition. Another, perhaps more practical,
condition that incorporates both
++
and
+

is

++
+
+

= 0. (5.41)
Here the mismatch in conformal weights has been made up by inserting worldsheet deriva-
tives, rather than other composite operators. This condition can be written in the following
covariant form,

n
D
n

m
2
m
D
m

+
= 0, (5.42)
demonstrating its compatibility with dieomorphism invariance.
Note that (5.41) correctly anticipates the dynamics of the gravitino: If we solve for
+

using (5.41) in the kinetic term

, we get

++
. Thus,
++
replaces
+

as the
conjugate partner of

, and turns it from a right-moving eld into a left-moving one. Note

also that both terms in (5.41) transform under supersymmetry as
+

+. . ., which implies
that the bosonic superghosts

, associated with this gauge xing will have a second-order
kinetic term,

_
d
2

+ . . .
_
. (5.43)
In the limit, this gauge can be expected to leave a residual fermionic gauge symmetry.
Various other classes of R

-type gauges can be considered. For example, one could study

combinations of superconformal gauge with the naive unitary gauge in which
+
is set to
zero, leading to gauge xing conditions such as
F
++
+ k
+

+
X
+
= 0, (5.44)
or
F
+
+
++
k
+

X
+
= 0. (5.45)
Another interesting possibility is

++
+
+
_

+
F
_
= 0, (5.46)
which would interpolate between superconformal gauge and a weaker form of unitary gauge,
in which the left-moving part of the goldstino is set to zero.
We have not investigated the worldsheet theory in these mixed gauges, but it would be
interesting to see if any of them shed some new light on the dynamics of tachyon condensation.
22
6. The Condensed Phase: Exploring the Worldsheet Theory at =
The focus of the present paper has been on developing worldsheet techniques that can eluci-
date the super-Higgs mechanism and the dynamics of the worldsheet gravitino in the process
of spacetime tachyon condensation. Here we comment briey on the structure of the world-
sheet theory in the regime where the tachyon has already condensed.
This condensed phase corresponds to the system at late X
+
. In the worldsheet theory, a
constant translation of X
+
rescales the value of the superpotential coupling , with the late
X
+
limit mapping to . In that limit, the worldsheet theory simplies in an interesting
way. First, we rescale the parameter of local supersymmetry transformation,

+
= F
+
. (6.1)
The supersymmetry variations then reduce in the = limit and in conformal gauge to
X
+
= 0,

= 4

+
,
X

= i

+
,

= 0, (6.2)
X
i
= 0,

= 0,

+
=
+
,
++
=
2

+
k
+

+
X
+

+
_
.
Note that X
+
is now invariant under supersymmetry. Consequently, the terms in the action
that originate from the superpotential,
2
2
exp(2k
+
X
+
) ik
+

, (6.3)
are now separately invariant under (6.2), in the strict = limit. This in turn implies that
we are free to drop the potential term
2
exp(2k
+
X
+
) without violating supersymmetry.
The resulting model is then described, in the alternative gauge of Section 5.3, by a free eld
action:
S
=
=
1

_
d
2

+
X
i

X
i
+
i
2

+
X
+

+
i
2

A
+

A
+
+ i

++

_
+ S
bc
+ S
e
e
. (6.4)
We argued in Section 5.3 that at nite , our alternative gauge (5.22) does not leave any
residual unxed supersymmetry, making the theory conformal but not superconformal. This
is to be contrasted with superconformal gauge, which leaves residual right-moving supercon-
formal symmetry. At = , however, it turns out that the alternative gauge (5.22) does
leave an exotic form of residual supersymmetry. Note rst that at any , the fermionic gauge
xing condition is respected by that satisfy

+ k
+

X
+

1
F

+
= 0. (6.5)
23
At nite , these apparent residual transformations do not preserve the bosonic part of
our gauge xing condition, e
m
a
=
a
m
. In the = limit, however, the supersymmetry
transformation of e
a
m
is trivial, and all solutions of (6.5) survive as residual supersymmetry
transformations.
Unlike in superconformal gauge, this residual supersymmetry is left-moving. Moreover,
the generator
+
of local supersymmetry is of conformal dimension (1/2, 0); hence, local
supersymmetry cannot be expected to square to a worldsheet conformal transformation. As
is clear from (6.2), this local supersymmetry is in fact nilpotent.
This residual supersymmetry (
+
) can be xed by a supplemental gauge choice, setting
a purely left-moving fermion to zero. The one that suggests itself is
+
, which in this gauge
satises the free equation of motion; moreover,
+
transforms very simply under supersym-
metry, and setting it to zero xes that symmetry completely. Once we add the condition

+
= 0 to the gauge choice
+

= 0, this gauge now becomes very similar to the naive unitary

gauge of Section 5.2. This is another way of seeing that the goldstino is not an independent
dynamical eld, since it has been absorbed into the dynamics of the other elds in the process
of setting the gravitino free.
Alternatively, one can leave the residual supersymmetry unxed, and instead impose the
constraint (5.39) which reduces in the = limit to

+
= 2

. (6.6)
Depending on whether the coupling of the quantum bosonic potential exp(2k
+
X
+
) is
tuned to zero or not, we have two dierent late-time theories: One whose equations of motion
expel all degrees of freedom to future innity along X

for some constant X

+
, the other
allowing perturbations to reach large X
+
. It is the latter theory which may represent a good
set of variables suitable for understanding the physics at late X
+
. If the potential is retained,
the physical string modes are pushed away to innity along X

before reaching too deeply

into the condensed phase, conrming that very few of the original stringy degrees of freedom
are supported there.
7. Conclusions
7.1 Overview
The main focus of this paper has been on examining the worldsheet theory of tachyon con-
densation in the E
8
heterotic string model. We studied the theory with a linear dilaton
V = V

, and a tachyon prole T (X

) = 2exp(k
+
X
+
). At rst glance, this tachyon pro-
le produces a worldsheet potential term which expels all degrees of freedom from the large
X
+
region, indicating a possible topological phase, conjecturally related to the nothing
phase in heterotic M-theory.
We found that the worldsheet dynamics of tachyon condensation involves a super-Higgs
mechanism, and that its analysis simplies when local worldsheet supersymmetry is xed in a
24
new gauge, specically conformal gauge augmented by
+

= 0. Following a detailed analysis

of one-loop measure eects, we found that exact quantum conformal invariance is maintained
throughout in this gauge. At late times, the worldsheet theory contains a free left-moving
propagating gravitino sector (obscured in superconformal gauge, as there the gravitino is set
to zero). The gravitino sector contributes 11 units of central charge to the left-movers. In
addition, the gauge xing leads to a set of left-moving ghosts with c = 1, and a spacelike
shifted linear dilaton V = V

+ k
+
X
+
.
In the process of making the gravitino dynamical, the worldsheet goldstino
+
has been
eectively absorbed into the rest of the system; more precisely, the constraint generated by
the alternative gauge can be solved by expressing
+
in terms of the remaining dynamical
degrees of freedom.
7.2 Further Analysis of the E
8
System
In this paper, we have laid the groundwork for an in-depth analysis of the late time physics of
the E
8
heterotic string under tachyon condensation. The emphasis here has been on develop-
ing the worldsheet techniques, aimed in particular at clarifying the super-Higgs mechanism.
The next step, which we leave for future work, would be to examine the spacetime physics
in the regime of late X
+
where the tachyon has condensed. There are signs indicating that
this phase contains very few conventional degrees of freedom; more work is needed to provide
further evidence for the conjectured relation between tachyon condensation in the E
8
string
and the spacetime decay to nothing in E
8
E
8
heterotic M-theory.
It would be interesting to use the standard tools of string theory, combined with the new
worldsheet gauge, to study the spectrum and scattering amplitudes of BRST invariant states
in this background, in particular at late times. The use of mixed R

gauges could possibly

extend the range of such an analysis, by interpolating between the superconformal gauge and
its alternative.
7.3 Towards Non-Equilibrium String Theory
In the process of the worldsheet analysis presented in this paper, we found two features which
we believe may be of interest to a broader class of time-dependent systems in string theory:
(1) in superconformal gauge, the spacetime tachyon condensate turns the worldsheet theory
into a logarithmic CFT; and (2) the worldsheet dynamics of some backgrounds may simplify
in alternative gauge choices for worldsheet supersymmetry.
We have only explored the rst hints of the LCFT story and its utility in the description
of string solutions with substantial time dependence. The new gauge choices, however, are
clearly applicable to other systems. As an example, consider the Type 0 model studied in [7].
We can pick a gauge similar to our alternative gauge (5.22) by setting
+

=
+
+
= 0, again
in addition to conformal gauge. We expect to simply double the gauge xing procedure in
Section 5.3, producing one copy of c = 1 superghosts and one copy of the propagating
gravitino sector in both the left and right movers. When the one-loop determinant eects
are included, the linear dilaton shifts by 2k
+
, resulting in additional 24 units of central
25
charge. Together with the two c = 11 sectors and the two c = 1 ghost sectors, this shift
again leads c
tot
= 0, similarly to the heterotic model studied in the present paper. Exact
conformal invariance is again maintained at the quantum level. This is in accord with the
results of [7]; however, our results are not manifestly equivalent to those of [7], as a result of a
dierent gauge choice. In addition, our results also suggest an interpretation of the somewhat
surprising appearance of fermionic rst-order systems with c = 11 in [7]: It is likely that
they represent the re-emergence of the dynamical worldsheet gravitino, in superconformal
gauge.
This picture suggest a new kind of worldsheet duality in string theory. Instead of viewing
one CFT in two dual ways while in superconformal gauge, the new duality is between two
dierent CFTs representing the same solution of string theory, but in two dierent worldsheet
gauges. The physics of BRST observables should of course be gauge independent, but this
does not require the CFTs to be isomorphic. In fact, the alternative gauge studied in the
body of this paper represents an example: It leads to a conformal, but not superconformal
theory, yet it should contain the same physical information as the SCFT realization of the
same string background in superconformal gauge.
We hope that this technique of using gauge choices other than superconformal gauge
will be applicable to a wider class of time dependent string backgrounds, beyond the case of
models with tachyon condensation studied here, and that it will increase our understanding
of non-equilibrium string theory.
Acknowledgments
One of us (PH) wishes to thank Allan Adams, Michal Fabinger, Uday Varadarajan and Bruno
Zumino for useful discussions in the early stages of this project, in 2001-02. This work has
been supported by NSF Grants PHY-0244900 and PHY-0555662, DOE Grant DE-AC03-
76SF00098, an NSF Graduate Research Fellowship, and the Berkeley Center for Theoretical
Physics.
A. Supersymmetry Conventions
In the NSR formalism, the tachyonic E
8
heterotic string is described by worldsheet super-
gravity with (0, 1) supersymmetry.
9
Here we list our conventions for worldsheet supergravity.
A.1 Flat worldsheet
In the worldsheet coordinates
a
= (
0
,
1
), the at Lorentz metric is

ab
=
_
1 0
0 1
_
, (A.1)
9
A useful source of information on two-dimensional supergravities is, e.g., [33].
26
with gamma matrices given by

0
=
_
0 1
1 0
_
,
1
=
_
0 1
1 0
_
, (A.2)
and satisfying
a
,
b
= 2
ab
. We dene the chirality matrix

0

1
=
_
1 0
0 1
_
. (A.3)
The two-component spinor indices are raised and lowered using the natural symplectic
structure,

=
_
0 1
1 0
_
, (A.4)
with

. This implies, for example, that

and

a
=
a
, for any two real spinors , .
Any two-component spinor

can be decomposed into its chiral components, dened via

_
,

=
1
2
(1 ). (A.5)
A.2 Local worldsheet supersymmetry
On curved worldsheets, we will distinguish the worldsheet index m from the internal Lorentz
index a. The spacetime index runs over 0, . . . D 1, typically with D = 10. The heterotic
string action with (0, 1) worldsheet supersymmetry is given by
S =
1
4

_
d
2
e
_

(h
mn

m
X

n
X

+ i

i
m

n
X

)
+ i
A

A
F
A
F
A
_
. (A.6)
The fermions and gravitino satisfy the following chirality conditions:

,
+
=
+
,
+
=
+
(A.7)
The action (A.6) is invariant under the supersymmetry transformations given by
X

= i

=
m

m
X

(A.8)
for the right-moving sector,

A
= F
A
, F
A
= i
m
D
m

A
(A.9)
for the left-movers, and
e
m
a
= i
a

m
,
m
=
2

D
m
(A.10)
27
in the supergravity sector. Note that we set = 2 in [11]. Of course,
m
= e
m
a

a
, and the
covariant derivative on spinors is
D
m
=
_

m
+
1
4

m
ab

ab
_
, (A.11)
with
ab
= [
a
,
b
]/2. In general, the spin connection
m
ab
in supergravity contains the piece
that depends solely on the vielbein,
m
ab
(e), plus a fermion bilinear improvement term. In
conformal (0, 1) supergravity in two dimensions as described by (A.6), however, the improve-
ment term vanishes identically, and we have
m
ab
=
m
ab
(e), with

m
ab
=
1
2
e
na
(
m
e
n
b

n
e
m
b
)
1
2
e
nb
(
m
e
n
a

n
e
m
a
)
1
2
e
na
e
pb
(
n
e
pc

p
e
nc
)e
m
c
. (A.12)
Note also that the susy variation of F
A
is sometimes written in the literature as F
A
=
i
m

D
m

A
, using the supercovariant derivative

D
m

A
+

_

m
+
1
4

m
ab

ab
_

A
+

m+
F
A
. (A.13)
This simplies, however, in several ways. First of all, the gravitino drops out from (A.13) if
the (0, 1) theory is independent of the superpartner of the Liouville eld, as is the case for
our heterotic worldsheet supergravity. Secondly, for terms relevant for the action, we get

m

D
m

A
+

A

A
+
. (A.14)
A.3 Lightcone coordinates
The worldsheet lightcone coordinates are

= , (A.15)
in which the Minkowski metric becomes

ab
=
_
0
1
2

1
2
0
_
, (A.16)
resulting in the lightcone gamma matrices

+
=
_
0 2
0 0
_
,

=
_
0 0
2 0
_
. (A.17)
On spin-vectors, such as the gravitino
m
, we put the worldsheet index rst, and the spinor
index second when required. We will use labels for spinor indices as well as lightcone
worldsheet and spacetime indices, as appropriate. The nature of a given index should be
clear from context. In lightcone coordinates, and in conformal gauge, the supersymmetry
transformations of the matter multiplets are given by
X

= i
+

= 2

+
,

A
+
= F
A

+
, F
A
= 2i
+

A
+
, (A.18)
28
and the linearized supersymmetry transformations of the supergravity multiplet are
e
+
= 2i
+

++
,
++
=
2

+
. (A.19)
Once we have picked conformal gauge, we can meaningfully assign a conformal weight to
each eld. The chart below lists these conformal dimensions for all relevant objects:
Field Symbol Conformal Weight
gravitino
++
(1,
1
2
)
goldstino
+
(
1
2
, 0)
fermion

(0,
1
2
)
boson X

(0, 0)
aux. eld F (
1
2
,
1
2
)
SUSY parameter (0,
1
2
)
w.s. derivative

(0, 1)
w.s. derivative
+
(1, 0)
(A.20)
In spacetime, similarly, we dene the lightcone coordinates thus:
X

= X
0
X
1
, (A.21)
and denote the remaining transverse dimensions by X
i
, so that the spacetime index decom-
poses as (+, , i).
When we combine the spacetime lightcone parametrization with the lightcone coordinate
choice on the worldsheet, the heterotic action becomes
S =
1

_
d
2

+
X
i

X
i
+
i
2

1
2

+
X
+

1
2

+
X

X
+

i
4

i
4

+
i
2

A
+

A
+
+
1
4
F
A
F
A
(A.22)

i
2

++
(2
i

X
i

X
+
)
_
,
where we have dened d
2

d
+
= 2d d.
B. Evaluation of the Determinants
We wish to calculate the determinant for the operator
D
F

1
F
D

F (B.1)
as it acts on elds of arbitrary half-integer spin j. F here is a shorthand for the tachyon
condensate F = 2exp(k
+
X
+
), as determined in (5.2); we will continue with this notation
through this appendix. As D
F
is a chiral operator, we will take the usual approach of
29
nding its adjoint and calculating the determinant of the corresponding Laplacian. The
actual contribution we are interested in will be the square root of the determinant of the
Laplacian. Throughout this appendix, we work in Euclidean signature; we will Wick rotate
our results back to Minkowski signature before adding the results to the body of the paper.
Also, we gauge x only worldsheet dieomorphisms by setting
h
mn
= e
2

h
mn
, (B.2)
where is the Liouville eld. For most calculations, we set the ducial metric

h
mn
to be the
at metric. In this gauge, we nd
D

= e
2
, (B.3)
independently of j.
We dene the adjoint D

F
of the Faddeev-Popov operator D
F
via
T
1
[
1
F
D

(FT
2
)) = D

F
T
1
[T
2
), (B.4)
where T
1
is a worldsheet tensor of spin j 1, T
2
is a tensor of spin j, and the inner product
on the corresponding tensors is the standard one, independent of F.
10
We nd the left hand side of (B.4) becomes
T
1
[
1
F
D

FT
2
) =
_
d
2
ze
2(2j)
T

1
_
1
F
D

F
_
T
2
=
_
d
2
ze
2(2j)
T

1
1
F
e
2
(FT
2
)
=
_
d
2
z
_
e
2(j1)
1
F
T

1
_
FT
2
=
_
d
2
ze
2(1j)
_
FD
+
1
F
T
1
_

T
2
. (B.5)
Thus, the adjoint operator is
D

F
= FD
+
1
F
, (B.6)
where
D
+
= e
2(j1)
e
2(j1)
(B.7)
when acting on a eld of spin j 1. As above, our conventions are such that D
+
and act
on everything to their right.
We are now interested in the determinant of the Laplace operator D

F
D
F
,
det
_

1
F
D
+
F
2
D

1
F
_
= det
_

1
F
2
D
+
F
2
D

_
. (B.8)
Determinants of such operators were carefully evaluated in [35];
11
Eqn. (3.2) of that paper
gives a general formula for the determinant of fg. For our case,
f = e
(2j2)+2k
+
X
+
, g = e
2j2k
+
X
+
, (B.9)
10
See Section II.E of [34] for more details on the corresponding inner products and the denition and
properties of dierential operators on Riemann surfaces.
11
Similar determinants have played a central role in other areas of CFT, perhaps most notably in the free-eld
Wakimoto realizations of WZW models; see, e.g., [36].
30
and we get
log det
_
D

F
D
F
_
=
1
24
_
d
2
e
_
[3(2j 1)
2
1]

h
mn

+ 12(2j 1)k
+

h
mn

n
X
+
+12k
2
+

h
mn

m
X
+

n
X
+
_
. (B.10)
B.1 J

The case of the Faddeev-Popov operator corresponds to spin j = 1/2. Since we are in fact
interested in the inverse square root of the determinant of the Laplacian, we nd that it
contributes to the eective Euclidean worldsheet action

1
48
_
d
2
e
_

h
mn

n
+ 12k
2
+

h
mn

m
X
+

n
X
+
_
. (B.11)
We can rewrite this contribution as a path integral over a bosonic ghost-antighost system,
1
det J

=
_
T T exp
_

_
d
2

1
F
D

(F)
_
. (B.12)
The antighost has the same conformal dimension as
+

, or (0, 1/2), while the ghost eld

has the dimension of , namely (0, 1/2). The path-integral measure of the ghost and
antighost elds, T T, is dened in the standard way, independently of X
+
. More precisely,
the standard measure on the uctuations f of a spin j eld f is induced from the covariant
norm
|f|
2
=
_
d
2
e
(22j)
f

f, (B.13)
written here in conformal gauge h
mn
= e
2

mn
.
The ghost elds and have a kinetic term suggesting that they should be purely
left-moving elds, but their conformal dimensions do not conform to this observation. It is
natural to introduce the rescaled elds

=

F
, = F. (B.14)
Because the conformal weight of F is (1/2, 1/2), both

and are now elds of conformal
dimension (1/2, 0). Moreover, in terms of these rescaled elds, the classical ghost action takes
the canonical form of a purely left-moving rst-order system of central charge c = 1,
S
e
e
=
_
d
2

. (B.15)
However, the rescaling (B.14) has an eect on the measure in the path integral. In terms
of the rescaled variables, the originally X
+
independent measure acquires a non-canonical,
explicit X
+
dependence. The new measure on

and is induced from
|

|
2
=
_
d
2
e

F
2

, | |
2
=
_
d
2
e

F
2

. (B.16)
31
In order to distinguish it from the standard X
+
independent measure, we denote the measure
induced from (B.16) by T
X
+

T
X
+
. It is convenient to replace this X
+
dependent measure
by hand with the standard measure T

T for left-moving elds

, of spin 1/2, dened
with the use of the standard norm (B.13) that is independent of X
+
. In order to do so, we
must correct for the error by including the corresponding Jacobian, which is precisely the X
+
dependent part of J

. Thus, we can now write

1
det J

=
_
T
X
+

T
X
+
exp
_

_
d
2

_
(B.17)
= exp
_
1
48
_
d
2
e
_
12k
2
+

h
mn

m
X
+

n
X
+
_
__
T

T exp
_

_
d
2

_
.
The canonical

, ghosts correctly reproduce the Liouville dependence of J

. Of course,
we now have a contribution to the eective action. Written in Minkowski signature, this
correction becomes
1
4
_
d
2
e
_
k
2
+

h
mn

m
X
+

n
X
+
_
. (B.18)
B.2

J
The evaluation of J

is not the only calculation that can contribute to the shift of the linear
dilaton. We also need to analyze the determinant involved in the change of variables that
turns the gravitino and its conjugate eld into manifestly left-moving elds. As we shall now
show, this transformation also contributes to the linear dilaton shift.
Consider the kinetic term between the conjugate pair of
++
and

. The relevant part

of the path integral, written here still in the Minkowski worldsheet signature, is
_
T
++
T

exp
_
iV

_
d
2

++

k
+

++

X
+

_
_
. (B.19)
This path integral would give the determinant of the operator D
F
acting on elds of spin j =
3/2. Wick rotating to Euclidean signature and using (B.10), we obtain for this determinant
exp
_

1
48
_
d
2
e
_
11

h
mn

n
+ 24k
+

h
mn

n
X
+
+ 12k
2
+

h
mn

m
X
+

n
X
+
_
_
.
(B.20)
As in the calculation of the Faddeev-Popov determinant above, it is again useful to rescale
the elds by an X
+
dependent factor. This rescaling was in fact introduced in Eqn. (5.24),
repeated here for convenience:

++
= F
++
,

F
. (B.21)
The rst term in the conformal anomaly can be interpreted as due to the purely left-moving
conjugate pair of elds
++
and

with the standard X

+
independent measure induced
32
from (B.13). In order to reproduce correctly the full determinant (B.20), we again have to
compensate for the X
+
dependence of the measure by including the rest of the gravitino
determinant (B.20) as an explicit conversion factor. Thus, the consistent transformation of
elds includes the measure change
T
X
+

++
T
X
+

=

J T
++
T

, (B.22)
with

J = exp
_

1
4
_
d
2
e
_
2k
+

h
mn

n
X
+
+ k
2
+

h
mn

m
X
+

n
X
+
_
_
. (B.23)
Together with the contribution from the Faddeev-Popov determinant (B.17), we see that
the contributions to the
+
X
+

X
+
term in fact cancel, and we are left with the following
one-loop correction to the Euclidean action due to the measure factors,
S
E
=
1
2
_
d
2
e k
+
h
mn

n
X
+
, (B.24)
together with the bosonic superghosts of spin 1/2 (and with the canonical path integral
measure) and central charge c = 1, plus the gravitino sector consisting of the canonical
conjugate pair
++
,

also with the canonical measure and central charge c = 11.

Integrating (B.24) by parts and using the following expression for the worldsheet scalar
curvature in conformal gauge,
eR = 2
mn

n
, (B.25)
we end up with
S
E
=
1
4
_
d
2
e k
+
X
+
R. (B.26)
Rotating back to Minkowski signature, we nd
S =
1
4
_
d
2
e k
+
X
+
R. (B.27)
This term represents an eective shift in the dilaton by V
+
= k
+
. The total central charge
of the combined matter and ghost system, with the one-loop correction (B.27) to the linear
dilaton included, is zero.
References
[1] A. Sen, Tachyon dynamics in open string theory, Int. J. Mod. Phys. A20 (2005) 55135656,
[hep-th/0410103].
[2] M. Headrick, S. Minwalla, and T. Takayanagi, Closed string tachyon condensation: An
overview, Class. Quant. Grav. 21 (2004) S1539S1565, [hep-th/0405064].
[3] E. Witten, Instability of the Kaluza-Klein vacuum, Nucl. Phys. B195 (1982) 481.
33
[4] M. Fabinger and P. Horava, Casimir eect between world-branes in heterotic M-theory, Nucl.
Phys. B580 (2000) 243263, [hep-th/0002073].
[5] S. Hellerman and I. Swanson, Cosmological solutions of supercritical string theory,
hep-th/0611317.
[6] S. Hellerman and I. Swanson, Dimension-changing exact solutions of string theory,
hep-th/0612051.
[7] S. Hellerman and I. Swanson, Cosmological unication of string theories, hep-th/0612116.
[8] S. Hellerman and I. Swanson, Charting the landscape of supercritical string theory,
arXiv:0705.0980 [hep-th].
[9] J. L. Karczmarek and A. Strominger, Closed string tachyon condensation at c = 1, JHEP 05
(2004) 062, [hep-th/0403169].
[10] P. Horava and C. A. Keeler, Noncritical M-theory in 2+1 dimensions as a nonrelativistic Fermi
liquid, JHEP 07 (2007) 059, [hep-th/0508024].
[11] P. Horava and C. A. Keeler, Closed-string tachyon condensation and the worldsheet super-Higgs
eect, arXiv:0709.2162 [hep-th].
[12] P. Horava and E. Witten, Heterotic and Type I string dynamics from eleven dimensions, Nucl.
Phys. B460 (1996) 506524, [hep-th/9510209].
[13] P. Horava and E. Witten, Eleven-dimensional supergravity on a manifold with boundary, Nucl.
Phys. B475 (1996) 94114, [hep-th/9603142].
[14] H. Kawai, D. C. Lewellen, and S. H. H. Tye, Classication of closed fermionic string models,
Phys. Rev. D34 (1986) 3794.
[15] S. P. De Alwis and A. T. Flournoy, Closed string tachyons and semi-classical instabilities, Phys.
Rev. D66 (2002) 026005, [hep-th/0201185].
[16] L. J. Dixon and J. A. Harvey, String theories in ten-dimensions without space-time
supersymmetry, Nucl. Phys. B274 (1986) 93105.
[17] P. Forg acs, Z. Horv ath, L. Palla, and P. Vecsernyes, Higher level Kac-Moody representations
and rank reduction in string models, Nucl. Phys. B308 (1988) 477.
[18] D. C. Lewellen, Embedding higher level Kac-Moody algebras in heterotic string models, Nucl.
Phys. B337 (1990) 61.
[19] S. Elitzur and A. Giveon, Connection between spectra of nonsupersymmetric heterotic string
models, Phys. Lett. B189 (1987) 52.
[20] P. Di Francesco, P. Mathieu, and D. Senechal, Conformal eld theory, Springer (1997).
[21] S. Chaudhuri, G. Hockney, and J. D. Lykken, Maximally supersymmetric string theories in d
10, Phys. Rev. Lett. 75 (1995) 22642267, [hep-th/9505054].
[22] S. Chaudhuri and J. Polchinski, Moduli space of CHL strings, Phys. Rev. D52 (1995)
71687173, [hep-th/9506048].
[23] N. Berkovits and C. Vafa, On the uniqueness of string theory, Mod. Phys. Lett. A9 (1994)
653664, [hep-th/9310170].
34
[24] F. Bastianelli, A locally supersymmetric action for the bosonic string, Phys. Lett. B322 (1994)
340343, [hep-th/9311157].
[25] H. Kunitomo, On the nonlinear realization of the superconformal symmetry, Phys. Lett. B343
(1995) 144146, [hep-th/9407052].
[26] I. N. McArthur, The Berkovits-Vafa construction and nonlinear realizations, Phys. Lett. B342
(1995) 9498, [hep-th/9411150].
[27] I. N. McArthur, Gauging of nonlinearly realized symmetries, Nucl. Phys. B452 (1995) 456468,
[hep-th/9504160].
[28] J. McGreevy and E. Silverstein, The tachyon at the end of the universe, JHEP 08 (2005) 090,
[hep-th/0506130].
[29] G. T. Horowitz and E. Silverstein, The inside story: Quasilocal tachyons and black holes, Phys.
Rev. D73 (2006) 064016, [hep-th/0601032].
[30] O. Aharony and E. Silverstein, Supercritical stability, transitions and (pseudo)tachyons, Phys.
Rev. D75 (2007) 046003, [hep-th/0612031].
[31] M. Flohr, Bits and pieces in logarithmic conformal eld theory, Int. J. Mod. Phys. A18 (2003)
44974592, [hep-th/0111228].
[32] M. R. Gaberdiel, An algebraic approach to logarithmic conformal eld theory, Int. J. Mod. Phys.
A18 (2003) 45934638, [hep-th/0111260].
[33] P. C. West, Introduction to supersymmetry and supergravity, World Scientic, Singapore (1990)
2nd ed.
[34] E. DHoker and D. H. Phong, The geometry of string perturbation theory, Rev. Mod. Phys. 60
(1988) 917.
[35] R. Kallosh and A. Y. Morozov, Green-Schwarz action and loop calculations for superstring, Int.
J. Mod. Phys. A3 (1988) 1943.
[36] A. Gerasimov, A. Morozov, M. Olshanetsky, A. Marshakov, and S. L. Shatashvili,
Wess-Zumino-Witten model as a theory of free elds, Int. J. Mod. Phys. A5 (1990) 24952589.
35