hep-th/0204143

Field Theory of Tachyon Matter

Ashoke Sen
arXiv:hep-th/0204143 v4 2 May 2002

Harish-Chandra Research Institute

Chhatnag Road, Jhusi, Allahabad 211019, INDIA
and
Department of Physics, Penn State University
University Park, PA 16802, USA
E-mail: asen@thwgs.cern.ch, sen@mri.ernet.in

Abstract

We propose a field theory for describing the tachyon on a brane-antibrane system

near the minimum of the potential. This field theory realizes two known properties of
the tachyon effective action: 1) absence of plane-wave solutions around the minimum,
and 2) exponential fall off of the pressure at late time as the tachyon field evolves from
any spatially homogeneous initial configuration towards the minimum of the potential.
Classical solutions in this field theory include non-relativistic matter with arbitrary spatial
distribution of energy.

1
 Dynamics of the tachyon on D-brane anti-D-brane system or unstable D-branes in
superstring theory and bosonic string theory have been investigated during the last few
years. Of the various known properties of the classical tachyon effective action,1 two
specific properties which specify the behaviour of the tachyon effective action around the
minimum of the potential are as follows:

1. Since the minimum of the potential describes a configuration where there are no
D-branes[1], around this minimum there are no physical open string excitations.
Translated to a property of the tree level tachyon effective action, this implies that
there are no plane wave solutions to the linearized equations of motion around the
minimum of the tachyon potential.

2. If we let the tachyon roll beginning with any spatially homogeneous initial configuration[2],
it evolves asymptotically towards its minimum instead of oscillating about the
minimum[3, 4]. The total energy density is conserved during the evolution but
the pressure evolves to zero. The information about the tachyon effective action
near the minimum of the potential is encoded in the late time behaviour of the
pressure. This is given by[3, 4]2
0
p = −Ke−αx , (1)

where K is an irrelevant normalization constant which can be changed by shifting

the origin of the time coordinate x0 , and α is given (in the α′ = 1 unit) by

α = 1 for bosonic string theory

√
= 2 for superstring theory . (2)

Several field theory models have been proposed for realising the first property. These
models are of two kind.3 In the first kind[7] the tachyon potential is non-singular at the
minimum of the potential but the absence of physical plane-wave solutions has its origin in
the structure of the kinetic term (which typically involves higher derivative terms). This
1
Although we shall use the language of tachyon effective action obtained by integrating out the other
open string modes, we should note that many of these results are derived using the technology of full
string field theory, containing all the infinite number of fields describing the dynamics of the open string.
2
Since we are interested in studying tree level properties of the tachyon potential, we shall ignore the
backreaction of gravity and other closed string fields on the evolution of the tachyon. Once the form of
the action is derived by this method, one can of course couple it to background gravitational field and
study the evolution of the coupled system. These effects have been studied in refs.[5]. For some earlier
attempts at the study of time dependent solutions for the tachyon field see [6].
3
Here we are stating the results in terms of properly redefined field φ for which the two derivative
term has the standard form − 21 η µν ∂µ φ∂ν φ.

2
possibility seems to be realized[8] in the explicit analysis of the string field theory based
on ∗-product interaction[9]. In the second kind of field theory models[10] the tachyon
potential is singular at the minimum. In particular the second derivative of the potential
blows up and hence the tachyon mass becomes infinite. Analysis based on boundary string
field theory[11] seems to support this view[12]. Of course these results are not mutually
incompatible, since an effective field theory of the first kind may be turned into one of
second kind and vice versa by a suitable field redefinition that involves (infinite number
of) derivatives of the tachyon field.
A field theory model that realises the evolution of the system to a pressureless gas was
proposed in [4] following earlier proposals in [13, 14]. The action was taken to be
Z √
S=− dp+1x V (T ) − det A , (3)

where
Aµν = ηµν + ∂µ T ∂ν T . (4)
However in this proposal the tachyon potential V (T ) was left undetermined. The only
property of V (T ) that was determined is that the minimum of V (T ) where V (T ) = 0
should be at T = ∞.
In this paper we shall show that by taking V (T ) ∝ e−αT /2 for large T in the action (3),
we can satisfy both the requirements given above: the absence of plane-wave solutions
around the tachyon vacuum, and the exponential fall off of the pressure for large x0 . The
constant of proportionality can be absorbed into a constant shift of T . Thus the proposed
tachyon effective action for large T on an unstable D-p-brane system is
Z √ Z q
p+1 −αT /2
S=− d xe − det A = − dp+1 x e−αT /2 1 + η µν ∂µ T ∂ν T , (5)

where Aµν is given by eq.(4). We expect this to be a valid approximation when the
second and higher order derivatives of T are small. As we shall see, the tachyon matter
does satisfy this condition.
On a brane-antibrane system where T is a complex field, the tachyon potential is
given by e−α|T |/2 . In this case the action (5) can be regarded as the restriction of the
full action to the case ℑ(T ) = 0. We shall carry out our analysis for this restricted
field configuration. The imaginary part of the tachyon, which can be regarded as the
Goldstone mode associated to the broken U(1) phase symmetry, is absorbed by the U(1)
gauge field due to Higgs mechanism, and is expected to decouple from the dynamics near
the minimum of the potential.
We shall first verify that the action (5) produces the correct large x0 behaviour of
the pressure. For this analysis we can restrict to spatially homogeneous, time dependent

3
field configurations. As discussed in [14, 4], for such configurations the conserved energy
density is given by:
T00 = e−αT /2 (1 − (∂0 T )2 )−1/2 . (6)
Since T00 is conserved, we see that for any given T00 , as T → ∞, ∂0 T → 1. In particular,
for large x0 the solution has the form:4
0 0
T = x0 + Ce−αx + O(e−2αx ) . (7)

In order to see that (7) gives the correct form of the solution we simply need to note
that the leading contribution to T00 computed from this configuration remains constant
in time:
1
T00 ≃ √ . (8)
2αC
The pressure associated with this configuration is given by[14, 4]:
√ 0
p = −e−αT /2 (1 − (∂0 T )2 )1/2 ≃ − 2αCe−αx . (9)

This is in precise agreement with (1).

Next we shall demonstrate the absence of plane-wave solutions. First let us expand
the action in powers of derivatives of T up to terms containing at most two powers of
derivatives. This gives
Z
1 µν
S=− dp+1x e−αT /2 (1 + η ∂µ T ∂ν T + . . .) , (10)
2
where . . . denotes terms with higher powers of derivatives. Defining

φ = e−αT /4 , (11)

we get
Z
16 1 µν α2 2
S= dp+1x
(− η ∂µ φ∂ν φ − φ + . . .) . (12)
α2 2 16
The minimum of the potential is at φ = 0. If we ignore the higher derivative terms then
√
the quantization of the action seems to lead to a scalar particle of mass α/2 2. This
is reflected in the fact that the equations of motion without the higher derivative terms
contain a plane-wave solution of the form:

φ = a eikµ x ,
µ
(13)
4
Note that since for this configuration ∂0n T for n > 1 falls off exponentially, the contribution from
possible higher derivative corrections like (∂µ ∂ µ T )2 inside the square root on the right hand side of (5)
will be suppressed.

4
with −η µν kµ kν = α2 /8 and a an arbitrary constant.
We shall now show that when we take into account the effect of the higher derivative
terms, (13) ceases to be a solution of the equations of motion for any kµ .5 To show this
we note that the full action (5) written in terms of φ defined in (11) takes the form:
s
Z
16 −2 µν
S=− dp+1 x φ2 1 + φ η ∂µ φ∂ν φ . (14)
α2
The action is homogeneous of degree 2 in φ. The equations of motion derived from this
action is given by:
 
∂ν φ α2 φ + α82 φ−1 η µν ∂µ φ∂ν φ
− η µν ∂µ q + q = 0. (15)
1+ 16 −2 µν
φ η ∂µ φ∂ν φ 8 1 + 162 φ−2 η µν ∂µ φ∂ν φ
α2 α

Substituting (13) into (15) we get:

a
q = 0. (16)
16 µν
1− α2
η kµ kν

Clearly this equation has no non-trivial solution for finite values of kµ . This establishes
the absence of plane-wave solutions.6 (Note that if we expand the left hand side of this
equation in powers of kµ and keep up to quadratic terms, we reproduce the mass-shell
condition −η µν kµ kν = α2 /8 derived earlier.)
Absence of plane wave solutions does not imply absence of other classical solutions
however. We have already seen the existence of solutions with constant (but arbitrary)
energy density. As emphasized in [14] (see also [15, 16]) the correct excitations of the
system can be found by working in the hamiltonian formalism. Defining the momentum
conjugate to T as:
δS
Π(x) = , (17)
δ(∂0 T (x))
we can construct the Hamiltonian H following [14]:
Z √ q
H= dp xH, H = T00 = Π2 + e−αT 1 + ∂i T ∂i T . (18)
5
Note that (13) describes a complex solution. Since the full equations of motion are homogeneous
of degree one, but not linear, even if we had a solution of the form (13), it is not clear how we could
construct a real solution since we cannot superpose the solutions. We shall show however that even this
complex solution does not exist.
6
Note again that since for the configuration (13) ∂µ ∂ν T vanishes, contribution to the equations of
motion from possible higher derivative corrections like (∂µ ∂ µ T )2 inside the square root on the right hand
size of (5) vanishes.

5
For large T we can ignore the e−αT term, and the equations of motion take the form:
 
δH ∂j T
∂0 Π(x) = − = ∂j Π(x) √ . (19)
δT (x) 1 + ∂i T ∂i T
δH q
∂0 T (x) = = 1 + ∂i T ∂i T , (20)
δΠ(x)
for Π(x) > 0. Thus we can get a solution to the equations of motion by taking ∂i T =
∂0 Π = 0, ∂0 T = 1. This gives

Π(x) = f (~x) , T (x) = x0 , (21)

where f (~x) is any arbitrary function of the spatial coordinates. The energy density as-
sociated with such a solution is proportinal to f (~x). This shows that the system admits
classical solutions with energy density which is time independent but has arbitrary de-
pendence on the spatial coordinates. Classically this energy density can be as low as we
like. The ability to create configurations with arbitrarily low energy density can be traced
to the scale invariance:

T (x) → λ−1 T (λx), Π(x) → λp+1 Π(λx) , (22)

under which the Poisson brackets and the equations of motion remain unchanged, and
H(x) → λp+1 H(λx).
Since the original theory is Lorentz invariant, it is clear that given a localized density
of tachyon matter we should be able to boost it to an arbitrary velocity. Intuitively we also
expect that is should be possible to construct configurations with different local velocities
at different points. This can be seen to be the case by rewriting the equations of motion
in a slightly different form. We define:

uµ = ∂µ T, ǫ(x) = Π(x)/∂0 T (x) . (23)

The equations of motion then take the suggestive form:

η µν uµ uν = −1, ∂µ (ǫ(x)uµ ) = 0 . (24)

The energy momentum tensor can be computed following [14]. Expressed in terms of
these new variables, Tµν take the form:

Tµν = ǫ(x)uµ uν . (25)

These are precisely the equations govering gradient flow of non-interacting dust, with uµ
interpreted as the local velocity vector. These equations are expected to be valid as long
as ∂µ uν is small in magnitude (in string units).

6
 Although in our analysis we have ignored the coupling of the tachyon to various
massless fields on the D-brane world-volume, these can be easily incorporated following
[17, 13, 14, 15]. We can generalize (5) to curved background space-time by replacing
ηµν in eq.(4) by the closed string metric gµν . If we also have anti-symmetric tensor field
background then ηµν should be replaced by gµν +Bµν . The dilaton φ couples via an overall
factor of e−φ multiplying the lagrangian density. For non-BPS D-p-brane of type II string
theories, another quantity of interest is the coupling of the tachyon to Ramond-Ramond
(RR) field C (p) . In particular it is known[18] that the D-p-brane world-volume theory has
a coupling: Z
dp+1 x f (T ) dT ∧ C (p) , (26)
where f (T ) is some function of the tachyon field T . We can determine the behaviour
of f (T ) for large T if we can find the source of the RR field that is generated by the
rolling tachyon field[2]. This can be computed from the boundary state associated with
the solution. Although we have not performed a systematic analysis of the problem along
the lines of [3, 4], general arguments based on symmetry and other considerations lead us
to guess that (in α′ = 1 unit) the source for the RR p-form field is proportional to:
√ √
 0 0 
e ex / 2 e−x / 2
sin(λπ) e
√ − e
√ , (27)
1 + sin2 (λπ)e 2x0 1 + sin2 (λπ)e− 2x0

e is the parameter labelling the total energy density of the system:

where λ
Tp e
T00 = (1 + cos(2λπ)) , (28)
2
Tp being the tension
√
of the non-BPS D-p-brane. Thus for large x0 the source is propor-
0
tional to e−x / 2 . Since T ≃ x0 for large x0 , this indicates that for large T , the coupling
of the RR p-form field to the tachyon has the form:
Z √
dp+1x e−T / 2
dT ∧ C (p) , (29)

up to an overall normalization constant.

There are many issues which call for further investigation. As we have seen, the
effective field theory describing tachyon matter has classical solutions with localized energy
densities. It will be interesting to construct the two dimensional conformal field theories
associated with these solutions along the lines of refs.[3, 4] where the conformal field
theories associated with spatially homogeneous energy densities were constructed. (Of
course, one possibility is to take an arbitrary spatial distribution of coincident D0-D̄0 pair
and construct the rolling tachyon solution on each such pair.) Another issue of importance
is the effect of quantum corrections on such background.

7
 Acknowledgement: I would like to thank L. Kofman, J. Maldacena and T. Pad-
manabhan for useful discussions. This work was supported in part by a grant from the
Eberly College of Science of the Penn State University.

References
[1] A. Sen, JHEP 9808, 010 (1998) [arXiv:hep-th/9805019]; JHEP 9808, 012 (1998)
[arXiv:hep-th/9805170]; JHEP 9912, 027 (1999) [arXiv:hep-th/9911116].

[2] M. Gutperle and A. Strominger, “Spacelike branes,” arXiv:hep-th/0202210.

[3] A. Sen, “Rolling Tachyon,” arXiv:hep-th/0203211.

[4] A. Sen, “Tachyon matter,” arXiv:hep-th/0203265.

[5] G. W. Gibbons, arXiv:hep-th/0204008; M. Fairbairn and M. H. Tytgat, arXiv:hep-

th/0204070; S. Mukohyama, arXiv:hep-th/0204084; A. Feinstein, arXiv:hep-
th/0204140; T. Padmanabhan, arXiv:hep-th/0204150; A. Frolov, L. Kofman and
A. A. Starobinsky, arXiv:hep-th/0204187; D. Choudhury, D. Ghoshal, D. P. Jatkar
and S. Panda, arXiv:hep-th/0204204; X. Li, J. Hao and D. Liu, arXiv:hep-
th/0204252; G. Shiu and I. Wasserman, arXiv:hep-th/0205003.

[6] C. Acatrinei and C. Sochichiu, arXiv:hep-th/0104263; S. H. Alexander, Phys. Rev.

D 65, 023507 (2002) [arXiv:hep-th/0105032]; A. Mazumdar, S. Panda and A. Perez-
Lorenzana, Nucl. Phys. B 614, 101 (2001) [arXiv:hep-ph/0107058]; S. Sarangi and
S. H. Tye, arXiv:hep-th/0204074.

[7] L. Brekke, P. G. Freund, M. Olson and E. Witten, Nucl. Phys. B 302, 365 (1988);
P. H. Frampton and Y. Okada, Phys. Rev. D 37, 3077 (1988); D. Ghoshal and A. Sen,
Nucl. Phys. B 584, 300 (2000) [arXiv:hep-th/0003278]; J. A. Minahan, JHEP 0103,
028 (2001) [arXiv:hep-th/0102071].

[8] V. A. Kostelecky and S. Samuel, Nucl. Phys. B 336, 263 (1990); A. Sen and
B. Zwiebach, JHEP 0010, 009 (2000) [arXiv:hep-th/0007153]; W. Taylor, JHEP
0008, 038 (2000) [arXiv:hep-th/0008033]; H. Hata and S. Teraguchi, JHEP 0105,
045 (2001) [arXiv:hep-th/0101162]; I. Ellwood and W. Taylor, Phys. Lett. B 512,
181 (2001) [arXiv:hep-th/0103085]; L. Rastelli, A. Sen and B. Zwiebach, arXiv:hep-
th/0012251.

[9] E. Witten, Nucl. Phys. B 268, 253 (1986). N. Berkovits, Nucl. Phys. B 450, 90
(1995) [Erratum-ibid. B 459, 439 (1995)] [arXiv:hep-th/9503099].

8
[10] J. A. Minahan and B. Zwiebach, JHEP 0009, 029 (2000) [arXiv:hep-th/0008231];
JHEP 0103, 038 (2001) [arXiv:hep-th/0009246]; JHEP 0102, 034 (2001) [arXiv:hep-
th/0011226].

[11] E. Witten, Phys. Rev. D 46, 5467 (1992) [arXiv:hep-th/9208027]; Phys. Rev. D 47,
3405 (1993) [arXiv:hep-th/9210065]; S. L. Shatashvili, Phys. Lett. B 311, 83 (1993)
[arXiv:hep-th/9303143].

[12] A. A. Gerasimov and S. L. Shatashvili, JHEP 0010, 034 (2000) [arXiv:hep-

th/0009103]; D. Kutasov, M. Marino and G. W. Moore, JHEP 0010, 045 (2000)
[arXiv:hep-th/0009148]; arXiv:hep-th/0010108.

[13] M. R. Garousi, Nucl. Phys. B 584, 284 (2000) [arXiv:hep-th/0003122];

E. A. Bergshoeff, M. de Roo, T. C. de Wit, E. Eyras and S. Panda, JHEP 0005,
009 (2000) [arXiv:hep-th/0003221]; J. Kluson, Phys. Rev. D 62, 126003 (2000)
[arXiv:hep-th/0004106].

[14] G. W. Gibbons, K. Hori and P. Yi, “String fluid from unstable D-branes,” Nucl.
Phys. B 596, 136 (2001) [arXiv:hep-th/0009061].

[15] O. Bergman, K. Hori and P. Yi, “Confinement on the brane,” Nucl. Phys. B 580, 289
(2000) [arXiv:hep-th/0002223]; J. A. Harvey, P. Kraus and F. Larsen, JHEP 0012,
024 (2000) [arXiv:hep-th/0010060]; F. Larsen, Int. J. Mod. Phys. A 16, 650 (2001)
[arXiv:hep-th/0010181].

[16] U. Lindstrom and R. von Unge, Phys. Lett. B 403, 233 (1997) [arXiv:hep-
th/9704051]; H. Gustafsson and U. Lindstrom, Phys. Lett. B 440, 43 (1998)
[arXiv:hep-th/9807064]; U. Lindstrom, M. Zabzine and A. Zheltukhin, JHEP 9912,
016 (1999) [arXiv:hep-th/9910159].

[17] A. Sen, JHEP 9910, 008 (1999) [arXiv:hep-th/9909062]; J. Math. Phys. 42, 2844
(2001) [arXiv:hep-th/0010240].

[18] E. Witten, JHEP 9812, 019 (1998) [arXiv:hep-th/9810188]; P. Horava, Adv. Theor.
Math. Phys. 2, 1373 (1999) [arXiv:hep-th/9812135]; A. Sen, arXiv:hep-th/9904207.