UFIFT-HEP-92-10

hep-ph/9210210

Wiggly Relativistic Strings

Jooyoo Hong∗ , Jaewan Kim and Pierre Sikivie

Department of Physics, University of Florida, Gainesville, FL 32611

arXiv:hep-ph/9210210v1 5 Oct 1992

We derive the equations of motion for general strings, i.e. strings with arbitrary

relation between tension τ and energy per unit length ǫ. The renormalization of τ

and ǫ that results from averaging out small scale wiggles on the string is obtained in

the general case to lowest order in the amount of wiggliness. For Nambu-Goto strings

we find deviations from the equation of state ǫτ = constant in higher orders. Finally

we argue that wiggliness may radically modify the gauge cosmic string scenario.

PACS numbers:

Typeset Using REVTEX

1
 Computer simulations [1] have shown that cosmic gauge strings acquire much more small
scale structure than had been thought previously [2]. The effect of small scale structure is
generally speaking to increase the energy per unit length of a gauge string and to decrease

its tension when a coarse grained description of the wiggly string is adopted [3,4]. While
this fact has been appreciated in the past, it appears that there is no general treatment
yet of the renormalization of string tension and energy per unit length due to small scale
wiggliness. It appears even that there is no relativistic description in place for the motion

of strings when the tension differs from the energy per unit length. The present paper aims
to fill this gap. The formalism developed below can be applied to all strings and we hope it
will prove useful in other contexts. An example which comes to mind is a gauge string with
fermion zero modes attached to it [5]. When a large number of particles are attached to a

string, they act collectively as a fluid adding to the energy per unit length but subtracting
from the tension.
Consider then a general string, i.e. an object whose stress-energy-momentum is localized
on a line in space. Let X µ (σ) be the location of the string worldsheet with respect to

a Lorentz reference frame. σ = (σ 0 , σ 1 ) are arbitrarily chosen coordinates parametrizing

points on the worldsheet. The associated 2-dim. metric is, as usual:

hab (σ) = ∂a X µ ∂b Xµ , a, b = 0, 1 . (1)

At point X µ (σ) on the string lives a 2-dim. stress-energy-momentum tensor

tab = (ǫ − τ )ua ub + τ hab , (2)

where ǫ(σ) is the energy per unit length of the string, τ (σ) its tension and ua (σ) is the fluid
velocity parallel to the string; ua ua = +1. The 4 dim. stress-energy-momentum tensor is
Z √  
T µν (x) = d2 σ −h tab (σ)∂a X µ (σ)∂b X ν (σ)δ 4 x − X(σ) , (3)

where h = det(hab ). This expression is both invariant under 2-dim. reparametrizations

and covariant under 4-dim. Lorentz transformations. Its validity in the neighborhood of

2
each point on the worldsheet can be verified explicitly by choosing a Lorentz frame which
is instantaneously at rest with respect to the string at that point. The motion of the string
must be such that ∂µ T µν (x) = 0. It is easy to show that this condition is equivalent to
√
" #
ab µ
∂a −h t (σ)∂b X = 0 (µ = 0, 1, 2, 3) . (4)

To provide a complete description of the string dynamics, Eq.(4) must be supplemented by

an equation of state:

τ = τ (ǫ) . (5)

~ ⊥ (σ), ǫ(σ), τ (σ) and β(σ) ≡

We then have five equations for the five unknowns X
~ ⊥ (σ) represents the two transverse degrees of freedom of the string. β(σ)
u1 (σ)/u0 (σ). X
is its longitudinal velocity. Eqs.(4) and (5) uniquely specify the motion of the string that
results from arbitrary initial conditions. Note that our description is Lorentz invariant as
well as generally covariant in the 2-dim. sense.

The case of the Nambu-Goto (NG) string is included in this description. The well-known
NG equations of motion are Eq.(4) with tab (σ) replaced by hab (σ). Let us first note that
two of these equations are merely mathematical identities since
√
" #
∂e Xµ ∂a −h hab (σ)∂b X µ = 0 (e = 0, 1) (6)

follows from Eq.(1) and nothing else. The other two NG equations specify the motion of
~ ⊥ (σ). Now, returning to our description of a general string, let us adopt the equation of
X
state τ = ǫ. It is easy to show, using Eqs.(6), that two of the Eqs.(4) are equivalent to

∂a τ (σ) = 0 a = 0, 1 . (7)

They therefore imply that τ must be a constant. The two remaining Eqs.(4) are the non-
~ ⊥ (σ). Thus we have learned that
trivial NG equations that determine the motion of X
τ = constant is the only consistent way to have τ = ǫ.
The equation of state appropriate to a particular kind of string must be derived from the
relevant microphysics. As in other studies of fluid dynamics, an average over a large number

3
of micro-configurations consistent with a given macroscopic description must be performed
to obtain the energy per unit length ǫ and the tension τ of the string. The equation of state
gives the relationship between ǫ and τ when the string is slowly stretched (i.e. the stretching

timescale is long compared to the microphysics timescales). The focus of our paper is the
renormalization of ǫ and τ due to small scale wiggles on a string with arbitrary equation of
2π
state. We must define an averaging length scale λ ≡ k
. ǫ(k) and τ (k) are the values of the
energy per unit length and tension of the string when all wiggles of wavelength shorter than

λ are averaged over. We will derive the renormalization group equations for ǫ and τ due to
transverse and longitudinal wiggles to second order in the amplitude of the wiggles.
Consider a string which lies on average along the x-axis. We choose t and x as worldsheet
   
coordinates. Thus X µ (σ) = t, x, y(t, x), z(t, x) . In this gauge, Eqs.(4) are:
√ 
∂a −h tab = 0 , (8a)

tab ∂a ∂b y = tab ∂a ∂b z = 0 . (8b)

Let us first discuss transverse wiggles. Consider a string of equation of state τ = τ (ǫ)

stretched along the x-axis. At rest (β = y = z = 0) the string has energy per unit length
ǫ0 and tension τ0 = τ (ǫ0 ). At time t = 0, the string is given a transverse velocity in the
y-direction: ẏ(0, x) = βT sin kx. We assume βT ≪ 1 and expand Eqs.(8) in powers of βT .
This yields to lowest order

ǫ0 ÿ − τ0 y ′′ = 0 (9a)
1
ǫ̈1 − vL2 ǫ′′1 = (−ǫ0 ∂t2 + τ0 ∂x2 ) (ẏ 2 + y ′2 ) (9b)
2
(ǫ0 − τ0 )β̇ = τ0 y ′(ÿ − y ′′ ) + vL2 ǫ′1 (9c)

for the time evolution of y, β and ǫ1 = ǫ − ǫ0 . By definition vL2 ≡ − dτ | . As usual, dots and
dǫ 0

primes denote derivatives with respect to t and x. Eq.(9a) implies y = (βT /w) sin kx sin ωt
with ω = k vT where vT = ( τǫ00 )1/2 is the phase velocity of transverse wiggles. Eqs.(9b) and
(9c) determine ǫ1 (σ) and β(σ), both of which are of order ẏ 2. To obtain the renormalized

4
values ǭT and τ̄T of the energy per unit length and tension, we calculate hTµν (x)i to second
order in ẏ. The result, including the contribution from wiggles in the x-z plane, is

ǭT = ǫ0 + ǫ0 (hẏ 2 i + hż 2 i) (10a)

" #
1 ǫ

τ̄T = τ0 − (hẏ 2 i + hż 2 i) τ0 + ǫ0 + vL2 ǫ0 1 − 0 .
2 τ0
(10b)

Next, let us discuss longitudinal wiggles. Again we start with a string which is at rest,
stretched along the x-axis. In this state, it has energy per unit length ǫ0 and tension τ0 =

τ (ǫ0 ). At time t = 0, the string is given a longitudinal velocity β(0, x) = βL sin kx. Provided
βL ≪ 1, Eqs.(9) are still valid but now y = z = 0. They imply that ǫ1 ∼ (ǫ0 − τ0 )vL−1 βL and
that vL = (− dτ | )1/2 is the phase velocity of longitudinal wiggles. For the renormalization
dǫ 0

of ǫ and τ due to longitudinal wiggles we find, up to second order in β:

ǭL = ǫ0 + 2(ǫ0 − τ0 )hβ 2i (11a)

" #
d ln vL
τ̄L = τ0 − (ǫ0 − τ0 )hβ 2 i 1 + vL2 + (ǫ0 − τ0 ) |0 .
dǫ
(11b)

Thus the promised renormalization group equations for ǫ(k) and τ (k) are
dǫ
− = WT (k)ǫ + 2WL (k)(ǫ − τ ) (12a)
d ln k " #
dτ 1 2 ǫ
− = − WT (k) τ + ǫ + vL ǫ(1 − ) (12b)
d ln k 2 τ
" #
2 d ln vL
−WL (k)(ǫ − τ ) 1 + vL + (ǫ − τ ) .
dǫ
where WT (k) and WL (k) are the spectral densities on a ln k scale of respectively hẏ 2i + hż 2 i
and hβ 2 i. Eqs.(12) relate the values of ǫ and τ at one scale to their values at a vastly
different scale provided that WL , WT ≪ 1 at all intermediate scales. Note that we have

not yet obtained how the equation of state itself changes from scale to scale. To do so we
need to analyze the response of the wiggles to adiabatic stretching of the string. We leave
this to a future publication which will also contain the details of the calculation that led to
Eqs.(10-12).

5
 Let us consider the case of wiggly Nambu-Goto strings. At the shortest distance scale
k0 we have ǫ = τ = µ, where µ is the bare string tension. At slightly longer distance scales
Eqs.(12) imply ǫ(k) = µ[1 + WT (k0 ) ln k0 /k] and τ (k) = µ[1 − WT (k0 ) ln k0 /k]. Therefore

in the neighborhood of k = k0 we have the equation of state ǫτ = µ2 . Moreover, a short

d(ǫτ )
calculation shows that Eqs.(12) imply d ln k
= 0 when the equation of state is ǫτ = constant.
This equation of state is therefore a fixed point of the renormalization group equations (12).
Thus in the particular case of wiggly Nambu-Goto strings, Eqs.(12) do by themselves es-

tablish the equation of state to be ǫτ = µ2 in lowest order. (As was already emphasized,
for a general string the renormalization group equations for ǫ and τ do not by themselves
provide enough information to determine the equation of state. The Nambu-Goto string is
an exception in this regard.) The equation of state ǫτ = µ2 was found earlier by Carter [3]

and Vilenkin [4]. It seems that we have found considerable support for it. However we will
now show that in general ǫτ 6= µ2 for wiggly Nambu-Goto strings although ǫτ = µ2 may
be an excellent approximation in many cases. The fact that ǫτ 6= µ2 in general does not
contradict what we have said so far because Eqs.(12) are valid only in lowest order.

It is well known that an arbitrary motion of a Nambu-Goto string [6] is given by x(t, σ) =
1
2
[a(σ − t) + b(σ + t)] where a and b are arbitrary functions subject only to the constraint
a′2 = b′2 = 1. To describe a wiggly NG string lying on average along the x-axis we write

h i
a(σ − t) = (σ − t)γ1 + f1 (σ − t), g1y (σ − t), g1z (σ − t)
h i
b(σ + t) = (σ + t)γ2 + f2 (σ + t), g2y (σ + t), g2z (σ + t) (13)

where γ1 and γ2 are constants and f1 , f2 , g1y , g1z , g2y and g2z are functions which average to
zero and which describe the wiggles on the string. These functions are not all independent
since they must obey the gauge condition a′2 = b′2 = 1. Let us choose (t, x) as the worldsheet

coordinates of the averaged string. It is easy to show that in these coordinates the 2-dim.
stress-energy-momentum tensor of the averaged string is given by

1 2
   
00
t =µ dx =µ
| dσ
| γ1 + γ2 + f1 (σ − t) + f2′ (σ + t)
′

6
 −γ1 + γ2 − f1′ (σ − t) + f2′ (σ + t)
 
01
t =µ (14)
γ1 + γ2 + f1′ (σ − t) + f2′ (σ + t)
(γ1 + f1′ (σ − t))(γ2 + f2′ (σ + t))
 
t11 = −2µ
γ1 + γ2 + f1′ (σ − t) + f2′ (σ + t)

Clearly ǫτ = −t00 t11 + (t01 )2 6= µ2 in general. For example, consider the particular case

where the wiggles are reflection-symmetric on average (γ1 = γ2 ≡ γ, hf1′p i = hf2′p i ≡ hf ′p i

for p = 2, 3, . . .). Then an expansion of tab in powers of f ′ yields:
" #
µ 1 1 1
ǫ =t = 00
1 + 2 hf ′2 i − 3 hf ′3 i + 4 (hf ′4 i + 3hf ′2i2 ) − · · ·
γ 2γ 4γ 8γ
" #
11 1 ′2 1 ′3 1 ′4 ′2 2
τ = −t = γµ 1 − 2 hf i + 3 hf i − 4 (hf i − hf i ) + · · · . (15)
2γ 4γ 8γ

Eqs.(15) show that ǫτ 6= µ2 although the 2d and 3d order terms in the expansion of ǫτ − µ2
vanish. That the 2d order term vanishes, we already knew from Eqs.(12). In general, one
has

ǫτ 1
= 1 + hf ′2 ihf2′2 i + · · · . (16)
µ2 4γ 2 1

Note that if the wiggles are purely transverse (f1′ = f2′ = 0) then ǫτ = µ2 to all orders

[4]. However, it is clear that one must allow f1′ , f2′ 6= 0. Physically this corresponds to
the possibility of longitudinal wiggles once the Nambu-Goto string has ǫ > τ because of
transverse wiggles.
Finally, we would like to speculate on the behavior of cosmic gauge strings. Let us assume

that higher order terms in the renormalization group Eqs.(12) do not play an important role.
The equation of state is then ǫτ = µ2 , and

dǫ µ2
− = WT ǫ + 2WL (ǫ − ) . (17)
d ln k ǫ

How large are WT and WL ? At cosmic time t, when the correlation length of the string

network is ξ(t), the strings carry wiggles which have been inherited from earlier times when
2π
the correlation length was shorter. For λ = k
somewhat shorter than ξ(t), wiggles are
abundant and the corresponding values of WT and WL are large, of order one. For λ < Gµt,
WT and WL are exponentially suppressed because the decay time of wiggles on the bare

7
string into gravitational radiation is of order (Gµ)−1 λ [7]. For Gµt < λ < ξ(t), the values
of WT and WL are the outcome of a number of competing processes [8] some of which
tend to increase and some of which tend to decrease the size of wiggles associated with the

corresponding length scales. Stretching of the strings (e.g. by Hubble expansion) and the
production of loops by self-intersection with reconnection tend to decrease the size of wiggles,
whereas shortening of the string after reconnections have occurred and the production of
kinks, also as a result of reconnections, tend to increase the size of wiggles. We will assume

here that WT and WL have approximately constant values for all λ: Gµt < λ < ξ(t). We
make this assumption not because we believe that it is necessarily correct but as a means to
explore the effect of small scale wiggliness on the cosmic gauge string scenario. With regard
to the computer simulations, it is unclear to us whether they are in disagreement with this

assumption. There is so far no published result describing unambiguously the spectrum of

wiggliness.
At any rate, if WT and WL are approximately constant for Gµt < λ < ξ(t), then from
Eq.(17),
µ2
WT +2WL
ξ(t)

ǫ = ǫ(ξ) = ∼µ . (18)
τ Gµt
The typical velocity of cosmic strings is then
1 WT +2WL
τ (t) Gµt
 
2
v(t) = ∼ . (19)
ǫ(t) ξ(t)
One expects the correlation length to be ξ(t) = v(t)t. This determines ξ(t) ∼ t(Gµ)α where
WT +2WL
α= WT +2WL +1
. The density of strings today is then given by:
ǫ
Ωstr ∼ (6πGt2 ) 2
∼ (6π)(Gµ)1−3α . (20)
ξ (t)
Because of limits on the anisotropy of the microwave background radiation, Ωstr must be
1
much smaller than one. This requires α < 3
or WT + 2WL < 0.5. Even if this condition is
satisfied, the limit on Gµ may be much more severe than it is in the usual scenario which
assumes α = 0. For example, if WT + 2WL = 0.2, Ωstr < 10−5 implies Gµ < 10−12 instead
of Gµ < 10−6 in the usual scenario.

8
 ACKNOWLEDGMENTS

We are grateful to A. Vilenkin, P. Ramond, C. Thorn, T. Kibble, E. Copeland and P.

Griffin for useful discussions. One of us (P.S.) would like to thank the Aspen Center for
Physics for its hospitality while he was working on part of this project. This work was
supported in part by the U.S. Department of Energy under contract DE-FG05-86ER40272,
and by the Korea Science and Engineering foundation.

9
 REFERENCES

∗
Permanent Address: Department of Physics, Hanyang University, Ansan, Kyôngki-Do,
South Korea.

[1] D. P. Bennett and F. R. Bouchet, Phys. Rev. Lett. 60, 257 (1988) and Phys. Rev. Lett.
63, 2776 (1989) and in Symposium on the formation and evolution of cosmic strings,

edited by G. W. Gibbons, S. W. Hawking, and T. Vachaspati (Cambridge Univ. Press,

Cambridge, 1989); B. Allen and E.P.S. Shellard, Phys. Rev. Lett. 64, 119 (1990). For
earlier work on cosmic string simulations, see A. Albrecht and N. Turok, Phys. Rev. Lett.
54, 1868 (1985) and Phys. Rev. D 40, 973 (1989).

[2] For a review on cosmic strings, see A. Vilenkin, Phys. Rep. 121, 263 (1985).

[3] B. Carter, Phys. Rev. D 41, 3869 (1990).

[4] A. Vilenkin, Phys. Rev. D 41, 3038 (1990).

[5] E. Witten, Phys. Rev. D 30, 272 (1984).

[6] P. Goddard, J. Goldstone, C. Rebbi and C. Thorn, Nucl. Phys. B56, 109 (1973); T. W.

B. Kibble and N. Turok, Phys. Lett. 116B, 141 (1982).

[7] M. Hindmarsh, Phys. Lett. 251B, 28 (1990); J. Quashnock and D. Spergel, Phys. Rev.

D 42, 2505 (1990); J. Quashnock and T. Piran, Phys. Rev. D D43, 3785 (1991).

[8] These processes are idscussed in T. W. B. Kibble and E. Copeland, Phys. Scripta T36,
153 (1991); B. Allen and R. Caldwell, Phys. Rev. D 43, 2457 (1991); E. Copeland, T.

W. B. Kibble and D. Austin, Phys. Rev. D 45, R1000 (1992).

10