PSU/TH/159

Warm Inflation

Arjun Berera
Department of Physics, The Pennsylvania State University, University Park, Pennsylvania,

16802, USA
arXiv:astro-ph/9509049v1 8 Sep 1995

(November 26, 2024)

Abstract

Assuming a first order phase transition during inflation, a model scenario is

described which does not require a tiny coupling constant. Thermal equilib-

rium is closely maintained as inflation commences. No large scale reheating

is necessary. Solutions for N > 70 are found for any δρ/ρ in the range

10−5 − 10−3 .

In press Physical Review Letters 1995

PACS numbers: [Link], 05.40.+j

Typeset using REVTEX

1
 The standard slow-roll inflation model separates expansion and reheating into two dis-
tinguished time periods. It is first assumed that exponential expansion from inflation places
the universe in a super-cooled phase. Subsequently thereafter the universe is reheated. Two

outcome arise from such a scenario. Firstly the required density perturbations in this cold
universe are left to be created by the quantum fluctuations of the inflaton. Secondly, the
temperature cliff after expansion requires a temporally localized mechanism that rapidly
distributes sufficient vacuum energy for reheating.

It was recognized by Berera and Fang [1] (hereafter referred to as (I) ) that meshing
these two isolated stages may resolve the disparities created by each separately. In (I) it was
shown that slow-roll inflation [2,3] is parametrically consistent with a thermal component.
The thermodynamic analysis there made plausible, but gave no explicit model scenario. In

this paper we present a model example in which thermal equilibrium is maintained during
the inflation stage. The result depends on the presence of a first order phase transition. In
such a case, below Tc there is a temperature barrier which separates the symmetry broken
and unbroken phases. Near Tc , the region from the top of the temperature barrier to the

symmetry broken minima is somewhat flat. Furthermore, beyond the inflection point, where
the curvature is positive, thermal fluctuations of the inflaton are damped. It appears that
these two conditions can support a slow-roll solution for unexceptional values of the coupling
constant, provided that there is also a dissipative component of sufficient size in the inflaton’s
equation of motion.

To demonstrate this, we will examine the finite-temperature one-loop Coleman-Weinberg

effective potential in the form [4,5],

1 φ2 1
V (φ, T ) = BM 4 + Bφ4 [ln( 2 ) − ] + CT 2 φ2 . (1)
2 M 2

At temperatures greater than the critical temperature Tc , the effective potential has a unique
minima at < φ >= 0. For 0 < T < Tc , there are two minima separated by a temperature
barrier. One is a metastable minima at < φ >= 0 and the other lies at < φ >≡ φB ∼ M.
These valleys are separated by a hill with the peak position defined as φT . Hereafter, we

2
will set the GUT scale M = 1 unless otherwise specified.
In (I) we had found the equation of motion of the inflaton with also a dissipative term
to be,

< (3H + Γ(φ))φ̇ + V ′ (φ, T ) >= 0 (2)

where the Hubble constant

s
M 8πV (0)
H = M( ) (3)
mpl 3M 4

with mpl = 1.2 × 1019 GeV and in the model of eq. (1) V (0) = BM 4 /2. Here < φ > means
the thermal average of the inflaton field φ. In the region of interest we will find < φ >≫ δφ.
Thus we can remove the averaging brackets and treat eq. (2) classically. We will also denote

Γφ ≡< Γ(φ) >, and Γ as the average with respect to dN/dφ of Γφ over the slow-roll interval
at fixed T.
The slow-roll conditions

φ̇2 /2 ≪ V (φ, T ) (4)

and

φ̈ ≪ (3H + Γφ )φ̇ (5)

will require respectively that

q
′
|V (φ, T )| ≪ 2V (φ, T )(3H + Γφ ) (6)

and

|V ′′ (φ, T )| ≪ (3H + Γφ )2 . (7)

When there is a sizable thermal component, eq. (6) is replaced by the condition,
1 2
2
φ̇ + ρr
R≡ ≪ 1. (8)
V (φ, T )

3
For later use, near the critical point, for T < Tc in the region about φB , the effective potential
behaves as,

V (φ, T ) ≈ c1 T 2 (φ − φB )2 (9)

where c1 and c2 have logarithmic dependence on T.

Within the slow-roll regime, the number of e-folds from φT < φi to φi < φf < φB is
Z φf 3H + Γφ
N =H dφ . (10)
φi −V ′ (φ, T )

The amplitude of density perturbation [2,3,6] at the onset of slow-roll, when there is also
thermal radiation, is from (I-25),

δρ −δφV ′ (φ, T )
∆≡ = (11)
ρ φ̇2 + 34 ρr
where following (I), for the region V ′′ ≫ H 2 ,
v
3 H2
u
u
δφ = t HT ′′ . (12)
4π V (φ, T )

Using

Γφ 2
ρr ≃ φ̇ (13)
4H

from (I-9), and

π2 ∗ 4
ρr = g T (14)
30

one obtains
s !1
9 10 1 Γφ 1 H 3 H2 2
∆(φ) = √ ( )2 ( )2 (15)
4 π3 g∗ H T V ′′ (φ, T )

and the auxiliary condition,

s s
15 Γφ |V ′ (φ, T )|
= 1. (16)
2π 2 g ∗H T 2 (3H + Γφ )

Here g ∗ is the effective number of degrees of freedom, which implicitly depends on T. When
T > MZ (ie. the mass of the Z particle), g ∗ = 423/4. By setting T ∼ 0.1M in eq. (9) in
order to satisfy eq. (7) it will require

4
 Γφ ≫ H. (17)

In this limit, substituting for Γφ from eq. (16) into eqs. (10) and (15) we get,

15 1 φf
Z
N =− V ′ (φ, T )dφ (18)
2π 2 g ∗ T 4 φi

and
s  23
45 3 1 H −V ′ (φ, T )

∆φ ≡ ∆(φ) = q (19)
4 π5 g∗ T T 2 V ′′ (φ, T )

For later, the notation ∆ with no subscript will refer to ∆φ averaged with respect to dN/dφ
over the slow-roll region. The slow-roll motion here is likened to viscous damping by Γφ
along a mildly cusped potential surface.
For our numerical study, we will use the values from the SU(5) Georgi-Glashow model
[7] in eq. (1) of [5],

5625 4
B= g (20)
1024π 2

and

75 2
C= g . (21)
16

The acceptable SU(5) range for the coupling constant is 0.5 < g < 0.6. In what follows, we
will set g=0.57.

In this model, the following scaling relations are satisfied,

N(xg, xT ) = N(g, T ), (22)

3
∆φ (xg, xT ) = x 2 ∆φ (g, T ). (23)

and

Γφ (xg, xT ) = x2 Γφ (g, T ). (24)

The relations (23) and (24) also hold for ∆ and Γ respectively. With respect to the model,

5
 π2 g∗T 4
R0 ≡ R(φ = 0) = (25)
15 BM 4

where B is given by eq. (20) and where we have explicitly written the M dependence. Since

R0 (xg, xT ) = R0 (g, T ) (26)

and

Tc (xg) = xTc (g), (27)

the critical point is an invariant point Rc in R0 for all g. For the parameters (20) and (21),
we find Rc ≈ 0.65 and Tc (g = 0.57) = 0.15.
By holding g fixed and ranging over T, one obtains a single universal curve from which
N can be computed for any choice of g and T. With this, the model is completely solved for

N. It is not possible to arbitrarily adjust the coupling constant in this model, since varying
g also alters the thermodynamic properties of the system. From eq. (22) it is equivalent to
holding g fixed and varying T. As such, in this thermal scenario, ”fine tuning” has meaning
of thermodynamics. The central problem becomes the common one of determining the

behavior of the system with respect to its intensive variable, here temperature.
In fig. (1) we have computed N for g= 0.57 and M = 1015 GeV in the region V ′′ > 0
and V ′ < 0 where eqs (6) and (7) hold. In the same figure, the dashed curve is αΓ with
α = 2.5 × 10−4 . In fig. (2), ∆ is given over the same range of temperatures.
From fig. (1) one finds that solutions consistent with observation are for T < 0.033. For

example at T=0.032 where R0 = 1.2 × 10−3 , we find N=73, ∆ ≃ 10−4 and Γ ≃ 3.6 × 105 .
There is a monotonic decrease in ∆φ and Γφ along the slow-roll trajectory. Their variation is
due to the respective decrease and increase of |V ′ (φ, T )| and V ′′ (φ, T ) in this region, which
is a common feature of free energy functions that describe first order transitions. However,

a factor 2-3 variation in ∆φ is within observational uncertainties [2,8].

Let us get an idea of how much ∆φ and Γφ change in the above case. By expressing ∆φ
and Γφ as functions of N rather than φ, we find that ∆(70) = 3 × 10−4 , ∆(1) = 0.8 × 10−5

6
whereas Γφ varies from 5 × 105 to 1.6 × 104 between the same limits. However Γφ changes
by less than 20% for the first 30 e-folds of inflation.
One can decrease these variations significantly by going to lower temperature. Here,

the observational constraints can be adequately satisfied within a region near the inflection
point, where V ′ and V ′′ vary the least. For example at T = 0.02 for about a quarter of
the slow-roll interval, N > 70 can be attained with less than a 30% variation in ∆φ . As
one case, from φi = 0.78 to φf = 0.81, 72 e-folds are generated with ∆(φi ) = 4.9 × 10−4 ,

∆(φf ) = 3.8 × 10−4 , Γφ (φi ) = 3.2 × 106 and Γφ (φf ) = 2.9 × 106 . Going even lower in
temperature to T = 0.01, solutions for N > 70 lie within a small interval of ∆φ ≡ φf − φi .
In this case, the variation in ∆φ and Γφ is reduced to less than 1% with ∆ ≃ 10−3 and
Γ ≃ 107 .

Although it is interesting to study the solutions contained in this model from figures
(1) and (2) and the relations (22)-(24), since gφ/T > 1, this model only semi-quantitatively
represents the finite temperature effects of the SU(5) theory [9]. In the relevant region which
is T ∼ Tc /5, the free energy function that must be computed is at the lower edge of the

sensitive critical region. Noting from eq. (18) that N only depends on the end points of
the slow-roll interval, one expects less significant changes of it in an improved treatment
of temperature effects. However the curvature V ′′ (φ, T ) and so the variation in ∆φ and Γφ
could be sizably different in a careful treatment of the exact finite temperature SU(5) theory.
In this paper we have formulated a simple-minded thermal scenario, which already agrees

reasonably well with observational bounds [10,11]. There are two immediate concerns with
our solution. The first is the variation in the ”constant” Γφ which changes by at most
a factor 10 for reasonable scenarios. When interpreted in terms of temperature from eq.
(13), this would imply a factor 1.7 variation in T. One could eliminate this variation by

including the ρ̇r term in eq. (13) and forego the limit of perfect equilibrium. Equilibrium
results within the above range of temperature remain in the set of observationally consistent
solutions for N and ∆. Using this as a guide, one expects this to give some estimate when
nonequilibrium dynamics is treated. One possibility for treating nonequilibrium dynamics

7
for the inflaton would be with a modified version of the KPZ-equation [12,13]. However,
at present we do not have a quantitative description of the dynamics for such a situation.
The similar problem of treating non-equilibrium dynamics also arises for reheating in the

standard inflation scenario. To our knowledge, there is no satisfactory resolution to this

problem there either. In the present case, since equilibrium estimates are already promising,
this problem now has primary importance in formulating a complete thermal scenario.
The second concern is in regards to the magnitude of Γφ . We can get some particle

physics estimates by treating φ(x) as an external source for particle creation. For example,
consider the Yukawa interaction,

LI = gy φ(x)ψ̄(x)ψ(x) (28)

where the fermions are light with mass µ. The probability P to create a fermion pair is [14],

gy2 Z d4 p 2 2 4µ2 3/2 2

P = 2 | φ̃(p)| p (1 − ) θ(p − 4µ2 ) (29)
8π (2π) 4 p 2

where
Z
φ̃(p) = d4 xφ(x)eip·x (30)

with

φ̃(p) ∼ MV T (31)

for the lowest mode p ∼ H. Using V ∼ 1/H 3 , the decay rate is,

P M
Γy = ∼ gy2 M, (32)
T H

which is of the same scale as what we found for Γφ . Although this numerical correspondence

is noteworthy, at present there is no formal connection between estimates such as Γy and

Γφ .
To summarize, this paper has demonstrated a class of equilibrium scenarios that para-
metrically satisfy the observational constraints. To formulate a complete thermal scenario,

8
the small variation in Γφ , which we believe is tied to nonequilibrium dynamics, must be
understood.
It is interesting to note an alternative interpretation of this model scenario† . The inclu-

sion of thermal effects acts similar to a mass term which breaks the scale symmetry of the
zero-temperature theory. In [15] it has been argued that the coupling constant fine-tuning
problem is closely associated to this scale symmetry. On general grounds breaking this scale
symmetry has been suggested there as a way to avoid the small-coupling constant problem.

Accepting their argument, the the thermal scenario presented here and more generally in
(I) can be regarded as an explicit realization of scale symmetry breaking.
Before concluding, three points should be clarified. Firstly, since the temperature in this
scenario is near Tc , the process of rapid cooling followed by rapid heating is replaced by a

smoothened dissipative mechanism. Secondly, since

√
< φ >∼ 30T ≫ HT ≫ δφ, (33)

our semiclassical treatment is justified. Finally in this thermal scenario there are no issues

about how quantum coherence is generated from an initially random field configuration.

I thank Professor L. Z. Fang for many helpful suggestions. Financial support was pro-

vided by the U. S. Department of Energy, Division of High Energy and Nuclear Physics.

†
I thank the referee for this observation.

9
 REFERENCES

[1] A. Berera and L. Z. Fang, Phys. Rev. Lett. 74, 1912 (1995).

[2] For a review of inflation please see [Link] and M.S. Turner, The Early Universe,
(Addison-Wesley, New York, 1990).

[3] J.M. Bardeen, P.J. Steinhardt, & M.S. Turner, Phys. Rev. D28, 679 (1983); A. Linde,
Phys. Lett. 129B, 177 (1983); R. Brandenberger & R. Kahn, Phys. Rev. D28, 2172
(1984); G. Mazenko, W. Unruh & R. Wald, Phys. Rev. D9, 3357 (1985).

[4] H. H. Jacobson and D. J. Amit, Ann. Phys. 133, 57 (1981); S. Coleman and E. Weinberg,
Phys. Rev. D 7, 1888 (1973).

[5] A. D. Linde, Phys. Lett. 108B, 389 (1982); G. P. Cook and K. T. Mahanthappa, Phys.
Rev. D23, 1321 (1980); R. H. Brandenberger, Rev. Mod. Phys. 57, 1 (1985).

[6] [Link], Principles of Physical Cosmology, (Princeton, 1993).

[7] H. Georgi and S. L. Glashow, Phys. Rev. Lett. 32, 438 (1974).

[8] P. J. Steinhardt and M. S. Turner, Phys. Rev. D29, 2162 (1984).

[9] L. Dolan and R. Jackiw, Phys. Rev. D9, 3320 (1974).

[10] G. Smoot et al., Astrophys. J. 396, L1 (1992).

[11] C.L. Bennett, et al., Astrophys. J. 436, 423 (1994).

[12] M. Kardar, G. Parisi, and Y. C. Zhang, Phys. Rev. Lett. 56, 342 (1986).

[13] A. Berera and L.Z. Fang, Phys. Rev. Lett. 72, 458 (1994).

[14] C. Itzykson and J. Zuber, Quantum Field Theory, (McGraw Hill, 1980).

[15] B. Ratra, Phys. Lett. 260B, 21 (1991).

10
 FIGURE CAPTIONS
figure 1: e-folds N of inflation (solid line) and αΓ (dashed line) versus temperature with
α = 2.5 × 10−4 , and for g=0.57 and M = 1015 GeV.

figure 2: ∆ versus temperature with g=0.57 and M = 1015 GeV.

11