Vacuum Decay in Real Time and Imaginary Time Formalisms

Mark P. Hertzberg1,2∗ and Masaki Yamada1†

1
Institute of Cosmology, Dept. of Physics and Astronomy, Tufts University, Medford, MA 02155, USA and
2
Department of Physics, Tokyo Institute of Technology, Ookayama, Meguro-ku, Tokyo 152-8551, Japan
(Dated: August 15, 2019)
We analyze vacuum tunneling in quantum field theory in a general formalism by using the Wigner
representation. In the standard instanton formalism, one usually approximates the initial false
vacuum state by an eigenstate of the field operator, imposes Dirichlet boundary conditions on the
initial field value, and evolves in imaginary time. This approach does not have an obvious physical
interpretation. However, an alternative approach does have a physical interpretation: in quantum
field theory, tunneling can happen via classical dynamics, seeded by initial quantum fluctuations in
both the field and its momentum conjugate, which was recently implemented in Ref. [1]. We show
that the Wigner representation is a useful framework to calculate and understand the relationship
arXiv:1904.08565v2 [hep-th] 13 Aug 2019

between these two approaches. We find there are two, related, saddle point approximations for the
path integral of the tunneling process: one corresponds to the instanton solution in imaginary time
and the other one corresponds to classical dynamics from initial quantum fluctuations in real time.
The classical approximation for the dynamics of the latter process is justified only in a system with
many degrees of freedom, as can appear in field theory due to high occupancy of nucleated bubbles,
while it is not justified in single particle quantum mechanics, as we explain. We mention possible
applications of the real time formalism, including tunneling when the instanton vanishes, or when
the imaginary time contour deformation is not possible, which may occur in cosmological settings.

I. INTRODUCTION Since the above method involves non-intuitive features,

namely a restriction to Dirichlet boundary conditions on
The subject of quantum mechanical tunneling is an the field and dynamics in imaginary time, it begs the
essential topic in modern physics, with a range of appli- question whether there may be other formulations of the
cations, including nuclear fusion [2], diodes [3], atomic tunneling process. Furthermore, if one moves to more
physics [4], quantum field theory [5], cosmological infla- general settings, such as in cosmology, there may not
tion [6], etc. In the context of a possible landscape of always be the usual instanton solution, so one wonders
classically stable vacua in field theory, motivated by con- whether other formulations can be employed instead. In
siderations in string theory [7], it is essential to deter- this paper we will investigate under what circumstances
mine the quantum tunneling rate from one vacuum to an alternative approach to tunneling, from classical evo-
the next. This has ramifications for the stability of our lution of fields whose initial conditions are drawn from
current electroweak vacuum [8], as well as for the viabil- some approximation to the initial wave-function, can pro-
ity of inflationary models [9], and may have ramifications vide an alternative formulation for decay.1
for the cosmological constant problem [10]. This work was motivated by the very interesting work
In ordinary non-relativistic quantum mechanics of a of Ref. [1]. In that work they numerically obtained a
single particle, quantum tunneling can be calculated in tunneling rate from a false vacuum in 1 + 1 dimensional
principle by a direct solution of the time dependent spacetime by solving for the classical dynamics of a scalar
Schrödinger equation. However, our interest here is that field starting from initial conditions generated by a Gaus-
of quantum field theory. In this case, a direct solution sian distribution. The method was to consider many re-
of the Schrödinger equation is notoriously difficult, and alizations of initial conditions and then to calculate the
so approximation schemes are needed. The most famous ensemble-averaged tunneling rate. For their choice of pa-
approximation method, which is analogous to the WKB rameters they found that the tunneling rate was similar
approximation in non-relativistic quantum mechanics, in- to the one calculated by the instanton method.
volves the computation of the Euclidean instanton solu- This leads to several natural questions: (i) what is
tion from one vacuum to another [11, 12]. This leads the relationship between these two approaches? One is
to the well known estimate for the decay rate per unit in an imaginary time formalism, the other is in a real
volume Γ ∝ e−SE , where SE is the bounce action of a time formalism; so how, if at all, are they related? (ii)
solution of the classical equations of motion in imaginary Under what circumstances are the rates comparable to
time. This method is generally thought to be accurate each other? (iii) Under what circumstances are these
when the bounce action SE is large; which evidently cor- approaches valid? It is known that the instanton method
responds to exponentially suppressed decay rates. requires the bounce action to be large to justify a semi-

∗ mark.hertzberg@tufts.edu 1 See Refs. [13, 14] for a different approach using complex classical
† masaki.yamada@tufts.edu trajectories.
 2

classical approximation, but what is the corresponding quantum fluctuation around the false vacuum and ap-
statement for the other real time method? proximate the energy eigenstate by the eigenstate for the
In this paper, we address these questions. We will operator φ̂. In this case, Z can be written as
argue that this real time analysis from classical dynam- Z φ(T )=φi
ics is not identical to, but is very closely related to, the
Z≈ Dφ e−SE [φ] . (4)
instanton tunneling process. We will show that for sim- φ(0)=φi
ple choices of parameters, the two rates are parametri-
cally similar. However, things are more complicated for The path integral can be approximated by the contri-
potentials with unusual features, which we will discuss, bution from a saddle point, which is known as the in-
and there can be advantages to the real time formula- stanton solution. One can also calculate the Gaussian
tion in special circumstances. We will make use of the integral for the perturbation around the instanton solu-
Wigner representation as it will provide a general for- tion. The result is given by the well-known formula:
malism to cleanly identify these two complementary ap-
ΓI ∼ ImK e−SE [φbounce ] , (5)
proaches. We will discuss under what circumstances the
classical dynamics is justified, explain why this would fail where φbounce is the so-called bounce solution in an
in single particle quantum mechanics, and discuss some “upside-down” potential, with V → −V . Therefore, the
cosmological applications. decay rate can be calculated from the path integral with
Our paper is organized as follows: In Section II we re- imaginary time T .
cap the standard instanton contribution to the decay. In Strictly speaking, however, Eq. (5) is not a tunnel-
Section III we present a more general formalism, using ing rate from the false vacuum energy eigenstate because
the Wigner representation, which allows us to describe the boundary condition for the path integral Eq. (4) im-
these two approaches within a single framework. In Sec- plies the transition between eigenstates for the opera-
tion IV we discuss the conditions under which the clas- tor φ̂. The difference between the energy eigenstate and
sical dynamics method is applicable. In Section V we the eigenstate for the operator φ̂ is negligible only if the
estimate and compare the tunneling rates. Finally, in zero-point fluctuation around the local minimum is much
Section VI we discuss our findings. smaller than the typical scale of the potential.
In quantum field theory, the number of effective de-
grees of freedom can be large and hence the quantum fluc-
II. STANDARD EUCLIDEAN FORMALISM tuations can accidentally overcome the potential barrier.
This accidental arrangement and subsequent barrier pen-
Let us begin by recapping the standard approach to etration was seen in the simulations of Ref. [1]. Hence,
vacuum decay, which occurs within the confines of a Eu- to only focus on initial conditions that are eigenstates
clidean, or imaginary time, formalism. In this approach of the field operator, as the usual instanton approach
the decay rate can be calculated from the imaginary part does, is not guaranteed to be the most natural choice of
of the vacuum energy E0 as boundary conditions. Therefore, we would like to utilize
a formalism that can accommodate general initial con-
ΓI = −2 ImE0 , (1) ditions on the fluctuations for a more complete analysis
of tunneling. In the next section, we analyze tunneling
where within the Wigner representation as it will allow us to
systematically study these different possibilities.
ln Z
E0 = −limT →∞ , (2)
T
III. MORE GENERAL FORMALISM
and Z is defined by

Z ≡ hφi | e−HT |φi i . (3) Since vacuum decay is a time-dependent process, it is

natural to calculate it by using a real time formalism (or
Here we denote the (approximate) energy eigenstate Schwinger-Keldysh formalism), which can describe the
around a false vacuum as |φi i.2 One may neglect the time evolution of observables. As we will see, this will
allow us to more systematically identify initial conditions
for the decay, rather than restricting to only those that
are useful in the standard imaginary time analysis.
2 One may think that the imaginary part of the vacuum energy The time evolution of the expectation value of an ob-
is absent because Z defined in Eq. (3) is real. This issue has servable Ô can in principle be calculated from
been investigated in detail in Ref. [15, 16]. They discussed that
the contour of path integral should be deformed along a steepest
D E h R 0
R 0
i
descent contour passing through the false vacuum. Fortunately, Ô(t) = Tr ρ̂ Ttime ei Hdt Ô e−i Hdt , (6)
the resulting tunneling rate can still be calculated by Eq. (5),
which is the standard formula to calculate the tunneling rate by where ρ̂ is an initial density operator and Ttime is
the instanton calculation. Schwinger’s time-ordered operator. The operator Ô may
 3

be taken to be an order parameter of the phase transi- Im t

tion. Instead, one may use an operator that gives zero
Forward
around the false vacuum and nonzero around the true Re t
vacuum. This equation implies that the contour for the 0
time integral is given by the one shown in Fig. 1. We can Backward

define two kinds of fields: forward (φf ) and backward

(φb ) fields, depending on the direction of time evolution.
It is convenient to then define
FIG. 1. Integration contour for the time variable in the real
1 time formalism.
φc = (φf + φb ) , (7)
2
1
πc = (πf + πb ) , (8) If the quantum fluctuations are much smaller than the
2
1 classical quantities, we can approximate HW as
φq = (φf − φb ) , (9)
2 HW (πc + πq , φc + φq , t) − HW (πc − πq , φc − φq , t)
1
πq = (πf − πb ) , (10) ∂HW (πc , φc , t) ∂HW (πc , φc , t)
2 ' 2πq + 2φq . (14)
∂πc ∂φc
where π is the canonical conjugates of the field φ. Here
Then we can perform the integrations over φq and πq ,
φc and πc are effectively classical fields, as we will ex-
which give delta functions of the form
plain shortly, while φq and πq are effectively quantum
fluctuations. The path integral can then be written as
   
dφc ∂HW dπc ∂HW
δ − δ + . (15)
D E Z Z dt ∂πc dt ∂φc
Ô(t1 ) = dφc,0 dπc,0 W0 (φc,0 , πc,0 )
Z This means that indeed {φc , πc } obey the classical equa-
tion of motion. The integrals over the fields are deter-
× Dφc (t)Dπc (t)Dφq (t)Dπq (t) OW (φc (t1 ), πc (t1 ), t1 )
mined by the delta functions and the result is given by
 Z t1 h D E Z
× exp i dt 2φq π̇c − 2πq φ̇c Ô ' dφc,0 dπc,0 W0 (φc,0 , πc,0 ) OW (φc , πc , t)|classical .
0
+HW (πc + πq , φc + φq , t) − HW (πc − πq , φc − φq , t)]] , (16)
(11)
If the initial quantum fluctuations are sufficiently small
where the Wigner function W0 is defined as the Weyl at the false vacuum, we can approximate the potential
transform of the density matrix ρ̂ in the field representa- by a quadratic form; we will return to discuss under
tion. It is given by what conditions this approximation is valid. We de-
note the mass parameter as m at the false vacuum, i.e.,
V 00 (0) = m2 . The ground state wave-function can then
Z
W0 (φc,0 , πc,0 ) = dφq ρ(φc,0 − φq , φc,0 + φq )e2iπc φq . be approximated as
(12) 1 2
Momentarily we focus on only one particular mode for ψ0 ' 1/4
e−ωk |φc,0 | /2 , (17)
(π/ωk )
notational simplicity, but will generalize shortly. The
√
functions HW and OW are the Hamiltonian H and ob- where ωk = m2 + k 2 . The initial Wigner distribution
servable O in the Wigner representation, respectively. In W0 can be then estimated by the one for a free field:
particular, Z
Z W0 = dφq,0 ψ ∗ (φc,0 + φq,0 ) ψ (φc,0 − φq,0 ) e2iπc,0 φq,0
OW (φc , πc ) = dφq dπq O(φc − φq , πc + πq ) e−2iφq πq ,   
1
(13) ∝ exp − ωk |φc,0 (k)|2 + |πc,0 (k)|2 . (18)
ωk
where O(φ, π) is the function obtained from the operator
Ô by direct substitution φ̂ → φ and π̂ → π. For related This can be regarded as a probability distribution for the
work, see Refs. [17, 18]. initial field values, with φ and π treated as independent
random variables. For quantum field theory in d + 1
spacetime dimension, the full result for the initial Wigner
A. Classical Approximation distribution is approximated as
 Z d  
d k 2 1 2
Now we shall rewrite the above path integral by using W0 ∝ exp − ωk |φc,0 (k)| + |πc,0 (k)| ,(19)
(2π)d ωk
some approximations. We will discuss the meaning and
justification of these assumptions in the next section. where we integrate over all k-modes.
 4

We can calculate the tunneling rate by evolving the are the actions for the forward and backward fields, re-
classical dynamics with an initial condition generated spectively. This can be rewritten as
from Eq. (19). Since the initial condition is generated D E Z
randomly by the Wigner distribution, we should perform Ô = Dφ OW (φ(t1 ), t1 ) exp [iS] , (24)
a large number of simulations to obtain a statistically
reasonable result. The tunneling rate is therefore given by defining t0 ∈ (−t1 , t1 ) and φ(t) = φf (t) for t ∈ (0, t1 )
by the statistical average of many realizations as done and φ(t) = φb (−t) for t ∈ (−t1 , 0). This path integral
in Ref. [1]. We note that the system has a translational can be calculated by the standard instanton method by
symmetry and there is no strong correlation between two deforming to imaginary time. Since we assume φc,0 =
distant points. This implies that we can replace the en- πc,0 = 0 in this calculation, this saddle point corresponds
semble average of many simulations by the spatial aver- to the transition from vanishingly initial classical fields.
age of a single simulation with a large simulation box. It also corresponds to vanishing initial quantum field for
The operator Ô should be taken such that it is nonzero φq = 0, though it leaves the initial condition for the
around the true vacuum and is zero around the false vac- quantum field π unspecified. This is identical to the one
uum. This can be realized by, e.g., taking O = θ(φ − φ∗ ) calculated by the instanton method discussed earlier in
with φ∗ being the field value at the other side of the Section II. It is therefore associated with going from a
potential barrier. Then we can count the number of nu- field eigenstate with Dirichlet boundary conditions and
cleated bubbles per unit space as an exponential function again returning, in imaginary time, to a field eigenstate
of time. The tunneling rate is the coefficient of the time with Dirichlet boundary conditions. It is the so-called
variable at the exponent [1]. bounce solution in imaginary time. Importantly, the dif-
ference from the saddle point solution corresponding to
Eq. (16) is the initial condition (or the boundary condi-
B. Relation to the Instanton Calculation tion at t = 0).
Let us comment on how to rotate the time variable
The same result can be obtained by the saddle point in the imaginary space. If we naively take φq = 0 in
approximation for Eq. (11). In the classical limit, the Eq. (11), the exponent vanishes. This is not consistent
path integral can be approximated by saddle points of with Eq. (24), where the action does not vanish and gives
the exponent of the integrand. Varying it with respect the Euclidean action in the imaginary time. This incon-
to φc , φq , πc , and πq , and eliminating πc and πq , we sistency comes from the naive analytic continuation of
obtain the time variable. We can use the epsilon prescription to
specify a possible way to change the integration contour
2φ̈c = −V 0 (φc + φq ) − V 0 (φc − φq ), (20) in the imaginary space. The Hamiltonian should include
0 0
2φ̈q = −V (φc + φq ) + V (φc − φq ). (21) an imaginary mass term that specify the way to change
the integration contour. Therefore the time variable for
One of the solutions to this equation is φq = 0 with the Hamiltonian for φf should be rotated in the opposite
φc being the solution to the usual classical equation of way to the one for φb . This is the reason that we obtain
motion, with initial conditions drawn from the initial a nonzero exponent even if we take φq = 0 in Eq. (11).
Wigner distribution. This saddle point corresponds to Later we will comment on more general situations, which
Eq. (16). We refer to this as the real time formalism may occur in cosmology, where this rotation to imaginary
from classical dynamics, seeded by non-trivial initial con- time may be more problematic.
ditions that ultimately arise from a choice for the initial
wave-function. Indeed we note that this process is ab-
sent if we were simply to assume trivial initial conditions C. Comparison
W0 = δ(φc,0 )δ(πc,0 ), which would be the “purely” classi-
cal behavior. In summary, there are two basic sets of initial con-
Now we show that there is another, related, contri- ditions one may utilize to implement the saddle point
bution to Eq. (11) that is non-zero even if we were to for the path integral Eq. (11). The first one is given by
set φc,0 = πc,0 = 0. Assuming W0 = δ(φc,0 )δ(πc,0 ), we Eq. (16), where the initial condition is given by some
rewrite Eq. (11) as approximation to the initial wave-function and the time
evolution is purely given by the classical equation of mo-
D E Z
tion. The second one is given by the saddle point of
Ô = Dφf Dφb OW ((φf + φb )/2, t1 )
Eq. (24), where the initial condition is φ = 0 and the
× exp [iSf − iSb ] , (22) time evolution is deformed into the complex plane to the
imaginary time axis.
where we assume that OW is independent of πc and At first sight it may seem surprising that the first
Z t1 should be associated with tunneling. But indeed tunnel-
ing can occur because of the non-trivial initial conditions
Sf,b = i dtLf,b , (23)
0
can make for rare events to take place even within the
 5

framework of classical dynamics. This is the tunneling process calculated by Eq. (16) can therefore describe the
process that was calculated in Ref. [1]. In this sense, this tunneling process itself as well as the dynamics of nucle-
contribution is complementary to the instanton contribu- ated bubble after the nucleation.
tion, though in appropriate regimes that we will discuss, Since the nucleated bubble obeys the classical equa-
they can approximate each other quite well. tions of motion, its behavior can be understood easily,
particularly for the thin-wall case. The bubble wall tends
to shrink to a point due to its tension while it tends to
IV. CONDITIONS FOR THE CLASSICAL expand due to the pressure of the vacuum energy. As we
APPROXIMATION evolve the field classically with an initial condition, a lot
of small bubbles are nucleated, but most of them do not
In this section we discuss conditions to calculate a tun- have enough pressure to overcome the tension of the wall.
neling rate by Eq. (16) in the context of quantum field In order for the bubble to expand after the nucleation,
theory. We first note that the distinction between quan- the pressure of the vacuum energy should overcome the
tum and classical mechanics comes from the commuta- tension of the bubble. For a thin-wall bubble, this re-
tion relation for quantum operators. In particular, the quires
commutation relation between creation and annihilation
operators is given by Ad−1 Rd−1 σ . Vd Rd , (26)

âi â†j − â†i âj = δij . (25) where R is the radius of the bubble, σ is the tension of
the wall, and  is the difference of the vacuum energy.
However, the effect of the right-hand
D E side is negligible Here we define area and volume factors in the unit d-
when the occupation numbers â†i âi are large. We then dimensional sphere:
expect that the high occupancy limit corresponds to the 2π d/2
classical limit of quantum systems. This implies that the Ad−1 ≡ , (27)
Γ(d/2)
approximation Eq. (14) is justified when the number of
particles in the system is extremely large. π d/2
Vd ≡ . (28)
In Ref. [19], we have shown that the expectation val- Γ(d/2 + 1)
ues of quantum operators are approximated by a corre-
sponding classical ensemble average over many classical A similar type of inequality is expected to be satisfied for
micro-states, with initial conditions drawn from the ini- a thick-wall bubble.
tial quantum wave-function. Eq. (16) is a mathematical
expression of this statement. It can be understood as
an extension of this discussion to the quantum regime, B. Occupation Number
where the initial state is not a high occupancy state, but
a (quasi) vacuum state with zero point fluctuations. Due Now we examine under what conditions the occupa-
to the possible production of bubbles, which arises due to tion number of the quanta describing nucleated bubbles
rare accidental arrangements from the non-trivial initial is much larger than unity. In this case the nucleated bub-
conditions, the occupation number can be large enough ble is essentially coherent and can be treated within the
to use the classical description. framework of classical field theory.
In this case, the approximation φq  φc is satisfied, We estimate the occupation number of nucleated bub-
except for the initial condition, and we can evolve φc by ble in two simple cases. First we consider the case where
the classical equation of motion. In the regime before the the scalar potential is described by typical values of cur-
tunneling, φq  φc may not be satisfied. However, we vature scale around vacua m, field value v, height of the
can still use Eq. (16) if the amplitude of fluctuations is potential barrier Vh , and the difference of the vacuum en-
small enough to neglect terms in the potential that are ergy  (see Fig. 2). We note that Vh must be smaller than
higher-order than quadratic. This is because the Wigner of order v 2(d+1)/(d−1) for d > 1 because of the unitarity
approximation is exact for the free-field theory. We will bound (e.g., in 3 + 1 dimensions this is related to the
discuss situations in which the neglecting of these higher familiar idea that the quartic coupling λ φ4 obeys λ . 1
order terms may not be valid. to be in a weakly coupled regime).

A. Tension and Pressure 1. Thin-Wall

Eq. (16) can describe the classical dynamics of the We assume   Vh and use the thin-wall approxima-
field after the bubble nucleation. This is different from tion for now. In this case, the wall tension is given by
the instanton method, where we need to connect the
Lorentzian and Euclidean regimes to describes the dy-
Z √ p
namics of the bubble after nucleation. The tunneling σ = dφ 2 V ∼ v Vh . (29)
 6

degenerate and the difference of the VEV is as small as

V( ) m. This implies that the bubble is a thick-wall type and
so we should re-examine the above analysis. Here the
scalar self coupling could be as large as O(1). We check
that the occupation number is larger than unity in this
extreme case, too. We consider the following potential:
"    3 #
2
1 φ λ3 φ
V (φ) = U0 + , (36)
2 Λ 3! Λ
✏ Vh
V 00 (0) = m2 where U0 and Λ (U0 . Λ2(d+1)/(d−1) ) are dimension-full
parameters and λ3 (. O(1)) is a dimensionless constant.
0 v Since φ = 0 is a false vacuum, it can tunnel to the other
side of the potential hill. The tunneling action and the
occupation number are given by
FIG. 2. Schematic picture of a typical potential and parame-  2(d+1)/(d−1) (d−1)/2
ters describing its shape. Λ
SE ' cS λ−2
3 A d , (37)
U0
 2(d+1)/(d−1) (d−1)/2
Using Eq. (26), we obtain a typical radius of the nucle- Λ
ated bubble as N ' cN λ−2
3 A d−1 , (38)
U0
σ
Rb ' d . (30) where the numerical constants are given by cS ' 4.1 × 10

and cN ' 1.0 × 102 for d = 3. Even in this extreme
Let us first report on the Euclidean instanton action
case, the tunneling action and the occupation number
associated with the bubble, as it is the standard quantity
are larger than O(10/λ23 ). This justifies that the nu-
to compute tunneling in the literature, as
cleated bubble can be described classically (or a scalar
2Rd σ 2dd σ d+1 condensate).
SE ' ' (31)
d+1 d + 1 d
 d  2(d+1)/(d−1) (d−1)/2
Vh v C. Single Particle Quantum Mechanics
∼ . (32)
 Vh
However, in order to justify the alternative real time One may wonder if these arguments can extend to
tunneling from classical dynamics, we need to compute the problem of single particle tunneling in ordinary non-
the bubble’s occupancy number. We define it by the relativistic quantum mechanics. In this case there is ob-
gradient energy of the bubble in the unit of m: viously no such thing as a “bubble” that can be formed.
1
Z 
1 2
 So there is no obvious sense in which there is any object
d
N ≡ d x (∇φ) (33) at high occupancy.
m 2 bubble
Nevertheless, we can formally view this problem as
Ad−1 Rd−1 σ quantum field theory in 0 + 1 dimensions. So, for the
' . (34)
m sake of completeness, let us formally take the result in
(if we were to reinstate factors of ~, the actual occupancy Eq. (35) and take d → 0. Then we formally obtain
number would be this divided by ~). It is roughly given 
by N ∼ . (39)
m
 d−1  2(d+1)/(d−1) d/2−1  2/(d−1)
Vh v v Now we should note that in this case m, which is the
N ∼ ,
 Vh m(d−1)/2 (square root) of curvature of the potential in quantum
(35) field theory at the false vacuum, is just the characteris-
Note that every factor on the most right-hand side tic frequency of oscillation of the particle ω0 around the
is larger than unity for weakly coupled field theories, meta-stable minimum in quantum mechanics.
so the occupation number can in fact be quite large However, what is important is that in quantum me-
N  1. (Note that for d = 1, it simplifies to N ∼ chanics, energy conservation tells us that the tunneling
v 2 (Vh /(m2 v 2 ))1/2 .) process requires that the particle tunnel to a point at
the same potential energy as its starting values. Hence,
 here should be the potential height difference, so it is
2. Thick-Wall in fact just  = 0. This implies N = 0. So there is no
sense in which one is at high occupancy. This means that
The above suggests that the occupation number can be this procedure of sampling from some Gaussian approxi-
as small as of order unity when the vacuum energy is not mation to the wave-function and using this to determine
 7
q
tunneling will typically fail in ordinary quantum mechan- where ωb = m2 + kb2 ∼ m2 + Rb−2 is some charac-
p
ics. Conversely, it can be applicable in field theory in teristic frequency, associated with the bubble associated
higher dimensions with bubbles of high occupancy, as we with a characteristic wavenumber kb ∼ 1/Rb .
discussed above.
However, this is not a sufficient condition for nucle-
Furthermore, in single particle mechanics, with a con-
ation, because if the bubble appears on the other side of
fining potential, the system does not exhibit ergodicity.
the barrier with arbitrarily low energy then it can col-
On the other hand, some ergodicity is exhibited in the
lapse. Suppose that a small bubble with radius Rini and
field theory, providing the adequate time evolution for d
kinetic energy Rini πc2 forms due to fluctuations. The ki-
the development of bubbles over time as the system ex-
netic energy must be larger than the energy of the bubble
plores phase space.
with radius Rb so that the bubble can expand after the
(th)
nucleation. This means we need πc > πc where
V. TUNNELING RATE  2
d
Rini πc(th) ∼ Rbd . (42)
Now we shall compare the tunneling rate (per unit vol-
ume) via the standard Euclidean instanton to the tun- Note that this πc
(th)
is the conjugate momentum in po-
neling rate via classical dynamics with initial conditions sition space. It can be written in terms of π(k) in the
drawn some approximation to the wave-function. The (th)
momentum space as πc = 1/(2π)d dd ke−ikx πc (k) ∼
R
latter one can be estimated in the following way.
(th) −d
Ideally the relevant initial fluctuations that ultimately πc (k ∼ 1/Rini )Rini . The probability Pb to have suffi-
lead to the formation of the bubble are sufficiently small cient energy can then again be estimated from the Wigner
that the potential can be approximated to be a quadratic distribution as
form around a false vacuum; we will revisit this shortly.
dd k 1 (th)  d
Z
Then the initial distribution of fluctuations is given by − ln Pb ∼ |πc (k)|2 ∼ R . (43)
the Wigner distribution of a free massive scalar field with d
(2π) ωk ωb b
mass m around the false vacuum Eq. (19). This distribu-
tion does not change much even if we allow the field to
classically evolve in time. The tunneling rate can there- A. Tunneling Rate
fore be estimated by the probability that φc (k) and πc (k)
are large enough to nucleate a classical bubble. A classi- According to the Wigner representation, the variables
cal bubble that expands after the nucleation must satisfy φc and πc are taken as independent random variables.
the condition Eq. (26) and hence its radius must be larger This says that the probability that both (a) and (b) oc-
than Rb ∼ σ/. cur is the product Pa Pb . This allows us to estimate the
Now in order to completely determine the probability tunneling rate (per unit volume) within this real time
for tunneling, one should perform a simulation of this formalism as
non-linear system of classical equations of motion, with
the appropriate initial conditions specified above. How- ΓR ∼ cR e−γR , (44)
ever, we can give an estimate of the probability of bubble
formation by utilizing the initial wave-function’s statisti- with
cal distribution as a guide; we will return to this shortly.
 d
In order to form a bubble there are two conditions that γR = a ωb v 2 Rbd + b R (45)
need to be satisfied: (a) the field need’s to be on the ωb b
far side of the barrier and (b) the bubble needs to have
where a and b are O(1) prefactors.
sufficient energy to avoid collapse. Let us estimate these
probabilities in turn We can compare this to the usual result for tunnel-
First, in order for the bubble to be on the other side ing using the Euclidean imaginary time formalism given
of the barrier, we need that the field value in position earlier in Eq. (5) as
space obeys φc & v. In the k-space representation this
(th) ΓI ∼ cI e−γI , (46)
condition means that we need φc (k) > φc (k) where
Z  with
(th) d ikx
∼ Rbd v,

φc (k) = d x e φ(x) (40)
bubble γI = SE ∼  Rbd+1 . (47)
for a bubble of radius Rn to be nucleated. The probabil-
ity Pa that φc (k) exceeds this threshold can be estimated The instanton rate is calculated for the thin-wall case,
from the Wigner distribution as but the result is not qualitatively different for the thick-
Z d wall case once we identify  as the difference of the vac-
d k uum energy between the false vacuum and the tunneling
− ln Pa ∼ ωk |φ(th)
c (k)|2 ∼ ωb v 2 Rbd , (41)
(2π)d point. Since this involves an extra factor of Rb compared
 8

to the scaling in γR , we need an estimate for the bubble (using λ ∼ 0.01 in this regime). This radius is much
radius, which is roughly much smaller than the Compton wavelength of the Higgs
√ which is m−1 ∼ 10−2 GeV−1 . Hence now we are in a
v Vh regime in which ωb ∼ 1/Rb . In this regime, both Pa and
Rb ∼ , (48)
 Pb are comparable, and they both give
This allows us to make the estimate
1
p
γI ∼ v Vh Rbd , (49) γR ∼ , (53)
λ
for the instanton tunneling exponent. In this final ex- (we naturally focus here on the physical case of 3+1
pression we have still kept a factor of Rbd for convenience, dimensions). This is comparable to the instanton rate
since this is a common factor that appears in Eq. (45) γI ∼ 1/λ, so we again have γI ∼ γR .
also.
We note that in this case with Rb  m, giving ωb  m,
and probing deep into the quartic term in the potential,
B. Examples it was not guaranteed that the Gaussian approximation
based on the free theory would suffice. However, para-
metrically it is of the right order.
We now use the above results to compare the tunnel-
ing rates that we have estimated in these different for-
malisms.
3. Flat Hill-Top
1. Weakly Broken Z2 Symmetry
Suppose the hill-top is very flat, moreso than it appears
Let us consider a potential of the form in Fig. 2. To be clear, let us imagine that it is so flat that
Vh  m2 v 2 , which would be the naive value based on di-
V (φ) = λ(φ2 − v 2 )2 + δV (φ), (50) mensional analysis. Such a potential is perhaps unusual
from the microscopic point of view, but it is allowed in
where δV (φ) is a term that weakly breaks the Z2 sym- principle. In this case the instanton gives an exponent
metry. This potential is similar to the kind of poten- (normalized to bubble volume) that is linear in the bar-
tial shown in Fig. 2 with Vh ∼ m4 /λ. In this case the rier width γI /Rbd ∝ v. On the other hand, if we turn
bubble thickness is approximately set by the Compton to the real time formalism we obtain different estimates.
wavelength as λC ∼ 1/m. However the bubble radius From Eq. (43) the contribution from the kinetic energy
is at least this large, i.e., m Rb & 1. This ensures the effect gives γR /Rbd ∝ v 0 , which is too small. On the other
frequency ωb can be approximated by the mass ωb ' m. hand, from Eq. (41) the contribution from the need to be
By noting that  is bounded to be of the order of or on the other side of the barrier gives γR /Rbd ∝ v 2 , which
much smaller than V ∼ v 2 m2 , we can conclude that the is too large.
probability Pa . Pb . Hence the rate γR is approximated
In this case, the Gaussian approximation for the initial
as
wave-function is not accurate, since it assume that the
γR ∼ m v 2 Rbd . (51) fields mass is m, but for such a potential, the effective
mass in the barrier is smaller. Instead we need to al-
Then from Eq. (49), with Vh /m ∼ m v 2 , we have γI ∼ γR . ter our simple estimates. We need to essentially replace
the frequency of the bubble by some appropriate effec-
tive mass, from√the effective curvature of the potential,
2. SM Higgs namely meff ∼ Vh /v. This leads to
p
As another example, let us consider the Higgs poten- − ln Pa → meff v 2 Rbd ∼ v Vh Rbd , (54)
tial in the minimal SM. Upon RG running of the Higgs
self-coupling λ, the top mass, and other couplings, one
which is indeed of the order of the instanton rate.
finds that the Higgs potential turns over and then goes
negative. This happens at around v ∼ 1011 GeV, or so. On the other hand if we persist with the original
In this example, the potential is dominated by the quar- Wigner distribution, we believe that it is plausible that
tic term near the tunneling point. In this case, we can a simulation can arrive at roughly the correct tunnel-
use the above formula by taking  → Vh ∼ λ v 4 . The ing rate anyhow. This is because even though the ini-
bubble radius is now of order or larger than tial distribution is not an accurate representation of the
false vacuum eigenstate, these initial conditions may be
1 partially washed away in the simulation, leading to the
Rb ∼ √ ∼ 10−10 GeV−1 , (52)
λv appropriate rate.
 9

VI. DISCUSSION However, our simple estimates for the tunneling rates
did involve a dependence on the mass of the field de-
We have used a general Wigner representation to es- fined around the initial false vacuum, as it affects the
tablish two formulations of tunneling with slightly differ- initial Gaussian approximation to the wave-function. So
ent boundary conditions and dramatically different dy- these simple estimates involve some sensitivity to initial
namics: in addition to the usual formulation of the imag- conditions, especially in the case of potentials with ex-
inary time saddle point contribution to the decay am- treme features. But in the case in which the bubble has
plitude with Dirichlet boundary conditions on the field, characteristic wavenumbers k-values (k . m), we are not
there is another real time formulation based on classical sensitive to the UV behavior of the initial conditions.
dynamics with initial conditions set by some estimate for Furthermore, more general estimates could be made in
the initial wave-function. While the former one is the more extreme situations also.
familiar one from the instanton action, the latter one
is an ensemble average of classical field theory dynam-
ics seeded by quantum zero point fluctuations. We note B. Applications
that this ensemble average can be practically realized by
a spatial average of a single simulation by appealing to As an application of these results, suppose there is an
a form of ergodic theorem. Since we use the Wigner ap- AdS vacuum between two dS vacua. The tunneling rate
proximation for the initial wavefunction in the real-time from a dS vacuum to the other dS vacuum cannot be cal-
approach, we do not expect that the resulting rate is culated by using the standard instanton method because
exactly the same as the one calculated in the imaginary- there is no instanton solution. However, the transition
time approach. However, we have checked that the ex- rate must be nonzero because anything can happen in
ponent in the tunneling rate is parametrically the same quantum theory according to the path integral expres-
in both approaches in several examples. sion [20]. In fact, the “classical tunneling” discussed in
this paper is expected to give a nonzero transition rate.
This is the only practical way we are aware of to calcu-
A. Classicality late the transition rate in such a case. This transition
process is complementary to the standard instanton tun-
In order to justify the classicality of the field in this lat- neling process. In this sense, the result gives a lower
ter approach, the quantum fluctuations have to organize bound on the tunneling rate.
into a bubble and the occupation number has to be much As another application, consider a dynamical setting,
larger than unity. This can be realized only if the degrees such as during preheating after inflation. In this case a
of freedom in the system are large enough, as is possible field may exhibit a strongly time dependent effective po-
in quantum field theory, as it is for the nucleated bub- tential from its interactions with the inflaton or the met-
bles. We note however that much of the universe would ric etc. If such a field is also trapped in a type of false
remain at low occupancy, so it is not entirely guaranteed vacuum then it may be highly non-trivial to implement
that the classical dynamics is extremely accurate, but the standard instanton tunneling procedure, as this re-
perhaps only roughly accurate. Furthermore, this ap- quires deforming the contour to the imaginary time axis.
proach is ordinarily not valid in single particle quantum If there is (quasi) periodic behavior in the time domain
mechanics as the notion of high occupancy there does not it will re-organize into growing exponential behavior in
seem to be valid. imaginary time, which may be an obstruction to an effi-
One may wonder if the tunneling rate depends sen- cient implementation of the Euclidean instanton analysis.
sitively on the initial fluctuations. This is actually the Furthermore, if there is some form of non-analytic struc-
case when the number of degrees of freedom in the sys- ture to the time dependence, such as from a step-like time
tem is not much larger than of order unity. However, we behavior, then this may be an obstruction to deforming
are interested in the tunneling process in quantum field the contour. In these cases it may be more intuitive and
theory, where the number of relevant degrees of freedom more practical to perform a real time analysis.
can be quite large. In this case, all relevant modes may
interact with each other somewhat chaotically and the
distribution will be randomized after an ergodic time.
So one expects dynamical evolution to wash away some
features of the initial condition. (We may have to wait ACKNOWLEDGMENTS
for a time scale longer than the time scale of oscillation
around a false vacuum so that some features of the initial We would like to thank Jonathan Braden, Alexander
condition are washed away. This condition is similar to Vilenkin, and Mohammad Hossein Namjoo for useful dis-
T /Tslosh → ∞ for the “direct method” that was intro- cussions. MPH is supported by National Science Foun-
duced in Refs. [15, 16].) dation grant PHY-1720332 and a JSPS fellowship.
 10

[1] J. Braden, M. C. Johnson, H. V. Peiris, A. Pontzen, and [12] C. G. Callan, Jr. and S. R. Coleman, Phys. Rev. D16,
S. Weinfurtner, (2018), arXiv:1806.06069 [hep-th]. 1762 (1977).
[2] A. B. Balantekin and N. Takigawa, Rev. Mod. Phys. 70, [13] N. Turok, New J. Phys. 16, 063006 (2014),
77 (1998), arXiv:nucl-th/9708036 [nucl-th]. arXiv:1312.1772 [quant-ph].
[3] L. Esaki, Physical Review 109, 603 (1958). [14] S. F. Bramberger, G. Lavrelashvili, and J.-L. Lehners,
[4] G. Binning, H. Rohrer, C. Gerber, and E. Weibel, Phys- Phys. Rev. D94, 064032 (2016), arXiv:1605.02751 [hep-
ical Review Letters 49, 57 (1982). th].
[5] P. H. Frampton, Phys. Rev. D 15, 2922 (1977). [15] A. Andreassen, D. Farhi, W. Frost, and M. D. Schwartz,
[6] A. H. Guth, Phys. Rev. D23, 347 (1981), [Adv. Ser. As- Phys. Rev. Lett. 117, 231601 (2016), arXiv:1602.01102
trophys. Cosmol.3,139(1987)]. [hep-th].
[7] M. R. Douglas, JHEP 05, 046 (2003), arXiv:hep- [16] A. Andreassen, D. Farhi, W. Frost, and M. D. Schwartz,
th/0303194 [hep-th]. Phys. Rev. D95, 085011 (2017), arXiv:1604.06090 [hep-
[8] M. Sher, Phys. Lett. B317, 159 (1993), [Addendum: th].
Phys. Lett.B331,448(1994)], arXiv:hep-ph/9307342 [hep- [17] A. Kamenev and A. Levchenko, Adv. Phys. 58, 197
ph]. (2009), arXiv:0901.3586 [cond-mat.other].
[9] A. H. Guth and Y. Nomura, Phys. Rev. D86, 023534 [18] A. Polkovnikov, Annals Phys. 325, 1790 (2010),
(2012), arXiv:1203.6876 [hep-th]. arXiv:0905.3384 [cond-mat.stat-mech].
[10] L. F. Abbott, Phys. Lett. 150B, 427 (1985). [19] M. P. Hertzberg, JCAP 1611, 037 (2016),
[11] S. R. Coleman, Phys. Rev. D15, 2929 (1977), [Erratum: arXiv:1609.01342 [hep-ph].
Phys. Rev.D16,1248(1977)]. [20] A. R. Brown and A. Dahlen, Phys. Rev. Lett. 107,
171301 (2011), arXiv:1108.0119 [hep-th].