OFFPRINT

Stochastic quantization and Casimir forces

P. Rodriguez-Lopez, R. Brito and R. Soto
EPL, 96 (2011) 50008
Please visit the new website
www.epljournal.org
A LETTERS JOURNAL EXPLORI NG
THE FRONTI ERS OF PHYSI CS
AN INVITATION TO
SUBMIT YOUR WORK
www.epljournal.org
The Editorial Board invites you to submit your letters to EPL
EPL is a leading international journal publishing original, high-quality Letters in all
areas of physics, ranging from condensed matter topics and interdisciplinary research
to astrophysics, geophysics, plasma and fusion sciences, including those with
application potential.
The high profile of the journal combined with the excellent scientific quality of the
articles continue to ensure EPL is an essential resource for its worldwide audience.
EPL offers authors global visibility and a great opportunity to share their work with
others across the whole of the physics community.
Run by active scientists, for scientists
EPL is reviewed by scientists for scientists, to serve and support the international
scientific community. The Editorial Board is a team of active research scientists with
an expert understanding of the needs of both authors and researchers.
I
M
P
A
C
T

F
A
C
T
O
R

2
.
7
5
3
*
*
A
s

r
a
n
k
e
d

b
y

I
S
I

2
0
1
0
www.epljournal.org
www.epljournal.org
A LETTERS JOURNAL EXPLORI NG
THE FRONTI ERS OF PHYSI CS
Quality The 40+ Co-Editors, who are experts in their fields, oversee the
entire peer-review process, from selection of the referees to making all final
acceptance decisions
Impact Factor The 2010 Impact Factor is 2.753; your work will be in the
right place to be cited by your peers
Speed of processing We aim to provide you with a quick and efficient
service; the median time from acceptance to online publication is 30 days
High visibility All articles are free to read for 30 days from online
publication date
International reach Over 2,000 institutions have access to EPL,
enabling your work to be read by your peers in 100 countries
Open Access Articles are offered open access for a one-off author
payment
Details on preparing, submitting and tracking the progress of your manuscript
from submission to acceptance are available on the EPL submission website
www.epletters.net.
If you would like further information about our author service or EPL in general,
please visit www.epljournal.org or e-mail us at [email protected].
Six good reasons to publish with EPL
We want to work with you to help gain recognition for your high-quality work through
worldwide visibility and high citations.
2.753
*
* As listed in the ISI 2010 Science
Citation Index Journal Citation Reports
IMPACT FACTOR
500 000
full text downloads in 2010
OVER
30 DAYS
16 961
average receipt to online
publication in 2010
citations in 2010
37% increase from 2007
1
2
3
4
5
6
www.epljournal.org
EPL is published in partnership with:
IOP Publishing EDP Sciences European Physical Society Societ Italiana di Fisica
Weve had a very positive
experience with EPL, and
not only on this occasion.
The fact that one can
identify an appropriate
editor, and the editor
is an active scientist in
the field, makes a huge
difference.
Dr. Ivar Martinv
Los Alamos National Laboratory,
USA
EPL Compilation Index
Visit the EPL website to read the latest articles published in
cutting-edge fields of research from across the whole of physics.
Each compilation is led by its own Co-Editor, who is a leading
scientist in that field, and who is responsible for overseeing
the review process, selecting referees and making publication
decisions for every manuscript.

Graphene

Liquid Crystals

High Transition Temperature Superconductors

Quantum Information Processing & Communication

Biological & Soft Matter Physics

Atomic, Molecular & Optical Physics

BoseEinstein Condensates & Ultracold Gases

Metamaterials, Nanostructures & Magnetic Materials

Mathematical Methods

Physics of Gases, Plasmas & Electric Fields

High Energy Nuclear Physics

If you are working on research in any of these areas, the Co-Editors would be
delighted to receive your submission. Articles should be submitted via the
automated manuscript system at www.epletters.net
If you would like further information about our author service or EPL
in general, please visit www.epljournal.org or e-mail us at
[email protected]
Biaxial strain on lens-shaped quantum rings of different inner
radii, adapted from Zhang et al 2008 EPL 83 67004.
Artistic impression of electrostatic particleparticle
interactions in dielectrophoresis, adapted from N Aubry
and P Singh 2006 EPL 74 623.
Artistic impression of velocity and normal stress profiles
around a sphere that moves through a polymer solution,
adapted from R Tuinier, J K G Dhont and T-H Fan 2006 EPL
75 929.
www.epl journal.org
A LETTERS JOURNAL
EXPLORI NG THE FRONTI ERS
OF PHYSI CS
Image: Ornamental multiplication of space-time figures of temperature transformation rules
(adapted from T. S. Br and P. Vn 2010 EPL 89 30001; artistic impression by Frdrique Swist).
December 2011
EPL, 96 (2011) 50008 www.epljournal.org
doi: 10.1209/0295-5075/96/50008
Stochastic quantization and Casimir forces
P. Rodriguez-Lopez
1
, R. Brito
1
and R. Soto
2
1
Departamento de Fsica Aplicada I and GISC, Universidad Complutense - 28040 Madrid, Spain, EU
2
Departamento de Fsica, FCFM, Universidad de Chile - Casilla 487-3, Santiago, Chile
received 28 July 2011; accepted in nal form 18 October 2011
published online 25 November 2011
PACS 05.40.-a Fluctuation phenomena, random processes, noise, and Brownian motion
PACS 03.70.+k Theory of quantized elds
PACS 42.50.Lc Quantum uctuations, quantum noise, and quantum jumps
Abstract In this paper we show how the stochastic quantization method developed by Parisi
and Wu can be used to obtain Casimir forces. Both quantum and thermal uctuations are taken
into account by a Langevin equation for the eld. The method allows the Casimir force to be
obtained directly, derived from the stress tensor instead of the free energy. It only requires the
spectral decomposition of the Laplacian operator in the given geometry. The formalism provides
also an expression for the uctuations of the force. As an application we compute the Casimir
force on the plates of a nite piston of arbitrary cross-section. Fluctuations of the force are also
directly obtained, and it is shown that, in the piston case, the variance of the force is twice the
force squared.
Copyright c EPLA, 2011
Fluctuation-induced (Casimir) forces [1,2] are currently
attracting renewed attention, probably because of the
accessability of small systems at the nano- and microscale,
as it is in these small systems where Casimir forces are
revealed, despite some macroscopic eects [3]. However,
techniques to calculate forces for complicated geometries,
beyond the usual ones with high degrees of symmetry, are
scarce. Recently, a powerful technique to calculate elec-
tromagnetic (EM) Casimir forces [4] has been proposed.
It is based on a multiscattering technique and has been
successfully applied to many congurations, such as plates,
cylinders, spheres, wedges, etc. [5].
Casimir forces have their origin in uctuations of EM
elds. Already Lifshitz in 1956 [6] use the concept of uc-
tuations (in this case of the electric currents inside a metal)
to compute such forces. With the idea of uctuations in
mind, some authors [7,8] used a method that takes as its
starting point the Langevin equation that describes the
evolution of systems subjected to uctuations (in their
case of thermal origin). Such method was applied to equi-
librium situations, but also to non-equilibrium ones [911],
in which the uctuation-disspiation theorem is not satis-
ed. However, application of the Langevin approach to
EM Casimir forces of quantum origin has not yet been
developed.
The goal of this paper is to apply the Langevin equa-
tion method to calculate forces of quantum origin. The
method is based on the stochastic quantization approach
developed by Parisi and Wu [12]. These authors construct
a Langevin equation for a given eld subjected to thermal
uctuations, in a non-physical, pseudo-time. Despite the
classical nature of uctuations we will show how quantum
uctuations arise naturally within that method. We will
follow the formalism developed in ref. [11], that takes as
a starting point the Langevin equation derived by Parisi
and Wu, and construct its solution by diagonalization of
the deterministic evolution operator with the appropriate
bounday conditions the immersed bodies and later
averaging over the thermal noise.
The resulting method has the advantage of computing
the Casimir force directly and can be directly applied
to compute torques and stresses over extended bodies.
Also, it is suitable for numerical or perturbation methods,
allowing to compute forces on realistic geometries. Finally,
the eects of nite temperature are included directly.
The Casimir force is calculated via the stress tensor, T
(see ref. [11]), which must be averaged over the probability
distribution of the elds . Such a probability distribution
is given by
P[] =Z
1
e
S[]/
, (1)
where S[] is the action of the scalar eld with zero mass,
Wick-rotated in the time variable [2] (t =i), i.e.,
S[] =
1
2
_

0
d
_
dr
_
1
c
2

2
+
2
_
, (2)
50008-p1
P. Rodriguez-Lopez et al.
where =1/k
B
T and Z is the partition function
Z =
_
De
S[]/
. For the bosonic case, the eld
must obey periodic boundary conditions in time, that
is, ( +, r) =(, r). The stress tensor is normally a
bilinear form in the eld T=TT [, , r], the expression
that denes the stress tensor operator TT .
Herein we use the formalism of Parisi and Wu [12,13] to
evaluate the average of the stress tensor in an alternative
way. The idea of Parisi and Wu consists in a formulation
of quantum mechanics or quantum eld theory in terms of
a stochastic process. More precisely, a Langevin equation
for the eld is written as an evolution equation in an
auxiliary time, which we will call pseudo-time, s. In this
description, the eld depends on the new pseudo-time
variable: (, r) (, r; s). The Langevin equation takes
the form

s
(, r; s) =
S[]

+(, r; s)
=
_
1
c
2

2
+
2
_
+(, r; s). (3)
The term (, r; s) is the source of uctuations, given by a
Gaussian white noise satisfying the uctuation-dissipation
relation [14]
(, r; s)= 0,
(4)
(, r; s) (

, r

; s

)= 2k
B
T(

)(r r

)(s s

).
Higher correlation of odd number of -functions vanish,
while correlations of an even number of -functions factor-
ize in all possible products of two -functions [15]. For
instance, the 4-correlation:
(1)(2)(3)(4) =(2k
B
T)
2
_
(1 2)(3 4)
+(1 3)(2 4) +(1 4)(2 3)

, (5)
where the symbol 1, 2, 3 or 4 stands for the complete set
, r, s, and the means the product of 3 -functions, as in
eq. (4).
The solution of the Langevin equation in the stationary
limit s reproduces the probability distribution given
by eq. (1) [13]. The pseudo-time s has not physical
meaning and the role of the evolution in s in solely to
sample over the probability distribution (1). Statistical
correlations of two or more elds, must be evaluated at
the same value of s, but otherwise the positions and times
can be arbitrary.
This approach, although similar, is qualitatively dier-
ent to the Lifshitz method. In Lifshitz method, uctu-
ating electric currents are introduced in the conductors
which produce the uctuating electromagnetic eld in
the vacuum. The uctuations of the current are thermal,
taking place in real time, and the quantum character of the
model is introduced by the use of the quantum uctuation-
dissipation relation [6]. In the Parisi-Wu method, the uc-
tuation process takes place in a pseudo-time and its role
is to reproduce the full quantum-thermal statistical prop-
erties of the eld. In particular, the complete quantum
correlations, including quantum coherence, are obtained.
The second dierence is that the Parisi-Wu approach does
not consider the uctuations in the macroscopic metal but
on the (electromagnetic) eld itself.
Having an expression for the stress tensor and a
Langevin equation for the eld, we can follow the proce-
dure recently developed in [11] to obtain the Casimir force.
The eld is written as (and a similar decomposition for
the noise with coecients
nm
):
(, r; s) =

n,m

nm
(s)g
m
()f
n
(r), (6)
where f
n
and g
m
are the eigenfunctions:

2
f
n
(r) =
2
n
f
n
(r),
1
c
2

2
g
m
() =
2
m
g
m
(), (7)
which are orthogonal under the L
2
scalar product in space
or time. The above expression indicates that f
n
(r) and
2
n
encode the spatial conguration of the system, that is,
the position of the bodies and their boundary conditions.
In a similar fashion, g
m
() and
2
m
contain the (Wick-
rotated) time dependence. As we are considering a bosonic
eld that obeys periodic boundary conditions in , the
eigenvalues are the known Matsubara frequencies
m
=
2m/c, mZ, and the eigenfunctions are g
m
() =
exp(i
m
). Then, from eq. (3), the coecients
nm
satisfy the dierential equation
d
nm
(s)
ds
=
_

2
n
+
2
m

nm
(s) +
nm
(s), (8)
which can be integrated to give

nm
(s) =
_
s

de
(
2
n
+
2
m
)(s)

nm
(), (9)
where the noise coecients satisfy, from eq. (4)

nm
(s)

m
(s

) =2k
B
T(s s

)
nn

mm
. (10)
The formalism developed in [11] allows the average stress
tensor to be calculated by substituting expression (6)
into the stress tensor and taking the average over the
uctuations, (, r; s), in the limit s :
T(r) = lim
s
TT [(s), (s), r] (11)
= lim
s

n
1
,m
1
n
2
,m
2

n
1
m
1
(s)

n
2
,m
2
(s)TT [f
n
1
, f

n
2
, r].
The two functions
nm
are replaced by their values in
terms of the uctuations, given by eq. (9), resulting into a
double integral over two pseudo-times. Then, the average
50008-p2
Stochastic quantization and Casimir forces
is carried out using eq. (10), that eliminates one
integral, with the nal result
T(r) =
1

nm
TT
nn
(r)

2
n
+
2
m
, (12)
where TT
nm
(r) =TT [f
n
, f

m
, r]. Moreover, the sum over the
Matsubara frequencies,
m
, can be carried out to obtain
the result
T(r) =
c
2

n
TT
nn
(r)

n
_
1 +
2
e
c
n
1
_
, (13)
which can be related with the quantum uctuation-
dissipation theorem for the EM eld [6,16]. Finally, to
obtain the Casimir force over a certain body, the stress
tensor must be integrated over the surface that denes
the object
F
C
=
_

T(r) dS. (14)

Equation (13) is the main result of this letter. It gives
an expression for the quantum Casimir force including the
eects of a nite, non-vanishing temperature, in terms
of the eigenvalues and eigenfunctions of the Laplacian
operator. This expression has been easily obtained using
the stochastic quantization approach to a quantum eld,
together with the formalism of the Langevin equation
calculation of uctuation-induced forces [11].
Let us note that, as it occurs in most of the calcula-
tions of the Casimir Force, the expression for T(r) in
eq. (13) is generally divergent at every point of space r
if the sum over eigenvalues runs to innity. However, the
expression for the force, that is obtained integrating over
the surface of the body, eq. (14), is nite. It means that
the integration over the body regularizes the divergences
of the averaged stress tensor. If we assume that such regu-
larization is carried out mode by mode, we can interchange
the integration over the surface and the summation over
eigenvalues, to obtain,
F
C
=
c
2

n
1

n
_
1 +
2
e
c
n
1
_ _

TT
nn
(r) dS. (15)
which is a nite result. Therefore, the interchange of
the integral and summation regularizes the Casimir force,
avoiding the use of ultraviolet cutos. Other regulariza-
tions, that in some cases may lead to non-universal forces
or uctuations are, for instance, the subtraction of the
vacuum stress tensor [17] or by averaging the stress tensor
over a nite area or a nite time [18]. Having regularized
the divergences, eq. (15), provides a new method to calcu-
late Casimir forces for a given geometry by diagonaliz-
ing the Laplace operator. So, this approach is suitable for
numerical calculations of Casimir forces in complicated,
realistic geometries. Moreover, this method provides the
force directly, not as a dierence of the free energy with
respect to a reference state, which in some congurations it
may be dicult to establish. Also, it can be used as a start-
ing point for a perturbative theory for, e.g., nonat geome-
tries, rough surfaces or similar problems. Other authors
have obtained expressions for the free energy in terms of
the eigenvalues of the spatial operator, but not for the
force [19,20].
The expression for the Casimir force, eq. (13), allows
one to evaluate the quantum limit, that is, by setting the
temperature equal to zero (or ), where the second
summand inside the brackets in eq. (13) vanishes, i.e.,
lim
T0
T(r) =
c
2

n
TT
nn
(r)

n
. (16)
In the opposite, classical limit, when 0, a Taylor
expansion of the square bracket in eq. (13) gives
lim
0
T(r) =
1

n
TT
nn
(r)

2
n
. (17)
This expression for the Casimir force has been used
for classical systems [11], such as liquid crystals [7] or
reactiondiusion systems [21]. The two limits, quantum
(16) and thermal (17), show that the driving force of
the uctuations has dierent origin. In the rst case, the
presence of the factor indicates the quantum nature of
the uctuations, whereas in the second case, the factor
reveals its thermal origin.
Fluctuations. The Casimir force has its origin in
uctuations, so it is a uctuating quantity itself. The net
force is calculated as the average of the uctuating one,
eq. (14). However, the formalism developed in this letter
allows the calculation of uctuations of the force, dened
as customary for the uctuations:

2
F
=
_

[T(r
1
) dS
1
][T(r
2
) dS
2
] F
2
C
. (18)
Calculation of
2
F
requires the evaluation of the product of
4 elds , as the tensor T is bilinear form of . Substitution
of in the tensor leads to the correlation of a product of
four noises, . Because of the Gaussian nature of , it
factorizes into three products of pairs of noises, as shown
in eq. (5). The precise expression for the product of the
two stress tensors in eq. (18) reads
T(r
1
)T(r
2
) =
(c)
2
4

nm
P(
n
)P(
m
)
[TT
nn
(r
1
)TT
mm
(r
2
) +2TT
nm
(r
1
)TT
mn
(r
2
)] , (19)
where
P() =
1

_
1 +
2
e
c
1
_
. (20)
The rst term in (19) exactly gives F
2
C
when substitued in
eq. (18), while the other term (where the factor 2 comes
because the two last terms in eq. (5) give equal terms)
50008-p3
P. Rodriguez-Lopez et al.
A
L L
F
C

x
Fig. 1: Geometry of the considered problem. The Casimir force
is evaluated for the plate, of area A and general geometry. The
net force is obtained as the force exerted by a plate at distance
L and another plate at distance L

. The cylinder is
oriented along the x axis.
give non-vanishing contributions to
2
F
. Again, there is no
need to introduce cuto in the eigenvalues, as interchang-
ing the summations with the integral over the bodies regu-
larize the divergences. Therefore, we nd a universal form
for the uctuations, as opposite to other authors [8,18].
The dierence has its origin in that we do not compute
the stress uctuations or the uctuations of the force on
each side of a plate. Rather, we rst compute the total
uctuating force on the body (which is nite) and then
we compute its variance.
Application to the calculation of the Casimir
force of a cylinder of arbitrary cross-section. In
order to show how the formalism developed in this letter
works, let us consider a simple case of a piston depicted
in g. 1. It consists of a piston of area A, but general
shape, made of a perfectly conducting metal surface [22].
Two at conducting plates of the same cross-section of
the piston are placed at a distance L apart along the
x direction. The plates are perpendicular to the surface
of the cylinder. We calculate the Casimir force, F
C
, for
this plate by evaluating the expressions eq. (13) and later
integrating over the surface. In order to obtain a nite
result, we need a third auxiliary plate located at innite
distance, L

.
First, we solve the eigenvalue problem for the EM
eld and apply eqs. (12) and (14) to obtain the force.
For the EM eld the normal component of the stress
tensor reads T
xx
=E
2
x
+B
2
x

1
2
E
2

1
2
B
2
, which has to
be averaged over the noise and then integrated over the
surface of the plates, as shown in eq. (14). For perfectly
conducting plates, the boundary conditions are: En =0
and B n =0, where n is the normal vector at the surface.
In this geometry, the EM eld can be decomposed into
transverse electric (TE) and transverse magnetic (TM)
modes, which are discussed independently [23].
For the TM modes, the magnetic eld is transversal to
the x-direction, and the vector potential A for TM modes
can be written as
A=(C
2

D,
y
D
x
C,
z
D
x
C)e
it
. (21)
Here the elds C(x) and D(r

) satisfy

2
x
C(x) = k
2
x
C(x) (Neumann BC on x =0, L),

D
n
(r

) =
2
n
D
n
(r

) (Dirichlet BC on S), (22)

where S is the surface of the cylinder and r

=(y, z).
For the TE set, the electric eld is transversal to x,
so the vector potential is A=(0, S
z
N, S
y
N)e
it
,
where the functions S(x) and N(r

) satisfy eqs. (22)

with the opposite boundary conditions: Dirichlet for S(x)
and Neumann for N(r

). However, in this case, we must

exclude the constant eigenfunction, with eigenvalue
2
n
equal to zero, as it gives that A=0 and then E=B=0.
Substitution of the TE modes into the expression for
the stress tensor, and integration over one side of the
plates gives, after a long but straightforward calculation,
the Casimir force as
_
1 side

T
TE
xx
_
dS
x
=
1
L

mZ

n
x
=1

n
k
2
x

2
m
+k
2
x
+
2
n
,
(23)
where k
2
x
=(n
x
/L)
2
, and
m
is dened after eq. (7). For
the TM modes, one obtains exactly the same expression,
but
2
n
are the eigenvalues of the two-dimensional (2D)
Laplacian with Neumann boundary conditions. We will
denote the complete set of eigenvalues of the Laplacian
with Neumann (excluding the zero eigenvalue) and Dirich-
let boundary conditions by the index p. The expression
above is the equivalent of eq. (12) when the spectrum
can be split into a longitudinal and transversal part, that
is,
2
n
=k
2
x
+
2
p
. These series are divergent, but the net
Casimir force, which is the dierence between the force
exerted from the plate at distance L, and the plate at
L

(see g. 1), is nite.

The sum over the variable n
x
in eq. (23) can be carried
out with the help of the ChowlaSelberg summation
formula [24]. This formula extracts the divergent, L-
independent part of the summation, which cancels when
the integral in eq. (23) is performed for both sides of the
piston. Equivalently, the L-independent contribution is the
same for the second plate at distance L or at distance L

,
resulting in
F
C
=
1

mZ
_
m
2

2
+
2
p
e
2L

m
2

2
+
2
p
1
. (24)
Here, =2/c is the inverse thermal wavelength. This
expression gives the nite or regularized Casimir force
between two plates at distance L, valid for any cross-
section and temperature. The precise geometry of the
plates enters into the double set of eigenvalues of the
Laplacian (with Neumann and Dirichlet boundary condi-
tions)
2
p
.
We can proceed to evaluate the Casimir force at T =0,
that is, the purely EM case without thermal corrections.
50008-p4
Stochastic quantization and Casimir forces
This case is obtained by noting that, when T 0 (equiva-
lent to the limit 0 in eq. (24)), the sum over m can be
rigorously replaced by an integral. Computing such inte-
gral the result is
F
C
=
c
2

n=1

2
p
[K
0
(2nL
p
) +K
2
(2nL
p
)] . (25)
Here K

(x) is the modied Bessel function of order . In

ref. [19] it was obtained a formula for the free energy of the
conguration considered here, that, after dierentiation
with respect to the distance between the plates, leads to
the force above.
For short distances, however, the summation above can
be calculated without explicitly knowing the eigenvalues
of the Laplacian. Such eigenvalues,
2
p
, must scale with the
inverse of the square of the typical size of the piston. So, for
distances L much smaller than the section of the piston,
we can replace the sum over the eigenvalues,
2
p
, by an
integral, using the asymptotic expression for the density of
states of the Laplacian in two dimensions
1
: (
p
) =
A
2

p
,
for each set of Dirichlet or Neumann. This result will
be universal, independent of the shape of the section of
the piston, as the density of states is itself universal. The
resulting integrals can be performed to obtain the Casimir
force as
F
T=0
=
c
2
240L
4
A, (26)
which is the well-known result of the EM Casimir force
for innite parallel plates [1]. In the opposite limit, when
L is much larger than the typical size of the plate, the
argument of the Bessel functions in eq. (25) is much
larger than one. Because of the exponential behavior
of the Bessel functions, only the smallest eigenvalue
2
1
contributes to the sum, with the result
F
T=0
=
c
2

L
g
1

3/2
1
e
2L
1
, (27)
being g
1
the degeneration of
1
. Here, a counterintuitive
result is obtained. One would expect that the thin piston
would tend to the known one-dimensional (1D) Casimir
force, but instead an exponential decay of the force is
found. The known 1D result would be obtained if the
zero eigenvalue were considered. However, this eigenvalue
is excluded, as it leads to a vanishing eigenfunction.
Therefore, the 1D Casimir force cannot be obtained as
the limit from three dimensions (3D) to 1D.
At intermediate distances, that is, L comparable to the
size of the plates, one must solve the eigenvalue problem.
To illustrate the intermediate behavior and the transition
from eq. (26) to (27), we study the case of a circular
cylinder of radius R. In this case, the eigenvalues are the
zeros of the Bessel function J

(r) and its derivative. The

1
The asymptotic DoS has corrections coming from the perimeter
of the piston and its curvature [22].
0.01 0.1 1 10
10
-6
10
-4
10
-2
10
0
10
2
10
4
10
6
N=10
N=100
N=1000

F
C
R
2

h
c
LR
Fig. 2: Casimir force on the plates of a cylindrical piston of
circular cross-section of radius R as a function of the distance L
between the plates. Dierent curves are obtained by summing
N eigenvalues of the Laplacian, using eq. (25), where N is
indicated in the legend. Solid line is the 3D Casimir force,
eq. (26) and dotted line is the far distance result eq. (27).
dependence with the radius of the cylinder is
2
p
(R) =

2
p
(R=1)/R
2
, so the Casimir force as expressed in eq. (25)
is a function of L/R when the force is multiplied by R
2
.
Figure 2 shows the results for the Casimir force, eq. (25),
when N =10, N =100 and N =1,000 eigenvalues are
summed (taking into account their degeneration). We
have also plotted the two limiting results: i) the 3D
Casimir force (26), valid for LR with an algebraic
behavior: F
C
R
2
(L/R)
4
(solid line), ii) the far distance
limit, given by (27) with the smallest eigenvalue
2
1
(R)
3.39/R
2
with degeneracy g
1
=2, that is valid for LR
(dotted line). The transition between both regimes is
observed at LR. As expected, when few eigenvalues
are summed, for instance N =10, the resulting force is
only valid in the limit of long distances. As the number
of eigenvalues increase, the numerical result approaches
the 3D Casimir force. For N =1,000 eigenvalues, we have
excellent results for L/R>0.05. We remark, however,
that the full curve can be obtained by only considering
N =100 eigenvalues for large distances and matching this
numerical result with the asymptotic expression (26) for
short distances, with a crossover distance L0.3R.
In a similar fashion, we can calculate the thermal
Casimir force when 0, or . Then, in eq. (24),
only the term with m=0 is dierent from zero, as a result,
F
=0
=
1

p
e
2L
p
1
. (28)
For short (LR) and long (LR) distances, (28) reads
F
=0
=
(3)
4L
3
A, F
=0
=
1

g
1

1
e
2L
1
. (29)
We nish the application to the cylinder by evaluating
the uctuation of the Casimir Force. For the piston
geometry considered in this letter, the uctuations of the
force can be obtained by evaluating eq. (18). In this case,
50008-p5
P. Rodriguez-Lopez et al.
and because of the geometry of the problem, each of the
summands that appear in the four-point correlations of the
noise gives F
2
C
. Therefore, the uctuations of the force, for
any temperature and cross-section, are simply

2
F
=2F
2
C
. (30)
Somehow similar uctuations have been obtained for a
purely thermal force [8] for each mode that enters in
the Casimir force. In a non-equilibrium hydrodynamical
system the uctuations have been measured in [25] by
means of numerical simulation. The fact that the uctua-
tion of the force is as large as the force itself is a signature
of uctuation-induced forces.
Summary. We have shown in this letter that
stochastic quantization together with the Langevin
formalism provides a new approach to calculate Casimir
forces in the quantum-electrodynamics (QED) case,
including thermal eects. The starting point is the calcu-
lation of the Casimir force via the stress tensor, which is a
function of the uctuating elds. Parisi and Wu derived a
Langevin equation to describe the dynamics of such elds,
which can be integrated to give an expression for the
force. The method presented herein is quite simple, and
avoids some technical complication of other approaches.
Moreover, it calculates the force directly, instead of the
free energy. It also can be extended to calculate torques in
asymmetric congurations. Another advantage is that it
provides a numerical method to calculate the Casimir force
in complicated geometries, such as those of interest for
microelectromechanical systems (MEMS) devices, alter-
native to others [26]. The method only requires spectral
decomposition of the Laplacian operator in the given
geometry, and summation of the eigenvalues and the inte-
gration of the eigenfunction along the boundary of the
object, as shown in eqs. (13) and (14). Quantum (T 0)
and classical ( 0) limits are recovered by eqs. (16) and
(17), respectively. The integration of the eigenfunction
over the surface of the body leads to a regularization of
the Casimir force, producing nite results for the force
and the uctuations.
This approach can be generalized to include non-
equilibrium situations. For instance, it can be used to
calculate the Casimir force between two bodies at dierent
temperatures [27,28]. Also, the method could be applied
to time evolving temperatures [29] and space-dependent
temperature T(r).

The authors would like to thank A. Mu noz-Sudupe
and A. N u nez for helpful discussions. PR-L and RB
are supported by the Spanish projects MOSAICO and
MODELICO. PR-Ls research is also supported by an
FPU MEC grant. RS is supported by Fondecyt grant
1100100 and Proyecto Anillo ACT 127.
REFERENCES
[1] Casimir H. B. G., Proc. K. Ned. Akad. Wet., 51 (1948)
793.
[2] Bordag M., Klimchitskaya G. L., Mohideen U. and
Mostepanenko V. M., Advances in the Casimir Eect
(Oxford University Press, Oxford) 2009.
[3] Klimchitskaya G. L., Mohideen U. and Mostepa-
nenko V. M., Rev. Mod. Phys., 81 (2009) 1827.
[4] Emig T. et al., Phys. Rev. Lett., 99 (2007) 170403; Rahi
S. J. et al., Phys. Rev. D, 80 (2009) 085021.
[5] Maghrebi M. F. et al., Proc. Natl. Acad. Sci. U.S.A.,
108 (2011) 6867 and references therein.
[6] Lifshitz E. M., Sov. Phys. JETP, 2 (1956) 73.
[7] Ajdari A. et al., J. Phys. II, 2 (1992) 487.
[8] Bartolo D. et al., Phys. Rev. Lett., 89 (2002) 230601.
[9] Najafi A. and Golestanian R., Europhys. Lett., 68
(2004) 776.
[10] Bartolo D. et al., Phys. Rev. E, 67 (2003) 061112.
[11] Rodriguez-Lopez P., Brito R. and Soto R., Phys.
Rev. E, 83 (2011) 031102.
[12] Parisi G. and Wu Y.-S., Sci. Sin., 24 (1981) 483;
Damgaard P. H. and H uffel H., Phys. Rep., 152
(1987) 227.
[13] Masujima M., Path Integral Quantization and Stochastic
Quantization (Springer, Berlin) 2009.
[14] Kubo R., Toda M. and Hashitsume N., Statistical
Physics II (Springer Verlag).
[15] Risken H., The Fokker-Planck Equation, 2nd edition
(Springer, Berlin) 1989, Chapt. 3.
[16] Weber J., Phys. Rev., 101 (1956) 1620.
[17] Wu C.-H. and Ford L. H., Phys. Rev. D, 64 (2001)
045010; Wu C.-H., Kuo C.-I. and Ford L. H., Phys.
Rev. A, 65 (2002) 062102.
[18] Barton G., J. Phys. A: Math. Gen., 24 (1991) 991; 5533.
[19] Marachevsky V. N., J. Phys. A: Math. Gen., 41 (2008)
164007.
[20] Abalo E. K., Milton K. A. and Kaplan L., Phys. Rev.
D, 82 (2010) 125007.
[21] Brito R., Marconi U. M. B. and Soto R., Phys. Rev.
E, 76 (2007) 011113.
[22] Hertzberg M. P. et al., Phys. Rev. D, 76 (2007) 045016.
[23] Jackson J. D., Classical Electrodynamics, 3rd edition
(Wiley) 1999, Chapt. 8.3.
[24] Elizalde E., J. Phys. A: Math. Gen., 27 (1994) 3775.
See also eq. (A.3) of ref. [11].
[25] Cattuto C. et al., Phys. Rev. Lett., 96 (2006) 178001.
[26] Rodriguez A. et al., Phys. Rev. Lett., 99 (2007) 080401.
[27] Antezza M., Pitaevskii L. P. and Stringari S., Phys.
Rev. Lett., 95 (2005) 113202.
[28] Kr uger M. et al., EPL, 95 (2011) 21002.
[29] Dean D. S. and Gopinathan A., Phys. Rev. E, 81 (2010)
041126.
50008-p6