Anda di halaman 1dari 16

Theoretical and Mathematical Physics, 158(3): 406–421 (2009)

QUANTUM STATISTICAL THEORY OF RADIATION FRICTION OF


A RELATIVISTIC ELECTRON
G. F. Efremov∗ and V. V. Sharkov∗

We comprehensively investigate the effect of quantum space–time nonlocality that accounts for retardation
of the electron interaction with both the electron’s own radiation field and the fluctuation field of the
electromagnetic vacuum. We rigorously show that the quantum nonlocality effect eliminates the self-
acceleration and the causality violation paradoxes that are inherent in the classical theory of radiation
friction.

Keywords: generalized radiation friction force, theory of open quantum systems, quantum statistics,
electromagnetic vacuum, Brownian motion

1. Introduction

The problem of radiation friction and the electromagnetic mass of the electron arose before the creation
of quantum theory in the early 20th century. Fundamental contributions to the classical theory of radiation
friction were made by Abraham, Lorentz [1], and Dirac [2]. The main results, which, for example, are
expounded in [3], [4], and [5], have been scrutinized in the literature over many decades (a vast bibliography
on the subject can be found in [5]).
The classical formula for the Abraham–Lorentz radiation friction force [1], [3]–[5]

2 e2 d3
F= r(t) (1)
3 c3 dt3

leads to the well-known self-acceleration paradox [3], [4]: under the action of force (1), the electron state
with a definite velocity becomes unstable, and the electron gains ultrarelativistic speed in a very short time
interval τ0 = 10−23 sec. In addition, radiation force (1) contradicts the causality principle [6].
The physical reason for the paradoxes and divergences of classical electrodynamics is determined by
the idealized concept of a charged particle (the electron) as a pointlike object. Because the classical
electron has no internal structure and is pointlike, retardation of the interaction between the electron and
its own radiation field is absent, whence the known friction radiation paradoxes arise. The absence of
retardation of the interaction between the electron and the radiation field also leads to divergences: in
classical electrodynamics, the electron mass is infinite.
Recently, there was a revival of interest in the problem of radiation friction. This pertains to both
classical and quantum theory. In classical theory, the primary interest is focused on radiation friction in


Lobachevsky Nizhni Novgorod State University, Nizhni Novgorod, Russia, e-mail: efremov@rf.unn.ru,
sharkov@rf.unn.ru.

Translated from Teoreticheskaya i Matematicheskaya Fizika, Vol. 158, No. 3, pp. 478–496, March, 2009. Original
article submitted November 11, 2007; revised June 27, 2008.

406 0040-5779/09/1583-0406 
c 2009 Springer Science+Business Media, Inc.
relativistic problems [7], [8]. In [7], a procedure was proposed for eliminating divergences in classical electro-
dynamics, and the problem of the backreaction of the radiation field of a relativistic pointlike electron was
solved exactly. In [8], as in [7], the radiation backreaction was considered with renormalization in classical
field theory with singular sources taken into account. In a number of works, the radiation backreaction
was considered in different models of classical field theory (see, e.g., [9]). A critical analysis of the existing
classical theory can be found, for example, in [10], where various modifications of this theory were also
proposed. Eliminating the radiation friction paradoxes in classical electrodynamics was attempted in [11].
But despite considerable achievements, it has not been possible to avoid paradoxes in the theory of radiation
friction in the framework of a systematic classical description of radiation backreaction.
The appearance of quantum theory did not lead to automatically eliminating the radiation friction
paradoxes. The problem in the quantum theory of radiation friction has persisted over many decades;
various approaches and approximations were proposed for solving it [6], [12]–[21]. A possible way to solve
the radiation friction problem, entailing going beyond the perturbation theory, is based on the methods
of the fluctuation–dissipation theory of nonlinear open quantum systems [22]–[24]. These methods are
an essential generalization of the linear theory of Brownian motion in quantum systems developed by
Schwinger [25] and Senitzky [26]. This theory does not assume that the interaction is Markovian and small.
The problem setting and the initial equations were formulated in a number of works for both the relativistic
field theory [27] and the one-particle model of statistical quantum electrodynamics (QED) [6].
The rigorous quantum mechanical expression for the radiation force obtained in [6] was analyzed in the
nonrelativistic limit, and the absence of paradoxes inherent in the classical theory was shown. A quantum
theory of radiation friction for the proper magnetic moment of the neutron was proposed in [20], [21].
We note that in the nonrelativistic limit, the contribution to the radiation friction from photons with
large transferred momenta is eliminated. But in the complete solution of the radiation friction problem,
the contribution of large momenta transferred from the radiation field to the electron must be taken into
account.
As is known, the relativistic electron has internal degrees of freedom, specified by the Dirac matrices.
The existence of an internal structure (degrees of freedom) of the Dirac electron leads to the quantum
nonlocality effect established in [28], which explicitly takes the retardation of the interaction between the
electron and its own radiation field into account. This eliminates the radiation friction paradoxes of the
classical theory. In [28], in the framework of local quantum theory, it was shown how the retardation of the
interaction between a field source (charge) and its radiation field occurs; we note that only one radiation
friction mechanism was considered in that work, the one due to the backreaction of the radiation field.
Here, we study the very important relativistic effect of retardation of the interaction with both the
charge’s own radiation field and the fluctuation field of the electromagnetic vacuum. The phenomena under
consideration therefore acquire a distinct quantum statistical flavor. In investigating the radiation effects
of an essentially statistical nature, such as the Lamb shift and the anomalous magnetic moment of the
electron, it is of principal importance that the contribution of the electromagnetic vacuum fluctuation field
to the interaction retardation is additionally taken into account because interaction retardation results in
eliminating divergences.

2. Generalized radiation friction force

We consider the interaction of a relativistic electron with the quantum radiation field and an external
potential field V (r, t). To study the radiation effects, it is convenient to use the Heisenberg picture of
motion. We first note that the Heisenberg equations of motion for dynamical variables of the electron in
free space admit an exact solution, first obtained by Schrödinger in 1930 [29]. In addition, there are exact
stochastic equations for dynamical variables of the electron [6]. In the Heisenberg picture of motion, all the

407
dynamical variables of the system under consideration are explicitly time dependent and enter the equations
of motion on equal footing. We therefore represent the initial Hamiltonian of the system as
 
e
H = cα(t) p(t) − A(r(t), t) + β(t)mc2 + eA0 (r(t), t) + V (r(t), t) + F (t), (2)
c

where F (t) is the Hamiltonian of the quantum radiation field, A0 (r(t), t) and A(r(t), t) are radiation field
potentials, and α(t) and β(t) are Dirac matrices, which satisfy the known commutation relation, for example,

αj (t)αl (t) + αl (t)αj (t) = 2δjl .

We now take advantage of the gauge symmetry by choosing the transverse gauge for the field potentials:

div A(r(t), t) = 0. (3)

In the Heisenberg representation, the field potentials are functions of the electron coordinates r(t), and they
can therefore be conveniently represented in terms of a Fourier decomposition:

d3 k ikr(t)
A0 (r(t), t) = e A0 (k, t), (4)
(2π)3

d3 k ikr(t)
Aj (r(t), t) = e Aj (k, t), (5)
(2π)3

where the Fourier modes Aj (k, t) have no explicit dependence on the electron variables. The current density
modes canonically conjugate to Aj (k, t) are found from Hamiltonian (2) in accordance with the relation

δH e
− = ṙj (t)eikr(t) .
δAj (k, t) c

We write the Heisenberg equations for the dynamical variables of the system under consideration. For
the Fourier modes Aj (k, t) of the field potentials in (2), we have the equations
 
2 1 d2 4π
k + 2 2 Aj (k, t) = eṙj (t)eikr(t) , (6)
c dt c

which are the Maxwell equations for the radiation field interacting with the electron. We also write the
Heisenberg equations for the dynamical variables of the electron. The system Hamiltonian in (2), in
particular, implies the quantum analogue of the Lorentz equation:

d e d e  
πj (t) + ∇j V (r(t), t) + ∇j A0 (r(t), t) = − Aj (r(t), t) + ∇j ṙα (t)Aα (r(t), t) , (7)
dt c dt c
where
e
πj (t) = pj (t) − Aj (r(t), t)
c
is the kinetic electron momentum and pj (t) is the canonical electron momentum.
A characteristic feature of the quantum analogue of the Lorentz force in the right-hand side of (7) (the
force acting on the electron from the radiation field) is that this force is given by a total derivative of the
vector potential and hence contains a simpler term that depends on the velocity operator ṙj (t) = cαj (t).
We assume that the interaction of the field source (the electron) with the radiation field is turned on
adiabatically slowly in the infinite past t = −∞. Before the interaction is turned on, we have a system of

408
noninteracting particles: the electron in some state ψ ∈ Γ and the radiation field in the ground state (with
all photon numbers nks = 0).
It is known that in the ground (vacuum) state, the temporal correlation function of the field potentials
0
Aj (k, t) in the chosen Coulomb gauge is
   
1 0  4πc kj kl
Mjl (k, t − t1 ) = [Aj (k, t), A0l (−k, t1 )]+ = cos(ck(t − t1 )) δjl − 2 .
2 2 k k

With the interaction turned on adiabatically, the solution of Eq. (6) has the form

e ∞
Aj (k, t) = A0j (k, t) + dt1 Djl (k, t − t1 )e−ikr(t) ṙl (t1 ), (8)
c −∞

where A0j (k, t) is the nonperturbed field of the electromagnetic vacuum. The second term, determined by
the photon Green’s function Djl (k, t − t1 ), is the radiation field of the electron. In the gauge chosen in (3),
the Green’s functions are given by [30]
 
i 0
Djl (k, t − t1 ) = [A (k, t), Al (−k, t1 )]− η(t − t1 ) =
0
 j
 
4πc kj kl
= sin(ck(t − t1 )) δjl − 2 η(t − t1 ), (9)
k k

where the unit Heaviside function η(t − t1 ) accounts for retardation of the electron interaction with the
radiation field. In the Coulomb gauge, the modes of the Green’s function Dj0 (k, τ ) and D00 (k, τ ), τ = t−t1 ,
are

Dj0 (k, τ ) = 0, D00 (k, τ ) = 2 δ(τ ).
k
With the above expression for D00 (k, τ ), it can be easily shown that the scalar potential A0 (r, t) does not
contribute to Eq. (7), and we omit it in what follows.
We substitute exact solution (8) in the expression for vector potential (5), which determines the quan-
tum Lorentz force in Eq. (7). After symmetrizing the product of mutually commuting operators Aj (k, t)
and eikr(t) , we obtain

d3 k 1 0
Aj (r(t), t) = [A (k, t), eikr(t) ]+ +
(2π)3 2 j
 
e ∞ d3 k 1
+ dt1 3
Djl (k, t − t1 ) [eikr(t) , ṙl (t1 )e−ikr(t1 ) ]+ . (10)
c −∞ (2π) 2

After substituting solution (8), we can write the second term in the Lorentz force in the right-hand side
of (7) in two ways: as
 
e   e d3 k 1 0
∇j ṙα Aα (r(t), t) = ikj [A (k, t), ṙα eikr(t) ]+ +
c c (2π)3 2 α

e2 ∞ 1 ikr(t) −ikr(t1 )
+ 2 dt1 Dαl (k, t − t1 ) [e ṙα (t), e ṙl (t1 )]+ (11)
c −∞ 2

and as
 
e   e d3 k 1 0
∇j ṙα Aα (r(t), t) = ∇ j [A (k, t), ṙα eikr(t) ]+ +
c c (2π)3 2 α

e2 ∞ 1
+ 2 dt1 Dαl (k, t − t1 ) [eikr(t) ṙα (t), e−ikr(t1 ) ṙl (t1 )]+ . (12)
c −∞ 2

409
The first term in (10) determines a parametric action of the vacuum field fluctuations on the electron.
This action gives a certain contribution to the dynamics along with the second term in (10); it also involves
the fluctuation source of the vacuum field. To isolate the contribution to the dynamics from fluctuations of
the electromagnetic vacuum, we must eliminate the vacuum field potentials from the equations of motion
for the electron. Such a procedure for eliminating phonon variables in the derivation of the generalized
kinetic equation for conductance electrons can be found in [24].
Following the methods proposed in [23], we have
 ∞  
e d3 k i
Aj (r(t), t) = dt1 Mjl (k, t − t1 ) [eikr(t) , ṙl (t1 )e−ikr(t1 ) ]− η(t − t1 ) +
c −∞ (2π) 3 

1
+ Djl (k, t − t1 ) [eikr(t) , ṙl (t1 )e−ikr(t1 ) ]+ + ξjA (t), (13)
2

where ξjA (t) is the contribution to the general fluctuation source from the first term in the friction force; this
term is determined only by the vector potential. Here, the unit Heaviside function η(t − t1 ) accounts for the
causality principle under the parametric action exerted on the electron by the electromagnetic vacuum field.
In agreement with the general theory [23], the fluctuation sources ξj (t) are well defined, and a prescription is
given for calculating their correlation functions of any order. As a result, we obtain an equation describing
Brownian motion of the relativistic Dirac electron in the electromagnetic vacuum field:

d
πj (t) + ∇j V (r, t) = Fj (t) + ξj (t), (14)
dt

where ξj (t) is a well-defined fluctuation source with zero mean with respect to the vacuum state of the field.
The generalized radiation friction force Fj (t) accounts for both the backreaction of the electron’s radiation
field and the action of the vacuum field fluctuations. With (11) taken into account, the rigorous expression
for the radiation force Fj (t) in (14) becomes
 ∞  
e2 d d3 k 1
Fj (t) = − dt1 Djl (k, t − t1 ) [eikr(t) , e−ikr(t1 ) ṙl (t1 )]+ +
c2 dt −∞ (2π)3 2

i
+ Mjl (k, t − t1 ) [eikr(t) , e−ikr(t1 ) ṙl (t1 )]− η(t − t1 ) +

  
e2 ∞ d3 k 1
+ 2 dt1 3
ikj Dαl (k, t − t1 ) [eikr(t) ṙα (t), e−ikr(t1 ) ṙl (t1 )]+ +
c −∞ (2π) 2

i
+ Mαl (k, t − t1 ) [eikr(t) ṙα (t), e−ikr(t1 ) ṙl (t1 )]− η(t − t1 ) . (15)


We note that in writing the second term of the Lorentz force in form (12), the coupling of A0j (k, t)
with the operator ∇j acting on it from the left must also be taken into account; this coupling is given in
one of our works. The occurrence of ∇j in expression (12) allows directly determining the renormalization
of the potential energy V (r) due to fluctuations of the electromagnetic vacuum and using this to evaluate
the Lamb shift. For comprehensively investigating the quantum mechanical effects of radiation friction, the
expression for the radiation force in form (15) is preferable.
The exact expression for radiation force in Eq. (15) involves two closely related radiation friction
mechanisms: the action exerted on the electron by fluctuations of the electromagnetic vacuum and the
action of the electron’s own radiation field. Jointly taking these physical mechanisms into account is of
principal value for analyzing different electrodynamic effects based on formulas (14) and (15). In addition,
the rigorous expression (15) implies that the mean radiation force is zero in the thermodynamic equilibrium

410
of the electron with the radiation field. We note that in accordance with [6], expression (15) is applicable
for the interaction of the electron with a thermal radiation field at a finite temperature. To prove the
above property, we must write the mean of (15) in the spectral form and then use the Callen–Welton
fluctuation–dissipation theorem.

3. Radiation friction coefficient of the electron

We study the features of the radiation friction effect that are caused by retardation of the interaction
with both the charge’s own radiation field and the fluctuation field of the electromagnetic vacuum. Let a
homogeneous electric field with a strength E(t) act on the electron whose time evolution is governed by
Eq. (14); the energy of the interaction with this field is

V (t) = −rj eEj (t) = −rj fj . (16)

We average Eq. (14) over the initial state of the system and recall that the mean of the fluctuation source
is zero; this yields
d
πj (t) = fj (t) + Fj (t). (17)
dt
From this equation, we find the response to the external force fj (t) in accordance with

rj (t) = dt1 ϕjl (t, t1 )fl (t1 ),

whence    
δrj (t) i
= ϕ (t, t ) = [r (t), r (t )] η(t − t2 ). (18)
δfl (t2 ) f =0
jl 2 j l 2 −

We assume that the system is in equilibrium before interaction (16) is turned on; as noted above, the mean
radiation force in equilibrium is zero. With this taken into account, we can write the mean radiation force
as   
δFj (t)
Fj (t) = dt2 fl (t2 ). (19)
δfl (t2 )
From expression (15) for the radiation force, we have
    
δFj (t) δrn (t1 )
= dt1 Ljn (t, t1 ) . (20)
δfl (t2 ) δfl (t2 )

In accordance with the nonlinear fluctuation–dissipation theorems [31], we can neglect the fluctuations of
the response because the system nonlinearity (related to the radiation force) is small. Therefore, we can
replace the response in (19) with its statistical average:
    
δFj (t) δrn (t1 )
= dt1 Ljn (t, t1 ) . (21)
δfl (t2 ) δfl (t2 )

As already noted, the leading contribution to the radiation friction is given by the first term in (15), which
we consider here. We express radiation force (15) explicitly with the second term omitted. For this, we
substitute the expression for photon Green’s function (9) in (15) and also substitute the retarded correlation
jl (k, τ ) = Mjl (k, τ )η(τ ), for which we can write the relation
function M

jl (k, τ ) = − 1  d Djl (k, τ ).


M (22)
ck 2 dt1

411
After substituting (22) in (15) and integrating by parts, we move the time derivative to the velocity
operator. We then use the same strategy in formula (9), rewritten as
 
4π d kj kl
Djl (k, τ ) = 2 (cos(ckτ )η(τ )) δjl − 2 . (23)
k dt1 k

We substitute the last two relations in (16) and integrate by parts over time, recalling the transversality of
the gauge, to obtain   
1 ...
Fj (t) = dt1 γjl (t − t1 ) rl (t1 ), (24)
m
where the friction coefficient γjl (t − t1 ), defined by formula (24), is
  
1 e2 1 kj kl
γjl (t − t1 ) = d k 2 δjl − 2
3
×
2π 2 mc2 k k
  
1
× cos(ck(t − t1 )) [eikr(t) , e−ikr(t1 ) ]+ −
2
  0
i ikr(t) −ikr(t1 )
− sin(ck(t − t1 )) [e ,e ]− η(t − t1 ). (25)
2 0

The brackets  · 0 here denote the average over the nonperturbed state of the full system. By the isotropy
of the initial state of the system, the averages of the commutator and anticommutator of the exponential
factors in (25) are independent of the direction of the wave vector k. Therefore, γjl (t − t1 ) = δjl γ(t − t1 ),
where
 ∞   
4 e2 1 ikr(t) −ikr(t1 )
γ(t − t1 ) = dk cos(ck(t − t1 )) [e ,e ]+ −
3π mc2 0 2
  0
i ikr(t) −ikr(t1 )
− sin(ck(t − t1 )) [e ,e ]− η(t − t1 ). (26)
2 0

We note that in the classical limit, the friction coefficient becomes

2 e2
γ(t − t1 ) = γ0 δ(t − t1 ) = δ(t − t1 ), (27)
3 mc3
which agrees with classical formula (1).
The radiation friction coefficient in (26) has an essentially nonlinear dependence on the electron vari-
ables r(t) and r(t1 ), taken at different times. The retardation of the interaction between the electron and
the radiation field is therefore taken into account in the quantum mechanical expression. To illustrate this,
we evaluate the linear susceptibility of the electron from Eq. (17) with the interaction retardation taken
into account, as defined by the radiation friction force given by formulas (24)–(26). For this, we rewrite (17)
in the electron rest frame:

d2 1 d3
2
rj (t) = f j (t) + dt1 γ(t − t1 ) 3 rj (t1 ). (28)
dt m dt1

From this equation, we find the response to a harmonic force fj (t) = fj (ω)e−iωt . By definition, we have

rj (ω) = χ(ω)fj (ω). (29)

Applying the Fourier transformation to (28), we obtain

1
−ω 2 rj (ω) = fj (ω) + iω 3 γ(ω)rj (ω).
m
412
In accordance with (29), the linear susceptibility has the form

1 1
χ(ω) = − . (30)
m ω 2 (1 + iωγ(ω))

In accordance with the causality principle, the susceptibility must be an analytic function in the upper
half-plane of the complex variable ω; but using classical form (27) instead of (30) yields an expression for
the susceptibility with a pole in the upper half-plane of ω,

1 1
χ(ω) = − 2
,
m ω (1 + iωγ0 )

which contradicts the causality principle.

4. Effects of quantum space–time nonlocality


In contrast to classical coefficient (27), quantum mechanical radiation friction coefficient (26) automat-
ically accounts for the causality principle and the retardation of the electron interaction with the radiation
field due to the finiteness of the light propagation speed and the noncommutativity of the Heisenberg op-
erators of electron coordinates taken at different times. The essentially nonlinear and nonlocal dependence
of the radiation force on the position operators r(t) and r(t1 ) is determined by the product of exponentials,
which can be written as
eikr(t) e−ikr(t1 ) = eik∆r e−iB , (31)

where ∆r = r(t) − r(t1 ) is the operator of coordinate increment over the time interval t − t1 . The unitary
operator e−iB occurs in (31) because the operators r(t) and r(t1 ) do not commute at different times; its
presence has a certain effect on the retardation of the interaction between the electron and the radiation
field.
Here, the contribution to the radiation friction from large transferred momenta is taken into account;
the asymptotic (k→∞) contribution to (26) then corresponds to small values of ∆r and therefore to small
time intervals. Hence, for calculating the asymptotic form of radiation friction coefficient (26), it suffices
to use the free-evolution approximation for the electron position operator rj (t); this evolution is governed
by the Hamiltonian
H = cαp + βε0 , ε0 = mc2 . (32)

From the Heisenberg equations, we have

1
ṙj (t) = cαj , α̇j (t) = [αj , H]. (33)
i

We note that the momentum p and electron Hamiltonian (32) are integrals of motion. From the commu-
tation relations for the Dirac α and β matrices, we also have the relations

αj H + Hαj = 2cpj . (34)

Solving system of equations (33) and using (34), we find


    
c2 p j  2H c 2H
rj (t) = rj (0) + t− sin t + αj (0) sin t +
H 2H 
2H 
  
2 c 2H
+ α̇j (0) 1 − cos t , (35)
4H 2 

413

where H 2 ≡ H 2 = c2 p2 + ε2 , and therefore H = c2 p2 + ε20 . This solution was first obtained by
0
Schrödinger [29].
As in [28], we use solution (35) to calculate the dispersion of the operator

F0 = k(r(t) − r(t1 )) = k∆r (36)

in the electron rest frame. Using the commutation relations for the matrices αj and α̇j , we obtain the
expression for the dispersion of increment (36):

F02 = k 2 λ2c sin2 (ω0 t) = x2 sin2 (ω0 t),

where ω0 = mc2 /, λc = 1/k0 is the Compton wavelength of the photon, and x = k/k0 . The dispersion of
the increment of the Dirac electron coordinate in the electron rest frame is finite, unlike in the case of a
relativistic spinless particle. Fast oscillations at large times, i.e., for t  τ0 = 1/ω0 , are inessential, and we
can assume that
1 1
F02 ≈ k 2 λ2c , ∆r ≈ √ λc . (37)
2 2
Hence, the dispersion of the coordinate increment r(t)−r(t1 ) for a Dirac electron is finite in the electron
rest frame because of the existence of the internal degrees of freedom represented by the Dirac matrices. A
finite value of the dispersion, as we show in what follows, leads to a retardation of the interaction between
the electron and its own radiation field, which eliminates the known radiation friction paradoxes and the
logarithmic divergence in QED. The retardation is determined by the time it takes light to travel over the
Compton wavelength of the electron:
λ0  1
τ0 = = 2
= . (38)
c mc ω0
Therefore, the quantum theory of radiation friction must be an essentially non-Markovian theory, taking
the retardation of the interaction between the electron and the radiation field into account. Because the
characteristic interaction retardation time (38) is determined by the rest-frame energy mc2 , the quantum
theory is an essentially relativistic theory.

5. Calculating an asymptotic expression for the radiation friction


coefficient of the electron in the relativistic case
We now rigorously justify the quantum nonlocality effect taking the retardation of the electron inter-
action by both its own radiation field and the electromagnetic vacuum fluctuation field into account. For
this, we find the radiation friction coefficient in the asymptotic limit k  k0 . To evaluate the commutator
and the anticommutator in (26), we write the product of the exponentials in form (31), where the factor
eik∆r , which can be evaluated from relation (35) exactly, in particular contains the spatial nonlocality ∆r
given in (37). Because the operator ∆r involves Dirac matrices in accordance with formula (35), we can
represent eik∆r as
ik∆r
eik∆r = cos(k∆r) + sin(k∆r), (39)
k∆r
where ∆r = |∆r| = λ0 | sin(ω0 τ )| does not contain Dirac matrices.
Some specific features occur in calculating the unitary operator e−iB because the electron coordinate
r(t) depends on the momentum and Dirac matrices (see formula (35)). As a result of calculations based on
the S-matrix theory, given in the appendix, we have

eikr(t) e−ikr(t1 ) = eik∆r e−iω0 τ (βx arctan x+nα(x−arctan x))/2 ,


 †
e−ikr(t1 ) eikr(t) = eikr(t) e−ikr(t1 ) k→−k = (40)
= e−iω0 τ (−βx arctan x+nα(x−arctan x))/2 eik∆r ,

414
where n is the unit vector aligned with k, τ = t − t1 , and k = k0 x. The obtained expressions allow
explicitly writing the commutator and the anticommutator involved in formula (26) for the radiation friction
coefficient.
We now use formulas (39) and (40) and the representation for ∆r in the electron rest frame following
from (35); after averaging over the ground state, we obtain
   
1 ikr(t) −ikr(t1 ) ω0 τ
[e ,e ]+ = cos(x| sin(ω0 τ )|) cos f (x) ,
2 2
    (41)
i ikr(t) −ikr(t1 ) x arctan x ω0 τ
[e ,e ]− = cos(x| sin(ω0 τ )|) sin f (x) ,
2 f (x) 2

where f (x) = (x − arctan x)2 + x2 arctan2 x. Substituting equalities (41) in (26) and performing some
very simple trigonometric transformations, we write the radiation friction coefficient as (τ = t − t1 )
 ∞
2
γ(τ ) = α dx cos(x sin(ω0 τ ))η(τ ) ×

 0      
f (x) f (x)
× cos ω0 τ x + + cos ω0 τ x − +
2 2
      
x arctan x f (x) f (x)
+ cos ω0 τ x + − cos ω0 τ x − . (42)
f (x) 2 2
In the expression for the friction force, the contributions of each of the two radiation friction mechanisms
(the self-action and the effect of the vacuum) are added, and the asymptotic form of both terms can therefore
be considered simultaneously. For this, we note that the singularities of γ(τ ) in the asymptotic limit are
determined by the function f (x), which we write as
 π
f (x) = x2 y 2 + (x − y)2 = xF (y), y = arctan x ≤ ,
 2
 2
y
F (y) = y 2 + 1 − .
tan y
The leading contribution to the asymptotic expression for the radiation friction coefficient is given by the
domain x > 1, where the function F (y) = F (arctan x) changes very slowly and is bounded by the value
F (π/2). We therefore find the average of this function

2 π/2
F = dy F (y) ≈ 0.85 (43)
π 0

and set f (x) = xF . Because the function x(arctan x)/f (x) = y/F (y) also changes very slowly, we use its
average

2 π/2 y
ξ= dy ≈ 0.95. (44)
π 0 F (y)
With (43) and (44), we rewrite (42) as
 ∞ 
4 1+ξ 1−ξ
γ(τ ) = α dx cos(x sin(ω0 τ ))η(τ ) cos(β1 xω0 τ ) + cos(β2 xω0 τ ) , (45)
3π 0 2 2

where β1 = 1 + F /2 ≈ 1.425 and β2 = 1 − F /2 ≈ 0.575. Integrating over x in (45), we obtain



4 1+ξ
γ(τ ) = α π[δ(β1 ω0 τ + sin(ω0 τ )) + δ(β1 ω0 τ − sin(ω0 τ ))] +
3π 4

1−ξ
+ π[δ(β2 ω0 τ + sin(ω0 τ )) + δ(β2 ω0 τ − sin(ω0 τ ))] η(τ ). (46)
4

415
Using the values of the parameters β1 and β2 and recalling the causality principle for τ > 0, we obtain the
expression
1  
γ(τ ) = α(1 − ξ)δ β2 ω0 τ − sin(ω0 τ ) η(τ ). (47)
3
Solving the transcendent equation β2 ω0 τ − sin(ω0 τ ) = 0, we obtain its root τ ∗ as ω0 τ ∗ ≈ 1.72, sin(ω0 τ ∗ ) =
σ ≈ 0.99. It hence follows that the replacement sin(ω0 τ ) → σ does not change the asymptotic limit of the
friction coefficient
1  
γ(τ ) = α(1 − ξ)δ β2 ω0 (τ − τ ∗ ) η(τ ).
3
 
The corresponding replacement cos x sin(ω0 τ ) → cos(σx) in (42), which follows from asymptotic limit (47),
leads to an expression where the asymptotic regimes of small transferred momenta and accordingly of small
frequencies are simultaneously taken into account:
 ∞       
2 f (x) f (x)
γ(τ ) = α dx cos(σx)η(τ ) cos ω0 τ x + + cos ω0 τ x − +
3π 0 2 2
      
x arctan x f (x) f (x)
+ cos ω0 τ x + − cos ω0 τ x − . (48)
f (x) 2 2
 
Indeed, in the small-frequency domain, we can neglect the fast oscillations in the dispersion cos x sin(ω0 τ )
of the coordinate increment in expression (42). Therefore, the established asymptotic form in the domain
of large transferred momenta is simultaneously taken into account in the entire domain of k values if we
replace
jl (k, τ ) = Djl (k, τ ) cos(σλc k),
Djl (k, τ ) → D
(49)
jl (k, τ ) = Mjl (k, τ ) cos(σλc k)
Mjl (k, τ ) → M

in the original expression for the radiation force in (15).


From the physical standpoint, in accordance with (49), the factor cos(σx) in (15) results in taking
the retardation of the interaction between the field source (electron) and the radiation field into account.
The interaction retardation time is then the time for light to travel over the Compton wavelength (because
σ ≈ 1). Because of an effective renormalization of the interaction due to the factor cos(σx), the radiation
friction paradoxes are resolved, and divergences in the calculation of radiation effects are eliminated.
Analysis of the consequences of our main result in this work, formulas (49), is beyond the scope of
this paper. Here, we only remark on the essentially quantum nature of the theory of radiation friction and
on the absence of paradoxes; specifically, we consider the contribution to the radiation friction from the
first term in (15) with (49) taken into account in the simplest, first order of smallness in the fine-structure
constant. In this case, we can use expression (48) for the friction coefficient in the electron rest frame.
We consider the behavior of the friction coefficient at the frequencies ω < ω0 /2 in detail. In this
frequency range, we can use the approximation arctan x ≈ x. Consequently, it follows from (48) that
 ∞   
4α x2
γ(τ ) = dx cos(σx) cos ω0 τ x + η(τ ). (50)
3π 0 2

It is easy to evaluate the dependence of the radiation friction coefficient on frequency. For the real part,
we have  ∞  
2 e2 1 ω 2
γ  (ω) = Re dτ eiωτ γ(τ ) =  cos σ  . (51)
−∞ 3 mc3 1 + 2|ω|/ω0 ω0 1 + 1 + 2|ω|/ω0

We note that the dependence γ  (ω) is essentially quantum. Indeed, the ratio ω/ω0 = ω/mc2 in this
expression contains the Planck constant explicitly. As  → 0, expression (50) passes into its classical
counterpart (27), whence follow the known paradoxes.

416
We elucidate the physical meaning of the obtained expression (50); in particular, we show that the
parameter τ ∗ introduced above is the characteristic interaction retardation time. For simplicity, we con-
sider the domain of lower frequencies and therefore small transferred momenta, where the x2 term can be
neglected and friction coefficient (50) can be written as
 ∞

γ(τ ) = dx cos(σx) cos(ω0 τ x)η(τ ).
3π 0

Integrating over the wave vector leads to the expression

2      2 α
γ(τ ) = α δ ω0 (τ − τ ∗ ) + δ ω0 (τ + τ ∗ ) η(τ ) = δ(τ − τ ∗ ),
3 3 ω0

which implies the dependence of the radiation friction coefficient on frequency

2 α iωτ ∗ 2 e2 iωτ ∗
γ(ω) = e = e . (52)
3 ω0 3 mc3

With (52), linear susceptibility (30) can be written as

 −1
1 2α ω iωτ ∗
χ(ω) = − 1+i e . (53)
mω 2 3 ω0

It is easy to see that this expression has no poles in the upper half-plane of the complex variable ω and
hence satisfies the causality principle. But the radiation friction coefficient that can be obtained from the
classical Abraham–Lorentz formula (1),
e2
γ 0 (ω) = ,
mc3
gives a pole in the expression for susceptibility (30) in the upper half-plane of ω, which is in fact the reason
for the self-acceleration paradox in the classical theory of radiation damping.
Formula (53) for the linear susceptibility implies the equation of motion for the electron in an external
electric field:
... e
r̈j (t) − γ0 rj (t − τ ∗ ) = Ej (t).
m
Because the radiation friction force is “turned on” with a delay of τ ∗ , we approximate that during this time
interval, the electron moves only under the action of the external field; hence, during that interval, we have
the equation of motion
mr̈j (t) = eEj (t).

For longer time intervals, the effect of the radiation friction force can be taken into account using the
bounded-in-time equation of motion
... 1 d
rj (t) = eEj (t).
m dt
For large time intervals, we thus obtain the well-known equation, commonly used for practical calculations,
but with one essential difference:

e e2 d
r̈j (t) = Ej (t) + eEj (t − τ ∗ ).
m mc3 dt

In accordance with the obtained equation, the electron interacts with the electric field with a delay by the

417
time for light to travel over the Compton wavelength. This effect is a consequence of the nonlocality effect
for the relativistic quantum electron, and it is entirely absent in classical theory.

6. Conclusions

We have proposed a principally important generalization of the quantum nonlocality effect established
in [28]; this effect amounts to a retardation of the interaction between the electron and its own radia-
tion field. The fluctuations of the Dirac electron density, which have an essentially relativistic quantum
nature, were taken into account, and the asymptotic contribution of large transferred momenta to the re-
tardation effect for the interaction of a Dirac electron with the radiation field was determined with both
radiation friction mechanisms, the radiation field backreaction and the action exerted on the electron by the
electromagnetic vacuum fluctuation field, taken into account. We showed that the interaction retardation
(quantum nonlocality) effect eliminates the self-acceleration and the causality violation paradoxes occurring
in classical theory.
We found the dependence of the radiation friction coefficient on the frequency with quantum effects
taken into account. The obtained effect accounts for the contribution to the interaction retardation of all
momenta transferred from radiation to the electron. In the domain of small frequencies and accordingly
of small transferred momenta, the interaction retardation is a consequence of the fact that fast oscillations
of the dispersion of the coordinate increment can be neglected in the expression for the radiation friction
coefficient. The contribution of large transferred momenta to the interaction retardation follows from
analyzing the asymptotic expression for the friction coefficient for k  k0 . With high precision, the
retardation time is equal to the time for light to travel over the Compton wavelength λc .
The proof of the quantum nonlocality effect becomes especially important because the interaction
retardation automatically eliminates divergences and it is no longer necessary to use the renormalization
procedure to calculate the fundamental QED effects such as the Lamb shift and the anomalous magnetic
moment of the electron. Removal of the divergences in the framework of the simplest model of statistical
QED allows hoping that the quantum nonlocality effect can be extended to a systematically constructed field
theory. Some original applications of the fluctuation–dissipation QED have been published in conference
proceedings [27], [32]. In field theory, the retardation of interaction between a charge and the radiation
field can also be related to fluctuations in the electron density. On dimensional grounds, it then follows
that there is only one characteristic time, τ0 = λc /c.
The one-particle QED has a sufficiently wide applicability domain [33]; in particular, it allows inves-
tigating a broad spectrum of radiation effects, for example, radiation friction, the Lamb shift, and the
anomalous magnetic moment of the electron. If we pass to a more rigorous description of radiation ef-
fects and take the interaction of the electromagnetic and electron–positron vacuums into account, then
also in that case, because their interaction is weak, the Gaussian statistics for potentials of the vacuum
field is preserved with a high degree of precision (a rigorous physical justification that vacuum processes
only weakly depart from being Gaussian follows from the nonlinear fluctuation–dissipation theorems). The
photon Green’s function must then change because of polarization of the electron–positron vacuum and can
be evaluated only based on field theory. The form of stochastic equations then remains the same, and the
numerical value of the nonlocality parameter that determines the renormalization of the photon Green’s
function due to the interaction retardation effect also remains the same with high precision.
The above physical arguments regarding the applicability of one-particle QED agree with the corre-
spondence principle. For QED, the correspondence principle states that the quantum field theory, being
a more general theory, contains the one-particle QED as its limit case. At the same time, the field the-
ory is constructed based on Dirac’s one-particle relativistic theory with recourse to the identity principle.
Accordingly, the one-particle QED passes into classical electrodynamics as  → 0. In accordance with the

418
correspondence principle, the quantum nonlocality effect established in one-particle QED can be carried
over to field theory using a renormalization of the photon Green’s function.

Appendix

To calculate the unitary operator e−iB , we introduce a fictitious Schrödinger equation

∂ψ
−i = (F0 + V )ψ(θ), (A.1)
∂θ
where ψ is some function, θ plays the role of the time variable, F0 = k∆r is the free evolution, and
V = kr(t1 ) is the interaction operator. The solution of this equation is given by

ψ(θ) = ei(F0 +V )θ ψ(0). (A.2)

Our problem is to find the radiation damping coefficient for the electron in external fields that vary slowly
in time, i.e., with a frequency ω < ω0 = mc2 /. Therefore, to calculate the operator B, we use the terms
that are asymptotic in time. It follows from expression (35) that in the case where ω0 (t − t0 ) = ω0 τ > 1,
the free evolution operator F0 = k(r(t) − r(t1 )) is given by

c2 kp
F0 = τ, H = cαp + βmc2 . (A.3)
H

We can easily write (A.1) in the interaction representation:

∂ ψ
−i = V (θ)ψ(θ), (A.4)
∂θ
where

ψ(θ) = e−iF0 θ ψ(θ), V (θ) = e−iF0 θ V eiF0 θ . (A.5)

We note that in the momentum representation, the interaction operator has the form V = ik∇p , and
therefore

V (θ) = V + e−iF0 θ ik eiF0 θ ≡ V + U (p, θ), (A.6)
∂p
and Eq. (A.4) can be rewritten as
∂ ψ
−i = (V + U (p, θ))ψ(θ). (A.7)
∂θ
Calculating the derivative in (A.6) encounters an issue because the free evolution operator F0 depends on the
Dirac matrices. The commutation relations for these matrices imply that F02 = F 02 , where F 0 = (c2 kp/H)τ
,
i.e., F0 is free of Dirac matrices. Therefore, we can write

F0
eiF0 θ = cos(F 0 θ) + i sin(F 0 θ).
F 0

Using this result and the explicit form of free evolution operator (A.3), we express the operator U (p, θ) as

τ c2 k 2 θ τ  c4 (kp)2 θ
U (p, θ) = − + . (A.8)
H H3

In (A.7), we use the interaction representation and seek the solution in the form


ψ(θ) = eiV (θ−1) ϕ(θ). (A.9)

419
We then obtain an equation for the function ϕ(θ),

∂ϕ (p, θ)ϕ(θ),
−i =U (A.10)
∂θ

where
(p, θ) = e−iV (θ−1) U (p, θ)eiV (θ−1) = e−ikr(t1 )(θ−1) U (p, θ)eikr(t1 )(θ−1) .
U

We note that in the coordinate representation, the momentum operator is p = −i∇. It is easy to prove
the equality
U (p, θ) = U (p +  k(θ − 1), θ). (A.11)

Because the operator U (p, θ) does not commute with U


(p, θ1 ) for θ = θ1 , the solution of (A.10) must be
written in accordance with the S-matrix theory:

  θ
ϕ(θ) = T exp i
dθ1 U (p, θ1 ) ϕ(0) = S(θ)ϕ(0), (A.12)
0

where T is the operator of chronological ordering with respect to the parameter θ.


We find a relation between the original function ψ(θ) and the obtained function ϕ(θ). It follows
from (A.5) and (A.9) that
ψ(θ) = eiF0 θ eiV (θ−1) ϕ(θ). (A.13)

Using (A.12) and (A.13), we see that Eq. (A.1) with θ = 1 is solved by ψ(1) = eiF0 S(1)eiV ψ(0); at the
same time, it follows from general solution (A.12) that ψ(1) = ei(F0 +V ) ψ(0). Equating these two results,
we find that ei(F0 +V ) = eiF0 S(1)eiV or, in the original notation,

eikr(t) = eik∆r S(1)eikr(t1 ) .

We finally obtain the sought product of exponentials (32):

  1
eikr(t) e−ikr(t1 ) = eik∆r T exp i dθU (p + k(θ − 1), θ) . (A.14)
0

Formulas (A.8) and (A.11) imply the weak noncommutativity of the operators U(p, θ) with different
θ, and we can therefore drop the chronological ordering operation in (A.14) without significant loss of
(p, θ) with different values of θ
precision. We note that for a spinless relativistic particle, the operators U
mutually commute and the S-matrix is evaluated exactly.
Using (A.8) and integrating in the reference frame of the electron, we can write the right-hand side
of (A.14) as

  1
T exp i dθ U ( k(θ − 1), θ) =
0
      
iωτ k k kα kα k
= exp − β arctan + − arctan .
2 k0 k0 k0 k k0

Acknowledgments. The authors are grateful to Yu. G. Shondin for the useful discussions of a number
of issues raised in this paper.

420
REFERENCES
1. M. Abraham, Theorie der Elektrizität: II. Elektromagnetische Theorie der Strahlung, Teubner, Leipzig (1905);
H. A. Lorentz, Theory of Electrons and Its Applications to the Phenomena of Light and Radiant Heat, Dover,
New York (1953).
2. P. A. M. Dirac, Proc. Roy. Soc. Lond. Ser. A, 167, 148–169 (1938).
3. L. D. Landau and E. M. Lifshits, Theoretical Physics [in Russian], Vol. 2, Theory of Fields, Nauka, Moscow
(1973); English transl. prev. ed.: The Classical Theory of Fields, Addison-Wesley, Cambridge, Mass. (1951).
4. J. D. Jackson, Classical Electrodynamics, Wiley, New York (1975).
5. V. L. Ginzburg, Theoretical Physics and Astrophysics (Internat. Ser. Natur. Philos., Vol. 99), Pergamon, Oxford
(1979).
6. G. F. Efremov, JETP, 83, 896–901 (1996).
7. G. F. Efremov, JETP, 87, 899–904 (1998).
8. P. O. Kazinski and A. A. Sharapov, Theor. Math. Phys., 143, 798–820 (2005).
9. B. P. Kosyakov, Theor. Math. Phys., 119, 493–505 (1999).
10. F. Rohrlich, Phys. Lett. A, 283, 276–278 (2001); 295, 320–322 (2002).
11. V. V. Lidsky, Theor. Math. Phys., 143, 583–598 (2005).
12. R. P. Feynman, Phys. Rev., 74, 939–946 (1948).
13. G. W. Ford, J. T. Lewis, and R. F. O’Connell, Phys. Rev. Lett., 55, 2273–2276 (1985).
14. P. M. Barone and A. O. Caldeira, Phys. Rev. A, 43, 57–63 (1991).
15. V. Hakim and V. Ambegaokar, Phys. Rev. A, 32, 423–434 (1985).
16. A. I. Nikishov and V. I. Ritus, JETP, 87, 421–425 (1998).
17. A. Komech, Phys. Lett. A, 241, 311–322 (1998).
18. A. C. R. Mendes and F. I. Takakura, Phys. Rev. E, 64, 056501 (2001).
19. T. Petrosky, G. Ordonez, and I. Prigogine, Phys. Rev. A, 68, 022107 (2003).
20. A. Yu. Smirnov, J. Phys. A, 30, 1135–1141 (1997).
21. A. Yu. Smirnov, Phys. Rev. E, 56, 1484–1489 (1997).
22. G. F. Efremov and V. A. Kazakov, Izv. Vuzov. Ser. Radiofiz., 22, 453 (1979).
23. G. F. Efremov and A. Yu. Smirnov, Sov. Phys. JETP, 53, 547–554 (1981).
24. N. N. Bogolyubov and N. N. Bogoljubov Jr., Theor. Math. Phys., 43, 283–292 (1980).
25. J. Schwinger, J. Math. Phys., 2, 407–432 (1961).
26. I. R. Senitzky, Phys. Rev., 119, 670–679 (1960); 124, 642–648 (1961).
27. G. F. Efremov, M. A. Novikov, and L. G. Mourokh, “Fluctuation–dissipation quantum electrodynamics,”
in: The Present Status of the Quantum Theory of Light (Fund. Theories Phys., Vol. 80, S. Jeffers, S. Roy,
J.-P. Vigier, and G. Hunter eds.), Kluwer, Dordrecht (1997), p. 97.
28. G. F. Efremov and V. V. Sharkov, JETP, 98, 171–180 (2004).
29. P. A. M. Dirac, The Principles of Quantum Mechanics, Clarendon, Oxford (1935).
30. A. A. Abrikosov, L. P. Gorkov, and I. E. Dzyaloshinski, Methods of Quantum Field Theory in Statistical Physics
[in Russian], Fizmatlit, Moscow (1962); English transl., Prentice-Hall, Englewood Cliffs, N. J. (1963).
31. G. F. Efremov, Sov. Phys. JETP, 28, 1232 (1969).
32. G. F. Efremov, M. A. Novikov, V. V. Ivanov, and A. G. Efremov, “Causality and nonlocality in problem of
radiative damping of electron,” in: Causality and Locality in Modern Physics (Fund. Theories Phys., Vol. 97,
G. Hunter, S. Jeffers, and J.-P. Vigier, eds.), Kluwer, Dordrecht (1998), pp. 87–95.
33. Bjorken J. D. and Drell S. D., Relativistic Quantum Mechanics, McGraw-Hill, New York (1964).

421

Anda mungkin juga menyukai