r
X
i
v
:
1
1
0
9
.
0
7
0
0
v
1
[
q
u
a
n
t

p
h
]
4
S
e
p
2
0
1
1
Could quantum mechanics, and even gravity, be all about a
correct resolution of the classical selfforce problem?
Yehonatan Knoll
Email: yonatan2806@gmail.com
Abstract
In a recent paper by the present author [5], using a novel mathematical construction,
the formalism of extended charge dynamics (ECD) was presented. In that Lorentz and
scale covariant framework, charges are represented by localized conserved currents,
while the electromagnetic ﬁeld is the classical Maxwellian ﬁeld. Despite this seem
ingly classical setting, and the reduction of ECD to classical electrodynamics in the
latter’s domain of validity, it is shown in the present paper that ensembles of ECD so
lutions could, in principle, reproduce the statistical predictions of quantum mechanics.
Exclusively quantum mechanical concepts, such as interference, violations of Bell’s in
equalities, spin and even photons (despite the use of a classical EM ﬁeld), all emerge as
mere statistical manifestations of the self interaction of ECD charges. Moreover, ECD
is not merely an interpretation of relativistic quantum mechanics, but rather holds an
independent testable content, possibly unifying all forms of interaction including grav
ity. Much of the imbroglio and stagnation of modern physics may be due to a ‘wrong
turn’ taken over a century ago with regard to the self force problem, and exponentiated
ever since.
1 Introduction
At the turn of the 20th century there was an absolute ruler to theoretical physics — clas
sical electrodynamics. It was remarkably compact in its formulation, rather accurate in its
domain of application, and even compatible with the newly discovered theory of relativity.
There remained, however, two problems with classical electrodynamics: It was not even a
theory and, when applied to small scales, it was not even wrong (to paraphrase Pauli). It
was not even a theory because of the selfforce problem — the ill deﬁned Lorentz selfforce at
the position of the charge — and it was not even wrong because, even when self interaction
is ignored, or rather because it is ignored (as ignoring the self interaction is just one possible
solution to the selfforce problem) there is no model of matter based on classical electrody
namics with which to confront experimental results, such as the strength of chemical bonds
or the spectrum emitted by a heated gas.
The apparent connection between the two problems notwithstanding, two distinct paths
were pursued in solving each. Quantum mechanics gave concrete and increasingly accurate
predictions in circumstances in which classical electrodynamics was silent, but at the same
time demolished the clear, and thoroughly tested ontology of the latter — the manifest cor
puscular character of elementary charges and the continuous nature of the electromagnetic
ﬁeld. The resolution of the selfforce problem followed a completely diﬀerent, largely peda
gogical, rout. Many extensions and deformations of classical electrodynamics were proposed
(and still are, with focus shifting to the gravitational selfforce problem) all retaining the
1
classical ontology of pointlike charges and a continuous EM ﬁelds. All modiﬁed theories,
however, result in some local perturbation to the Lorentz force, and are therefore incapable
of explaining the nonlocal aspects of QM.
A century after the above bifurcation took place, equipped with the vast mathematical
arsenal accumulated thereafter, we return in this paper to those early 20th century days.
Guided by the principle of scale covariance — a hidden symmetry of classical electrodynamics
on equal footing with Poincar´e covariance — it is shown that the two fundamental problems
of classical electrodynamics should, and apparently can, be solved at once. This conclusion
stems from a deeper analysis of a recent work by the current author (with Yavneh; [5]),
proposing a Lorentz and scale covariant deformation of classical electrodynamics, depending
on a small dimensionless quantum parameter. In that deformation, dubbed extended charge
dynamics (ECD), particles enjoy a dual local nonlocal character. The local aspect explains,
for example, the thin traces left by charges in bubble chambers, and yet leads to a well deﬁned
self interaction. The nonlocal part — an inevitable consequence of Lorentz covariance —
accounts for quantum mechanical nonlocality which, for example, enables an ECD particle
passing through one slit in a doubleslit apparatus, to ‘remotely sense’ the status of the other
one. This feeble remote sensing mechanism, it is speculated, is also behind gravitational
interaction, ampliﬁed this time by a huge mass rather than by the huge distance between
the doubleslit and the detection screen. From yet a wider perspective, gravitation may be
just another facet of entanglement.
The possibility of unifying electrodynamics with such a seemingly unrelated theory as
gravity, extends also to small scales. The self force in classical electrodynamics is generally
ignored as a ﬁrst approximation, not because it is small (in fact, in the vicinity of the world
line of a point charge, the EM ﬁeld becomes arbitrarily strong) but rather because it leads
to fairly accurate results when applied to charges in a slowly varying EM ﬁeld. The colossal
failure of this approximation at small scales, rather than implying the breakdown of the
approximation, was taken as an indication for the breakdown of classical electrodynamics
altogether at small scales, inviting other modes of interaction, such as the strong force,
into the arena of small scale physics. It therefore seems only logical to ﬁrst have a fully
consistent classical electrodynamics, free of selfinteraction problems, and only then confront
it with small scale observations. One possibility, for example, suggested by ECD is that two
suﬃciently close, positively charged ECD particles, need not repel each other, rendering the
strong force just a small scale feature of ECD.
It is argued that quantum mechanics (QM), including quantum ﬁeld theory, only describes
certain statistical aspects of ensembles of ECD solutions. And indeed, QM, with its builtin
unitarity, deals mainly with statistical questions (Smatrices, reaction cross sections, thermal
properties of matter etc.). Over the years, one must acknowledge, some more ‘deterministic’
applications of QM have emerged — determining the strength of a chemical bond, or the
mass of a subatomic particle — but it seems that these deterministic uses were ‘forced’ on a
statistical theory, in the absence of any alternative method. In those applications, QM does
not really provide a satisfactory ontology for the microworld, but rather a set of heuristics
surviving experimental tests. ECD, on the other hand, just like classical electrodynamics,
2
is a singlesystemtheory, dealing explicitly with such deterministic question and, as will
transpire, holds the potential for a wide range of predictions. ECD, then, is not another
interpretation of QM, but rather an independent theory, supposedly compatible with the
statistical predictions of QM.
Section 3 of this paper deals with this compatibility conjecture. The classical ontology
of ECD notwithstanding, it is demonstrated in representative cases, how various features of
ECD render possible the realization of the quantum mechanical predications by means of an
ensemble of ECD solutions. The exact nature of the ensemble relevant to the experiment, it
is argued, is an independent law of nature, on equal footing with ECD itself. QM therefore
describes some statistical attributes of those ensembles, and like the nature of the ensemble
itself, enjoys the status of an independent law, complementing ECD rather than rivaling it.
In section 2 (and in the appendix), the mathematical structure of ECD is analyzed in
greater depth. The ECD equations derived in [5], originate from a brutally formal La
grangian, involving ‘deltafunction potentials moving in Minkowski’s space’. Such formal
objects never come equipped with a precise meaning which is to be determined only by the
global consistency of the mathematical structure resulting from their deﬁnition. Indeed, such
global considerations lead to a ‘reﬁned’ deﬁnition of the ECD equations, compared with their
form in [5]. One consequence of this reﬁnement, along with another central theme of ECD
— scale covariance — is the ability of ECD solutions to ‘drift in scale’. Such a scale drift
should be extremely slow on our native timescale, but could manifest itself on cosmological
scales. This speculation, as well as many other complementary remarks, can be found in the
many footnotes appearing in this paper, and may be ignored on ﬁrst reading.
Although acquaintance with [5] may provide some intuition into the origin of ECD, the
present paper is entirely self contained, requiring mainly standard graduate level background
for its understanding.
2 Extended Charge Dynamics
Throughout this paper the labeling of spacetime is chosen so as to have the speed of light
equal 1. The metric convention in Minkowski space, M, is g := diag(1, −1, −1, −1) , and u v
stands for g
µν
u
µ
v
ν
.
2.1 Manifestly scale covariant classical electrodynamics
There are two components in classical electrodynamics of n interacting charges. One is the
Lorentz force, governing the motion of a charge in a ﬁxed EM ﬁeld
¨ γ
µ
= q F
µ
ν
˙ γ
ν
, (1)
with γ(s) ≡ γ
s
: R → M the world line of a charge, parametrized by the Lorentz scalar s, q
a coupling constant and F
µν
= ∂
µ
A
ν
−∂
ν
A
µ
the antisymmetric Faraday tensor. Multiplying
both sides by ˙ γ
µ
and using the antisymmetry of F, we get that
d
ds
˙ γ
2
= 0, hence ˙ γ
2
is conserved
3
by the sevolution. This is a direct consequence of the sindependence of the Lorentz force,
and can also be expressed as the conservation of a ‘masssquared current’
b(x) =
_
∞
−∞
ds δ
(4)
(x −γ
s
) ˙ γ
2
s
˙ γ
s
. (2)
Deﬁning m =
__
d
3
xb
0
_
1/2
≡
_
˙ γ
2
≡
dτ
ds
with τ =
_
s
_
(dγ)
2
the propertime, equation (1)
takes the familiar form
m¨ x
µ
= q F
µ
ν
˙ x
ν
, (3)
with x(τ) = γ (s(τ)) above standing for the same worldline parametrized by propertime.
We see that the (conserved) eﬀective mass m emerges as a constant of motion associated with
a particular solution rather than entering the equations as a ﬁxed parameter. Equation (1),
however, is more general than (3), and supports solutions conserving a negative ˙ γ
2
(tachyons
— irrespective of their questionable reality).
The second ingredient of classical electrodynamics is Maxwell’s inhomogeneous equations,
prescribing an EM potential given the worldlines of all charges
∂
ν
F
νµ
≡ A
µ
−∂
µ
(∂ A) =
n
k=1
k
j
µ
, (4)
with
k
j(x) = q
_
∞
−∞
ds δ
(4)
_
x −
k
γ
s
_
k
˙ γ
s
(5)
the electric current associated with charge k, which is conserved,
∂
µ
j
µ
= q
_
∞
−∞
ds ∂
µ
δ
(4)
(x −γ
s
) ˙ γ
µ
s
= −q
_
∞
−∞
ds ∂
s
δ
(4)
(x −γ
s
) = 0 .
The selfforce problem of classical electrodynamics, to which we shall return in section
2.2.3, refers to the fact that the EM ﬁeld generated by (4) diverges everywhere on the world
line, ¯ γ ≡ ∪
s
γ
s
, traced by γ, rendering the Lorentz force (1) ill deﬁned (Another troubling
aspect of the selfforce problem is the divergence of formally conserved quantities such as
energy and momenta.)
The above unorthodox formulation of classical electrodynamics highlights its scale co
variance, meaning that the scaled variables
A
′
(x) = λ
−1
A(λ
−1
x) , γ
′
(s) = λγ(λ
−2
s) , (6)
also solve (1)
1
and (4), without scaling of any parameter. Like the Poincar´e symmetry, the
scaling symmetry (6) admits both an active and a passive interpretation. In the active
sense, it relates between diﬀerent solutions of the theory, for a given labeling (unitlength
1
An addition of the LorentzDirac radiation reaction force, written in our convention as
2
3
q
2
_
...
γ
˙ γ
2
−
˙ γ·
...
γ ˙ γ
( ˙ γ
2
)
2
_
,
still preserves the symmetry (6).
4
convention) of spacetime, from which one can read the representation under which each
variable, and products thereof, transform in a scale transformation — its ‘scaling dimension’:
[x] = [γ] = 1; [s] = 2; [A] = [m] = −1; [j] = −3 and, by deﬁnition, [q] = 0.
In the passive interpretation, the symmetry (6) prescribes how one must rescale A (more
generally, functions deﬁned on spacetime) when relabeling (scaling the unitlength) space
time. In this sense, A can be measured in units of length, and its scaling dimension may
just as well be named its length dimension, or simply dimension.
The simplicity in which scale covariance emerges in classical electrodynamics is due to
the representation of a charge by a mathematical point, obviously invariant under scaling of
spacetime. As we shall see, achieving scale covariance with extended charges is a lot more
diﬃcult, as no dimensionful parameter may be introduced into the theory from which the
charge may inherit its typical scale.
Associated with symmetry (6) is an interesting conserved ‘dilatation current’
ξ
ν
= p
νµ
x
µ
−
n
k=1
_
ds δ
(4)
_
x −
k
γ
s
_
s
k
˙ γ
2
s
k
˙ γ
ν
s
, (7)
with
p
νµ
(x) =
1
4
g
νµ
F
2
+ F
νρ
F
µ
ρ
+
n
k=1
k
m
νµ
, (8)
the (formally) conserved energymomentum (em) tensor associated with translation covari
ance, and
m
νµ
=
_
∞
−∞
ds ˙ γ
ν
˙ γ
µ
δ
(4)
_
x −γ
s
_
, (9)
the ‘matter’ em tensor, formally satisfying
∂
ν
m
νµ
= F
µν
j
ν
, (10)
∂
ν
m
νµ
=
_
ds ˙ γ
ν
˙ γ
µ
∂
ν
δ
(4)
_
x −γ
s
_
= −
_
ds ˙ γ
µ
∂
s
δ
(4)
_
x −γ
s
_
=
_
ds ¨ γ
µ
δ
(4)
_
x −γ
s
_
=
_
ds qF
µν
˙ γ
ν
δ
(4)
_
x −γ
s
_
= F
µν
j
ν
.
Note that the conserved dilatation charge,
_
d
3
xξ
0
, depends on the choice of origin for both
spacetime, and the n parameterizations of
k
γ, and is therefore diﬃcult to interpret.
2.2 Extended Charge Dynamics
In a nutshell, the transition from classical electrodynamics to ECD, involves two modiﬁca
tions. The ﬁrst grants the electric current (5) a nonsingular support in a way respecting all
the symmetries of classical electrodynamics — scale covariance in particular. To this end we
5
add to the representation of each charge an auxiliary complex (more generally spinor valued;
see appendix B) ‘wavefunction’
k
φ(x, s) : MR →C, and modify the current (5) to read
k
j
µ
(x) =
_
∞
−∞
ds
iq
2
_
k
φ
_
D
µ k
φ
_
∗
−
k
φ
∗
D
µ k
φ
_
≡
_
∞
−∞
ds q Im
k
φ
∗
D
µ k
φ , (11)
with
D
µ
=
¯
h∂
µ
−iqA
µ
(12)
the gauge covariant derivative and
¯
h some real dimensionless ‘quantum parameter’, not to
be confused with . Note the similar structure of (11) and (5). In (5) it is the trace in
Minkowski’s space of a singular vectorvalued distribution, δ
(4)
_
x −
k
γ
s
_
k
˙ γ
µ
s
, generating the
current, whereas in (11), the corresponding distribution is Im
k
φ
∗
D
µ k
φ, and need not be
singular. Despite this similarity between the ECD current (11) and the classical current (5),
there is a striking diﬀerence between the two: the EM potential A enters the deﬁnition of
the current (through D’s dependence on it) which, in turn, depends on all charges. It will
be demonstrated how this interdependence, along with an implicit dependence of φ on A,
described next, leads to quantum mechanical ‘entanglement’.
Summarizing, each ECD charge is now represented by a pair ¦φ, γ¦ but, of course, φ is
not independent of γ, as described next.
2.2.1 The ‘reﬁned’ central ECD system
The second component of ECD is the central ECD system — the counterpart of the Lorentz
force equation (1) — prescribing the set of permissible pairs ¦φ, γ¦ for a given A. This
system is composed of two coupled equations. The ﬁrst reads
φ(x, s) = −2π
2
¯
h
2
ǫi
_
s−ǫ
−∞
ds
′
G(x, γ
s
′ ; s −s
′
)φ(γ
s
′ , s
′
) (13)
+ 2π
2
¯
h
2
ǫi
_
∞
s+ǫ
ds
′
G(x, γ
s
′ ; s −s
′
)φ(γ
s
′ , s
′
)
≡ −2π
2
¯
h
2
ǫi
_
∞
−∞
ds
′
G(x, γ
s
′ ; s −s
′
)φ(γ
s
′ , s
′
)(ǫ; s −s
′
) ,
with (ǫ; σ) = θ(σ −ǫ) −θ(−σ −ǫ) ,
and the second equation is
∂ [φ(γ
s
, s)[
2
≡ ∂ [φ(x, s)[
2
¸
¸
x=γs
= 0 . (14)
Above, G(x, x
′
; s) is the propagator of a propertime Schr¨ odinger equation (also known as a
ﬁve dimensional Schr¨ odinger equation, or Stueckelberg’s equation),
_
i
¯
h∂
s
−1(x)
_
G(x, x
′
; s) = 0 , with 1 = −
1
2
D
2
, (15)
6
satisfying the initial condition (in the distributional sense),
G(x, x
′
; s) −→
s→0
δ
(4)
(x −x
′
) . (16)
Finally, ǫ is a parameter of dimension 2, ultimately taken to zero (thereby eliminating the sin
gle dimensionful parameter of ECD). Although advertised as the counterpart of the Lorentz
force, it is clear that the system also applies to chargeless particles, viz. particles with a
vanishing monopole.
Both equations, (13) and (14), involve a delicate ǫ → 0 limit, requiring clariﬁcations
which were not fully given in [5]. Focusing ﬁrst on (13), we see that, for ﬁxed γ and G, it is
in fact an equation for a function f
R
(s) ≡ φ(γ
s
, s). Indeed, plugging an ansatz for f
R
into
the r.h.s. of (13), one can compute φ(x, s) ∀s, x, and in particular for x = γ
s
, which we call
f
L
(s). The linear map f
R
→ f
L
(which, using G(x
′
, x; s) = G
∗
(x, x
′
; −s), can be shown to be
formally selfadjoint) must therefore send f
R
to itself, for (13) to have a solution. Now, the
universal (viz. Aindependent) i/(2π
¯
hs)
2
divergence of G(x, x
′
, s), implies f
R
→ f
R
+O(ǫ),
so the nontrivial content of (13) is in this O(ǫ) term, which we write as ǫf
r
(‘r’ for residue),
with f
r
= O(1).
In [5], lim
ǫ→0
f
r
= 0 was implied as the content of (13). While this may turn out to be
true for some speciﬁc solutions (a freely moving particle, for example), the equation should
take a more relaxed form
Im
_
lim
ǫ→0
f
r∗
_
f
R
= 0 , (17)
where, as usual, ‘Im’ is the imaginary part of the entire expression to its right.
Moving next to the second ECD equation, (14), conveniently rewritten as
Re
¯
h∂φ(γ
s
, s)φ
∗
(γ
s
, s) = 0 , (18)
a similar isolation of the nontrivial content exists. For further use, however, we ﬁrst want
to isolate the contribution of the i/(2π
¯
hs)
2
divergence of G(x, x
′
, s) to φ(x, s), for a general
x other than γ
s
. To this end, we need the smalls form of the propagator G. Plugging the
ansatz
G(x, x
′
, s) = G
f
e
iΦ(x,x
′
,s)/
¯
h
(19)
into (15), with
G
f
(x, x
′
; s) =
i
(2π
¯
h)
2
e
i(x−x
′
)
2
2
¯
hs
s
2
sign(s) , (20)
the free propagator computed for A ≡ 0, and expanding Φ (not necessarily real) in powers of
s, Φ(x, x
′
, s) = Φ
0
(x, x
′
) + Φ
1
(x, x
′
)s + . . ., higher orders of Φ
k
can recursively be computed
with Φ
0
alone incorporating the initial condition (16) in the form Φ
0
(x
′
, x
′
) = 0 (note the
manifest gauge covariance of this scheme to any order k). For our purpose, Φ
0
is enough. A
simple calculation gives the gauge covariant phase
Φ
0
(x, x
′
) = q
_
x
x
′
dξ A(ξ) , (21)
7
where the integral is taken along the straight path connecting x
′
with x. Substituting (19)
into (13), and expanding the integrand around s to ﬁrst order in s
′
−s: γ
s
′ ∼ γ
s
+ ˙ γ
s
(s
′
−s),
Φ
0
(x, γ
s
′ ) ∼ Φ
0
(x, γ
s
), φ(γ
s
′ , s
′
) ∼ f
R
(s), leads to a gauge covariant deﬁnition of the singular
part of φ
φ
s
(x, s) = f
R
(s)e
i
_
Φ
0
(x,γs)+˙ γs·ξ
_
/
¯
h
sinc
_
ξ
2
2
¯
hǫ
_
(22)
with ξ ≡ x − γ
s
. Consequently, the residual (or regular) wavefunction is deﬁned via the
gauge covariant equation
ǫφ
r
(x, s) = φ(x, s) −φ
s
(x, s) . (23)
Using ∂Φ
0
(x, γ
s
)[
x=γs
= A(γ
s
), we have
φ
s
(γ
s
, s) = f
R
(s) ,
¯
h∂φ
s
(γ
s
, s) = i
_
˙ γ
s
+ A(γ
s
)
¸
f
R
(s) , (24)
and (18) is automatically satisﬁed up to an O(ǫ), gauge invariant term
ǫ Re
¯
h∂
_
φ
r
(γ
s
, s)φ
s
(γ
s
, s)
∗
¸
= ǫ Re Dφ
r
(γ
s
, s)φ
s
(γ
s
, s)
∗
, (25)
where the above equality follows from (24), φ
r
(γ
s
, s) = f
r
(s) and (17). The reﬁned deﬁnition
of (14) is therefore
lim
ǫ→0
Re Dφ
r
(γ
s
, s)φ
s
(γ
s
, s)
∗
= 0 . (26)
Using the above deﬁnitions, (17) can also be written as
lim
ǫ→0
Im φ
r
(γ
s
, s)φ
s
(γ
s
, s)
∗
= 0 . (27)
More insight into this reﬁnement of the central ECD system, shall be given in the sequel.
For the time being, let us just note that it is invariant under the original symmetry group
of ECD. In particular, the system is invariant under
φ
s
→ Cφ
s
, φ
r
→ Cφ
r
, C ∈ C, (28)
under a gauge transformation
A → A + ∂Λ, G(x, x
′
, s) → Ge
i [qΛ(x)−qΛ(x
′
)]/
¯
h
, φ
s
→ φ
s
e
iqΛ/
¯
h
, φ
r
→ φ
r
e
iqΛ/
¯
h
, (29)
and under scaling of spacetime
A(x) → λ
−1
A(λ
−1
x) , ǫ → λ
2
ǫ , γ(s) → λγ
_
λ
−2
s
_
,
φ
s
(x, s) → λ
−2
φ
s
_
λ
−1
x, λ
−2
s
_
, φ
r
(x, s) → λ
−2
φ
r
_
λ
−1
x, λ
−2
s
_
, (30)
directly following from the transformation of the propagator under scaling
A(x) → λ
−1
A(λ
−1
x) ⇒ G(x, x
′
; s) → λ
−4
G
_
λ
−1
x, λ
−1
x
′
; λ
−2
s
_
.
Regarding this last symmetry, two points should be noted. First, for a ﬁnite ǫ it relates
between solutions of diﬀerent theories, indexed by diﬀerent values of ǫ. It is only because ǫ
is ultimately eliminated from all results, via an ǫ → 0 limit, that scaling can be considered
a symmetry of ECD. The second point concerns the scaling dimension, −2, of φ
s
and φ
r
.
By the symmetry (28), that dimension can be arbitrarily chosen. However, the central ECD
system is but a part of the ECD formalism, which dictates this special choice of dimension
to comply with scale covariance (see next section).
8
2.2.2 Regularized currents
The ǫ → 0 limit of the current (11), as also the limits of all other currents in ECD, vanishes
everywhere (except on the world line, ¯ γ ≡ ∪
s
γ
s
, traced by γ, where it is ﬁnite), trivializing
ECD. To correct this situation, two steps are taken. Utilizing the symmetry (28), we ﬁrst
rescale φ → ǫ
−1
φ, granting j in (11) a nonsingular support. As we show next, however,
the resultant current diverges everywhere in the limit ǫ → 0. To ﬁx this new problem, we
substitute φ → ǫ
−1
φ
s
+φ
r
in (11), and note that this divergence, as well as those of all other
ECD currents, can be traced to gauge invariant contributions of bilinears in φ
s
, which in this
case read
_
∞
−∞
ds q Im
φ
s
ǫ
∗
D
µ
φ
s
ǫ
≡ j
s
. (31)
Indeed, by (22) we get
j
s
(x) =
q
¯
h
_
ds
_
˙ γ
s
−qA(x) + ∂
x
Φ(x, γ
s
)
_ ¸
¸
f
R
(s)
¸
¸
2 1
ǫ
2
sinc
2
_
(x −γ
s
)
2
2
¯
hǫ
_
. (32)
Using (in the distributional sense) ǫ
−1
sinc
2
(ǫ
−1
y) −→ πδ(y) for ǫ → 0, we see that j
s
contains
an ǫ
−1
term in its ǫexpansion. Taking further into account the ﬁnite width of ǫ
−1
sinc
2
(ǫ
−1
y)
(as oppose to a delta distribution) and its evenness, it can be shown that the next higher
power in the expansion is ǫ
1
. This leads to the deﬁnition of the regular current, j
r
— a gauge
invariant expression deﬁned as the free coeﬃcient in the ǫexpansion of j, or equivalently,
j
r
= lim
ǫ→0
(j −j
s
) . (33)
This regular current is the electric current ultimately associated with an ECD charge, enter
ing as a source into Maxwell’s equations (4). By (30), j
r
has dimension −3, consistent with
the scaling dimension of A, namely, (4) is invariant under
A(x) → λ
−1
A
_
λ
−1
x
_
, j
r
(x) → λ
−3
j
r
_
λ
−1
x
_
, (34)
establishing the scale covariance of ECD. Finally, we note that for A ≡ 0 and a freely moving
γ (γ
s
= us), j
r
vanishes, as it must on self consistency grounds.
In appendix A we prove that the regular current, (33), is conserved for x / ∈ ¯ γ. The
conservation of a current deﬁned on M/¯ γ, however, does not imply the time independence
of the associated charge Q =
_
d
3
xj
r0
(x
0
, x), due to a possible ‘leakage’ of charge into a
sink of j
r
on ¯ γ or ‘emergence’ of charge from a source thereon. Remarkably, the reﬁned ECD
equation (27), turns out to be exactly the condition guaranteeing that no such leakage occurs.
Likewise, the second reﬁned ECD equation, (26), guarantees that no energy or momenta leak
into sinks of the conserved energymomentum tensor on ∪
k
k
¯ γ. It is therefore natural to add
to the central ECD system the proviso that the electric charge of each particle, as well as
the collective em of the system, do not ‘leak to inﬁnity’ (although it is possible that at least
the ﬁrst condition is automatically satisﬁed).
Carefully applying Noether’s theorem, both charge and energymomentum conservation,
can be shown to follow from continuous symmetries of ECD, (see appendix A). The converse,
9
however, is not true, namely, not every continuous symmetry of ECD leads to a conservation
law, due to a possible leakage of the corresponding charge to sinks on
k
¯ γ. The counterparts
of the ‘masssquared current’, (2), associated with stranslation invariance, as well as the
counterpart of (7) corresponding to scale covariance, fall into this category (see appendix
A).
2.2.3 The self consistent potential
As in classical electrodynamics, so also in ECD, the EM potential, A, must satisfy a self
consistent ‘loop’:
(a) Start with A and n pairs ¦
k
φ,
k
γ
s
¦ satisfying the central ECD system (26), (27) (or the
Lorentz force equation, (1), in classical electrodynamics);
(b) From these, using (11) (or (5) in classical electrodynamics), compute n
k
j
r
’s, plug them
into the r.h.s. of (4) and, ﬁnally,
(c) verify that the the l.h.s. agrees with the original A.
To the above loop one should add the proviso that the electric and masssquared charges
of each particle, as well as the total energy and momentum of the system, must be ﬁnite.
As shown in appendix A, this proviso involves at most the asymptotic behavior of A and φ
away from γ, and not the classical selfenergy divergence which is automatically eliminated.
The common loop notwithstanding, two important diﬀerences should be noted. First,
in classical electrodynamics, the loop is only formal, due to the selfforce problem — the ill
deﬁned Lorentz selfforce at the position of a pointcharge. Resolving the selfforce problem
calls for extensions/deformations of classical electrodynamics which, while eliminating the
original problem, also introduce drastic changes to other, desirable, features of it
2
. An even
2
Attempts to resolve the classical selfforce problem follow two distinct approaches. In the ﬁrst, it
is postulated that the ﬁeld produced by a charge does not act on the charge itself, thereby allowing to
preserve the point structure of elementary charges, along with their scale covariance (a point remains a
point following the scaling of spacetime). The AbrahamLorentzDirac equation, or actionatadistance
electrodynamics, [8], are typical examples of this approach, both carrying a heavy pricetag in the form of
expressions for the (conserved) energy and momenta which are radically diﬀerent from the (illdeﬁned but
rather intuitive) corresponding expressions in the usual formulation of electrodynamics (in fact, in the case
of the AbrahamLorentzDirac equation, it is not even clear what is conserved in a system of interacting
charges; no Lagrangian for such a system is known).
In the second approach, the singular selfforce is resolved by ‘extending’ the charge so as to associate with
it a nonsingular current producing a smooth self ﬁeld. Prominent examples include Lorentz’s early attempt
to regard the electron as a small uniformly charged sphere, and Feynman’s covariant regularization of the self
LienardWiechert potential [4]. Done in the framework of actionatadistance electrodynamics, Feynman
directly regularizes the LienardWiechert potential rather than deriving it from a regular current. However,
his method can be readily adopted to the regularization of a current. Deﬁning
j(x) = ∂
2
_
∞
−∞
ds
1
r
2
f
_
(x −γ
s
)
2
r
2
_
˙ γ
s
for some integrable function f, we note, with Feynman, that j is everywhere conserved and negligible for
distances from the charge much greater than r (a covariant alternative, more in the spirit of what is done in
this paper, is, rather than operating with the d’Alembertian on the integral, taking its derivative ∂/∂(r
2
).).
This latter approach, however, introduces a privileged scale (e.g. r in the above example) into otherwise
10
greater shortcoming is the fact that those extensions/deformations of classical electrody
namics do not survive experimental tests when applied to small scales. This is the domain
of QM, along with its distinct ontology. In ECD, on the other hand, only φ
r
(x, s) needs to
be diﬀerentiable on ¯ γ. This condition easily tolerates discontinuities of the EM ﬁeld on ¯ γ
(which, in fact, exist) freeing ECD, as is, from the selfforce problem.
The second diﬀerence in the role played by the above loop is that, in ECD, the very
existence of an ECD charge is due to a solution for the loop. That is, a non vanishing
A must be found even for a single static charge in an otherwise void universe — diﬀerent
such solutions naturally corresponding to diﬀerent elementary particles. This is a nontrivial
requirement, possibly leading to constraints on the nature of fundamental ECD charges.
Charge quantization is one such possibility, as the magnitude of the total charge of a solution
is invariant under the full symmetry group of ECD. Another possibility is that, as in other
eigenvalue problems, only for certain values of the ECD parameters, viz.
¯
h, q, and g (for
spin
1
2
ECD), does there exist a solution.
The self consistent loop is further responsible for the emergence of a privileged scale
(e.g. the size/mass of the charge) in a scale covariant theory, containing no dimensionful
parameter. Of course, the scale selected is but one of the inﬁnite possibilities, corresponding
to diﬀerent λ in the scaling transformation (30).
A closely related aspect of the loop concerns the fact that A is generated by the collective
current of all the particles. As A enters the expressions of all the currents associated with
an ECD charge (both explicitly and implicitly via φ), a dense cloud of particles must there
fore be strongly entangled by this constraint, as apparently manifested in low temperature
experiments. Moreover, even attributes of individual particles, such as the mass or charge,
are, in principle, entangled. (This may explain why a continuum of scaled interacting atoms
are never seen — a self consistent A may not exist for the combined system. We shall have
a lot more to say about role of the consistency loop with regard to entanglement of remote
particles.)
2.2.4 Antiparticles
ECD can be shown to be invariant under a ‘CPT’ transformation
A(x) → −A(−x) , γ(s) → −γ(−s) φ(x, s) → φ
∗
(−x, −s)
⇒ j
r
(x) → −j
r
(−x) . (35)
scale covariant electrodynamics, dramatically diminishing its symmetry group, and also introduces inﬁnitely
many tunable constants — the function f — into singleparameter classical electrodynamics. Taking the
limit r → 0 introduces into the dynamics of γ a term of the form C¨ γ with C diverging in that limit,
trivializing the dynamics (the charges cannot move due to an inﬁnite eﬀective mass). To get a nontrivial
theory, one must then ‘renormalize the mass’ — a rather contrived procedure.
11
In fact, scalar ECD, as well as classical electrodynamics
3
, enjoys an even larger symmetry
group, C: A(x) → −A(x), j
r
(x) → −j
r
(x); and PT: A(x) → A(−x), j
r
(x) → j
r
(−x).
However, the spin
1
2
ECD, presented in appendix B, enjoys the CPT symmetry only. This
symmetry has some remarkable consequences. First, it implies that our naive notion of time
reversal — ‘running the movies backward’ — is not a symmetry of microphysics. Secondly, it
predicts the existence of an antiparticle for each particle (viz., a bound solution of one or more
elementary ECD charges) of opposite charge and equal selfenergy. Pair creation/annihilation
may then have a simple geometrical interpretation when γ ‘reverses its direction in time’ (see
picture).
pair annihilation pair creation
✲
✻
t
x
As a particle and its antiparticle have opposite signs for both their electric charges, and
their masssquared charges (expression (95), the counterpart of the classical (2)), such an
nihilation/creation scenarios respect the conservation law of both electric and masssquared
charges. The energy of a particle, however, equals that of its antiparticle. In such annihi
lation processes, EM radiation must be released in order to respect energy conservation. It
may even be possible for the particle and its antiparticle to belong to distinct γ’s. This, how
ever, is more speculative, as the two may also form a bound neutral state, waiting dormant
to be reseparated by, say, a strong EM pulse.
2.2.5 The classical limit of the central ECD system
The analysis of the classical, or
¯
h → 0 limit, of the central ECD system, is facilitated
by a simple representation of the propagator, G, in that limit, known as the semiclassical
propagator
4
G
sc
(x, x
′
; s) =
i sign(s)
(2π
¯
h)
2
β
T
β
(x, x
′
; s)e
iI
β
(x,x
′
;s)/
¯
h
. (36)
3
Maxwell’s equations and the Lorentz force are also symmetric under under T: γ(t) → γ(−t), E(x, t) →
E(x, −t), B(x, t) → −B(x, −t), and under P: γ(t) → −γ(t), E(x, t) → E(−x, t), B(x, t) → −B(−x, t).
However, if one includes in the deﬁnition of classical electrodynamics, a deﬁnite Green’s function, other than
the halfadvancedplushalfretardedLienardWiechertpotential, then T is no longer a symmetry.
4
The nature of the approximation involved in the use of the semiclassical propagator can be read from
the path integral representation of the propagator. The classical paths dominate that representation and the
semiclassical approximation amounts to misweighing paths that deviate signiﬁcantly from classical paths.
These, however, enter with a nearly random phase anyway.
12
Here, β runs over the diﬀerent classical paths, viz. paths solving (1) for the ﬁxed A, such
that β(0) = x
′
and β(s) = x; I
β
is the corresponding action of the path,
I
β
=
_
s
0
dσ
1
2
˙
β
2
σ
+ qA(β
σ
)
˙
β
σ
, (37)
and T — the so called VanVleck determinant — is the gaugeinvariant classical quantity,
given by the determinant
T(x, x
′
; s) =
¸
¸
−∂
xµ
∂
x
′
ν
I
β
(x, x
′
; s)
¸
¸
1/2
. (38)
Let us next show that, in the limit
¯
h → 0, the reﬁned central ECD system is solved by any
classical γ (in the given EM potential A), and by a corresponding ansatz of the form
f
R
(s
′
) = Ce
iIγ(γ
s
′ ,γ
0
,s
′
)/
¯
h
, (39)
where C ∈ C is arbitrary. Substituting in (13), G → G
sc
, x
′
→ γ
s
′ and x → γ
s
, we ﬁrst note
that one of the β’s, appearing in G
sc
, connecting γ
s
′ with γ
s
, must coincide with γ (as γ is
a classical path in A, connecting γ
s
′ with γ
s
). There are, in general, other oneparameter
families of indirect paths,
s
′
β(σ), parametrized by s
′
, connecting γ(s
′
) with γ(s) not via ¯ γ
(e.g. bouncing oﬀ of a remote potential). Focusing ﬁrst on this direct contribution, and
using
I
γ
(γ
s
, γ
s
′ , s −s
′
) I
γ
(γ
s
′ , γ
0
, s
′
) = I
γ
(γ
s
, γ
0
, s) (40)
we get
φ(γ
s
, s) =
ǫC
2
e
iIγ(γs,γ
0
,s)/
¯
h
_
∞
−∞
ds
′
T
γ
(γ
s
, γ
s
′ ; s −s
′
) sign(s −s
′
)(ǫ; s −s
′
)
⇒ φ
r
(γ
s
, s) =
C
2
e
iIγ(γs,γ
0
,s)/
¯
h
_
R(s, ǫ) −
2
ǫ
_
, (41)
with
R(s) =
_
∞
−∞
ds
′
T
γ
(γ
s
, γ
s
′ ; s −s
′
) sign(s −s
′
)(ǫ; s −s
′
) (42)
some real functional of the EM ﬁeld and its ﬁrst derivative (its local neighborhood in an
exact analysis) on ¯ γ, such that lim
ǫ→0
[R(s, ǫ) −2/ǫ] is ﬁnite, implying that (27) is satisﬁed.
Moving next to the second reﬁned ECD equation, (26), and pushing ∂ into the integral
in (13),
¯
h∂φ(γ
s
, s) =
ǫC
2
e
iIγ(γs,γ
0
,s)/
¯
h
_
∞
−∞
ds
′
_
i∂
x
I
γ
(x, γ
s
′ ; s −s
′
)
¸
¸
x=γs
T
γ
(γ
s
, γ
s
′ ; s −s
′
) (43)
+
¯
h∂
x
T
γ
(x, γ
s
′ ; s −s
′
)
¸
¸
x=γs
_
sign(s −s
′
)(ǫ; s −s
′
) .
13
The second term in (43) drops in the limit
¯
h → 0. Using a relativistic variant of the
HamiltonJacobi theory (see appendix B in [5]), we can write
∂
x
I
γ
(γ
s
, γ
s
′ , s −s
′
) = p(s) ≡ ˙ γ
s
+ qA(γ
s
)
which is independent of s
′
. The ﬁrst term in (43) therefore gives
¯
h∂φ(γ
s
, s) = ip(s)φ(γ
s
, s) ⇒
¯
h∂φ
r
(γ
s
, s) = ip(s)φ
r
(γ
s
, s)
⇒ lim
ǫ→0
Re Dφ
r
(γ
s
, s)f
R
∗
(s) = −˙ γ
s
lim
ǫ→0
Im φ
r
(γ
s
, s)f
R
∗
(s) , (44)
which vanishes by (27), hence (26) is satisﬁed.
We return now to the contribution of the indirectpaths, β, in the sum over classical
paths, appearing in the deﬁnition of G
sc
. The phase of the corresponding integrand in (13)
reads
¯
h
−1
I
β
(γ
s
, γ
s
′ , s −s
′
) I
γ
(γ
s
′ , γ
0
, s
′
) . (45)
As distinct
s
′
β and
s
′
γ see diﬀerent potentials, and do not lie on the same massshell, (45)
does depend on s
′
— the ‘conspirative’ s
′
independence, manifested in (40), is a privilege of
β = γ. In the limit
¯
h → 0, the contributions of the indirect paths are therefore suppressed
by the strong oscillation of the phase (45). For a ﬁnite
¯
h, however, the indirect paths appear
as an explicit mechanism by which information about distant potentials is incorporated into
φ
r
(γ
s
, s), aﬀecting the dynamics of γ.
The regular current, j
r
, also takes a simple form in the classical limit. Using (44), we
have Dφ(γ
s
, s) = i ˙ γ
s
φ(γ
s
, s). Moreover, for x ,= γ
s
, the corresponding phase in the integrand
of (13) becomes inﬁnitely oscillatory, as the classical path connecting γ
s
′ with x, is distinct
from γ. It follows that the gauge invariant integrand in the deﬁnition of j, (11), formally
reduces to its classical counterpart appearing in (5), and j
r
is supported on ¯ γ. Using similar
arguments, the formal reduction of the ECD electric current to its classical counterpart, also
carries to all other ECD currents.
3 Qualitative discussion of ECD
The ECD formalism presented in the previous section has a rather unusual structure. Ex
plicit solutions, relevant to physically interesting cases, are diﬃcult to solve, apparently
necessitating an extensive use of numerical calculations. However, the stage is completely
set for such detailed analysis. Isolated, self consistent ECD solutions or bound states of
any number of them, can be sought, possibly (and most desirably in the author’s opinion)
showing that all elementary particles are just diﬀerent solutions of the same set of equations
(signiﬁcantly reducing the number of tunable constants). The eﬀective mass and binding
energies of such solutions can be computed using the expression for the energy momentum
tensor derived in the appendices; detailed internal structures of such particles can be an
alyzed, possibly suggesting novel methods of ‘cracking’ (or fusing) subatomic particles. In
short, one can potentially have a clear, scale covariant ontology, based on interacting ECD
14
particles alone, rendering additional forces and particles superﬂuous. ECD, then, is clearly
not merely an interpretation of QM, but rather a complementary theory with independent
testable predictions.
To motivate such an endeavor, this section argues the case for the compatibility of ECD
with current, well tested, physics. ECD, essentially retaining the ontology of classical elec
trodynamics, is apparently susceptible to the same objections and ‘no go theorems’ standing
in the way of other hidden variables models. In particular, since the EM ﬁeld is just the
classical Maxwellian ﬁeld, one may rightfully wonder where does the ‘photon’ come from?
We shall demonstrate to the contrary, that in a wide range of cases in which the statistical
predictions of QM clearly cannot be realized by an ensemble of classical solutions, the unique
features of ECD could render possible such a realization by an ensemble of ECD solutions,
resulting in a rather prosaic ‘demystiﬁcation’ of QM — the photon included.
For simplicity, only the scalar case is analyzed. The spin of an ECD particle merely labels
diﬀerent ways of covariantly obtaining extended currents — ordinary currents, transforming
as fourvectors. An example of spin
1
2
ECD is covered in appendix B.
3.1 Singlebody ECD
Singlebody ECD deals with the ECD equations of a single particle in the presence of an
external EM potential, where the use of the term ‘particle’, rather than charge, reﬂects
the possibility for an elementary ECD solution to have a vanishing monopole. Speciﬁcally,
we assume the existence of an external potential, A
ext
, satisfying Maxwell’s equations (4)
for some ﬁxed current, j
ext
, generated by the rest of the particles in the universe, assumed
independent of the particle in question. This is clearly a simpliﬁcation of the real situation
to which we return in section 3.2, dealing with manybody ECD. Next, we ‘feed’ A
ext
+A
sel
into the self consistent loop of section 2.2.3, solving φ in the presence of A
ext
+ A
sel
, and
close the loop by requiring from the self potential to satisfy
A
sel
µ
−∂
µ
(∂ A
sel
) = j
r µ
, (46)
with j
r
computed from φ and the combined potential A
ext
+ A
sel
.
In section 2.2.5 we saw that in the limit
¯
h → 0, all ECD currents formally converge to
their classical counterparts, becoming entirely supported on ¯ γ. In moving to a non vanishing
¯
h, the currents extend beyond the support of ¯ γ. To see more closely how this extended
support aﬀects the dynamics of an ECD charge in an external EM ﬁeld F
ext
, we ﬁrst need
the following. In appendix A, a relation, (90), is derived, formally identical to its classical
counterpart (10), relating the energy momentum tensor, m, associated with a particle, to its
conserved electric current j (omitting the regularization label
r
)
∂
ν
m
νµ
= F
µν
j
ν
≡ (F
ext
µν
+ F
sel
µν
) j
ν
, (47)
where F
sel
is the selfﬁeld derived from A
sel
via (46). Let Σ(s) be a oneparameter family of
non intersecting timelike surfaces, each intersecting ¯ γ at γ
s
, C a fourcylinder containing ¯ γ
15
and p
µ
(s) the corresponding fourmomenta
p
µ
=
_
Σ(s)∩C
dΣ
ν
m
νµ
, (48)
where dΣ is the Lorentz covariant directed surface element, orthogonal to Σ(s). Let also
C(s, δ) ∈ C be the volume enclosed between Σ(s) and Σ(s + δ), and T(s, δ) its spacelike
boundary (see ﬁgure 1 for a 1 + 1 counterpart).
❅
❅
❅
❅
❅
❅
❅
❅
❅
❅
ret.
adv.
Σ(s)
Σ(s + δ)
γ
T T
✲
✻
t
x
ﬁgure 1
Integrating (47) over C(s, δ), and applying Stoke’s theorem to the l.h.s., we get
p
µ
(s + δ) −p
µ
(s) +
_
T
dT
ν
m
νµ
=
_
C(s,δ)
d
4
x
_
F
ext
µν
+ F
sel
µν
_
j
ν
. (49)
with dT the outward pointing directed surface element on T. For a point charge with m and
j given by (10) and (5) resp., the term
_
T
dT
ν
m
νµ
vanishes, p = ˙ γ, and upon taking the limit
δ → 0 and dividing by δ, (49) formally becomes just the Lorentz force equation (1) with
F = F
ext
+ F
sel
. As noted before, however, the self force is illdeﬁned in the classical case
(hence the reservation implied in ‘formally’). In moving from a singular electric current to
an extended one, the ﬁrst beneﬁt is that now the selfforce appearing in (49) is well deﬁned.
For a static charge, for example, the only non vanishing component of the electric current is
j
0
(x) from which the purely electrostatic F
sel
inherits its spherical symmetry, leading to a
vanishing self force. The simplest complication of the static case, then, is when the currents
retain an approximate spherical symmetry (in the rest frame of γ) and F
sel
is approximately
a non radiating spherical electrostatic ﬁeld, contributing a negligible selfforce only. Under
this assumption we can write p = α ˙ γ with α some positive constant, and
lim
δ→0
δ
−1
_
C(s,δ)
d
4
xF
ext
µν
j
ν
= Q¸F
µν
¸
s
˙ γ
ν
, (50)
with Q =
_
Σ(s)
dΣ j the sindependent electric charge (here we assume that Σ(s) ∩ C
supports the lion’s share of the charge) and ¸F
µν
¸
s
is the average ﬁeld in Σ(s), weighted by
the normalized charge density. (The above equalities are most conveniently established in
16
the restframe of γ where j
0
and m
00
are the only non vanishing components, Σ(s) is taken
to be x
0
= γ
0
s
threespace, the Lorentz force density is purely electrostatic, and dT
0
= 0 ⇒
dT
ν
m
νµ
= 0.) Equation (49) then leads to
α¨ γ
µ
= Q¸F
µν
¸
s
˙ γ
ν
, (51)
and the constant α is identiﬁed with
_
p
2
/˙ γ
2
, where p
2
is the Lorentz invariant restenergy
of the charge. We therefore reach the important conclusion: Whenever an ECD charge
maintains an approximate spherical symmetry, its dynamics must be classical.
It is instructive to compare the above treatment of an ECD charge with Lorentz’s model
ing of the electron as a rigid, uniformly charged sphere, enabling him to obtain a well deﬁned
expression for the selfforce, without going through a ﬁshy massrenormalization procedure
(as in later treatments, preserving the point structure of the charge). As a relativistic rigid
extended body is a meaningless concept, Lorentz’s sphere model is valid at most for a suﬃ
ciently uniform motion — the larger the sphere, the more uniform the motion must be. A
rapidly varying external ﬁeld on the scale of the sphere therefore signals the breakdown of
Lorentz’s selfforce analysis. Likewise, a non uniformly moving ECD particle cannot main
tain an exactly spherical charge distribution in the rest frame of every point along γ, and
a rapidly varying γ on the scale of the ball holding the lion’s share of the charge, dubbed
the core, needs not even resemble a classical path. In this respect, ECD can be seen as a
fully covariant extension of Lorentz’s analysis of the selfforce. It is argued below that, in
principle, all of QM can be traced to the breakdown of the spherical core approximation.
Assuming ECD indeed governs the microscopic world, the above spherical core model
explains at once the reductions of the QED KleinNishina formula for the cross section in
Compton scattering, to the classical Thompson formula, at wavelengths greatly exceeding
the electron’s Compton length, and sets the Compton length as the order of magnitude of
the core. Indeed the Thompson formula is obtained by simple averaging over the radiation
produced by point charges oscillating in an external plane wave. For wavelengths much longer
than the scale of the core, we have in (51) ¸F¸
s
≈ F(γ
s
), and point dynamics is reproduced.
The spherical core approximation further accounts for another conspicuous coincidence —
the agreement between the nonrelativistic classical and quantum cross sections for Coulomb
scattering, both given by the independent Rutherford formula
5
. The fact that the Coulomb
potential (being an electrostatic potential) is a harmonic function, implies that the potential
of a spherically symmetric core in it, equals that of the center of the core, viz. that of γ,
hence no ﬁnitecoresize corrections to point dynamics are observed for this special potential.
3.1.1 The breakdown of the spherical core
Equation (47) and its integral form (49), somewhat artiﬁcially divide the change in the
momentum of a particle into a work of the Lorentz force, plus a ‘radiative’ contribution,
_
T
dT
ν
m
νµ
, of the associated em density m. A more symmetric treatment of ‘matter’ and
5
The author thanks Prof. Shimon Levit for sharing with him a long time ago his wonderment over this
point.
17
the EM ﬁeld is provided by the conservation of the ECD counterpart of (8), p = Θ+
k
k
m
(see appendix A.1) where Θ is the canonical EM energymomentum tensor (93). Applying
Stoke’s theorem to ∂p = 0, and using the same construction as in ﬁgure 1, we get
p
µ
(s + δ) −p
µ
(s) = −
_
T
dT
ν
p
νµ
, (52)
with
p
µ
=
_
Σ(s)∩C
dΣ
ν
p
νµ
, (53)
the total fourmomentum content of Σ(s) ∩ C. Although p
µν
is due to all particles in the
system, in the vicinity of a suﬃciently isolated particle k
′
, p is dominated by
k
′
m and the self
ﬁeld generated by
k
′
j. This all leads to the conclusion that the conservation of energy and
momentum associated with an isolated particle (EM em included) can only be breached by
an energymomentum ﬂux penetrating T. This ﬂux is composed of the classical Poynting
vector, plus a ‘quantum’ piece associated with
k
′
m (recall from section 2.2.5 that this latter
piece vanishes in the
¯
h → 0 limit, hence ‘quantum’). For a ﬁnite
¯
h, however, this ‘electro
weak’ division is entirely artiﬁcial, as
k
′
m also depends on A both explicitly and implicitly
(via φ). Moreover, whenever the core breaks down, producing such matter em ﬂux over
T, a corresponding
¯
hdependent electric ﬂux also forms, which locally modiﬁes the Pointing
ﬂux (note that this ‘radiative component’ of j may be negligible in terms of charge capacity,
and still generate a strong EM ﬁeld if it strongly ﬂuctuates).
Summarizing our ﬁndings regarding a suﬃciently isolated particle, the integral of the em
ﬂux,
_
T
dT
ν
p
νµ
, does not vanish only if γ is non uniform. This, of course, is a standard result
of classical electrodynamics, the quantiﬁcation of which leads to the celebrated Lorentz
Dirac equation (but not before ‘renormalizing’ the mass of the charge — a rather contrived
procedure whose sole motivation is to render the result non trivial), and can be directly
derived from the the expression for the LienardWiechert potential generated by a moving
charge
Aret
adv
(x) = q
_
ds δ
_
(x −γ
s
)
2
¸
˙ γ
s
θ
_
x
0
∓γ
0
s
_
. (54)
The advanced and retarded potentials are the traces of densities, δ
_
(x −γ
s
)
2
¸
˙ γ
s
θ (x
0
∓γ
0
s
),
supported on the lightcone of γ
s
(schematically depicted by adv. and ret. in ﬁgure 1).
These ﬁelds are then further processed, yielding a Θ which can easily be shown to contain
a radiative component (viz. dropping as r
−2
from the charge) at a point x only if γ is non
uniformly moving in the neighborhood of points lying on the intersections of the lightcone
of x with ¯ γ.
The ECD counterpart of (54) can be expected to read
Aret
adv
(x) = q
_
ds
__
d
4
y δ
_
(x −y)
2
¸
θ
_
x
0
∓y
0
_
lim
ǫ→0
J
r
(s, y)
_
, (55)
with J
r
= Im φ
∗
Dφ removed of bilinears in φ
s
, and the ǫ → 0 limit is to be understood in the
distributional sense. Indeed, A appearing in (55) solves Maxwell’s equations (4) with j
r
as a
18
source, and upon plugging in (55) the corresponding classical density J
r
(s, y) → δ
(4)
(y−γ
s
) ˙ γ
s
,
(54) is reproduced. Equation (55), however, ignores a crucial diﬀerences between the above
densities. Unlike its classical counterpart, J
r
depends on A. Equation (55), unlike (54), is
therefore not a prescription for Aret
adv
but rather an equation for it. A solution in which only
A
adv
or A
ret
enter J
r
may not (and probably does not) exist. The ‘correct’ radiation ﬁeld,
containing both advanced and retarded components, so to speak, is therefore dictated by the
speciﬁcs of the radiation process. In classical electrodynamics, on the other hand, a solution
of Maxwell’s equation (4) is deﬁned only up to a solution of the homogeneous equation
∂
ν
F
νµ
= 0. For this reason, the motivation for the (almost universal) choice of the retarded
potential is not in the equations proper, but rather in the desire to conform with observations
concerning large scale radiation phenomena, involving huge numbers of particles (See more
in section 3.2.2). We are led, then, to the important conclusion that in ECD, advanced EM
ﬂux, combined with a corresponding advanced mechanical ﬂux associated with m, must be
considered. We shall see in 3.2.1 apparent experimental signatures left by such advanced
eﬀects.
3.1.2 The aether and its manifestations
The picture emerging from the previous section is that of a single conserved em ﬁeld,
p
µν
(x), organically integrating, or fusing, radiation and matter. The distribution of p in
M is highly nonuniform, with the lion’s share of the charges concentrated in small three
cylinders centered around
k
¯ γ (and integrably singular there; see appendix A). Timelike cross
sections of these tubes were dubbed ‘cores’ of particles. Amidst the densely charged three
cylinders resides a (mostly) weak, locally conserved p, smoothly merging with the dense
cylinders, reducing to Θ in the limit
¯
h → 0. This interparticle p which, in the absence
of a better name shall be referred to as the aether, cannot be decisively attributed to any
single particle as A, entering every term in p, is generated by the combined current of all
the particles. Nevertheless, as we have seen, its dynamics at a point is strongly correlated
with that of neighboring particles.
Consider now a scattering experiment in which a neutral particle passes through a small
aperture in a screen. As only the collective p — cores plus aether — is conserved, there is
a mutual inﬂuence between the particle and the (cores composing the) screen, mediated by
the aether. As noted above, since the particles are assumed neutral, in the limit
¯
h → 0 the
ether vanishes, and the particle does not suﬀer any deviation from a straight classical path.
But for a small value of
¯
h, small ‘quantum corrections’ to classical paths are expected, and it
is only via the ampliﬁcation brought about by the huge distance between the target and the
detection screen, that minute corrections to the classical cross section are detectable. One
can also understand why attempts to measure the position of the particle during its ﬂight, by
shining light on it, destroy the delicate interference pattern. This is because the external EM
ﬁeld applied to the particle, interacts with the core. This is essentially a classical interaction,
coming with a variability which greatly surpasses the feeble quantum corrections to classical
paths, responsible for the fringes in the ﬁrst place.
It is conjectured that gravity is yet another manifestation of such ether mediated interac
19
tion between remote cores, ampliﬁed this time by huge numbers/mass rather than by a huge
distance. In particular, the bending of the trajectory of an EM wavepacket in a gravita
tional ﬁeld, can bee seen as a perturbation to a nonuniform static aether conﬁguration (just
like mechanical waves bend in a nonuniform medium). In the context of ECD, therefore,
gravitational theories such as general relativity (GR), only play a role similar to that of QM,
as an eﬀective statistical theory, respecting all the symmetries of ECD, but applicable to
totally diﬀerent experimental settings.
Let us brieﬂy see what such a statistical theory must look like. Scale covariance naturally
suggests we represent the cores by mathematical points, obviously invariant under scaling
of spacetime. This, as we know, creates a self force problem similar to that plaguing
classical electrodynamics. GR, therefore, cannot be considered a completely satisfactory
theory and, as we saw in the case of ECD, the resolution of this problem may result in
radical changes to GR, not merely minute correction. Self force problem aside, we can
appreciate why GR is a reasonable candidate. The em tensor derived from the metric,
is locally conserved, transforming in a scale transformation as p, and carries ‘em waves’
generated by non uniformly moving bodies – much like p.
3.1.3 Interferometers
In the previous section we mentioned two ways of amplifying the small aetherinduced devi
ations of the cores from classical paths: Huge distances and huge numbers/mass. A distinct
third way, implemented in neutron interferometers, relies on the ability of chaotic systems
to amplify small perturbations. In a MachZehnder conﬁguration, (a), the beamsplitters
(BS) and mirrors are crystals of macroscopic thickness, forming a huge lattice of scatterers
in which a particle undergoes multiple scatterings before exiting.
BS1
✁
✁
✁
✁
✁
✁
✁
✁
✁
γ
✁
✁✕
(a)
BS2
❆
❆
❆
❆
❆
❆
✁
✁
✁
BS1
✁
✁
✁
✁
✁
✁
✁
✁
✁
γ
✁
✁✕
(b)
❆
❆
❆
❆
❆
❆
❆
❆
✧✦
★✥
R
BS1
✁
✁
✁
✁
✁
✁
✁
✁
✁
✁
✁✕
(c)
❆
❆
❆
❆
❆
✁
✁
✁
✧✦
★✥
R
Even at the classical level, the dynamics in such a maze is highly chaotic, meaning, in
particular, that the standard procedure of averaging over the impact parameter in order to
obtain the scattering cross section, is utterly meaningless. There is no classical cross section
when the target supports chaotic dynamics, and the call for a complementary statistical theory
for ECD is rooted already in classical dynamics.
To each scattering event in the crystals there corresponds a radiative perturbation to the
20
aether, containing both advanced and retarded components. These ‘aether waves’ propagate
inside and around the interferometer, slightly perturbing the (locally almost) classical path
of the cores, in a way which globally depends on the conﬁguration of the interferometer. As
the dynamics of the cores inside the crystals is chaotic, the aether excitations, their small
local eﬀect notwithstanding, have a dramatic eﬀect on the ﬁnal scattering direction of the
particle.
The chaoticity of the underlaying classical dynamics is crucial for the operation of the
interferometer. Suppose we remove BS2 from the apparatus (b). The inﬂuence of the aether
excitations on the dynamics of a particle passing in region R is now negligible, and the
particle continues its straight classical path, almost unperturbed, as follows from momentum
conservation. This should be contrasted with (c), ‘surrealistic’ trajectories predicted by
Bohmian mechanics, taking the other direction [2].
We have focused our discussion on a crystal BS as the arena for this chaotic dynamics but,
in fact, it is not chaoticity itself — a classical notion — which is essential for the operation
of the interferometer, but rather the sensitivity of chaotic dynamics to perturbations. All
interferometers, whether electronic or atomic, use BS’s in the form of highly sensitive devices
(usually involving the spin of the particle) facilitating the ampliﬁcation of the small aether
induced perturbations to the core’s dynamics.
3.1.4 The ensemble current
The above description of interferometers invites a troubling question. As is well known,
interferometers can be tuned to produce nearly deterministic results, with one detector ﬁring
some 99% of the times and the other only 1%. If the local dynamics of the particles are
so nearly classical, viz. locally deﬁned, then how do they acquire this destiny, of arriving
predominantly at one detector rather than the other? (or exit the crystals at the Bragg
angles only?)
To answer this question, we ﬁrst need to see what a scattering experiment is, in the
context of ECD. Let j ∈ c be the regular electric current associated with a solution realized
in the experiment (we drop the
r
superscript in this section), and c the ensemble of all such
currents. An experiment is seen as a realization of a measure dµ(j) deﬁned on c, namely, we
assume that as the number, n, of scattered particles goes to inﬁnity, the number of solutions
realized in any subset Σ ∈ c approaches nµ(Σ) ≡ n
_
Σ
dµ(j). The reader can verify that the
scattering cross section as well as any other measurable statistical expression produced by
singlebody QM, such as the spectrum of atoms, can be read from an ordinary, conserved,
21
fourcurrent — the ensemble current
6
,
j
ens
=
_
E
dµ(j) j . (57)
Stated in the above terminology, then, the question of interest is why does this current
have such an asymmetric form? The answer to the question does not lie in fourdimensional
Minkowski’s spacetime, on which j
ens
is deﬁned, but rather in inﬁnitedimensional c, the
domain of µ. A single ‘point’ in c — a current j — is such a complex, non locally deﬁned
object, that we lack any intuition regarding sensible distributions thereof. Why is 99% —
1% less intuitive than 50% — 50%? Likewise, why is the nonuniform shape of the Hydrogen
atom spectrum counter intuitive? Note that in both cases, no classsical counter proposal even
exists. In the scatterng case, the standard procedure of averaging over the impact parameter
leads to a meaningless result when applied to chaotic systems. As to the spectrum — a
classical Hydrogen atom is a meaningless concept to begin with.
In fact, the measure µ should be regarded as an independent law of nature, on equal
footing with ECD itself, constrained only by compatibility requirements with ECD and the
experimental settings (try thinking what would constitute a natural µ?). For this reason,
QM enjoys a similar status of an independent law.
3.1.5 Relativistic wave equations
Singleparticle QM, as argued above, describes very coarse aspects of the measure µ — very
‘low order moments’ of that inﬁnite dimensional distribution. It should not come as too great
a surprise that, assuming ECD is indeed the physics prevailing at the atomic scale, QM could
have been anticipated independently of ECD, with the latter’s very unique content. We shall
next show why relativistic wave equations, such as the ﬁrst or second order Dirac equation,
and the KleinGordon equation, are a natural tool for guessing those moments, in certain
cases of a single particle moving in an external ﬁeld.
6
The diﬀerential scattering crosssection to a given solid angle dΩ around Ω, for example, is easily de
ducible from the ensemble current (57). It is just
1
QdΩ
lim
x
0
→∞
_
C
d
3
xj
ens
0
, (56)
with Q =
_
d
3
xj
0
the conserved charge of the particle, and C = C(dΩ, Ω) the cone in three space deﬁned
by dΩ and Ω. This follows upon inserting expression (57) into (56). In the limit x
0
→ ∞, every j
0
(x
0
, x) is
entirely supported in C, or in its complement. The x integration then extracts Qχ
Σ
(j) with Σ = Σ(C) ∈ c
the subset of solutions scattering to cone C, and χ
Σ
() its characteristic function. The result is therefore
(dΩ)
−1
_
E
dµ(j) χ
Σ
(j) ≡ (dΩ)
−1
µ(Σ) which is the deﬁnition of the diﬀerential crosssection.
As yet another example, consider the EM spectrum emitted by a heated gas. For a suﬃciently dilute gas,
the currents associated with the bound electrons (those generating the radiation) can safely be assumed to
constitute an incoherent ensemble c. By the linearity of Maxwell’s equation, and the incoherence assumption,
the spectrum produced by the ensemble current equals the sum of spectra produced by the individual currents
in c. Equivalently, c can comprise diﬀerent, suﬃciently remote time segments, of the current associated with
a single atom. The spectral peaks, then, appear simply as dominant frequencies in the dipole radiation of
j
ens
, representing statistically more common frequencies in the dipole radiation of members in the ensemble.
22
Consider, then, the ECD solution of a single particle in the external ﬁeld F
ext
. Let this
solution be indexed by the electric current j, associated with the particle, and let
j
m be the
corresponding em tensor. From (47) we have
∂
ν
j
m
νµ
=
_
F
ext
µν
+
j
F
sel
µν
_
j
ν
, (58)
with
j
F
sel
the selfﬁeld generated by j,
∂
ν
j
F
sel
νµ
= j
µ
. (59)
Multiplying (58) by dµ(j) and integrating over c, we ﬁrst make the assumption that the
contribution of the selfﬁelds,
_
E
dµ(j)
j
F
sel
µν
(x)j
ν
(x) , (60)
can be neglected, compared with that of the external ﬁeld. This is a reasonable assumption
for a suﬃciently incoherent ensemble, as then the self contribution of diﬀerent members in
the ensemble, to the self force at a point x, enters with a nearly random orientation. For this
to happen, however, the diﬀerent charges must not radiate (advanced or retarded ﬁelds) in
preferred directions, hence the limitation of the ensemble current approach and of relativistic
wave equations in particular (see more in section 3.2.1). Note, nonetheless, that the self
ﬁelds are dominant in the individual currents j and
j
m, even when their contributions to
the integral over the ensemble have been neglected, guaranteeing that selfforce eﬀects are
not eliminated in that process. In particular, the eﬀective mass and charge of the particles,
strongly depend on that self ﬁeld.
With the above approximation, we get the following four relations
F
ext
µ
ν
j
ens
ν
= ∂
ν
m
ens
νµ
, with m
ens
=
_
E
dµ(j)
j
m, (61)
and a conservation constraint
∂
ν
j
ens
ν
= 0 , (62)
inherited from the conservation of the individual j. The Lorentz vector and second rank
tensor, j
ens
and m
ens
resp., must obviously transform like their constituents in any symmetry
transformation belonging to the symmetry group of ECD.
Consider now a low energy scattering experiment. As shown above, the scattering cross
section can be computed from j
ens
. However, a similar construction applied to m
ens
can also
produce the cross section, which must coincide with that computed using j
ens
. This relation
adds up to (61), (62) and the symmetry group of ECD, producing a very restrictive condition
on the set of permissible pairs ¦j
ens
, m
ens
¦, regardless of the details of the ECD dynamics.
A systematic way of producing such constrained pairs, enjoying the full symmetry group
23
of ECD, is via relativistic wave equations
7
. In the scalar case, the relevant equation is the
KleinGordon equation
_
D
2
+ ˆ m
2
_
ψ = 0 , (65)
with the gauge covariant derivative
D =
ˆ
h∂ −iˆ qA, (66)
where
ˆ
h, ˆ q and ˆ m are some constants, and A the external EM potential. The expressions for
the ensemble electric current
j
ens
µ
= ˆ q Imψ
∗
D
µ
ψ , (67)
and the ensemble em tensor
m
ens
νµ
= g
νµ
_
1
2
ˆ m
2
ψψ
∗
−
1
2
_
D
λ
ψ
_
∗
D
λ
ψ
_
+
1
2
_
D
ν
ψ (D
µ
ψ)
∗
+ c.c.
_
, (68)
satisfy all the above compatibility conditions — eq. (61) in particular. Eq. (61), when
restricted to a ﬁeldfree region, imposes certain relations between the parameters of (65),
and the conserved electric charge and mass of the particles comprising the ensemble.
The wavefunction ψ, then, labels an ‘irreducible ensembles’, µ
ψ
, to which there corre
sponds an ‘irreducible pair’, ¦j
ens
, m
ens
¦
ψ
. A generic experiment, however, involves a few
irreducible ensembles, which are sampled with diﬀerent weights. This is the meaning of a
‘statistical mixture’ of wavefunctions in QM. The collapse postulate of measurement the
ory, merely represents a transition from one ensemble to another. For example, to a beam
of particles escaping a hot oven there correspond one ensemble. When the beam is further
split into two by a (e.g. SternGerlach) polarizer, each part must obviously be represented
by a diﬀerent ensemble, when scattered oﬀ of a spin sensitive target. If, however, the two
parts of the original beam are recombined in an interferometer, the original ensemble is the
relevant one.
A major historical diﬃculty associated with the KG ﬁeld is also resolved in this frame
work. The nonpositivity of j
ens
0
(motivating the Dirac equation) simply reﬂects the non
positivity of the individual j
0
comprising j
ens
. It is only the space integral over those indi
vidual components, representing the total charge, that is guaranteed to remain constant.
7
The more general way is via the conservation laws associated with solutions, ψ, of the ﬁve dimensional
Schr¨odinger equation (15). Its unitarity implies
∂
s
[ψ[
2
= ∂ J ≡ ∂ (Im ψ
∗
Dψ) , (63)
while the Ehrenfest relations give
∂
s
J
µ
= F
µν
J
ν
−∂
ν
M
νµ
(64)
≡ F
µν
J
ν
−∂
ν
_
g
νµ
_
i
¯
h
2
(ψ
∗
∂
s
ψ −∂
s
ψ
∗
ψ) −
1
2
_
D
λ
ψ
_
∗
D
λ
ψ
_
+
1
2
_
D
ν
ψ (D
µ
ψ)
∗
+ c.c.
_
_
Integrating (63) and (64) from s = −∞ to s = ∞, we get two candidates, j
ens
=
_
dsJ and m
ens
=
_
dsM,
satisfying all our requirements. These, however, correspond to ensembles with a continuum of masses, and
are therefore more diﬃcult to relate to actual experiments, involving a single particle species.
24
3.2 Manybody ECD
In the previous section, the EM potential was divided into an external potential, generated
by all particles but one, plus a self potential, due entirely to this one, privileged, particle.
This division is legitimate on the premise that the self potential of the privileged particle does
not alter the solutions of the rest of the particles, which is not the case when the privileged
particle interacts with the rest of the particles, either ‘electrostatically’, viz. at close range,
or ‘radiatively’, via longrange aether waves.
Let us begin with the ﬁrst case. As explained in section 2.2.3, the self consistent potential
entangles closely interacting particles in such a way that one can no longer regard matter as
a composition of individual particles but, instead, as some ‘self consistent matterradiation
condensate’. What may seem surprising at ﬁrst is that long after their separation, and at
arbitrarily remote locations, two particles which have closely interacted in the past, ‘bear
the memory’ of their encounter.
8
Consider, for example, two nucleons, escaping a nucleus, arriving each at a polarimeter
(P1 and P2).
✍✌
✎☞
P1
✍✌
✎☞
P2
✲
✻
t
x
If the two polarimeters are positioned suﬃciently far apart, then the ECD system, self
potential included, can be solved independently for each particle. This is a consequence
of lim
s→∞
G(x, x
′
, s) = 0, suppressing the large [s − s
′
[ contribution to the s
′
integral in
8
As entanglement is transitive, all the particles in a dense cloud are entangled, and should be solved as a
single space time structure. If a large scale astronomical object, such as a galaxy, passes through an epoch of
a dense ‘ﬁreball’ (not necessarily anything as dramatic as a big bang or big crunch — whatever that means)
it could reasonably be that its morphology and dynamics at much latter stages, reﬂect that entangled epoch.
As it is also reasonable to assume that galaxies indeed pass through some ﬁreball epoch, gravitation on
galactic scales (and possibly above) could therefore be due to such a primordial entanglement, and need not
even resemble solar scale gravity (the largest scale in which general relativity has been directly conﬁrmed).
This would have obvious implications on the current interpretation of astronomical data. In this regard, we
should also notice another possibility opened by ECD — scale drift. As shown in appendix A, both the mass
of individual ECD particles, as well as the scale charge of their combined solution, may slowly drift over
time. This oﬀers an alternative explanation for the source of galactic redshifts. In fact, a universe collectively
increasing its scale leads to a Hubblelike relation, as light collected from remote galaxies is emitted at an
epoch of lower mass (hence longer wavelength) which is proportional to the distance of the emitter to the
observer, for all observers.
25
(13). There is therefore nothing unique about the dynamics of each particle, giving away
their histories, which is washed away over macroscopic scales. The above independence
notwithstanding, if one tries to continue those independent solutions into the past, then at
some point, when the two particles come close, it could become impossible to combine the
two solutions into a single, self consistent one, unless each of the two independent solutions
is restricted to a certain subset of the full set of independent solutions — a subset which
obviously depends on the orientation of both polarimeters. Given the orientations of the two
polarimeters, therefore, the combined solution of the two particles in the above example,
must be solved as a whole — as a single spacetime structure.
To each orientation choice, p
1
, p
2
, for P1 and P2 resp., there corresponds a diﬀerent
ensemble, c, of twoparticle ECD solutions, equipped with its own measure, µ. This con
tradicts Bell’s assumption in deriving his celebrated inequalities, which maintains that the
same ensemble of hidden variables be used, irrespective of the orientations of the polarimeters
(roughly corresponding to the fact that the particles should not ‘anticipate’ the orientations
of the polarimeters before encountering them).
9
Like its singlebody counterpart, the measure, µ, enjoys the status of an independent
law of nature, on equal footing with ECD itself. However, a simple generalization of the
ensemble current to the case of manybody ECD, probably doesn’t exist, hence the enormous
complication of manybody relativistic QM — quantum ﬁeld theory — involving both matter
and the EM potential.
Finally, let us note that the same discussion holds also for two particles, initially sep
arated, which later bind together. This scenario, however, does not correspond to known
experiments.
3.2.1 The conspiracy of the photon
Perhaps the strongest motivation for the introduction of photons, is the salvation of energy
momentum conservation. Indeed, the photoelectric and Compton’s eﬀects are manifestly in
violation of classical energymomentum conservation. More speciﬁcally, equation (52), ex
pressing the change in the fourmomentum of a particle as a function of the integrated em
ﬂux across a spacelike surface surrounding the particle, can formally be applied to the cor
responding classical currents as well. In the case of the photoelectric or Compton’s eﬀects,
the em ﬂux across T, identical with the Poynting vector, is computed from the external
trigger, F
ext
, and a possible retarded outgoing wave, generated in the jolting of the charge.
As both eﬀects are observed even for extremely feeble triggers, the contribution of F
ext
to
the Poynting vector may be neglected, while that of the retarded wave can be shown to be
positive. We may then get an arbitrarily large energy deﬁcit at times following the jolting of
the charge. In ECD, on the other hand, we saw that advanced em waves must be included
in the analysis. Thus, for example, ‘photon absorption’ by a molecule, should correspond to
predominantly incoming advanced em waves, converging on the molecule and increasing its
internal energy (or ionizing it as in the photoelectric eﬀect). In ‘spontaneous emission’, it is
9
The above mechanism, accounting for violations of Bell’s inequalities, is in the spirit of so called ‘retro
causal’ models. See, e.g. [1] and extensive references therein.
26
rather outgoing retarded waves removing energy from the molecule. In other situations —
Compton scattering for example — both waves play an equally important role.
❅
❅
❅
❅
❅
❅
❅
❅
❅
❅
❅
❅
✁
✁
✁
✁
✁
✁
❅
❅
❅
❅
❅
❅
❅
❅
❅
❅
❅
❅
✲
✻
t
x
absorption emission scattering
The above two features, viz. automatic selection of the correct radiation ﬁeld (guaranteeing
energymomentum conservation), along with its incorporation into the dynamics of the par
ticle (no selfforce problem), are missing from classical electrodynamics, hence the need for
ECD to implement this, otherwise, classical idea.
Being highly nonlocal and nonlinear, there are plenty of pairs ¦φ, γ¦, solving the ECD
system (self potential included) for a given external trigger. The distribution of the cor
responding currents can be read from the appropriate ensemblecurrent (see section 3.1.4)
which, in the case of the photoelectric eﬀect, gives the well known
10
result that the electron
either jolts with energy, λω, proportional to the frequency, ω, of the incident radiation (ig
noring for simplicity the binding energy), or else does not jolt at all. This binary response
of electrons, typical of all ‘photodetectors’ by deﬁnition (or else they are called calorimeters,
antennas, etc.), is the historical reason for the introduction of photons. It is as if a ‘light
corpuscle’ of energy λω has struck the jolted electron. Yet another standard result emerg
ing from the analysis of the ensemble current, is that the probability for a jolting event is
proportional to the amplitude squared of the incident wave, implying that the probability
drops as the inverse of the distance squared between source and detector — just as if a ﬂux
of particles in erupting from the emitter.
But the analogy with other particles goes even further. A typical example involves a
so called ‘singlephoton source’ or more generally an nphoton source (Fock state source),
e.g. a molecule excited by a femtoseconds laser pulse, and then allowed to spontaneously
decay. If the source is surrounded by a large sphere, consisting of independently operating
10
This calculation is usually preformed with the nonrelativistic Schr¨odinger equation, considering the
incident wave as a small perturbation. However, for wavelength much smaller then the electron’s Compton
length, the Dirac equation gives identical results.
The success of the ensemble current formalism in the case of the photoelectric eﬀect, is due to the isotropic
distribution of the direction of the ejected particle, hence also of the corresponding self force, justifying the
omission of the term (60). In contrast, this formalism fails when applied to Compton scattering, i.e. an
external EM trigger in the form a plane wave, but without a heavy trap holding the particle. Momentum
conservation — ignored in the photoelectric eﬀect due to the large mass of the trap — dictates that the
direction of the ejected charges, must be strongly correlated with that of the incident wave, and (60) cannot
be neglected.
27
photodetectors (which can further be prevented from crosstalking by, e.g. partitions) then
the above results of the ensemble current, imply that the average number of photodetections
does not depend on the radius of the sphere, and is entirely an attribute of the source (note
that, as the expectation value is additive even for dependent random variables, this result is
not altered when, latter, we argue that the photodetectors are not independently operating).
This, again, is consistent with a scenario of a release of a ﬁxed number of particles in each
decay of the molecule. However, the independence of the diﬀerent photodetectors also imply
that the number of detected photons should ﬂuctuate around its mead with a standard
deviation proportional to the squareroot of the mean. For a large mean, this ﬂuctuation
may be ignored, but for a small one, it is signiﬁcantly greater than the observed value which
is more consistent with a particle scenario having zero mean.
As implied above, the loophole in the analysis is in the assumption of independence of the
photodetections. While it is possible to prevent diﬀerent photodetectors from crosstalking,
it is, by deﬁnition, impossible to prevent each of them from cross talking with the source
if advanced waves are present in the radiation ﬁelds of the absorbing charges
11
. In actual
experiments, e.g. [6], the retarded ﬁeld of the source is relayed to the detecting charges
by other charges, comprising mirrors, beamsplitters, ﬁberoptics etc. The crucial point is
that, whatever optical path exists between the source and the detector, by means of retarded
ﬁelds, there must necessarily exist a reverse path leading from the detector to the source via
advanced ﬁelds. The source therefore serves as a hub for indirect crosstalking between the
absorbing charges, leading to statistical dependence in their responses. As to why the actual
ﬂuctuation around the mean is much smaller, rather than larger, than that expected on the
premise of independence — this is a statistical eﬀect, not to be sought in ECD alone. This
is the realm of QM — QED to be speciﬁc. Violations of Bell’s inequalities in photons pair
measurements etc., are presumably all manifestations of that indirect crosstalking.
source absorber 1 absorber 2
✆
✆
✆
✆
✆
✆
✆
✆
✆
✆
✆
✆
✆
✆
✆
✆
❊
❊
❊
❊
❅
❅
❅
❅
❅
❅ ■
❅
❅
❅
❅
❅
❅❘
✒
✠
✲
✻
t
x
We see how various features of ECD have conspired to bring about the illusion that
‘light particles’ must be involved in radiation processes. The real moral, however, lies in
11
The use of advanced solutions in order to explain the non classical statistics of photons, latter receiving
the name ‘the transactional interpretation of QM”, is described in [3]. Using point charges, however, that
proposal does not explicitly deal with energymomentum balance, nor with the mechanism causing a charge
to jolt.
28
the geometry of Minkowski’s space and in the unity of the em ﬁeld p. We have previously
argued that the self consistency loop entangles, in the statistical sense, two particles whose
associated worldcylinders, supporting the lions share of their charges, have a signiﬁcant
overlap in M. This can be seen as a manifestation of the fact that, fundamentally, the
value of p at a point cannot be attributed to any single particle, not even inside the dense
cylinders associated with the particle. The conclusion deduced from that example generalizes
to the observation that any connected volume in M, of a suﬃciently high em density must
be treated as a single spacetime structure. In particular, the following densely charged
connected structure, is typical of all emitterabsorber ‘transactions’.
B
✲
✻
t
x
The term ‘transaction’ is deliberately borrowed from [3] as it highlights the symmetric role
played by both charges appearing in the structure, viz. the absorber may just as well be seen
as the cause, triggering the emission via advanced waves, rather than the eﬀect, triggered
by the retarded waves of the emitter. Note that in ECD this blurring between cause and
eﬀect goes even further than in [3], as p, in particular on the ‘bridge’, B, between the two
particles, cannot be decomposed into advanced plus retarded contributions, and is therefore
a genuine attribute of the structure as a whole.
3.2.2 Advanced waves
The central role played by advanced waves in explaining the illusion of a photon, calls
for a closer look at these disputable objects. There is a strong, largely unjustiﬁed, bias
against the inclusion of advanced waves — advanced solutions of Maxwell’s equations in
particular — into the description of physical reality. The main objection draws parallels
with ‘contrived’ advanced solutions of other physical wave equations (e.g. surface waves in a
pond converging on a point and ejecting a pebble.). This parallelism, however, is a blatant
repetition of the historical mistake, which led to the invention of the aether (the historical
29
aether, not to be confused with that used in this paper). The formal mathematical similarity
between the d’Alembertian (the only Lorentz invariant secondorder diﬀerential operator)
and other (suitably scaled) wave operators, is no more than a misfortunate coincident (had
this coincidence had some real substance to it, then application of the Lorentz transformation
to the wave equation describing the propagation of sound, for example, would have yielded
a meaningful result).
Yet another argument against advanced solutions is their alleged involvement in causal
paradoxes.
❇
❇
❇
❇
❇
❇
❇
❇
❇
❇
❇
❇
❇
❇
❇
❇
relay bomb
❅
❅
❅
❅
B
S
R
✲
✻
t
x
Indeed, if advanced waves could be generated just like their retarded counterparts, then the
following paradoxical situation could occur. A device consisting of a bomb, a transmitter, a
receiver and a timer, is set to send a retarded signal at S. The signal is relayed at R, received
at B, and triggers the fuse of the bomb. But if the bomb goes oﬀ at B, then no signal is
sent at S. Why then did the bomb explode? On the other hand, if the bomb doe not go
oﬀ at B, then a signal must be sent at S, detonating the bomb at B. Either way we get a
contradiction.
The resolution of the paradox should not be sought in ECD proper. Indeed, if ECD is a
valid theory, then the CPT image of a radio transmitter sending retarded waves, is a radio
transmitter made of antimatter sending advanced waves. A radio transmitter, however, can
not be seen as an autonomous entity. Its generated waves are eventually absorbed by other
particles and, as argued above, the emission of waves cannot be separated from their ab
sorption (as in the WheelerFeynman absorber theory,[7]). The privileged status of retarded
waves in all macroscopic radiation processes, is therefore an attribute of the speciﬁc solution
of the ECD equations (selected, among else, by the anthropic principal), representing the
local part of the universe we live in, and is intimately connected with the excess of matter
over antimatter around us.
It seems, then, that the strongest case against advanced solutions is observational. While
spontaneous emission or absorption may be seen as direct evidences to the contrary, if
advanced solutions played a dominant role in any photoabsorption (as implied e.g. in
30
the ﬁgure on page 27) then their prevalence should have matched that of their retarded
counterparts, leaving a striking signature on all radiation processes. Let us then see why this
needs not be the case. Recall that our primary motivation for introducing advanced solutions
was to salvage energymomentum conservation. In the example from the previous section of
a single photon source surrounded by photodetectors, the integrated ﬂux of energy falling on
any single photodetector, must be smaller than λω (assuming the detectors are suﬃciently
far from the source). Advanced waves must then be invoked to account for the ﬁring of a
photodetector, as this amounts to increasing the energy of an electron by λω. However, if
the single photon source is replaced by a continuous light source of arbitrary intensity, then
the following process, not involving advanced waves, can be envisaged. Retarded waves,
originating from the source, arrive at the photodetector, which slowly absorbs them. In the
process of absorption the electrons in the device radiate in the same direction as that of
the incident radiation, but in a phase which interferes destructively with the latter, slowly
‘removing’ energy from the incident wave. This is essentially the classical description of
radiation absorption, only in ECD, the energy extracted from the incident wave needs not
appear instantly as kinetic energy in the absorbing charges. The extended support of the
ECD energy density, in conjunction with its dynamical evolution, support a scenario in which
energy is gradually accumulated by the charge in the form of latent ‘internal’ energy, and is
rapidly converted into kinetic energy only when a threshold, equal to λω, has been crossed.
That the conversion of latent energy into kinetic energy happens at the λω threshold can,
again, be read from the ensemble current which, as remarked before, is indiﬀerent to the
mechanism shooting the individual electrons.
Remarkably, it is known that the statistics of photodetection also changes when shifting
to a continuous source. When the readings of two photodetectors are correlated (as in
[6]), the anticorrelation consistent with a particle scenario, turns into the expected positive
correlation when the single photon source is replaced by a continuous light source of thermal
origin, or to (the equally intuitive) vanishing correlation when strongly attenuated laser light
is used. It appears, therefore, that advanced waves play a dominant role only in suﬃciently
‘delicate’ radiation process, involving energy transfer on the order of λω. Such processes are
overwhelmed by ordinary radiation processes involving a huge number of particles, such as
the burning of a candle, or in lasing devices.
Acknowledgments
The author wishes to thank Irad Yavneh for his useful comments.
Appendices
A Conservation of ECD currents
To prove the conservation of the regular current, j
r
, deﬁned in (33), we ﬁrst need the following
lemma, whose proof is obtained by direct computation.
31
Lemma. Let f(x, s) and g(x, s) be any (not necessarily square integrable) two solutions of
the homogeneous Schr¨ odinger equation (15), then
∂
∂s
(fg
∗
) = ∂
µ
_
i
2
_
D
µ
fg
∗
−(D
µ
g)
∗
f
_
_
. (69)
This lemma is just a diﬀerential manifestation of unitarity of the Schr¨ odinger evolution—
hence the divergence.
Turning now to equation (13), written for the rescaled wavefunction ǫ
−1
φ
φ(x, s) = −2π
2
¯
h
2
i
_
∞
−∞
ds
′
G(x, γ
s
′ ; s −s
′
)f
R
(s
′
)(ǫ; s −s
′
) , (70)
and its complex conjugate,
φ
∗
(x, s) = 2π
2
¯
h
2
i
_
∞
−∞
ds
′′
G
∗
(x, γ
s
′ ; s −s
′′
)f
R
∗
(s
′′
)(ǫ; s −s
′′
) , (71)
we get by direct diﬀerentiation
q
∂
∂s
_
−2π
2
¯
h
2
i
_
∞
−∞
ds
′
f
R
(s
′
) 2π
2
¯
h
2
i
_
∞
−∞
ds
′′
f
R
∗
(s
′′
) (72)
(ǫ; s −s
′
)G(x, γ
s
′ ; s −s
′
) (ǫ; s −s
′′
)G
∗
(x, γ
s
′′ ; s −s
′′
)
_
= −2qπ
2
¯
h
2
i
_
∞
−∞
ds
′
f
R
(s
′
) 2π
2
¯
h
2
i
_
∞
−∞
ds
′′
f
R
∗
(s
′′
)
∂
s
_
G(x, γ
s
′ ; s −s
′
) G
∗
(x, γ
s
′′ ; s −s
′′
)
¸
(ǫ; s −s
′
)(ǫ; s −s
′′
)
+
_
∂
s
(ǫ; s −s
′
)(ǫ; s −s
′′
) +(ǫ; s −s
′
)∂
s
(ǫ; s −s
′′
)
¸
G(x, γ
s
′ ; s −s
′
) G
∗
(x, γ
s
′′ ; s −s
′′
) .
Focusing on the ﬁrst term above, we note that, as G is a homogeneous solution of Schr¨ odinger’s
equation, we can apply our lemma to that term, which therefore reads
−2qπ
2
¯
h
2
i
_
∞
−∞
ds
′
f
R
(s
′
) 2π
2
¯
h
2
i
_
∞
−∞
ds
′′
f
R
∗
(s
′′
) (73)
∂
µ
_
i
2
_
D
µ
G(x, γ
s
′ ; s −s
′
)G
∗
(x, γ
s
′′ ; s −s
′′
) −
_
D
µ
G(x, γ
s
′′ ; s −s
′′
)
_
∗
G(x, γ
s
′ ; s −s
′
)
_
_
(ǫ; s −s
′
)(ǫ; s −s
′′
) .
Integrating (72) with respect to s, the lefthand side vanishes (we can safely assume it goes
to zero for all x, s
′
, s
′′
as [s[ → ∞), and the derivative ∂
µ
can be pulled out of the triple
integral in the ﬁrst term. The reader can verify that this triple integral is just ∂
µ
j
µ
, with j
given by (11) and φ, φ
∗
are explicated using (70), (71). The regular current, (33), is therefore
32
conserved, provided the s integral over the second term in (72) is missing an ǫ
0
term in its
ǫexpansion.
Let us then show that, in the distributional sense, this is indeed the case. Integrating
the second term with respect to s, and using
∂
s
(ǫ; s −s
′
) = δ(s −s
′
−ǫ) + δ(s −s
′
+ ǫ), that term reads
−2qπ
2
¯
h
2
i
_
∞
−∞
ds
′
f
R
(s
′
) 2π
2
¯
h
2
i
_
∞
−∞
ds
′′
f
R
∗
(s
′′
) (74)
(ǫ; s
′
−ǫ −s
′′
)G(x, γ
s
′ ; −ǫ)G
∗
(x, γ
s
′′ ; s
′
−ǫ −s
′′
)
+(ǫ; s
′
+ ǫ −s
′′
)G(x, γ
s
′ ; +ǫ)G
∗
(x, γ
s
′′ ; s
′
+ ǫ −s
′′
)
+(ǫ; s
′′
−ǫ −s
′
)G(x, γ
s
′ ; s
′′
−ǫ −s
′
)G
∗
(x, γ
s
′′ ; −ǫ)
+(ǫ; s
′′
+ ǫ −s
′
)G(x, γ
s
′ ; s
′′
+ ǫ −s
′
)G
∗
(x, γ
s
′′ ; +ǫ) .
Using (70) and (71), this becomes
Re −4qπ
2
¯
h
2
i
_
∞
−∞
ds
′
f
R
(s
′
)
_
φ
∗
(x, s
′
−ǫ)G(x, γ
s
′ ; −ǫ) + φ
∗
(x, s
′
+ ǫ)G(x, γ
s
′ ; ǫ)
_
. (75)
Writing φ = ǫ
−1
φ
s
+ φ
r
above, and using the shorts propagator (19) plus the explicit form,
(22), of φ
s
, one can obtain the ǫexpansion of (75). Expanding ﬁrst φ
s∗
(x, s ± ǫ) in powers
of ǫ, the part of the integrand involving φ
s
can be shown to comprise an ǫindependent term
multiplying ǫ
−2
f
s
(ξ
2
/2
¯
hǫ), with f
s
(y) = sinc(y) cos(y) = f
s
(−y), and another ǫindependent
term multiplying ǫ
−3
f
a
(ξ
2
/2
¯
hǫ), with f
a
(y) = sinc(y) sin(y) = −f
a
(−y). Using the evenness
and oddness of f
s
and f
a
resp. , the ﬁrst term behaves for small ǫ like ǫ
−1
δ (ξ
2
) +O(ǫ), while
the second — as ǫ
−1
δ
′
(ξ
2
) +O(ǫ), both, therefore, do not involve the ǫ
0
coeﬃcient, which is
due entirely to φ
r
. Using (16), the latter’s contribution reads in the limit ǫ → 0
−8qπ
2
¯
h
2
_
∞
−∞
ds Re i f
R
(s)φ
r∗
(γ
s
, s)δ
(4)
(x −γ
s
) =
8qπ
2
¯
h
2
_
∞
−∞
ds Im f
R
(s)φ
r∗
(γ
s
, s)δ
(4)
(x −γ
s
) . (76)
This is a distribution, supported on ¯ γ, which vanishes by virtue of (27). We have therefore
shown that ∂ j
r
= 0 in the distributional sense. This is enough to establish the time
independence of the charge, as one only needs to integrate ∂ j
r
= 0 over a volume in
Minkowski’s space, and apply Stoke’s theorem, to get a conserved quantity. But, in fact, it
is easily shown that j
r
is a smooth function in the limit ǫ → 0, implying a pointwise identity
∂ j
r
= 0.
To gain a more explicit geometrical insight into the meaning of a ‘line sink in Minkowski’s
space’, consider a small spacelike threetube, T, surrounding ¯ γ, the construction of which
proceeds as follows. Let β(τ) = γ (s(τ)) be the world line ¯ γ, parametrized by proper time
τ =
_
s
_
(dγ)
2
, and let x → τ
r
be the retarded lightcone map deﬁned by the relations
η
2
≡ (x −β
τr
)
2
= 0 , and η
0
> 0 . (77)
33
Let the ‘retarded radius’ of x be
r = η
˙
β
τr
. (78)
Taking the derivative of (77), treating τ
r
as an implicit function of x, and solving for ∂τ
r
, we
get
∂τ
r
=
η
r
⇒ ∂r =
˙
β
τr
−
_
1 +
¨
β
τr
η
_
η
r
. (79)
The (retarded) threetube of radius ρ is deﬁned as the spacelike three surface
T
ρ
= ¦x ∈ M: r(x) = ρ¦ .
It can be shown in a standard way that the directed surface element normal to x ∈ T
ρ
is
d
µ
T
ρ
= ∂
µ
r
¸
¸
r=ρ
ρ
2
dτ dΩ, (80)
where dΩ is the surface element on the twosphere.
Let Σ
1
and Σ
2
be two timelike surfaces, intersecting T
ρ
and T
R
. Applying Stoke’s theorem
to the interior of the three surface composed of T
ρ
, T
R
, Σ
1
and Σ
2
, and using ∂ j
r
= 0 there,
we get
_
Σ
2
dΣ
2
j
r
+
_
Σ
1
dΣ
1
j
r
= −
_
Tρ
dT
ρ
j
r
−
_
T
R
dT
R
j
r
. (81)
Realistically assuming that the second term on the r.h.s. of (81) vanishes for R → ∞, we
get that the ‘leakage’ of the charge,
_
Σ
2
dΣ
2
j
r
−
_
Σ
1
dΣ
1
j
r
, equals to −lim
ρ→0
_
Tρ
dT
ρ
j
r
.
As dT
ρ
= O(ρ
2
), the leakage only involves the piece of j
r
diverging as r
−2
. This piece,
reads
2q
¯
h
2
_
ds Im φ
r∗
(x, s)f
R
(s)∂
1
2
¯
hǫ
sinc
_
ξ
2
2
¯
hǫ
_
−→
ǫ→0
2q
¯
h
2
π
_
ds Im φ
r∗
(x, s)f
R
(s)∂δ
_
ξ
2
_
∼ 2q
¯
h
2
π ∂
_
ds Im φ
r∗
(γ
s
, s)f
R
(s)δ
_
ξ
2
_
= q
¯
h
2
π
s=sr,sa
Im φ
r∗
(γ
s
, s)f
R
(s) ∂
1
[ξ ˙ γ
s
[
,
where s
r
= s (τ
r
)), and γ
sa
is the corresponding advanced point on ¯ γ, deﬁned by
ξ
2
≡ (x −γ
sa
)
2
= 0 , ξ
0
< 0 .
Focusing ﬁrst on the contribution of s
r
, and using a technique similar to that leading to (79),
we get
∂
1
ξ ˙ γ
sr
= −
˙ γ
sr
(ξ ˙ γ
sr
)
2
+
_
˙ γ
2
sr
+ ¨ γ
sr
ξ
_
ξ
(ξ ˙ γ
sr
)
3
∼
ξ→0
−
˙
β
τr
mr
2
+
η
mr
3
, (82)
where m = dτ/ds needs not be constant. In the limit ρ → 0, using ∂
1
ξ· ˙ γsr
∂r[
r=ρ
→ m
−1
,
the contribution of s
r
to the ﬂux across T
ρ
is most easily computed
_
Tρ
dT
ρ
j
r
= q
¯
h
2
π
_
dΩ
_
dτ
r
m
−1
Im φ
r∗
(β
τr
, τ
r
)f
R
(τ
r
)
= 4q
¯
h
2
π
2
_
ds
r
Im φ
r∗
(β
sr
, s
r
)f
R
(s
r
) . (83)
34
The contribution of s
a
to the ﬂux of j
r
is more easily computed across a diﬀerent, (advanced)
T
ρ
, and gives the same result in the limit ρ → 0. The fact that ρ can be taken arbitrarily
small, in conjunction with the conservation of j
r
(x) for x / ∈ ¯ γ, implies that the ﬂux of j
r
across any threetube, T = ∂C, with C a threecylinder containing ¯ γ, equals twice the value
in (83), when C is shrunk to ¯ γ. Changing the dummy variable s
r
→ s in (83), the formal
content of (76) receives a clear meaning using Stoke’s theorem
_
C
d
4
x∂ j
r
= 8q
¯
h
2
π
2
_
ds Im φ
r∗
(β
s
, s)f
R
(s)
_
C
d
4
xδ
(4)
(x −γ
s
) =
_
T
dT j
r
,
which vanishes by virtue of (27).
A.1 Energymomentum conservation
The conservation of the ECD energy momentum tensor can be established by the same
technique used in the previous section. To explore yet another technique, as well as to
illustrate the role played by symmetries of ECD in the context of conservation laws, consider
the following functional
/[ϕ] =
_
∞
−∞
ds
_
M
d
4
x
i
¯
h
2
(ϕ
∗
∂
s
ϕ −∂
s
ϕ
∗
ϕ) −
1
2
_
D
λ
ϕ
_
∗
D
λ
ϕ, (84)
and let φ(x, s) be given by (70) for some ﬁxed A(x) and γ
s
. Using
_
i∂
s
−1
_
φ = 2π
2
¯
h
2
_
G(x, γ
s−ǫ
; +ǫ) f
R
(s −ǫ) + G(x, γ
s+ǫ
; −ǫ) f
R
(s + ǫ)
_
, (85)
directly following from the deﬁnition of φ, we calculate /[φ + δφ] and, after some integrations
by parts, we get for the ﬁrst variation
δ/ = Re
_
∞
−∞
ds
_
M
d
4
x4π
2
¯
h
2
_
G(x, γ
s−ǫ
; +ǫ) f
R
(s−ǫ) +G(x, γ
s+ǫ
; −ǫ) f
R
(s+ǫ)
_
δφ . (86)
Choosing δφ = ∂φ a, corresponding to φ(x, s) → φ(x + a, s), with inﬁnitesimal a(x, s),
vanishing suﬃciently fast for large [x[ and [s[, so as to render δ/ well deﬁned, we get in a
standard way
δ/ =
_
∞
−∞
ds
_
M
d
4
x
_
∂
ν
m
νµ
−F
µ
ν
j
ν
_
a
µ
=
by eq. (86)
(87)
_
∞
−∞
ds
_
M
d
4
xRe 4π
2
¯
h
2
_
G(x, γ
s−ǫ
; +ǫ) f
R
(s −ǫ) + G(x, γ
s+ǫ
; −ǫ) f
R
(s + ǫ)
¸
∂
µ
φ
∗
(x, s) a
µ
,
with j given by (11) and
m
νµ
=
_
∞
−∞
g
νµ
_
i
¯
h
2
(φ
∗
∂
s
φ −∂
s
φ
∗
φ) −
1
2
_
D
λ
φ
_
∗
D
λ
φ
_
+
1
2
_
D
ν
φ (D
µ
φ)
∗
+ c.c.
_
ds . (88)
35
The integrand above, as also the integrands appearing in the deﬁnitions of all other ECD
currents, can be shown to be a distribution which becomes increasingly focused on the light
cone of γ
s
for increasing distance from γ
s
. Writing φ = ǫ
−1
φ
s
+ φ
r
in (87), and using the
shorts propagator (19) plus the explicit form, (22), of φ
s
, one can obtain the ǫexpansion
of (87). The regular part of the second line (viz., coeﬃcient of ǫ
0
) only involves φ
r
. In the
limit ǫ → 0 it reads
8π
2
¯
h
2
_
∞
−∞
ds
_
M
d
4
xRe f
R
(s)δ
(4)
(x −γ
s
)∂φ
r∗
(γ
s
, s) a(x, s) =
8π
2
¯
h
2
_
∞
−∞
ds Re f
R
(s)∂φ
r∗
(γ
s
, s) a(γ
s
, s) (89)
which vanishes by virtue of (26) for any a. The arbitrariness of a implies that the regular
part of the expression in brackets, in the ﬁrst line of (87), vanishes in the distributional
sense,
∂
ν
m
r νµ
−F
µ
ν
j
r ν
= 0 , (90)
with the regular ‘matter em tensor’, m
r
, deﬁned by the same procedure as j
r
, viz. the
coeﬃcient of ǫ
0
in its ǫexpansion. Just like the electric current j
r
, the matter em tensor can
easily be shown to be a smooth function of x, implying pointwise equality in (90). Equation
(26), by which (89) vanishes, appears therefore as the condition that no mechanical energy
or momentum leak into a sink on ¯ γ.
Not surprisingly, m
r
is not conserved, due to broken translation covariance induced by
A(x). To compensate for this, using Noether’s theorem, we construct an ‘equally non con
served’ radiation em tensor, and subtract the two. Consider, then, the following functional
of A(x), for ﬁxed
k
j
r
, (k labels the diﬀerent particles)
o
_
A
¸
=
_
M
d
4
x
1
4
F
µν
F
µν
+
k
k
j
r
A. (91)
By the Euler Lagrange equations, we get Maxwell’s equations, (4), with
k
k
j
r
as a source.
As before, inﬁnitesimally shifting the argument of an extremal A, viz. A(x) → A(x + a) ⇒
δA
µ
= ∂
ν
A
µ
a
ν
, and following a standard symmetrization procedure of the resultant em
tensor (adding a conserved chargeless piece ∂
λ
_
F
νλ
A
µ
_
) leads to
∂
ν
Θ
νµ
+ F
µ
ν
k
k
j
r ν
= 0 , (92)
with Θ
νµ
=
1
4
g
νµ
F
2
+ F
νρ
F
µ
ρ
(93)
the canonical (viz. symmetric and traceless) ‘radiation em tensor’. Summing (90) over
the diﬀerent particles, k, and adding to (92), we get a conserved, symmetric em tensor,
∂
ν
p
νµ
= 0 , with
p = Θ +
k
k
m
r
. (94)
36
A.2 Charges leaking into worldline sinks
Both methods used above, can be applied to prove the conservation of the regular part of
the masssquared current — the counterpart of (2)
b(x) =
_
ds B(x, s) ≡
_
ds Re
¯
h∂
s
φ
∗
Dφ , for x / ∈ ¯ γ . (95)
In the ﬁrst method, used to establish the conservation of j
r
, the counterpart of (69) is
∂
s
(g
∗
1f) = ∂ (Re ∂
s
g
∗
Df), corresponding to the invariance of the Hamiltonian (in the
Heisenberg picture) under the Schr¨ odinger evolution. In the variational approach, the con
servation follows from the (formal) invariance of (84) φ(x, s) → φ(x, s + s
0
). However, the
leakage to the sink on ¯ γ, between γ
s
1
and γ
s
2
, is given by
8π
2
¯
h
3
_
s
2
s
1
ds Re ∂
s
φ
r∗
(γ
s
, s)f
R
(s) , (96)
is not guaranteed to vanish. Note that this leakage (whether positive or negative) is a ‘highly
quantum’ phenomenon — proportional to
¯
h
2
(the term ∂
s
φ
r
generally diverges as
¯
h
−1
).
Similarly, associated with the formal invariance of (84) under
A(x) → λ
−1
A
_
λ
−1
x
_
, φ(x, s) → λ
−2
φ
_
λ
−1
x, λ
−2
s
_
,
is a locally conserved dilatation current, the counterpart of the classical current (7),
ξ
µ
= p
µν
x
ν
−
k
2
_
∞
−∞
ds s
k
B, with B deﬁned in (95) . (97)
The leakage to the sinks on
k
¯ γ is due to the second term, involving the masssquared of the
particles. A leakage of mass, therefore, also modiﬁes the scalecharge of a solution.
B Spin
1
2
ECD
In a spin
1
2
version of ECD, the following modiﬁcations are made. The wavefunction φ is a
bispinor (C
4
valued), transforming in a Lorentz transformation according to
ρ (e
ω
) φ ≡ e
−i/4 σµνω
µν
φ , for e
ω
∈ SO(3, 1) , (98)
where σ
µν
=
i
2
[γ
µ
, γ
ν
], with γ
µ
Dirac matrices (not to be confused with γ the trajectory).
The propagator is now a complex, 4 4 matrix, transforming under the adjoint repre
sentation, satisfying
i
¯
h∂
s
G(x, x
′
, s) =
_
1+
g
2
σ
µν
F
µν
(x)
_
G(x, x
′
, s) , (99)
37
with the initial condition (16) at s → 0 reading δ
(4)
(x − x
′
)δ
αβ
, where δ
αβ
is the identity
operator in spinorspace, and g is some dimensionless ‘gyromagnetic’ constant of the theory.
The transition to spin
1
2
ECD is rendered easy by the observation that all expressions in
scalar ECD are sums of bilinears of the form a
∗
b, which can be seen as a Lorentz invariant
scalar product in C
1
. Deﬁning an inner product in spinor space (instead of C
1
)
(a, b) ≡ a
†
γ
0
b , (100)
with γ
0
the Dirac matrix diag(1, 1, −1, −1) (again, not to be confused with γ the trajectory)
and substituting a
∗
b → (a, b) in all bilinears, all the results of scalar ECD are retained. The
Lorentz invariance of (100) follows from the Hermiticity of σ
µν
with respect to that inner
product, viz. (σ
µν
)
†
= γ
0
σ
µν
γ
0
, and from (γ
0
)
2
= 1.
Let us illustrate this procedure for important cases. By a direct calculation of the short
s propagator of (99), as in section 2.2.1, the spin can be show to aﬀect the O(s) terms in
the expansion of Φ, leading to an equally simple φ
s
, the counterpart of (22), from which
the regular part of all ECD currents can be obtained. The action, (84), from which all
conservation laws can be derived, gets an extra spin term
/
s
[ϕ] =
_
∞
−∞
ds
_
M
d
4
x
i
¯
h
2
_
(ϕ, ∂
s
ϕ) −(∂
s
ϕ, ϕ)
¸
−
1
2
_
D
λ
ϕ, D
λ
ϕ
_
+
g
2
_
ϕ, F
λρ
σ
λρ
ϕ
_
, (101)
while the counterpart of the electric current, (11), derived from φ, is now a sum of an ‘orbital
current’ and a ‘spin current’
j
µ
(x) ≡ j
orbµ
+ j
spnµ
=
_
ds qIm(φ, D
µ
φ) −g∂
ν
(φ, σ
νµ
φ) , for x / ∈ ¯ γ . (102)
Expanding (102) in powers of ǫ, the coeﬃcient of ǫ
0
is the regular current, j
r
, associated with
a particle. Each of the terms composing j
r
is individually conserved and gauge invariant.
The conservation of the orbital current follows from the U(1) invariance of (101), while
conservation of the spin current follows directly from the antisymmetry of σ. This current
has an interesting property that its monopole vanishes identically. Calculating in an arbitrary
frame, using the antisymmetry of σ, and assuming j
spn i
(x) → 0 for [x[ → ∞
_
d
3
xj
spn0
=
_
d
3
x
_
ds ∂
0
(φ, σ
00
φ) −∂
i
(φ, σ
i0
φ) = 0 −0 = 0 . (103)
The counterpart of (90) becomes (omitting the
r
identiﬁer, as regularization is implied
henceforth)
∂
ν
_
k
m
orb νµ
+ g
νµ k
l
_
= F
µ
ν
k
j
orb ν
+
g
2
_
ds
_
k
φ, σ
λρ k
φ
_
∂
µ
F
λρ
, for x / ∈ ¯ γ , (104)
with m
orb
the same as (88) with a
∗
b → (a, b) in all bilinears, and
l(x) =
g
2
_
ds
_
φ, F
λρ
σ
λρ
φ
_
.
38
Note the ‘spin force’ density, vanishing for a constant F, which adds up to the Lorentz force
density.
Similarly, adding
_
M
d
4
xl(x) to the functional in (91), equation (92) becomes
∂
ν
Θ
νµ
+
k
F
µ
ν
k
j
orb ν
+
g
2
_
ds
_
k
φ, σ
λρ k
φ
_
∂
µ
F
λρ
+ ∂
ν
g
_
ds
_
k
φ, σ
ν
λ
F
λµ k
φ
_
= 0 . (105)
Summing (104) over k, and adding to (105), we get the locally conserved em tensor
Θ
νµ
+
k
k
m
orb νµ
+ g
νµ k
l + g
_
ds
_
k
φ, σ
ν
λ
F
λµ k
φ
_
, x / ∈ ∪
k
k
¯ γ , (106)
from which the timeindependence of the associated charges follows as in the scalar case, as
the extra terms involving spin, do not contain derivatives of φ.
References
[1] N. Argaman. Bells theorem and the causal arrow of time. Am. J. Phys., 78:1007–1013,
2010.
[2] G. S¨ ussman B.G. Englert, M.O. Scully and H. Walther. Surrealistic bohm trajectories.
Zeitschrift f¨ ur Naturforschung, 47a:1175–1186, 1992.
[3] John G. Cramer. Generalized absorber theory and the EinsteinPodolskyRosen paradox.
Phys. Rev. D, 22:362–376, 1980.
[4] R. P. Feynman. A relativistic cutoﬀ for classical electrodynamics. Phys. Rev., 74:939–
946, 1948.
[5] Y. Knoll and I. Yavneh. Scale covariant physics: a ’quantum deformation’ of classical
electrodynamics. J. Phys. A, 43:055401, 2010.
[6] A. Aspect P. Grangier, G. Roger. Experimental evidence for a photon anticorrelation
eﬀect on a beam splitter: A new light on single photon interferences. Europhys. Lett.,
1:173–179, 1986.
[7] J. A. Wheeler and R. P. Feynman. Interaction with Absorber as the Mechanism of
Radiation. Rev. Mod. Phys., 17:157–181, 1945.
[8] J. A. Wheeler and R. P. Feynman. Classical Electrodynamics in Terms of Direct Inter
particle Action. Rev. Mod. Phys., 21:425–433, 1949.
39
classical ontology of pointlike charges and a continuous EM ﬁelds. All modiﬁed theories, however, result in some local perturbation to the Lorentz force, and are therefore incapable of explaining the nonlocal aspects of QM. A century after the above bifurcation took place, equipped with the vast mathematical arsenal accumulated thereafter, we return in this paper to those early 20th century days. Guided by the principle of scale covariance — a hidden symmetry of classical electrodynamics on equal footing with Poincar´ covariance — it is shown that the two fundamental problems e of classical electrodynamics should, and apparently can, be solved at once. This conclusion stems from a deeper analysis of a recent work by the current author (with Yavneh; [5]), proposing a Lorentz and scale covariant deformation of classical electrodynamics, depending on a small dimensionless quantum parameter. In that deformation, dubbed extended charge dynamics (ECD), particles enjoy a dual local nonlocal character. The local aspect explains, for example, the thin traces left by charges in bubble chambers, and yet leads to a well deﬁned self interaction. The nonlocal part — an inevitable consequence of Lorentz covariance — accounts for quantum mechanical nonlocality which, for example, enables an ECD particle passing through one slit in a doubleslit apparatus, to ‘remotely sense’ the status of the other one. This feeble remote sensing mechanism, it is speculated, is also behind gravitational interaction, ampliﬁed this time by a huge mass rather than by the huge distance between the doubleslit and the detection screen. From yet a wider perspective, gravitation may be just another facet of entanglement. The possibility of unifying electrodynamics with such a seemingly unrelated theory as gravity, extends also to small scales. The self force in classical electrodynamics is generally ignored as a ﬁrst approximation, not because it is small (in fact, in the vicinity of the world line of a point charge, the EM ﬁeld becomes arbitrarily strong) but rather because it leads to fairly accurate results when applied to charges in a slowly varying EM ﬁeld. The colossal failure of this approximation at small scales, rather than implying the breakdown of the approximation, was taken as an indication for the breakdown of classical electrodynamics altogether at small scales, inviting other modes of interaction, such as the strong force, into the arena of small scale physics. It therefore seems only logical to ﬁrst have a fully consistent classical electrodynamics, free of selfinteraction problems, and only then confront it with small scale observations. One possibility, for example, suggested by ECD is that two suﬃciently close, positively charged ECD particles, need not repel each other, rendering the strong force just a small scale feature of ECD. It is argued that quantum mechanics (QM), including quantum ﬁeld theory, only describes certain statistical aspects of ensembles of ECD solutions. And indeed, QM, with its builtin unitarity, deals mainly with statistical questions (Smatrices, reaction cross sections, thermal properties of matter etc.). Over the years, one must acknowledge, some more ‘deterministic’ applications of QM have emerged — determining the strength of a chemical bond, or the mass of a subatomic particle — but it seems that these deterministic uses were ‘forced’ on a statistical theory, in the absence of any alternative method. In those applications, QM does not really provide a satisfactory ontology for the microworld, but rather a set of heuristics surviving experimental tests. ECD, on the other hand, just like classical electrodynamics,
2
is a singlesystemtheory, dealing explicitly with such deterministic question and, as will transpire, holds the potential for a wide range of predictions. ECD, then, is not another interpretation of QM, but rather an independent theory, supposedly compatible with the statistical predictions of QM. Section 3 of this paper deals with this compatibility conjecture. The classical ontology of ECD notwithstanding, it is demonstrated in representative cases, how various features of ECD render possible the realization of the quantum mechanical predications by means of an ensemble of ECD solutions. The exact nature of the ensemble relevant to the experiment, it is argued, is an independent law of nature, on equal footing with ECD itself. QM therefore describes some statistical attributes of those ensembles, and like the nature of the ensemble itself, enjoys the status of an independent law, complementing ECD rather than rivaling it. In section 2 (and in the appendix), the mathematical structure of ECD is analyzed in greater depth. The ECD equations derived in [5], originate from a brutally formal Lagrangian, involving ‘deltafunction potentials moving in Minkowski’s space’. Such formal objects never come equipped with a precise meaning which is to be determined only by the global consistency of the mathematical structure resulting from their deﬁnition. Indeed, such global considerations lead to a ‘reﬁned’ deﬁnition of the ECD equations, compared with their form in [5]. One consequence of this reﬁnement, along with another central theme of ECD — scale covariance — is the ability of ECD solutions to ‘drift in scale’. Such a scale drift should be extremely slow on our native timescale, but could manifest itself on cosmological scales. This speculation, as well as many other complementary remarks, can be found in the many footnotes appearing in this paper, and may be ignored on ﬁrst reading. Although acquaintance with [5] may provide some intuition into the origin of ECD, the present paper is entirely self contained, requiring mainly standard graduate level background for its understanding.
2
Extended Charge Dynamics
Throughout this paper the labeling of spacetime is chosen so as to have the speed of light equal 1. The metric convention in Minkowski space, M, is g := diag(1, −1, −1, −1) , and u · v stands for gµν uµ v ν .
2.1
Manifestly scale covariant classical electrodynamics
There are two components in classical electrodynamics of n interacting charges. One is the Lorentz force, governing the motion of a charge in a ﬁxed EM ﬁeld γµ = q F µ γν , ¨ ν˙ (1)
with γ(s) ≡ γs : R → M the world line of a charge, parametrized by the Lorentz scalar s, q a coupling constant and Fµν = ∂µ Aν − ∂ν Aµ the antisymmetric Faraday tensor. Multiplying d both sides by γµ and using the antisymmetry of F , we get that ds γ 2 = 0, hence γ 2 is conserved ˙ ˙ ˙ 3
. rendering the Lorentz force (1) ill deﬁned (Another troubling ¯ aspect of the selfforce problem is the divergence of formally conserved quantities such as energy and momenta. prescribing an EM potential given the worldlines of all charges n ∂ν F with k νµ ≡ A − ∂ (∂ · A) = k=1 ∞ µ µ k µ j . Equation (1). which is conserved. traced by γ. The second ingredient of classical electrodynamics is Maxwell’s inhomogeneous equations. Like the Poincar´ symmetry. (6) also solve (1)1 and (4). 4 . for a given labeling (unitlength 1 An addition of the LorentzDirac radiation reaction force. it relates between diﬀerent solutions of the theory.2. still preserves the symmetry (6).. In the active sense. however. meaning that the scaled variables A′ (x) = λ−1 A(λ−1 x) . written in our convention as 2 q 2 3 γ γ2 ˙ .3. (4) j(x) = q −∞ ds δ (4) x − kγs k γs ˙ (5) the electric current associated with charge k.. γ ′ (s) = λγ(λ−2 s) . This is a direct consequence of the sindependence of the Lorentz force. is more general than (3). ˙2 ˙ dτ ds (2) Deﬁning m = d3 x b0 takes the familiar form 1/2 ≡ γ2 ≡ ˙ with τ = s (dγ)2 the propertime.) The above unorthodox formulation of classical electrodynamics highlights its scale covariance. and supports solutions conserving a negative γ 2 (tachyons ˙ — irrespective of their questionable reality). equation (1) (3) m¨µ = q F µ xν .by the sevolution. The selfforce problem of classical electrodynamics. to which we shall return in section 2.. γ ≡ ∪s γs . . − ˙ γ· γ γ ˙ (γ 2 )2 ˙ . x ν˙ with x(τ ) = γ (s(τ )) above standing for the same worldline parametrized by propertime. the e scaling symmetry (6) admits both an active and a passive interpretation. ∞ ∞ ∂µ j µ = q −∞ ds ∂µ δ (4) (x − γs )γs = −q ˙µ −∞ ds ∂s δ (4) (x − γs ) = 0 . without scaling of any parameter. We see that the (conserved) eﬀective mass m emerges as a constant of motion associated with a particular solution rather than entering the equations as a ﬁxed parameter. refers to the fact that the EM ﬁeld generated by (4) diverges everywhere on the world line. and can also be expressed as the conservation of a ‘masssquared current’ ∞ b(x) = −∞ ds δ (4) (x − γs ) γs γs .
2 Extended Charge Dynamics In a nutshell. depends on the choice of origin for both spacetime. as no dimensionful parameter may be introduced into the theory from which the charge may inherit its typical scale. 2. and its scaling dimension may just as well be named its length dimension. In the passive interpretation. As we shall see. ∂ν mνµ = = ds γ ν γ µ ∂ν δ (4) x − γs = − ˙ ˙ ds γ µ δ (4) x − γs = ¨ ds γ µ ∂s δ (4) x − γs ˙ (10) ds qF µν γν δ (4) x − γs = F µν jν . (8) the (formally) conserved energymomentum (em) tensor associated with translation covariance.convention) of spacetime. by deﬁnition. The ﬁrst grants the electric current (5) a nonsingular support in a way respecting all the symmetries of classical electrodynamics — scale covariance in particular. obviously invariant under scaling of spacetime. the symmetry (6) prescribes how one must rescale A (more generally. [s] = 2. from which one can read the representation under which each variable. ˙ Note that the conserved dilatation charge. [q] = 0. d3 x ξ 0 . achieving scale covariance with extended charges is a lot more diﬃcult. the transition from classical electrodynamics to ECD. involves two modiﬁcations. ˙2 ˙ν (7) with 1 p (x) = g νµ F 2 + F νρ Fρ µ + 4 νµ n k k=1 mνµ . and the n parameterizations of kγ. [A] = [m] = −1. functions deﬁned on spacetime) when relabeling (scaling the unitlength) spacetime. The simplicity in which scale covariance emerges in classical electrodynamics is due to the representation of a charge by a mathematical point. ˙ ˙ (9) the ‘matter’ em tensor. To this end we 5 . [j] = −3 and. and products thereof. Associated with symmetry (6) is an interesting conserved ‘dilatation current’ n ξ = p xµ − k=1 ν νµ ds δ (4) x − kγs s kγs kγs . formally satisfying ∂ν mνµ = F µν jν . or simply dimension. and is therefore diﬃcult to interpret. A can be measured in units of length. transform in a scale transformation — its ‘scaling dimension’: [x] = [γ] = 1. and ∞ m νµ = −∞ ds γ ν γ µ δ (4) x − γs . In this sense.
(14) Above. s) : M × R → C. as described next.2. Note the similar structure of (11) and (5). s − s′ )φ(γs′ . Despite this similarity between the ECD current (11) and the classical current (5). s) = 0 . σ) = θ(σ − ǫ) − θ(−σ − ǫ) . 2 (15) . leads to quantum mechanical ‘entanglement’. the corresponding distribution is Im kφ∗ D µ kφ. = 0. s − s′ )φ(γs′ . δ (4) x − kγs kγs . γs′ . s′ ) −∞ ∞ (13) ds′ G(x. s′) s+ǫ ∞ ds′ G(x. This system is composed of two coupled equations. γs′ . s)2 ≡ ∂ φ(x. there is a striking diﬀerence between the two: the EM potential A enters the deﬁnition of the current (through D’s dependence on it) which. γ} but. o ¯ ih∂s − H(x) G(x. whereas in (11). 6 1 with H = − D 2 . see appendix B) ‘wavefunction’ kφ(x. −∞ U(ǫ. s′ )U(ǫ. and modify the current (5) to read ∞ k µ j (x) = −∞ ds iq 2 k φ D µ kφ ∗ ∞ − kφ∗ D µ kφ ≡ −∞ ds q Im kφ∗ D µ kφ . or Stueckelberg’s equation). G(x. depends on all charges.1 The ‘reﬁned’ central ECD system The second component of ECD is the central ECD system — the counterpart of the Lorentz force equation (1) — prescribing the set of permissible pairs {φ. described next.add to the representation of each charge an auxiliary complex (more generally spinor valued. not to be confused with . It will be demonstrated how this interdependence. s) = −2π 2 h2 ǫi ¯ + 2π 2 h2 ǫi ¯ ≡ −2π 2 h2 ǫi with and the second equation is ∂ φ(γs . in turn. and need not be singular. s)2 x=γs s−ǫ ds′ G(x. s − s′ ) . 2. along with an implicit dependence of φ on A. s) is the propagator of a propertime Schr¨dinger equation (also known as a o ﬁve dimensional Schr¨dinger equation. γ} for a given A. x′ . γs′ . s − s′ )φ(γs′ . x′ . The ﬁrst reads ¯ φ(x. of course. (11) with ¯ Dµ = h∂µ − iqAµ (12) ¯ the gauge covariant derivative and h some real dimensionless ‘quantum parameter’. each ECD charge is now represented by a pair {φ. In (5) it is the trace in Minkowski’s space of a singular vectorvalued distribution. Summarizing. φ is not independent of γ. generating the ˙µ current.
s). which we call f L (s). G(x. s→0 (16) Finally. Although advertised as the counterpart of the Lorentz force. s) = G∗ (x. for (13) to have a solution. with ¯ i e 2hs Gf (x. (18) a similar isolation of the nontrivial content exists. Both equations. for a general x other than γs . limǫ→0 f r = 0 was implied as the content of (13). which we write as ǫf r (‘r’ for residue). In [5]. The linear map f R → f L (which. x. Indeed. we ﬁrst want ¯ to isolate the contribution of the i/(2π hs)2 divergence of G(x. Aindependent) i/(2π hs)2 divergence of G(x. ǫ is a parameter of dimension 2. Moving next to the second ECD equation. x′ ) + Φ1 (x. s) = Φ0 (x.s)/h (19) into (15).. however. can be shown to be formally selfadjoint) must therefore send f R to itself. we see that. the ¯ universal (viz. x′ . For our purpose. Focusing ﬁrst on (13). it is in fact an equation for a function f R (s) ≡ φ(γs . for example). plugging an ansatz for f R into the r. requiring clariﬁcations which were not fully given in [5]. s) −→ δ (4) (x − x′ ) . x′ . s) = Gf eiΦ(x. Plugging the ansatz ′ ¯ G(x. we need the smalls form of the propagator G.x . Φ0 is enough. −s). (14).h. For further use. x′ . x′ ) = 0 (note the manifest gauge covariance of this scheme to any order k). .s. and expanding Φ (not necessarily real) in powers of s. x′ )s + . (20) the free propagator computed for A ≡ 0. (13) and (14). x. s) = 0 . ǫ→0 where. s) = ¯ (2π h)2 s2 i(x−x′ )2 ′ sign(s) . s). conveniently rewritten as ¯ Re h∂φ(γs . and in particular for x = γs . higher orders of Φk can recursively be computed with Φ0 alone incorporating the initial condition (16) in the form Φ0 (x′ . x′ .satisfying the initial condition (in the distributional sense). of (13). x′ . so the nontrivial content of (13) is in this O(ǫ) term. s) to φ(x. 7 (21) . using G(x′ . one can compute φ(x. While this may turn out to be true for some speciﬁc solutions (a freely moving particle. particles with a vanishing monopole. . s). for ﬁxed γ and G. viz. x′ ) = q x′ dξ · A(ξ) . ultimately taken to zero (thereby eliminating the single dimensionful parameter of ECD). x . x′ . s)φ∗ (γs . A simple calculation gives the gauge covariant phase x Φ0 (x. Φ(x. involve a delicate ǫ → 0 limit. the equation should take a more relaxed form (17) Im lim f r∗ f R = 0 . To this end. ‘Im’ is the imaginary part of the entire expression to its right. implies f R → f R + O(ǫ). with f r = O(1). Now. as usual. s) ∀s. it is clear that the system also applies to chargeless particles.
φr (x. that scaling can be considered a symmetry of ECD. (26) ǫ→0 Using the above deﬁnitions. indexed by diﬀerent values of ǫ. ′ φr → Cφr . s) = f (s)e sinc (22) ¯ 2hǫ with ξ ≡ x − γs . s) → λ−4 G λ−1 x. s)φs (γs . s)∗ . φ(γs′ . leads to a gauge covariant deﬁnition of the singular part of φ ξ2 ¯ i Φ0 (x. s)φs (γs . (17) can also be written as lim Im φr (γs . The reﬁned deﬁnition of (14) is therefore lim Re Dφr (γs . ′ R Φ0 (x. via an ǫ → 0 limit. ¯ (29) and under scaling of spacetime A(x) → λ−1 A(λ−1 x) . The second point concerns the scaling dimension. x′ . However. which dictates this special choice of dimension to comply with scale covariance (see next section). ¯ h∂φs (γs . s) → λ−2 φr λ−1 x. for a ﬁnite ǫ it relates between solutions of diﬀerent theories. s ) ∼ f (s). (30) φs (x. Substituting (19) ˙ into (13). s)φs (γs . λ−2 s . let us just note that it is invariant under the original symmetry group of ECD. s) . λ−1 x′ . (23) Using ∂Φ0 (x. that dimension can be arbitrarily chosen. s) − φs (x.where the integral is taken along the straight path connecting x′ with x. under a gauge transformation A → A + ∂Λ . where the above equality follows from (24). s)∗ = ǫ Re Dφr (γs . ¯ C ∈ C. s)∗ = 0 .γs )+γs ·ξ /h ˙ s R φ (x. 8 . By the symmetry (28). two points should be noted. It is only because ǫ is ultimately eliminated from all results. γs′ ) ∼ Φ0 (x. s) = f r (s) and (17). the residual (or regular ) wavefunction is deﬁned via the gauge covariant equation ǫφr (x. shall be given in the sequel. ¯ (28) φr → φr eiqΛ/h . ˙ (24) (25) and (18) is automatically satisﬁed up to an O(ǫ). γs )x=γs = A(γs ). Regarding this last symmetry. φs → φs eiqΛ/h . s)∗ = 0 . s) = f R (s) . γs ). ǫ→0 (27) More insight into this reﬁnement of the central ECD system. s) → G ei [qΛ(x)−qΛ(x )]/h . First. s) → λ−2 φs λ−1 x. the central ECD system is but a part of the ECD formalism. In particular. we have φs (γs . and expanding the integrand around s to ﬁrst order in s′ − s: γs′ ∼ γs + γs (s′ − s). s) = i γs + A(γs ) f R (s) . γ(s) → λγ λ−2 s . of φs and φr . λ−2 s . s)φs (γs . Consequently. gauge invariant term ¯ ǫ Re h∂ φr (γs . the system is invariant under φs → Cφs . −2. x′ . s) = φ(x. λ−2 s . directly following from the transformation of the propagator under scaling A(x) → λ−1 A(λ−1 x) ⇒ G(x. ǫ → λ2 ǫ . For the time being. G(x. φr (γs .
due to a possible ‘leakage’ of charge into a ¯ sink of j r on γ or ‘emergence’ of charge from a source thereon. and note that this divergence. by (22) we get q j s (x) = ¯ h ds γs − qA(x) + ∂x Φ(x. however. (26). turns out to be exactly the condition guaranteeing that no such leakage occurs. j r has dimension −3. granting j in (11) a nonsingular support. as also the limits of all other currents in ECD. vanishes everywhere (except on the world line. ǫ→0 (33) This regular current is the electric current ultimately associated with an ECD charge. It is therefore natural to add to the central ECD system the proviso that the electric charge of each particle.2. both charge and energymomentum conservation. The / ¯ conservation of a current deﬁned on M/¯ . the resultant current diverges everywhere in the limit ǫ → 0. we substitute φ → ǫ−1 φs + φr in (11). as it must on self consistency grounds. Finally. x). The converse. Remarkably. To ﬁx this new problem. guarantees that no energy or momenta leak ¯ into sinks of the conserved energymomentum tensor on ∪k kγ . γs ) f R (s) ˙ 2 1 sinc2 ǫ2 (x − γs )2 ¯ 2hǫ . is conserved for x ∈ γ . can be shown to follow from continuous symmetries of ECD. consistent with the scaling dimension of A. (4) is invariant under A(x) → λ−1 A λ−1 x . (34) establishing the scale covariance of ECD. (32) Using (in the distributional sense) ǫ−1 sinc2 (ǫ−1 y) −→ πδ(y) for ǫ → 0. 9 . j r = lim (j − j s ) . j r (x) → λ−3 j r λ−1 x . trivializing ¯ ECD. where it is ﬁnite). (33). j r vanishes. as well as those of all other ECD currents. This leads to the deﬁnition of the regular current. we note that for A ≡ 0 and a freely moving γ (γs = us). By (30). two steps are taken. or equivalently. we see that j s contains an ǫ−1 term in its ǫexpansion. as well as the collective em of the system. it can be shown that the next higher power in the expansion is ǫ1 . Utilizing the symmetry (28). does not imply the time independence γ 3 r0 0 of the associated charge Q = d x j (x . the second reﬁned ECD equation. which in this case read ∞ φs ∗ µ φs D ds q Im ≡ js . To correct this situation. traced by γ. the reﬁned ECD equation (27). (31) ǫ ǫ −∞ Indeed.2 Regularized currents The ǫ → 0 limit of the current (11). j r — a gauge invariant expression deﬁned as the free coeﬃcient in the ǫexpansion of j. we ﬁrst rescale φ → ǫ−1 φ. can be traced to gauge invariant contributions of bilinears in φs . entering as a source into Maxwell’s equations (4). As we show next. In appendix A we prove that the regular current. Taking further into account the ﬁnite width of ǫ−1 sinc2 (ǫ−1 y) (as oppose to a delta distribution) and its evenness. (see appendix A). namely. do not ‘leak to inﬁnity’ (although it is possible that at least the ﬁrst condition is automatically satisﬁed). γ ≡ ∪s γs .2. however. Carefully applying Noether’s theorem. Likewise.
2.). as well as the total energy and momentum of the system. the EM potential. associated with stranslation invariance. due to a possible leakage of the corresponding charge to sinks on kγ . [8].h. is. taking its derivative ∂/∂(r2 ). both carrying a heavy pricetag in the form of expressions for the (conserved) energy and momenta which are radically diﬀerent from the (illdeﬁned but rather intuitive) corresponding expressions in the usual formulation of electrodynamics (in fact. not every continuous symmetry of ECD leads to a conservation law. (b) From these. However. The counterparts ¯ of the ‘masssquared current’. and Feynman’s covariant regularization of the self LienardWiechert potential [4]. and not the classical selfenergy divergence which is automatically eliminated.3 The self consistent potential As in classical electrodynamics. To the above loop one should add the proviso that the electric and masssquared charges of each particle.h. is not true. (1). plug them into the r. due to the selfforce problem — the ill deﬁned Lorentz selfforce at the position of a pointcharge. it is not even clear what is conserved in a system of interacting charges.s. 2. ﬁnally. The common loop notwithstanding. An even Attempts to resolve the classical selfforce problem follow two distinct approaches. two important diﬀerences should be noted. in classical electrodynamics).g. it is postulated that the ﬁeld produced by a charge does not act on the charge itself. must be ﬁnite. rather than operating with the d’Alembertian on the integral. as well as the counterpart of (7) corresponding to scale covariance. Feynman directly regularizes the LienardWiechert potential rather than deriving it from a regular current. (2). along with their scale covariance (a point remains a point following the scaling of spacetime). while eliminating the original problem. Done in the framework of actionatadistance electrodynamics. (c) verify that the the l. however. or actionatadistance electrodynamics. desirable. thereby allowing to preserve the point structure of elementary charges. In the ﬁrst. are typical examples of this approach. agrees with the original A. (27) (or the Lorentz force equation. using (11) (or (5) in classical electrodynamics). features of it2 . we note. r in the above example) into otherwise 10 . in the case of the AbrahamLorentzDirac equation. In the second approach. namely. Deﬁning ∞ 2 j(x) = ∂ 2 −∞ ds (x − γs )2 1 f 2 r r2 γs ˙ for some integrable function f . First. compute n kj r ’s. kγs } satisfying the central ECD system (26).s. As shown in appendix A. so also in ECD. the singular selfforce is resolved by ‘extending’ the charge so as to associate with it a nonsingular current producing a smooth self ﬁeld. of (4) and. the loop is only formal. This latter approach. also introduce drastic changes to other. must satisfy a self consistent ‘loop’: (a) Start with A and n pairs {kφ. this proviso involves at most the asymptotic behavior of A and φ away from γ. introduces a privileged scale (e. no Lagrangian for such a system is known). Prominent examples include Lorentz’s early attempt to regard the electron as a small uniformly charged sphere. Resolving the selfforce problem calls for extensions/deformations of classical electrodynamics which. fall into this category (see appendix A). his method can be readily adopted to the regularization of a current. more in the spirit of what is done in this paper. in classical electrodynamics. with Feynman. The AbrahamLorentzDirac equation. that j is everywhere conserved and negligible for distances from the charge much greater than r (a covariant alternative.however. A.
in principle. such as the mass or charge. Another possibility is that. as in other ¯ eigenvalue problems. ¨ trivializing the dynamics (the charges cannot move due to an inﬁnite eﬀective mass). the very existence of an ECD charge is due to a solution for the loop. The second diﬀerence in the role played by the above loop is that. exist) freeing ECD. a dense cloud of particles must therefore be strongly entangled by this constraint. q. in fact. from the selfforce problem.g. containing no dimensionful parameter.2. viz. possibly leading to constraints on the nature of fundamental ECD charges. as the magnitude of the total charge of a solution is invariant under the full symmetry group of ECD. corresponding to diﬀerent λ in the scaling transformation (30). in ECD. (This may explain why a continuum of scaled interacting atoms are never seen — a self consistent A may not exist for the combined system. s) → φ∗ (−x. on the other hand. as apparently manifested in low temperature experiments. h. As A enters the expressions of all the currents associated with an ECD charge (both explicitly and implicitly via φ). even attributes of individual particles. Of course. a non vanishing A must be found even for a single static charge in an otherwise void universe — diﬀerent such solutions naturally corresponding to diﬀerent elementary particles. We shall have a lot more to say about role of the consistency loop with regard to entanglement of remote particles.) 2. Moreover. −s) (35) scale covariant electrodynamics. does there exist a solution. and also introduces inﬁnitely many tunable constants — the function f — into singleparameter classical electrodynamics. dramatically diminishing its symmetry group. the size/mass of the charge) in a scale covariant theory. only φr (x. and g (for 1 spin. The self consistent loop is further responsible for the emergence of a privileged scale (e. the scale selected is but one of the inﬁnite possibilities. as is. A closely related aspect of the loop concerns the fact that A is generated by the collective current of all the particles.greater shortcoming is the fact that those extensions/deformations of classical electrodynamics do not survive experimental tests when applied to small scales. Taking the limit r → 0 introduces into the dynamics of γ a term of the form C γ with C diverging in that limit. one must then ‘renormalize the mass’ — a rather contrived procedure. That is. φ(x.4 Antiparticles ECD can be shown to be invariant under a ‘CPT’ transformation A(x) → −A(−x) . This is the domain of QM. This condition easily tolerates discontinuities of the EM ﬁeld on γ ¯ ¯ (which. are. s) needs to be diﬀerentiable on γ . along with its distinct ontology. In ECD. Charge quantization is one such possibility. only for certain values of the ECD parameters. γ(s) → −γ(−s) ⇒ j r (x) → −j r (−x) . This is a nontrivial requirement.2 ECD). entangled. To get a nontrivial theory. 11 .
In such annihilation processes. is facilitated by a simple representation of the propagator. s)eiIβ (x. presented in appendix B. the spin. t).x .5 The classical limit of the central ECD system ¯ The analysis of the classical. Pair creation/annihilation may then have a simple geometrical interpretation when γ ‘reverses its direction in time’ (see picture). enjoys the CPT symmetry only. j r (x) → j r (−x). a deﬁnite Green’s function. however. a strong EM pulse. E(x. This. 4 The nature of the approximation involved in the use of the semiclassical propagator can be read from the path integral representation of the propagator.In fact. and PT: A(x) → A(−x). t) → −B(−x. −t). t) → E(−x. −t). x′ . 3 12 .2 ECD. G. as the two may also form a bound neutral state. 1 However.s)/h . 2. however. B(x. scalar ECD. waiting dormant to be reseparated by. C: A(x) → −A(x). t) → E(x. however. such annihilation/creation scenarios respect the conservation law of both electric and masssquared charges. and under P: γ(t) → −γ(t). as well as classical electrodynamics3 . The energy of a particle. However. and their masssquared charges (expression (95). t) → −B(x. It may even be possible for the particle and its antiparticle to belong to distinct γ’s. t).. These. enter with a nearly random phase anyway. then T is no longer a symmetry. the counterpart of the classical (2)). it predicts the existence of an antiparticle for each particle (viz. it implies that our naive notion of timereversal — ‘running the movies backward’ — is not a symmetry of microphysics. other than the halfadvancedplushalfretardedLienardWiechertpotential. This symmetry has some remarkable consequences. s) = Fβ (x. First. B(x. EM radiation must be released in order to respect energy conservation.2. (36) ¯ 2 (2π h) β Maxwell’s equations and the Lorentz force are also symmetric under under T: γ(t) → γ(−t). say. t ✻ x ✲ pair annihilation pair creation As a particle and its antiparticle have opposite signs for both their electric charges. x′ . known as the semiclassical propagator 4 i sign(s) ′ ¯ Gsc (x. Secondly. of the central ECD system. E(x. or h → 0 limit. The classical paths dominate that representation and the semiclassical approximation amounts to misweighing paths that deviate signiﬁcantly from classical paths. if one includes in the deﬁnition of classical electrodynamics. a bound solution of one or more elementary ECD charges) of opposite charge and equal selfenergy. enjoys an even larger symmetry group. in that limit. is more speculative. equals that of its antiparticle. j r (x) → −j r (x).
γs′ . such that limǫ→0 [R(s.γ0 .Here. s − s′ ) (43) sign(s − s′ )U(ǫ. γs′ . and by a corresponding ansatz of the form f R (s′ ) = CeiIγ (γs′ . must coincide with γ (as γ is a classical path in A. viz.s)/h ∞ ′ ¯ φ(γs . x′ . such that β(0) = x′ and β(s) = x. β runs over the diﬀerent classical paths. γ0 . s) = eiIγ (γs . the reﬁned central ECD system is solved by any classical γ (in the given EM potential A). s Iβ = 0 1 ˙2 ˙ dσ βσ + qA(βσ ) · βσ . given by the determinant F (x. γs′ . paths solving (1) for the ﬁxed A. s − s′ ) Iγ (γs′ . ǫC iIγ (γs .γ0 . (26). in the limit h → 0. s) = −∂xµ ∂x′ν Iβ (x. s − s′ ) sign(s − s′ )U(ǫ. γs′ . Substituting in (13). γs′ . s − s′ ) (42) some real functional of the EM ﬁeld and its ﬁrst derivative (its local neighborhood in an exact analysis) on γ . ǫ) − 2/ǫ] is ﬁnite. implying that (27) is satisﬁed. appearing in Gsc . connecting γs′ with γs . s − s′ ) −∞ x=γs Fγ (γs . x′ . γ0. s − s′ ) 2 −∞ 2 C ¯ . connecting γs′ with γs ). s β(σ). connecting γ(s′ ) with γ(s) not via γ ¯ (e. and pushing ∂ into the integral in (13). other oneparameter ′ families of indirect paths. ǫ) − 2 ǫ with R(s) = −∞ ∞ (41) ds′ Fγ (γs .g. 2 (37) and F — the so called VanVleck determinant — is the gaugeinvariant classical quantity. we ﬁrst note that one of the β’s. s) = e 2 ¯ + h∂x Fγ (x. s − s′ ) . Focusing ﬁrst on this direct contribution. G → Gsc . s) 1/2 . There are.s)/h R(s. and using (40) Iγ (γs . parametrized by s′ .γ0 . x′ → γs′ and x → γs .γ0 . ⇒ φr (γs . Iβ is the corresponding action of the path. in general. 13 . (38) ¯ Let us next show that. s − s′ ) x=γs ∞ ds′ i∂x Iγ (x. bouncing oﬀ of a remote potential). s) we get ǫC iIγ (γs . s − s′ ) sign(s − s′ )U(ǫ. s) = e ds Fγ (γs . s′ ) = Iγ (γs .s )/h . ¯ Moving next to the second reﬁned ECD equation. ′ ¯ (39) where C ∈ C is arbitrary.s)/h ¯ ¯ h∂φ(γs . γs′ .
in the sum over classical paths. possibly suggesting novel methods of ‘cracking’ (or fusing) subatomic particles. possibly (and most desirably in the author’s opinion) showing that all elementary particles are just diﬀerent solutions of the same set of equations (signiﬁcantly reducing the number of tunable constants). s). (45) does depend on s′ — the ‘conspirative’ s′ independence. ˙ (44) which vanishes by (27). aﬀecting the dynamics of γ. also carries to all other ECD currents. s) = iγs φ(γs . based on interacting ECD 14 . self consistent ECD solutions or bound states of any number of them. γs′ . Using a relativistic variant of the HamiltonJacobi theory (see appendix B in [5]). The eﬀective mass and binding energies of such solutions can be computed using the expression for the energy momentum tensor derived in the appendices. s). However. and do not lie on the same massshell. The ﬁrst term in (43) therefore gives ¯ h∂φ(γs . s − s′ ) = p(s) ≡ γs + qA(γs ) which is independent of s′ . Moreover. γs′ . however. β. s) ∗ ǫ→0 lim Re Dφr (γs . for x = γs . hence (26) is satisﬁed. can be sought. s) = ip(s)φ(γs . are diﬃcult to solve. is a privilege of ¯ β = γ. and j r is supported on γ . The phase of the corresponding integrand in (13) reads ¯ (45) h−1 Iβ (γs . formally reduces to its classical counterpart appearing in (5). γ0 . s) = ip(s)φr (γs . Explicit solutions. We return now to the contribution of the indirectpaths. In the limit h → 0. we can write ˙ ∂x Iγ (γs .¯ The second term in (43) drops in the limit h → 0. we have Dφ(γs . s)f R (s) . It follows that the gauge invariant integrand in the deﬁnition of j. the corresponding phase in the integrand ˙ of (13) becomes inﬁnitely oscillatory. ′ ′ 3 Qualitative discussion of ECD The ECD formalism presented in the previous section has a rather unusual structure. the indirect paths appear as an explicit mechanism by which information about distant potentials is incorporated into φr (γs . apparently necessitating an extensive use of numerical calculations. also takes a simple form in the classical limit. Using similar ¯ arguments. manifested in (40). relevant to physically interesting cases. detailed internal structures of such particles can be analyzed. s′ ) . Using (44). Isolated. s)f R (s) = −γs lim Im φr (γs . as the classical path connecting γs′ with x. (11). appearing in the deﬁnition of Gsc . As distinct s β and s γ see diﬀerent potentials. s) ⇒ ∗ ǫ→0 ⇒ ¯ h∂φr (γs . the formal reduction of the ECD electric current to its classical counterpart. the contributions of the indirect paths are therefore suppressed ¯ by the strong oscillation of the phase (45). is distinct from γ. scale covariant ontology. s − s′ ) Iγ (γs′ . The regular current. In short. For a ﬁnite h. j r . one can potentially have a clear. the stage is completely set for such detailed analysis.
then. 3.2. where the use of the term ‘particle’.3. this section argues the case for the compatibility of ECD with current. relating the energy momentum tensor. To motivate such an endeavor. formally identical to its classical counterpart (10). is clearly not merely an interpretation of QM. becoming entirely supported on γ . In particular. ECD.particles alone. we assume the existence of an external potential.1 Singlebody ECD Singlebody ECD deals with the ECD equations of a single particle in the presence of an external EM potential. m. C a fourcylinder containing γ ¯ ¯ 15 . jext . In appendix A. (90).2 ECD is covered in appendix B. The spin of an ECD particle merely labels diﬀerent ways of covariantly obtaining extended currents — ordinary currents. ECD. rather than charge.5 we saw that in the limit h → 0. we ﬁrst need the following. all ECD currents formally converge to their classical counterparts. (47) where Fsel is the selfﬁeld derived from Asel via (46). only the scalar case is analyzed. is derived. Let Σ(s) be a oneparameter family of non intersecting timelike surfaces. An example of spin.2. the currents extend beyond the support of γ . generated by the rest of the particles in the universe. one may rightfully wonder where does the ‘photon’ come from? We shall demonstrate to the contrary. well tested. and close the loop by requiring from the self potential to satisfy Asel µ − ∂ µ (∂ · Asel ) = j r µ . is apparently susceptible to the same objections and ‘no go theorems’ standing in the way of other hidden variables models. reﬂects the possibility for an elementary ECD solution to have a vanishing monopole. resulting in a rather prosaic ‘demystiﬁcation’ of QM — the photon included. ¯ In section 2. This is clearly a simpliﬁcation of the real situation to which we return in section 3. we ‘feed’ Aext + Asel into the self consistent loop of section 2. that in a wide range of cases in which the statistical predictions of QM clearly cannot be realized by an ensemble of classical solutions. each intersecting γ at γs . For simplicity. the unique features of ECD could render possible such a realization by an ensemble of ECD solutions. Aext . since the EM ﬁeld is just the classical Maxwellian ﬁeld. Next. essentially retaining the ontology of classical electrodynamics.2. assumed independent of the particle in question. (46) with j r computed from φ and the combined potential Aext + Asel . transforming 1 as fourvectors. dealing with manybody ECD. satisfying Maxwell’s equations (4) for some ﬁxed current. a relation. physics. but rather a complementary theory with independent testable predictions. Speciﬁcally. To see more closely how this extended ¯ support aﬀects the dynamics of an ECD charge in an external EM ﬁeld Fext . solving φ in the presence of Aext + Asel . associated with a particle. In moving to a non vanishing ¯ ¯ h. to its conserved electric current j (omitting the regularization label r ) ∂ν mνµ = F µν jν ≡ (Fext µν + Fsel µν ) jν . rendering additional forces and particles superﬂuous.
s. (49) with dT the outward pointing directed surface element on T . weighted by the normalized charge density. Let also C(s. As noted before. ❅ t ✻ ✲ x γ Σ(s) ﬁgure 1 Integrating (47) over C(s. and T (s. (The above equalities are most conveniently established in 16 . For a point charge with m and j given by (10) and (5) resp. and ˙ lim δ −1 C(s. p = γ. the term T dTν mνµ vanishes.h. leading to a vanishing self force.δ) d4 x Fext µν + Fsel µν jν .. (48) where dΣ is the Lorentz covariant directed surface element. for example.δ) δ→0 d4 x Fext µν jν = Q F µν s γν . the only non vanishing component of the electric current is j 0 (x) from which the purely electrostatic Fsel inherits its spherical symmetry. (49) formally becomes just the Lorentz force equation (1) with F = Fext + Fsel . however. δ) ∈ C be the volume enclosed between Σ(s) and Σ(s + δ). δ). In moving from a singular electric current to an extended one. and applying Stoke’s theorem to the l.❅ ❅ ❅ ❅ ❅ T T ❅ ❅ ❅ ❅ adv.. The simplest complication of the static case. Under this assumption we can write p = α γ with α some positive constant. ˙ (50) with Q = Σ(s) dΣ · j the sindependent electric charge (here we assume that Σ(s) ∩ C supports the lion’s share of the charge) and F µν s is the average ﬁeld in Σ(s). then. the ﬁrst beneﬁt is that now the selfforce appearing in (49) is well deﬁned. For a static charge. δ) its spacelike boundary (see ﬁgure 1 for a 1 + 1 counterpart). the self force is illdeﬁned in the classical case (hence the reservation implied in ‘formally’). is when the currents retain an approximate spherical symmetry (in the rest frame of γ) and Fsel is approximately a non radiating spherical electrostatic ﬁeld. and upon taking the limit ˙ δ → 0 and dividing by δ.and pµ (s) the corresponding fourmomenta pµ = Σ(s)∩C dΣν mνµ . orthogonal to Σ(s). Σ(s + δ) ret. we get pµ (s + δ) − pµ (s) + T dTν mνµ = C(s. contributing a negligible selfforce only.
1 The breakdown of the spherical core Equation (47) and its integral form (49). dubbed the core. It is argued below that. dTν mνµ . where p2 is the Lorentz invariant restenergy ˙ of the charge. the more uniform the motion must be. 17 . The fact that the Coulomb potential (being an electrostatic potential) is a harmonic function. both given by the independent Rutherford formula5 . The spherical core approximation further accounts for another conspicuous coincidence — the agreement between the nonrelativistic classical and quantum cross sections for Coulomb scattering. enabling him to obtain a well deﬁned expression for the selfforce.) Equation (49) then leads to α¨ µ = Q F µν s γν . of the associated em density m. Likewise. A more symmetric treatment of ‘matter’ and T 5 The author thanks Prof. Σ(s) is taken 0 to be x0 = γs threespace. Lorentz’s sphere model is valid at most for a suﬃciently uniform motion — the larger the sphere. Shimon Levit for sharing with him a long time ago his wonderment over this point.the restframe of γ where j 0 and m00 are the only non vanishing components. implies that the potential of a spherically symmetric core in it. we have in (51) F s ≈ F (γs ). hence no ﬁnitecoresize corrections to point dynamics are observed for this special potential. without going through a ﬁshy massrenormalization procedure (as in later treatments. As a relativistic rigid extended body is a meaningless concept. the Lorentz force density is purely electrostatic. that of γ. all of QM can be traced to the breakdown of the spherical core approximation. uniformly charged sphere. a non uniformly moving ECD particle cannot maintain an exactly spherical charge distribution in the rest frame of every point along γ.1. and dT0 = 0 ⇒ dTν mνµ = 0. needs not even resemble a classical path. ECD can be seen as a fully covariant extension of Lorentz’s analysis of the selfforce. the above spherical core model explains at once the reductions of the QED KleinNishina formula for the cross section in Compton scattering. its dynamics must be classical. preserving the point structure of the charge). γ ˙ (51) and the constant α is identiﬁed with p2 /γ 2 . viz. and a rapidly varying γ on the scale of the ball holding the lion’s share of the charge. equals that of the center of the core. and point dynamics is reproduced. to the classical Thompson formula. at wavelengths greatly exceeding the electron’s Compton length. and sets the Compton length as the order of magnitude of the core. In this respect. in principle. plus a ‘radiative’ contribution. It is instructive to compare the above treatment of an ECD charge with Lorentz’s modeling of the electron as a rigid. 3. For wavelengths much longer than the scale of the core. somewhat artiﬁcially divide the change in the momentum of a particle into a work of the Lorentz force. A rapidly varying external ﬁeld on the scale of the sphere therefore signals the breakdown of Lorentz’s selfforce analysis. We therefore reach the important conclusion: Whenever an ECD charge maintains an approximate spherical symmetry. Indeed the Thompson formula is obtained by simple averaging over the radiation produced by point charges oscillating in an external plane wave. Assuming ECD indeed governs the microscopic world.
is a standard result of classical electrodynamics.2. (53) the total fourmomentum content of Σ(s) ∩ C. and ret. does not vanish only if γ is non uniform. For a ﬁnite h. (52) with pµ = Σ(s)∩C dΣν pνµ . the integral of the em ﬂux. as m also depends on A both explicitly and implicitly (via φ). and still generate a strong EM ﬁeld if it strongly ﬂuctuates). and can be directly derived from the the expression for the LienardWiechert potential generated by a moving charge Aret (x) = q adv 0 ˙ ds δ (x − γs )2 γs θ x0 ∓ γs . (54) 0 The advanced and retarded potentials are the traces of densities. δ (x − γs )2 γs θ (x0 ∓ γs ). hence ‘quantum’). Moreover. this ‘electrok′ weak’ division is entirely artiﬁcial.5 that this latter ¯ ¯ piece vanishes in the h → 0 limit. in the vicinity of a suﬃciently isolated particle k ′ . whenever the core breaks down.the EM ﬁeld is provided by the conservation of the ECD counterpart of (8). p is dominated by k m and the self ′ ﬁeld generated by k j. dropping as r −2 from the charge) at a point x only if γ is non uniformly moving in the neighborhood of points lying on the intersections of the lightcone of x with γ . plus a ‘quantum’ piece associated with k m (recall from section 2. ǫ→0 (55) with J r = Im φ∗ D φ removed of bilinears in φs . A appearing in (55) solves Maxwell’s equations (4) with j r as a 18 . Summarizing our ﬁndings regarding a suﬃciently isolated particle. y) . This ﬂux is composed of the classical Poynting ′ vector. which locally modiﬁes the Pointing ﬂux (note that this ‘radiative component’ of j may be negligible in terms of charge capacity. This all leads to the conclusion that the conservation of energy and momentum associated with an isolated particle (EM em included) can only be breached by an energymomentum ﬂux penetrating T . ¯ The ECD counterpart of (54) can be expected to read Aret (x) = q adv ds d4 y δ (x − y)2 θ x0 ∓ y 0 lim J r (s. These ﬁelds are then further processed. yielding a Θ which can easily be shown to contain a radiative component (viz. of course. and the ǫ → 0 limit is to be understood in the distributional sense. Although pµν is due to all particles in the ′ system. the quantiﬁcation of which leads to the celebrated LorentzDirac equation (but not before ‘renormalizing’ the mass of the charge — a rather contrived procedure whose sole motivation is to render the result non trivial). This. and using the same construction as in ﬁgure 1. p = Θ + k km (see appendix A.1) where Θ is the canonical EM energymomentum tensor (93). Indeed. however. in ﬁgure 1). ˙ supported on the lightcone of γs (schematically depicted by adv. we get pµ (s + δ) − pµ (s) = − T dTν pνµ . Applying Stoke’s theorem to ∂p = 0. T dTν pνµ . producing such matter em ﬂux over ¯ T . a corresponding hdependent electric ﬂux also forms.
as we have seen. Amidst the densely charged threecylinders resides a (mostly) weak. As only the collective p — cores plus aether — is conserved. there is a mutual inﬂuence between the particle and the (cores composing the) screen. by shining light on it. interacts with the core. As noted above. Timelike cross ¯ sections of these tubes were dubbed ‘cores’ of particles. Unlike its classical counterpart. Equation (55). its dynamics at a point is strongly correlated with that of neighboring particles. then. with the lion’s share of the charges concentrated in small threecylinders centered around kγ (and integrably singular there.1. We are led. ignores a crucial diﬀerences between the above densities. 3. This interparticle p which. In classical electrodynamics.2). a solution of Maxwell’s equation (4) is deﬁned only up to a solution of the homogeneous equation ∂ν F νµ = 0. This is essentially a classical interaction. on the other hand. mediated by ¯ the aether. and it is only via the ampliﬁcation brought about by the huge distance between the target and the detection screen. and upon plugging in (55) the corresponding classical density J r (s. smoothly merging with the dense ¯ cylinders. since the particles are assumed neutral. We shall see in 3. One can also understand why attempts to measure the position of the particle during its ﬂight. reducing to Θ in the limit h → 0. coming with a variability which greatly surpasses the feeble quantum corrections to classical paths. locally conserved p. that minute corrections to the classical cross section are detectable. responsible for the fringes in the ﬁrst place. or fusing. This is because the external EM ﬁeld applied to the particle.2. involving huge numbers of particles (See more in section 3.2 The aether and its manifestations The picture emerging from the previous section is that of a single conserved em ﬁeld. in the absence of a better name shall be referred to as the aether. J r depends on A. Consider now a scattering experiment in which a neutral particle passes through a small aperture in a screen. is therefore not a prescription for Aret but rather an equation for it. combined with a corresponding advanced mechanical ﬂux associated with m. to the important conclusion that in ECD. in the limit h → 0 the ether vanishes. is generated by the combined current of all the particles. the motivation for the (almost universal) choice of the retarded potential is not in the equations proper.source. cannot be decisively attributed to any single particle as A. pµν (x). so to speak.2. small ‘quantum corrections’ to classical paths are expected. but rather in the desire to conform with observations concerning large scale radiation phenomena. unlike (54). must be considered. is therefore dictated by the speciﬁcs of the radiation process. For this reason. ¯ But for a small value of h. y) → δ (4) (y−γs )γs . advanced EM ﬂux. Nevertheless. radiation and matter. ˙ (54) is reproduced. The distribution of p in M is highly nonuniform. however. containing both advanced and retarded components. The ‘correct’ radiation ﬁeld. destroy the delicate interference pattern. Equation (55). A solution in which only adv Aadv or Aret enter J r may not (and probably does not) exist.1 apparent experimental signatures left by such advanced eﬀects. see appendix A). organically integrating. entering every term in p. and the particle does not suﬀer any deviation from a straight classical path. It is conjectured that gravity is yet another manifestation of such ether mediated interac19 .
In the context of ECD. as an eﬀective statistical theory. and carries ‘em waves’ generated by non uniformly moving bodies – much like p. in particular. GR. that the standard procedure of averaging over the impact parameter in order to obtain the scattering cross section. is utterly meaningless. only play a role similar to that of QM. ampliﬁed this time by huge numbers/mass rather than by a huge distance. Scale covariance naturally suggests we represent the cores by mathematical points. In a MachZehnder conﬁguration. forming a huge lattice of scatterers in which a particle undergoes multiple scatterings before exiting. the beamsplitters (BS) and mirrors are crystals of macroscopic thickness. not merely minute correction. Let us brieﬂy see what such a statistical theory must look like. To each scattering event in the crystals there corresponds a radiative perturbation to the 20 . as we saw in the case of ECD. ✁ ✁ ✁ BS2 ❆ ❆ ❆ ❆ ❆ ✁ ✁ ✁γ ✁ ✁ BS1 (a) ★✥ ❆ ❆ ❆ R ❆ ✧✦ ❆ ❆ ❆ ❆ ✁ ✁ ✁ ✁ ✁ BS1 ✁ ✕ ✁✁ ✁✁ (b) ✁ ★✥ ✁ ✁ R ❆ ✧✦ ❆ ❆ ❆ ❆ ✁ ✁ ✁ ✁ ✁ BS1 ✁ ✁✁ ✕ ✁✁ (c) ✁ ✕ ✁✁ ✁✁ Even at the classical level. (a). A distinct third way. can bee seen as a perturbation to a nonuniform static aether conﬁguration (just like mechanical waves bend in a nonuniform medium). This.3 Interferometers In the previous section we mentioned two ways of amplifying the small aetherinduced deviations of the cores from classical paths: Huge distances and huge numbers/mass. is locally conserved. gravitational theories such as general relativity (GR). respecting all the symmetries of ECD. Self force problem aside. but applicable to totally diﬀerent experimental settings. we can appreciate why GR is a reasonable candidate. transforming in a scale transformation as p. meaning. the resolution of this problem may result in radical changes to GR. the bending of the trajectory of an EM wavepacket in a gravitational ﬁeld. There is no classical cross section when the target supports chaotic dynamics. therefore. obviously invariant under scaling of spacetime.tion between remote cores. the dynamics in such a maze is highly chaotic. cannot be considered a completely satisfactory theory and. implemented in neutron interferometers. therefore. In particular. creates a self force problem similar to that plaguing classical electrodynamics. as we know.1. and the call for a complementary statistical theory for ECD is rooted already in classical dynamics. The em tensor derived from the metric. relies on the ability of chaotic systems to amplify small perturbations. 3.
we assume that as the number. locally deﬁned. of arriving predominantly at one detector rather than the other? (or exit the crystals at the Bragg angles only?) To answer this question. whether electronic or atomic. have a dramatic eﬀect on the ﬁnal scattering direction of the particle.1. interferometers can be tuned to produce nearly deterministic results. and E the ensemble of all such currents. with one detector ﬁring some 99% of the times and the other only 1%. conserved. such as the spectrum of atoms. We have focused our discussion on a crystal BS as the arena for this chaotic dynamics but. the aether excitations. the number of solutions realized in any subset Σ ∈ E approaches nµ(Σ) ≡ n Σ dµ (j). All interferometers. in fact. Let j ∈ E be the regular electric current associated with a solution realized in the experiment (we drop the r superscript in this section). The inﬂuence of the aether excitations on the dynamics of a particle passing in region R is now negligible. The chaoticity of the underlaying classical dynamics is crucial for the operation of the interferometer.4 The ensemble current The above description of interferometers invites a troubling question. use BS’s in the form of highly sensitive devices (usually involving the spin of the particle) facilitating the ampliﬁcation of the small aether induced perturbations to the core’s dynamics. As the dynamics of the cores inside the crystals is chaotic. in the context of ECD. viz. as follows from momentum conservation. namely. ‘surrealistic’ trajectories predicted by Bohmian mechanics. An experiment is seen as a realization of a measure dµ (j) deﬁned on E. of scattered particles goes to inﬁnity. slightly perturbing the (locally almost) classical path of the cores. almost unperturbed. n. 3. containing both advanced and retarded components. and the particle continues its straight classical path. 21 . it is not chaoticity itself — a classical notion — which is essential for the operation of the interferometer. taking the other direction [2]. but rather the sensitivity of chaotic dynamics to perturbations. can be read from an ordinary.aether. their small local eﬀect notwithstanding. This should be contrasted with (c). As is well known. we ﬁrst need to see what a scattering experiment is. These ‘aether waves’ propagate inside and around the interferometer. If the local dynamics of the particles are so nearly classical. The reader can verify that the scattering cross section as well as any other measurable statistical expression produced by singlebody QM. then how do they acquire this destiny. in a way which globally depends on the conﬁguration of the interferometer. Suppose we remove BS2 from the apparatus (b).
We shall next show why relativistic wave equations. 3. such as the ﬁrst or second order Dirac equation. Equivalently. assuming ECD is indeed the physics prevailing at the atomic scale. appear simply as dominant frequencies in the dipole radiation of jens . as argued above. the currents associated with the bound electrons (those generating the radiation) can safely be assumed to constitute an incoherent ensemble E. As to the spectrum — a classical Hydrogen atom is a meaningless concept to begin with. the standard procedure of averaging over the impact parameter leads to a meaningless result when applied to chaotic systems. and the incoherence assumption. QM enjoys a similar status of an independent law. Ω) the cone in three space deﬁned by dΩ and Ω. the spectrum produced by the ensemble current equals the sum of spectra produced by the individual currents in E. QM could have been anticipated independently of ECD. and C = C(dΩ. The spectral peaks. why is the nonuniform shape of the Hydrogenatom spectrum counter intuitive? Note that in both cases. E can comprise diﬀerent. the domain of µ. In the scatterng case. in certain cases of a single particle moving in an external ﬁeld. Why is 99% — 1% less intuitive than 50% — 50%? Likewise. on equal footing with ECD itself. then. for example. representing statistically more common frequencies in the dipole radiation of members in the ensemble. then. x) is entirely supported in C. with the latter’s very unique content. are a natural tool for guessing those moments. C 6 (56) with Q = d3 x j 0 the conserved charge of the particle. or in its complement. 22 . non locally deﬁned object. every j 0 (x0 . constrained only by compatibility requirements with ECD and the experimental settings (try thinking what would constitute a natural µ?). and χΣ (·) its characteristic function. For this reason. It is just 1 lim QdΩ x0 →∞ d3 xjens 0 . A single ‘point’ in E — a current j — is such a complex. on which jens is deﬁned. the question of interest is why does this current have such an asymmetric form? The answer to the question does not lie in fourdimensional Minkowski’s spacetime. but rather in inﬁnitedimensional E. By the linearity of Maxwell’s equation. In fact. As yet another example. no classsical counter proposal even exists. and the KleinGordon equation. the measure µ should be regarded as an independent law of nature. This follows upon inserting expression (57) into (56). (57) Stated in the above terminology. It should not come as too great a surprise that. of the current associated with a single atom.fourcurrent — the ensemble current 6 . suﬃciently remote time segments. describes very coarse aspects of the measure µ — very ‘low order moments’ of that inﬁnite dimensional distribution.5 Relativistic wave equations Singleparticle QM. jens = E dµ(j) j . consider the EM spectrum emitted by a heated gas. In the limit x0 → ∞.1. The result is therefore (dΩ)−1 E dµ(j) χΣ (j) ≡ (dΩ)−1 µ(Σ) which is the deﬁnition of the diﬀerential crosssection. The x integration then extracts Q χΣ (j) with Σ = Σ(C) ∈ E the subset of solutions scattering to cone C. The diﬀerential scattering crosssection to a given solid angle dΩ around Ω. For a suﬃciently dilute gas. is easily deducible from the ensemble current (57). that we lack any intuition regarding sensible distributions thereof.
enjoying the full symmetry group with mens = E dµ(j) jm . we ﬁrst make the assumption that the contribution of the selfﬁelds. With the above approximation. hence the limitation of the ensemble current approach and of relativistic wave equations in particular (see more in section 3. strongly depend on that self ﬁeld. (62) and the symmetry group of ECD. However. compared with that of the external ﬁeld. Consider now a low energy scattering experiment. This relation adds up to (61). dµ(j) jFsel µν (x)jν (x) . E (60) can be neglected. must obviously transform like their constituents in any symmetry transformation belonging to the symmetry group of ECD. regardless of the details of the ECD dynamics. As shown above. In particular. A systematic way of producing such constrained pairs. (61) 23 . (59) (58) Multiplying (58) by dµ(j) and integrating over E.. Let this solution be indexed by the electric current j. mens }. and let jm be the corresponding em tensor. This is a reasonable assumption for a suﬃciently incoherent ensemble.Consider. For this to happen. From (47) we have ∂ν jmνµ = Fext µν + jFsel µν jν . as then the self contribution of diﬀerent members in the ensemble. nonetheless.2. the ECD solution of a single particle in the external ﬁeld Fext . associated with the particle. jens and mens resp.1). ∂ν jFsel νµ = j µ . which must coincide with that computed using jens . the scattering cross section can be computed from jens . producing a very restrictive condition on the set of permissible pairs {jens . we get the following four relations Fext µν jens ν = ∂ν mens νµ . (62) inherited from the conservation of the individual j. with jFsel the selfﬁeld generated by j. then. the diﬀerent charges must not radiate (advanced or retarded ﬁelds) in preferred directions. guaranteeing that selfforce eﬀects are not eliminated in that process. the eﬀective mass and charge of the particles. that the selfﬁelds are dominant in the individual currents j and jm. however. a similar construction applied to mens can also produce the cross section. Note. even when their contributions to the integral over the ensemble have been neglected. The Lorentz vector and second rank tensor. to the self force at a point x. and a conservation constraint ∂ν jens ν = 0 . enters with a nearly random orientation.
however. and are therefore more diﬃcult to relate to actual experiments. The nonpositivity of jens 0 (motivating the Dirac equation) simply reﬂects the nonpositivity of the individual j 0 comprising jens . 2 (68) satisfy all the above compatibility conditions — eq. ψ. This is the meaning of a ‘statistical mixture’ of wavefunctions in QM. however. Its unitarity implies o ∂s ψ = ∂ · J ≡ ∂ · (Im ψ ∗ D ψ) . that is guaranteed to remain constant. the two parts of the original beam are recombined in an interferometer. (61) in particular.c. The more general way is via the conservation laws associated with solutions. 24 . and A the external EM potential. If. involves a few irreducible ensembles. involving a single particle species.c. jens = dsJ and mens = dsM . the original ensemble is the relevant one. {jens . The collapse postulate of measurement theory. and the conserved electric charge and mass of the particles comprising the ensemble. imposes certain relations between the parameters of (65).g. when restricted to a ﬁeldfree region. ˆ (65) with the gauge covariant derivative ˆ D = h∂ − iˆA . . mens }ψ . representing the total charge. of the ﬁve dimensional Schr¨dinger equation (15). In the scalar case. labels an ‘irreducible ensembles’. Eq. For example. correspond to ensembles with a continuum of masses. SternGerlach) polarizer. It is only the space integral over those individual components. The wavefunction ψ. which are sampled with diﬀerent weights. to which there corresponds an ‘irreducible pair’. while the Ehrenfest relations give ∂s J µ = F µν Jν − ∂ν M νµ ¯ 1 ih ∗ (ψ ∂s ψ − ∂s ψ ∗ ψ) − Dλ ψ ≡ F µν Jν − ∂ ν g νµ 2 2 (64) ∗ 2 7 (63) Dλ ψ + 1 ν ∗ D ψ (Dµ ψ) + c. (61). When the beam is further split into two by a (e. to a beam of particles escaping a hot oven there correspond one ensemble. the relevant equation is the KleinGordon equation D 2 + m2 ψ = 0 . satisfying all our requirements. however. when scattered oﬀ of a spin sensitive target. is via relativistic wave equations7 .of ECD. we get two candidates. ˆ (67) and the ensemble em tensor mens νµ = g νµ 1 2 ∗ 1 Dλψ m ψψ − ˆ 2 2 ∗ Dλ ψ + 1 ν D ψ (D µ ψ)∗ + c. each part must obviously be represented by a diﬀerent ensemble. A generic experiment. q and m are some constants. A major historical diﬃculty associated with the KG ﬁeld is also resolved in this framework. These. then. merely represents a transition from one ensemble to another. µψ . q (66) ˆ ˆ where h. The expressions for ˆ the ensemble electric current jens µ = q Im ψ ∗ D µ ψ . 2 Integrating (63) and (64) from s = −∞ to s = ∞.
suppressing the large s − s′  contribution to the s′ integral in As entanglement is transitive. 8 Consider. and should be solved as a single space time structure. If a large scale astronomical object. This would have obvious implications on the current interpretation of astronomical data. a universe collectively increasing its scale leads to a Hubblelike relation. as some ‘self consistent matterradiation condensate’. particle. This division is legitimate on the premise that the self potential of the privileged particle does not alter the solutions of the rest of the particles. due entirely to this one. As shown in appendix A. the self consistent potential entangles closely interacting particles in such a way that one can no longer regard matter as a composition of individual particles but. two nucleons. and at arbitrarily remote locations. 8 25 . privileged. at close range. as light collected from remote galaxies is emitted at an epoch of lower mass (hence longer wavelength) which is proportional to the distance of the emitter to the observer. may slowly drift over time. As explained in section 2. both the mass of individual ECD particles. gravitation on galactic scales (and possibly above) could therefore be due to such a primordial entanglement.3.2 Manybody ECD In the previous section. In this regard. plus a self potential. self potential included. all the particles in a dense cloud are entangled. such as a galaxy. as well as the scale charge of their combined solution. or ‘radiatively’. we should also notice another possibility opened by ECD — scale drift. viz. reﬂect that entangled epoch. s) = 0. and need not even resemble solar scale gravity (the largest scale in which general relativity has been directly conﬁrmed). for all observers. then the ECD system. ‘bear the memory’ of their encounter. escaping a nucleus. In fact.3. which is not the case when the privileged particle interacts with the rest of the particles. instead. either ‘electrostatically’. generated by all particles but one. can be solved independently for each particle. the EM potential was divided into an external potential. via longrange aether waves. ✎☞ P1 ✍✌ ✎☞ P2 ✍✌ t ✻ ✲ x If the two polarimeters are positioned suﬃciently far apart.2. two particles which have closely interacted in the past. x′ . This is a consequence of lims→∞ G(x. for example. passes through an epoch of a dense ‘ﬁreball’ (not necessarily anything as dramatic as a big bang or big crunch — whatever that means) it could reasonably be that its morphology and dynamics at much latter stages. This oﬀers an alternative explanation for the source of galactic redshifts. arriving each at a polarimeter (P1 and P2). What may seem surprising at ﬁrst is that long after their separation. As it is also reasonable to assume that galaxies indeed pass through some ﬁreball epoch. Let us begin with the ﬁrst case.
it could become impossible to combine the two solutions into a single. when the two particles come close. is in the spirit of so called ‘retrocausal’ models. for P1 and P2 resp. which maintains that the same ensemble of hidden variables be used. the contribution of Fext to the Poynting vector may be neglected. identical with the Poynting vector.1 The conspiracy of the photon Perhaps the strongest motivation for the introduction of photons. probably doesn’t exist. In ECD. for example. the combined solution of the two particles in the above example.9 Like its singlebody counterpart. We may then get an arbitrarily large energy deﬁcit at times following the jolting of the charge. accounting for violations of Bell’s inequalities. let us note that the same discussion holds also for two particles. To each orientation choice. on equal footing with ECD itself. on the other hand. In the case of the photoelectric or Compton’s eﬀects. must be solved as a whole — as a single spacetime structure. See. then at some point. [1] and extensive references therein. it is The above mechanism.. which later bind together. E. can formally be applied to the corresponding classical currents as well. expressing the change in the fourmomentum of a particle as a function of the integrated em ﬂux across a spacelike surface surrounding the particle. converging on the molecule and increasing its internal energy (or ionizing it as in the photoelectric eﬀect). ‘photon absorption’ by a molecule. should correspond to predominantly incoming advanced em waves. Fext . Finally. equation (52). More speciﬁcally. of twoparticle ECD solutions. the measure. This scenario. e. is the salvation of energymomentum conservation. The above independence notwithstanding.g. is computed from the external trigger. the em ﬂux across T . µ. irrespective of the orientations of the polarimeters (roughly corresponding to the fact that the particles should not ‘anticipate’ the orientations of the polarimeters before encountering them). enjoys the status of an independent law of nature. As both eﬀects are observed even for extremely feeble triggers. self consistent one. and a possible retarded outgoing wave. which is washed away over macroscopic scales. a simple generalization of the ensemble current to the case of manybody ECD. initially separated. the photoelectric and Compton’s eﬀects are manifestly in violation of classical energymomentum conservation. This contradicts Bell’s assumption in deriving his celebrated inequalities. p2 . Thus. In ‘spontaneous emission’. if one tries to continue those independent solutions into the past. Given the orientations of the two polarimeters. 3. therefore.2. generated in the jolting of the charge. 9 26 . There is therefore nothing unique about the dynamics of each particle. However. equipped with its own measure. however. unless each of the two independent solutions is restricted to a certain subset of the full set of independent solutions — a subset which obviously depends on the orientation of both polarimeters. while that of the retarded wave can be shown to be positive. giving away their histories. µ. does not correspond to known experiments. p1 . hence the enormous complication of manybody relativistic QM — quantum ﬁeld theory — involving both matter and the EM potential. there corresponds a diﬀerent ensemble. we saw that advanced em waves must be included in the analysis.(13). Indeed.
typical of all ‘photodetectors’ by deﬁnition (or else they are called calorimeters. In contrast. a molecule excited by a femtoseconds laser pulse. considering the o incident wave as a small perturbation.). of the incident radiation (ignoring for simplicity the binding energy). If the source is surrounded by a large sphere. and then allowed to spontaneously decay. otherwise. antennas.e. In other situations — Compton scattering for example — both waves play an equally important role. viz. hence also of the corresponding self force. justifying the omission of the term (60). ω. but without a heavy trap holding the particle. The success of the ensemble current formalism in the case of the photoelectric eﬀect. is that the probability for a jolting event is proportional to the amplitude squared of the incident wave. implying that the probability drops as the inverse of the distance squared between source and detector — just as if a ﬂux of particles in erupting from the emitter. ❅ ❅ ❅ ❅ ❅ ❅ ❅ ✁ ❅ ✁ ❅ ✁ ❅ ✁ ❅ ✁ ❅ ✁ ❅ ❅ ❅ ❅ ❅ ❅ scattering t ✻ ✲x ❅ ❅ ❅ ❅ ❅ ❅ absorption emission The above two features. However. γ}.g. is due to the isotropic distribution of the direction of the ejected particle. solving the ECD system (self potential included) for a given external trigger. Yet another standard result emerging from the analysis of the ensemble current. i. This binary response of electrons. for wavelength much smaller then the electron’s Compton length. The distribution of the corresponding currents can be read from the appropriate ensemblecurrent (see section 3. A typical example involves a so called ‘singlephoton source’ or more generally an nphoton source (Fock state source). is the historical reason for the introduction of photons. Momentum conservation — ignored in the photoelectric eﬀect due to the large mass of the trap — dictates that the direction of the ejected charges. gives the well known10 result that the electron either jolts with energy. consisting of independently operating This calculation is usually preformed with the nonrelativistic Schr¨dinger equation. or else does not jolt at all. proportional to the frequency. Being highly nonlocal and nonlinear. It is as if a ‘light corpuscle’ of energy λω has struck the jolted electron. hence the need for ECD to implement this. must be strongly correlated with that of the incident wave. e. there are plenty of pairs {φ.rather outgoing retarded waves removing energy from the molecule. But the analogy with other particles goes even further.4) which. etc. and (60) cannot be neglected. the Dirac equation gives identical results. 10 27 . automatic selection of the correct radiation ﬁeld (guaranteeing energymomentum conservation). are missing from classical electrodynamics. classical idea. this formalism fails when applied to Compton scattering.1. along with its incorporation into the dynamics of the particle (no selfforce problem). λω. an external EM trigger in the form a plane wave. in the case of the photoelectric eﬀect.
This. In actual experiments. whatever optical path exists between the source and the detector. e. This is the realm of QM — QED to be speciﬁc. beamsplitters. is consistent with a scenario of a release of a ﬁxed number of particles in each decay of the molecule. this ﬂuctuation may be ignored. However. While it is possible to prevent diﬀerent photodetectors from crosstalking. is described in [3].g. are presumably all manifestations of that indirect crosstalking. the loophole in the analysis is in the assumption of independence of the photodetections. but for a small one. this result is not altered when. 11 28 . as the expectation value is additive even for dependent random variables. imply that the average number of photodetections does not depend on the radius of the sphere.photodetectors (which can further be prevented from crosstalking by. however. leading to statistical dependence in their responses. lies in The use of advanced solutions in order to explain the non classical statistics of photons. than that expected on the premise of independence — this is a statistical eﬀect. it is. again. it is signiﬁcantly greater than the observed value which is more consistent with a particle scenario having zero mean. rather than larger. As to why the actual ﬂuctuation around the mean is much smaller. The crucial point is that. The real moral. For a large mean.g. comprising mirrors. partitions) then the above results of the ensemble current. ﬁberoptics etc. e. by means of retarded ﬁelds. not to be sought in ECD alone. latter receiving the name ‘the transactional interpretation of QM”. latter. however. nor with the mechanism causing a charge to jolt. Using point charges. and is entirely an attribute of the source (note that.. we argue that the photodetectors are not independently operating). The source therefore serves as a hub for indirect crosstalking between the absorbing charges. Violations of Bell’s inequalities in photons pair measurements etc. by deﬁnition. the independence of the diﬀerent photodetectors also imply that the number of detected photons should ﬂuctuate around its mead with a standard deviation proportional to the squareroot of the mean. As implied above. the retarded ﬁeld of the source is relayed to the detecting charges by other charges. absorber 1 ✆ ✆ ✆ ✆ source ✆ ✆ ✆ absorber 2 ❊ ❊ ❊ ✆ ✆ ❊ ■ ❅ ✒ ✆ ❅ ❅ ✆ ❅ ❅ ✆ ❅ ❅ ✆ ❅ ❅ ✆ ❅ ❅ ✆ ❘ ❅ ✠ t ✆ ✻ ✲x We see how various features of ECD have conspired to bring about the illusion that ‘light particles’ must be involved in radiation processes. there must necessarily exist a reverse path leading from the detector to the source via advanced ﬁelds. impossible to prevent each of them from cross talking with the source if advanced waves are present in the radiation ﬁelds of the absorbing charges11 . [6]. that proposal does not explicitly deal with energymomentum balance.
calls for a closer look at these disputable objects. There is a strong. and is therefore a genuine attribute of the structure as a whole.g. two particles whose associated worldcylinders. of a suﬃciently high em density must be treated as a single spacetime structure. In particular.). cannot be decomposed into advanced plus retarded contributions. We have previously argued that the self consistency loop entangles. supporting the lions share of their charges. B t ✻ ✲ x The term ‘transaction’ is deliberately borrowed from [3] as it highlights the symmetric role played by both charges appearing in the structure. largely unjustiﬁed. viz. B. not even inside the dense cylinders associated with the particle. The conclusion deduced from that example generalizes to the observation that any connected volume in M. rather than the eﬀect.2. fundamentally. bias against the inclusion of advanced waves — advanced solutions of Maxwell’s equations in particular — into the description of physical reality. the absorber may just as well be seen as the cause. triggering the emission via advanced waves. is typical of all emitterabsorber ‘transactions’.the geometry of Minkowski’s space and in the unity of the em ﬁeld p. is a blatant repetition of the historical mistake. have a signiﬁcant overlap in M. Note that in ECD this blurring between cause and eﬀect goes even further than in [3]. as p. the following densely charged connected structure. This parallelism. which led to the invention of the aether (the historical 29 . in the statistical sense. The main objection draws parallels with ‘contrived’ advanced solutions of other physical wave equations (e. triggered by the retarded waves of the emitter. This can be seen as a manifestation of the fact that. however. 3.2 Advanced waves The central role played by advanced waves in explaining the illusion of a photon. in particular on the ‘bridge’. surface waves in a pond converging on a point and ejecting a pebble. the value of p at a point cannot be attributed to any single particle. between the two particles.
A device consisting of a bomb. The privileged status of retarded waves in all macroscopic radiation processes. detonating the bomb at B. The signal is relayed at R. Indeed. S ❇ ❇ R❇ ❅ ❇ ❅ ❇ ❅ ❇ ❅ B ❇ ❇ ❇ ❇ relay ❇ bomb ❇ ❇ ❇ ❇ ❇ t ✻ ✲ x Indeed. then the following paradoxical situation could occur. for example. But if the bomb goes oﬀ at B. if the bomb doe not go oﬀ at B. The formal mathematical similarity between the d’Alembertian (the only Lorentz invariant secondorder diﬀerential operator) and other (suitably scaled) wave operators. Yet another argument against advanced solutions is their alleged involvement in causal paradoxes. Its generated waves are eventually absorbed by other particles and. the emission of waves cannot be separated from their absorption (as in the WheelerFeynman absorber theory. then a signal must be sent at S.[7]). representing the local part of the universe we live in. however.aether. A radio transmitter.g. received at B. then the CPT image of a radio transmitter sending retarded waves. is no more than a misfortunate coincident (had this coincidence had some real substance to it. Why then did the bomb explode? On the other hand. and is intimately connected with the excess of matter over antimatter around us. a receiver and a timer. among else. then application of the Lorentz transformation to the wave equation describing the propagation of sound. While spontaneous emission or absorption may be seen as direct evidences to the contrary. then. not to be confused with that used in this paper). that the strongest case against advanced solutions is observational. The resolution of the paradox should not be sought in ECD proper. if advanced solutions played a dominant role in any photoabsorption (as implied e. as argued above. would have yielded a meaningful result). in 30 . then no signal is sent at S. and triggers the fuse of the bomb. if ECD is a valid theory. a transmitter. is therefore an attribute of the speciﬁc solution of the ECD equations (selected. if advanced waves could be generated just like their retarded counterparts. is set to send a retarded signal at S. Either way we get a contradiction. by the anthropic principal). It seems. is a radio transmitter made of antimatter sending advanced waves. cannot be seen as an autonomous entity.
Remarkably. the energy extracted from the incident wave needs not appear instantly as kinetic energy in the absorbing charges. arrive at the photodetector. Advanced waves must then be invoked to account for the ﬁring of a photodetector. as remarked before. Such processes are overwhelmed by ordinary radiation processes involving a huge number of particles. support a scenario in which energy is gradually accumulated by the charge in the form of latent ‘internal’ energy. and is rapidly converted into kinetic energy only when a threshold. as this amounts to increasing the energy of an electron by λω. be read from the ensemble current which. must be smaller than λω (assuming the detectors are suﬃciently far from the source). This is essentially the classical description of radiation absorption. leaving a striking signature on all radiation processes. therefore. The extended support of the ECD energy density. Let us then see why this needs not be the case. has been crossed. involving energy transfer on the order of λω. that advanced waves play a dominant role only in suﬃciently ‘delicate’ radiation process. slowly ‘removing’ energy from the incident wave. we ﬁrst need the following lemma. or in lasing devices. It appears. j r . if the single photon source is replaced by a continuous light source of arbitrary intensity. Appendices A Conservation of ECD currents To prove the conservation of the regular current.the ﬁgure on page 27) then their prevalence should have matched that of their retarded counterparts. However. or to (the equally intuitive) vanishing correlation when strongly attenuated laser light is used. That the conversion of latent energy into kinetic energy happens at the λω threshold can. then the following process. Recall that our primary motivation for introducing advanced solutions was to salvage energymomentum conservation. the integrated ﬂux of energy falling on any single photodetector. When the readings of two photodetectors are correlated (as in [6]). In the example from the previous section of a single photon source surrounded by photodetectors. but in a phase which interferes destructively with the latter. whose proof is obtained by direct computation. Retarded waves. equal to λω. in conjunction with its dynamical evolution. can be envisaged. not involving advanced waves. In the process of absorption the electrons in the device radiate in the same direction as that of the incident radiation. is indiﬀerent to the mechanism shooting the individual electrons. 31 . Acknowledgments The author wishes to thank Irad Yavneh for his useful comments. turns into the expected positive correlation when the single photon source is replaced by a continuous light source of thermal origin. which slowly absorbs them. only in ECD. originating from the source. such as the burning of a candle. it is known that the statistics of photodetection also changes when shifting to a continuous source. deﬁned in (33). the anticorrelation consistent with a particle scenario. again.
γ . s − s′ )U(ǫ. γs′ . s) = −2π 2 h2 i and its complex conjugate. we can apply our lemma to that term. which therefore reads − 2qπ h i −∞ 2¯ 2 ∞ ′ R ′ ds f (s ) 2π h i −∞ 2¯ 2 ∞ ds′′ f R (s′′ ) ∗ (73) i µ ∗ D G(x. Focusing on the ﬁrst term above. γs′ . s − s′ ) 2 U(ǫ. (33). (69) This lemma is just a diﬀerential manifestation of unitarity of the Schr¨dinger evolution— o hence the divergence. Turning now to equation (13). s − s′′ ) + ∂s U(ǫ. s) = 2π h i −∞ ∗ 2¯2 ∞ ∞ ds′ G(x. γs′ . s) be any (not necessarily square integrable) two solutions of the homogeneous Schr¨dinger equation (15). we note that. s − s′ ) U(ǫ. γs′′ . γs′ . the lefthand side vanishes (we can safely assume it goes to zero for all x. Let f (x. The reader can verify that this triple integral is just ∂µ j µ . s − s′′ ) − D µ G(x. s − s′′ ) + U(ǫ. ∗ (71) we get by direct diﬀerentiation q ∂ ∂s ¯ − 2π 2 h2 i ∞ −∞ ¯ ds′ f R (s′ ) 2π 2 h2 i ∞ −∞ ds′′ f R (s′′ ) ∗ (72) U(ǫ. s − s′ ) G∗ (x. s − s′′ ) . (71). s − s′′ ) ¯ = −2qπ 2 h2 i s′ ∞ −∞ ′ ∗ s′′ ¯ ds′ f R (s′ ) 2π 2 h2 i ∞ −∞ ′′ ds′′ f R (s′′ ) ∗ ∂s G(x. s − s′′ ) . The regular current. with j given by (11) and φ. γ . s − s′ )U(ǫ. s′′ as s → ∞). γs′′ . γs′′ . s − s′ )U(ǫ. γs′′ .Lemma. is therefore 32 . s − s′′ ) G(x. s − s′ )G∗ (x. then o ∂ i µ ∗ D f g − (D µ g)∗ f (f g ∗) = ∂µ ∂s 2 . s − s ) G (x. γs′ . −∞ (70) ds′′ G∗ (x. s) and g(x. s − s′′ ) . φ∗ are explicated using (70). γs′ . s − s′ )G(x. ∂µ Integrating (72) with respect to s. as G is a homogeneous solution of Schr¨dinger’s o equation. s − s′′ )G∗ (x. and the derivative ∂µ can be pulled out of the triple integral in the ﬁrst term. s′ . s − s′ )∂s U(ǫ. s − s′ ) . written for the rescaled wavefunction ǫ−1 φ ¯ φ(x. s − s′′ ) G(x. s − s′′ )f R (s′′ )U(ǫ. s − s′ )f R (s′ )U(ǫ. φ (x. s − s ) U(ǫ.
s′′ + ǫ − s′ )G(x. it is easily shown that j r is a smooth function in the limit ǫ → 0. γs′ . s′ + ǫ)G(x. Expanding ﬁrst φs ∗ (x. But. γs′′ . Using (16). Integrating the second term with respect to s. Let us then show that. parametrized by proper time ¯ s 2 . to get a conserved quantity. −∞ (75) Writing φ = ǫ−1 φs + φr above. implying a pointwise identity ∂ · j r = 0. −ǫ) + φ∗ (x. the ﬁrst term behaves for small ǫ like ǫ−1 δ (ξ 2) + O(ǫ). Let β(τ ) = γ (s(τ )) be the world line γ . 33 and η 0 > 0 . do not involve the ǫ0 coeﬃcient. s′ − ǫ − s′′ ) +U(ǫ. γs′ . this is indeed the case. This is enough to establish the timeindependence of the charge. of φs . +ǫ)G∗ (x. while the second — as ǫ−1 δ ′ (ξ 2 ) + O(ǫ).conserved. that term reads − 2qπ h i −∞ ′′ 2¯ 2 ∞ ′ R ′ ds f (s ) 2π h i −∞ 2¯ 2 ∞ ds′′ f R (s′′ ) ∗ (74) U(ǫ. s′ − ǫ)G(x. s′′ + ǫ − s′ )G∗ (x. T . s′ − ǫ − s )G(x. γs′ . supported on γ . in the distributional sense. γs′′ . s)δ (4) (x − γs ) = −∞ ds Im f R (s)φr∗ (γs . and let x → τ be the retarded lightcone map deﬁned by the relations τ= (dγ) r η 2 ≡ (x − βτr )2 = 0 . the construction of which ¯ proceeds as follows. therefore. . (77) . with fs (y) = sinc(y) cos(y) = fs (−y). and apply Stoke’s theorem. which is due entirely to φr . s′ + ǫ − s′′ ) +U(ǫ. s′′ − ǫ − s′ )G∗ (x. (22). which vanishes by virtue of (27). −∞ (76) This is a distribution. both. consider a small spacelike threetube. and using the shorts propagator (19) plus the explicit form. γs′ . and using ∂s U(ǫ. −ǫ) +U(ǫ. as one only needs to integrate ∂ · j r = 0 over a volume in Minkowski’s space. s − s′ ) = δ(s − s′ − ǫ) + δ(s − s′ + ǫ). s ± ǫ) in powers of ǫ. γs′′ . provided the s integral over the second term in (72) is missing an ǫ0 term in its ǫexpansion. surrounding γ . −ǫ)G∗ (x. this becomes ¯ Re − 4qπ 2 h2 i ∞ ds′ f R (s′ ) φ∗ (x. s′ + ǫ − s′′ )G(x. We have therefore ¯ r shown that ∂ · j = 0 in the distributional sense. s)δ (4) (x − γs ) . Using the evenness and oddness of fs and fa resp. +ǫ) . γs′ . γs′ . and another ǫindependent −3 ¯ term multiplying ǫ fa (ξ 2 /2hǫ). one can obtain the ǫexpansion of (75). γs′′ . the part of the integrand involving φs can be shown to comprise an ǫindependent term ¯ multiplying ǫ−2 fs (ξ 2 /2hǫ). To gain a more explicit geometrical insight into the meaning of a ‘line sink in Minkowski’s space’. Using (70) and (71). with fa (y) = sinc(y) sin(y) = −fa (−y). ǫ) . s′′ − ǫ − s′ )G(x. the latter’s contribution reads in the limit ǫ → 0 ¯ − 8qπ 2 h2 ¯ 8qπ 2 h2 ∞ ∞ ds Re i f R (s)φr∗ (γs . in fact.
This piece. Applying Stoke’s theorem to the interior of the three surface composed of Tρ . of (81) vanishes for R → ∞. 34 . we get dΣ2 · j r + Σ2 Σ1 dΣ1 · j r = − Tρ dTρ · j r − TR dTR · j r .sa Im φr∗ (γs . and solving for ∂τr . ξ0 < 0 . reads ¯ 2q h2 1 ds Im φr∗ (x. As dTρ = O(ρ2 ).s. and using ∂ · j r = 0 there. using ∂ ξ·γs · ∂rr=ρ → m−1 . s)f R (s)∂δ ξ 2 1 . (79) r r The (retarded) threetube of radius ρ is deﬁned as the spacelike three surface Tρ = {x ∈ M : r(x) = ρ} . (80) where dΩ is the surface element on the twosphere. (81) Realistically assuming that the second term on the r. we get η η ¨ ˙ ∂τr = ⇒ ∂r = βτr − 1 + βτr · η . intersecting Tρ and TR . the leakage only involves the piece of j r diverging as r −2 . s)f R (s)∂ ¯ sinc 2hǫ ξ2 ¯ 2hǫ ¯ −→ 2q h2 π ǫ→0 ds Im φr∗ (x. s)f R (s)δ ξ 2 ¯ = q h2 π s=sr . TR . equals to − limρ→0 Tρ dTρ · j r . Σ2 dΣ2 · j r − Σ1 dΣ1 · j r . we get that the ‘leakage’ of the charge. Focusing ﬁrst on the contribution of sr . Let Σ1 and Σ2 be two timelike surfaces. ξ · γs  ˙ ¯ ∼ 2q h2 π ∂ ds Im φr∗ (γs . we get ˙ γ 2 + γsr · ξ ξ ˙ ¨ 1 η γsr ˙ βτ ∂ . sr )f R (sr ) . and γsa is the corresponding advanced point on γ .h. ˙ r the contribution of sr to the ﬂux across Tρ is most easily computed ¯ dTρ · j r = q h2 π Tρ dΩ dτr m−1 Im φr∗ (βτr . (78) Taking the derivative of (77). In the limit ρ → 0. and using a technique similar to that leading to (79). It can be shown in a standard way that the directed surface element normal to x ∈ Tρ is dµ Tρ = ∂ µ r r=ρ ρ2 dτ dΩ . treating τr as an implicit function of x.Let the ‘retarded radius’ of x be ˙ r = η · βτr . (82) =− + sr ∼ − r2 + 2 3 ξ→0 ξ · γsr ˙ mr mr 3 (ξ · γsr ) ˙ (ξ · γsr ) ˙ 1 where m = dτ /ds needs not be constant. Σ1 and Σ2 . τr )f R (τr ) (83) ¯ = 4q h2 π 2 dsr Im φr∗ (βsr . s)f R (s) ∂ ¯ where sr = s (τr )). deﬁned by ξ 2 ≡ (x − γsa )2 = 0 .
γs+ǫ . as well as to illustrate the role played by symmetries of ECD in the context of conservation laws. T = ∂C. when C is shrunk to γ . with inﬁnitesimal a(x. A. +ǫ) f R (s − ǫ) + G (x. Using ¯ i∂s − H φ = 2π 2 h2 G (x. s) be given by (70) for some ﬁxed A(x) and γs . To explore yet another technique. (84) and let φ(x. so as to render δA well deﬁned. in conjunction with the conservation of j r (x) for x ∈ γ . (86) = (87) ds −∞ M ¯ d4 x Re 4π 2 h2 G (x. and gives the same result in the limit ρ → 0. equals twice the value ¯ in (83). s) aµ . which vanishes by virtue of (27). −ǫ) f R (s + ǫ) δφ . +ǫ) f R (s − ǫ) + G (x. consider the following functional ∞ A[ϕ] = −∞ ds M d4 x ¯ ih ∗ 1 Dλϕ (ϕ ∂s ϕ − ∂s ϕ∗ ϕ) − 2 2 ∗ Dλ ϕ . −ǫ) f R (s + ǫ) . the formal ¯ content of (76) receives a clear meaning using Stoke’s theorem ¯ d4 x ∂ · j r = 8q h2 π 2 C ds Im φr∗ (βs . we calculate A [φ + δφ] and. Changing the dummy variable sr → s in (83). γs−ǫ . (85) directly following from the deﬁnition of φ. vanishing suﬃciently fast for large x and s. we get in a standard way ∞ δA = −∞ ∞ ds M d4 x ∂ν mνµ − F µ j ν aµ ν by eq. γs−ǫ . after some integrations by parts.1 Energymomentum conservation The conservation of the ECD energy momentum tensor can be established by the same technique used in the previous section. s)f R (s) C d4 x δ (4) (x − γs ) = T dT · j r . (advanced) Tρ .c. −ǫ) f R (s + ǫ) ∂ µ φ∗ (x. s). s) → φ(x + a. implies that the ﬂux of j r / ¯ across any threetube. (88) 2 . γs+ǫ . we get for the ﬁrst variation ∞ δA = Re −∞ ds M ¯ d4 x 4π 2 h2 G (x.The contribution of sa to the ﬂux of j r is more easily computed across a diﬀerent. ds . γs−ǫ . with j given by (11) and ∞ m νµ = −∞ g νµ ¯ 1 ih ∗ Dλφ (φ ∂s φ − ∂s φ∗ φ) − 2 2 35 ∗ Dλ φ + 1 ν D φ (D µ φ)∗ + c. γs+ǫ . (86) Choosing δφ = ∂φ · a. corresponding to φ(x. +ǫ) f R (s − ǫ) + G (x. with C a threecylinder containing γ . The fact that ρ can be taken arbitrarily small. s).
(90) ν with the regular ‘matter em tensor’. Writing φ = ǫ−1 φs + φr in (87). s) which vanishes by virtue of (26) for any a. Equation (26). Summing (90) over the diﬀerent particles. implying pointwise equality in (90). A(x) → A(x + a) ⇒ δAµ = ∂ν Aµ aν . as also the integrands appearing in the deﬁnitions of all other ECD currents. (22). we construct an ‘equally non conserved’ radiation em tensor. of φs . and subtract the two. In the limit ǫ → 0 it reads ¯ 8π 2 h2 ¯ 8π 2 h2 ∞ ds −∞ ∞ −∞ M d4 x Re f R (s)δ (4) (x − γs )∂φr∗ (γs . appears therefore as the condition that no mechanical energy or momentum leak into a sink on γ . (k labels the diﬀerent particles) S A = 1 d4 x Fµν F µν + 4 M k r k j · A. one can obtain the ǫexpansion of (87). s) · a(γs . due to broken translation covariance induced by A(x). we get a conserved. in the ﬁrst line of (87). To compensate for this. the matter em tensor can easily be shown to be a smooth function of x. deﬁned by the same procedure as j r . we get Maxwell’s equations. As before. ∂ν pνµ = 0 . coeﬃcient of ǫ0 ) only involves φr . and using the shorts propagator (19) plus the explicit form. viz. Consider. can be shown to be a distribution which becomes increasingly focused on the lightcone of γs for increasing distance from γ s . (4). using Noether’s theorem. symmetric em tensor. vanishes in the distributional sense. s) · a(x.. then. viz. ¯ Not surprisingly. The regular part of the second line (viz. symmetric and traceless) ‘radiation em tensor’. with k r m . inﬁnitesimally shifting the argument of an extremal A. mr . Just like the electric current j r . (92) 1 Θνµ = g νµ F 2 + F νρ Fρ µ (93) 4 the canonical (viz. The arbitrariness of a implies that the regular part of the expression in brackets. and following a standard symmetrization procedure of the resultant em tensor (adding a conserved chargeless piece ∂λ F νλ Aµ ) leads to ∂ν Θνµ + F µ ν k k rν j = 0. by which (89) vanishes.The integrand above. (94) p=Θ+ with k 36 . and adding to (92). (91) By the Euler Lagrange equations. ∂ν mr νµ − F µ j r ν = 0 . s) = (89) ds Re f R (s)∂φr∗ (γs . k. with k kj r as a source. the coeﬃcient of ǫ0 in its ǫexpansion. mr is not conserved. for ﬁxed kj r . the following functional of A(x).
used to establish the conservation of j r . In the variational approach. (97) The leakage to the sinks on kγ is due to the second term. s) → λ−2 φ λ−1 x. (98) i where σµν = 2 [γµ . λ−2 s . φ(x. transforming in a Lorentz transformation according to ρ (eω ) φ ≡ e−i/4 σµν ω φ . the counterpart of (69) is ∂s (g ∗ Hf ) = ∂ · (Re ∂s g ∗Df ). for x ∈ γ . can be applied to prove the conservation of the regular part of the masssquared current — the counterpart of (2) b(x) = ds B(x. is given by ¯ ¯ 8π 2 h3 s2 s1 ds Re ∂s φr∗ (γs . involving the masssquared of the ¯ particles. corresponding to the invariance of the Hamiltonian (in the Heisenberg picture) under the Schr¨dinger evolution. transforming under the adjoint representation. the cono servation follows from the (formal) invariance of (84) φ(x. Note that this leakage (whether positive or negative) is a ‘highly ¯ ¯ quantum’ phenomenon — proportional to h2 (the term ∂s φr generally diverges as h−1 ). with γµ Dirac matrices (not to be confused with γ the trajectory). γν ]. also modiﬁes the scalecharge of a solution.1 ECD 2 1 In a spin. /¯ (95) In the ﬁrst method. However. s) → φ(x. the following modiﬁcations are made. the counterpart of the classical current (7). between γs1 and γs2 . x′ . The wavefunction φ is a bispinor (C4 valued). is a locally conserved dilatation current. µν for eω ∈ SO(3. s + s0 ). satisfying g ¯ ih∂s G(x.A. s) . ∞ ξ µ = pµν xν − k 2 −∞ ds s kB . (96) is not guaranteed to vanish. The propagator is now a complex. 2 37 (99) . s) ≡ ¯ ds Re h∂s φ∗ Dφ .2 version of ECD. s) = H + σµν F µν (x) G(x. the leakage to the sink on γ . x′ . B Spin. s)f R (s) . Similarly. 4 × 4 matrix. therefore. 1) .2 Charges leaking into worldline sinks Both methods used above. with B deﬁned in (95) . associated with the formal invariance of (84) under A(x) → λ−1 A λ−1 x . A leakage of mass.
r (103) The counterpart of (90) becomes (omitting the henceforth) ∂ν kmorb νµ + g νµ kl = F µ kj orb ν + ν g 2 ds k identiﬁer. and g is some dimensionless ‘gyromagnetic’ constant of the theory. 38 . ∂s ϕ) − (∂s ϕ. Each of the terms composing j r is individually conserved and gauge invariant. (101) 2 2 2 while the counterpart of the electric current. b) in all bilinears. Calculating in an arbitrary frame. σ νµ φ) . which can be seen as a Lorentz invariant scalar product in C1 . and assuming j spn i (x) → 0 for x → ∞ d3 x j spn0 = d3 x ds ∂0 (φ. j r . (100) with γ 0 the Dirac matrix diag(1. as regularization is implied φ. gets an extra spin term ∞ As [ϕ] = −∞ ds M d4 x ¯ ih 1 g (ϕ. Deﬁning an inner product in spinor space (instead of C1 ) (a. using the antisymmetry of σ. viz. the counterpart of (22). (σ µν )† = γ 0 σ µν γ 0 . This current has an interesting property that its monopole vanishes identically. /¯ (102) Expanding (102) in powers of ǫ.2 ECD is rendered easy by the observation that all expressions in scalar ECD are sums of bilinears of the form a∗ b. for x ∈ γ . leading to an equally simple φs . and from (γ 0 ) = 1. The action. (84). from which all conservation laws can be derived. (11). Fλρ σ λρ φ . σ 00 φ) − ∂i (φ. By a direct calculation of the shorts propagator of (99). σ λρ kφ ∂ µ Fλρ . σ i0 φ) = 0 − 0 = 0 . from which the regular part of all ECD currents can be obtained. derived from φ. b) ≡ a† γ 0 b . D µ φ) − g∂ν (φ. associated with a particle. as in section 2. b) in all bilinears. not to be confused with γ the trajectory) and substituting a∗ b → (a. 1 The transition to spin. is now a sum of an ‘orbital current’ and a ‘spin current’ j µ (x) ≡ j orbµ + j spnµ = ds qIm (φ.2. Let us illustrate this procedure for important cases. and l(x) = g 2 ds φ. The Lorentz invariance of (100) follows from the Hermiticity of σ µν with respect to that inner 2 product. for x ∈ γ . the coeﬃcient of ǫ0 is the regular current. Dλ ϕ + ϕ. all the results of scalar ECD are retained.with the initial condition (16) at s → 0 reading δ (4) (x − x′ )δαβ . the spin can be show to aﬀect the O(s) terms in the expansion of Φ.1. −1) (again. ϕ) − D λ ϕ. 1. /¯ (104) with morb the same as (88) with a∗ b → (a. −1. Fλρ σ λρ ϕ . The conservation of the orbital current follows from the U(1) invariance of (101). where δαβ is the identity operator in spinorspace. while conservation of the spin current follows directly from the antisymmetry of σ.
2010. [8] J. Grangier. D. Phys.. as the extra terms involving spin. Mod. σ νλ F λµ kφ = 0 . Feynman. 39 . (105) Summing (104) over k. Scully and H. adding M d4 x l(x) to the functional in (91). Argaman. we get the locally conserved em tensor Θνµ + k k morb νµ + g νµ kl + g ds k φ.O. 74:939– 946. which adds up to the Lorentz force density. [5] Y. equation (92) becomes ∂ν Θνµ + k F µ kj orb ν + ν g 2 ds k φ. Phys. do not contain derivatives of φ. 17:157–181. 1980.Note the ‘spin force’ density. J.. Classical Electrodynamics in Terms of Direct Interparticle Action. x ∈ ∪k kγ . Aspect P. u Zeitschrift f¨r Naturforschung. [4] R. G. and adding to (105).. Feynman. Scale covariant physics: a ’quantum deformation’ of classical electrodynamics. Feynman. [7] J. [6] A. Phys. Bells theorem and the causal arrow of time. A. 47a:1175–1186. Rev. 1:173–179. Cramer. Walther. 43:055401. [2] G.. / ¯ (106) from which the timeindependence of the associated charges follows as in the scalar case. Am. vanishing for a constant F . Roger.. A relativistic cutoﬀ for classical electrodynamics. 1948. M. Generalized absorber theory and the EinsteinPodolskyRosen paradox. A.G. S¨ ssman B. u [3] John G. References [1] N. Phys. Lett. Phys. 22:362–376. Rev. 1992. Rev. Rev. Interaction with Absorber as the Mechanism of Radiation. Wheeler and R. Europhys. A. Similarly. 1945. Experimental evidence for a photon anticorrelation eﬀect on a beam splitter: A new light on single photon interferences. 1986. Mod. J. P. 1949. Knoll and I. 21:425–433. P. Englert. σ νλ F λµ kφ . Surrealistic bohm trajectories. 78:1007–1013. 2010. Yavneh. Phys. Wheeler and R. P. σ λρ kφ ∂ µ Fλρ + ∂ν g ds k φ.