Another particle physicist's view and my comments
-------------------------------------------------
The particle physicist Gordon Kane wrote in the Scientific American
article (directed to lay people), available at
http://www.scientificamerican.com/article.cfm?id=are-virtual-particles-rea
the following:
''Virtual particles are indeed real particles. Quantum theory predicts
that every particle spends some time as a combination of other
particles in all possible ways. These predictions are very well
understood and tested.''
Unfortunately, the article fails to give references to support
the claim that virtual particles are indeed real particles.
There is not a single verified measurement of a virtual particle.
Measured particles are always on-shell.
The tests mentioned are not tests of the reality of virtual
particles, but tests of QED and the standard model in general,
together will talk about the behavior of virtual particles that
may be a common fantasy of many physicists, but has no support at
all by experiments.
The Lamb shift and the Casimir effect can be calculated in a number
of different ways, not all requiring the concept of virtual particles.
High precision calculations of the Lamb shift using NRQED (and these
are the most accurate calculations available) are not based on virtual
photons. How can it then be experimental proof of the latter?
Even when virtual particles are involved, their alleged properties
are very different in different approximation techniques for the same
quantum field theory. For example, the virtual particles in light
front calculations have nothing in common with the virtual particles
in the standard covariant presentations of perturbation theory.
On the Z-boson, the article writes, ''the value agreed with that
obtained from the virtual particle analysis, providing a dramatic
test of our understanding of virtual particles.''
This virtual particle analysis (when followed through - no
references are given) consists in perturbative calculations
that start with Feynman diagrams involving virtual particles.
But these are unmeasurable things with an infinite mass and
infinite interactions, both physically meaningless. They are turned
into predictions only by a perturbative renormalization process
that restores physics to the ill-conceived bare interpretation.
After renormalization, when everything is finite, there is not
the smallest trace left of virtual particles. Virtual particles are
associated with individual Feynman diagrams (which diverge), not with
their convergent, renormalized sum. In the sum, there is no
recognizable space-time behavior of anything that could travel.
Moreover, modern nonperturbative renormalization techniques via
lattice gauge theory or similarity flows give comparable predictions
for quantum field theories, without encountering virtual particles
anywhere.
It is clear to anyone who learnt more than one approach to calculations
with quantum fields that the intermediate expressions that arise cannot
have any reality since they occur only in a particular approach.
Giving virtual reality to intermediate expressions in a technique to
calculate numbers with physical meaning is like saying the terms in
the series pi/4=1 - 1/3 + 1/5 - 1/7 + 1/9 - ... represent virtual
pieces and antipieces of the quarter-circle. But someone else computes
pi from pi^2/12=1 - 1/4 + 1/9 - 1/16 + 1/25 - ... and gets virtual
pieces with completely different properties! Of course, we have no
reason to take such virtuality seriously. It lives in quantum field
theory for the sole reason that the formulas are abstract and messy,
and couching them in a pictorial language with comprehensible names
and an illusion of reality makes them easier to remember and to refer
to.
It is the perturbative treatment -- and only that -- that generates
virtual particles as internal lines in diagrams that represent
multidimiensional integrals. Different perturbative approaches
of course give different integral representations and hence virtual
particles with different properties. And nonperturbative approaches do
not see any of these ghosts. Look at an _arbitrary_ book on lattice
gauge theory and try to find virtual particles. You'll not find them,
except perhaps when some results are compared to corresponding results
from perturbation theory.
If virtual particles were not just mathematical artifacts, the following
three questions would need a convincing answer:
- Why then do they behave so differently depending on the perturbation
scheme used?
- Why then is the mass of virtual electrons infinite?
This has never been observed.
- Why then do the values of all diagrams in QED involving virtual
particles come out infinite?
The only places where one can perhaps assign virtual particles some
physical meaning (though I think even this is misguided) are
- (i) in nonrelativistic quantum mechanics, where electromagnetic
fields are represented either by explicit Coulomb terms or by
external e/m potentials, and there are no photons,
- (ii) in a tree approximation to relativistic QFT; but this is
not enough for any of the great successes of QED.
Coulomb forces, Casimir forces, and the like (i.e., electromagnetic
fields) are real, the photons used to describe them in perturbation
theory cannot be.
None of the many very successful calculations comparing theory with
experiment depends on the reality of such virtual particles.
Photons are transverse excitations of the e/m field, i.e.,
deviations from the mean value of the field. To consider the
Coulomb field, say, as being composed of virtual photons is a
superficial misunderstanding of quantum field theory.
The Coulomb field has nothing to do with photons. It arises as an
interaction term in the QED Hamiltonian in the Coulomb gauge,
by the standard machinery that leads from an action to a Hamiltonian.
(One can read this in the quantum field theory
book of Bjorken and Drell,
http://en.wikipedia.org/wiki/James_Bjorken
which was for many years the textbook from which people learnt the
ropes, before path integral techniques took over.)
The correct way -- which is in full accord with renormalized
quantum field theory and remains valid nonperturbatively -- to think
about particle creation, annihilation, or scattering processes is
that particles are excitations of fields. The latter create and absorb
particles in a similar way as flowing water creates and absorbs
water waves.
For example, a charged pion is surrounded by (and in fact
part of) a weak-force field, which absorbs the pion and then creates
two leptons. During the interaction described as the decay of a
charged pion via a virtual W minus boson into a two leptons,
there is neither pion nor W-boson nor leptons, since the identification
of individual excitations requires the latter to be well enough
separated so that the asymptotic S-matrix regime applies. Only then
the particle nature of the field is apparent.
For more discussion on virtual particles, bare feynman diagrams, and
the like, see Chapter A7 of the theoretical Physics FAQ at
http://www.mat.univie.ac.at/~neum/physfaq/physics-faq.html