Definition: The freedom to vary the phase of the field operators is the gauge symmetry of qed.
The term
gauge symmetry» is something of a misnomer. It was introduced by
Herman Weyl», as part of an attempt to extend the local scale invariance of general relativity to unification with electrodynamics. That attempt failed, but later Weyl,
Vladimir Fock» and
Fritz London» adapted the idea and applied it to phase symmetry in quantum theory, and it is to phase symmetry that the term now applies. The relation to the of this phase symmetry to a corresponding symmetry in classical electrodynamics is shown below.
In the absence of interactions, there is no issue with local gauge freedom. The phase of an electron wave function is fixed at the point of creation and becomes simply the global symmetry of the one particle theory, in which kets can be multiplied by constant phase without altering their meaning in
quantum logic. When interactions are introduced the result is that the evolution of the wave function does not match the evolution of the field operator which created it, and which is defined on the non-interacting space. A difficulty arises because the momentum observable in the non-interacting theory,
extracts the frequency and wavelength of the the wave function. When the simplified notation,
i∂a, is used as an operator on ket space, the integral, the bra and the ket are implicit. We would like to use
Ehrenfest’s theorem to calculate the classical force due to the interaction, by differentiating the expectation of momentum,
but states,

evolve according to the full Hamiltonian, whereas the creation operators are defined on the Fock space of non-interacting particles, and create states obeying the Dirac equation. There is a real phase shift corresponding to change in momentum, which must be distinguished from the arbitrary phase in the definition of field operators.
To ensure that creation operators and states evolve identically, we define the field picture,
In the field picture states evolve as in the
Schrödinger picture for non-interacting particles. The momentum operator in the field picture is
In the semi-classical correspondence, for small t, this may be treated as a perturbation to the evolution of a non-interacting particle, in which the interaction is replaced with an expectation. For a classical particle with position x and velocity

, the classical current is

. The expectation of the interaction Hamiltonian is
This may be done more precisely, taking spin and antiparticle states into account, using the
Foldy-Wouthuysen Transformation. For a general discussion, see
Costella & McKellar Am.J.Phys. 63 (1995) 1119. For the present treatment, I will merely show the Lorentz force law for particles, ignoring spin.
Replacing the interaction Hamiltonian with its expectation, the momentum operator in the field picture is
Thus the expectation,

, of the operator which creates and annihilates photons, acts in the manner of a classical vector field, modifying energy and momentum. This is the standard formula for momentum in the presence of a field, but normally it would be assumed, on phenomenological grounds, i.e. because we know it works. For that one must also assume a concept of a classical field defined on a spacetime background. The treatment here has assumed only particles, electrons and photons, and has arrived at the classical field in an approximation in which it is an expectation of photon creation/annihilation.
The Interacting Dirac Equation
With the replacement
The Dirac equation
becomes the <a href=
http://en.wikipedia.org/wiki/Dirac_equation#Coupling_to_an_electromagnetic_field>interacting» Dirac equation</a>
The Interacting Dirac Equation:
Theorem: The Lorentz Force law,
Gauge Invariance
The
interacting Dirac equation,
can be written, in terms of creation operators, acting on any ket
A local gauge transformation applied to the creation operators
gives
So,
which is identical to the original form of the interacting Dirac equation apart from the replacement
But the Faraday tensor is also unchanged by this replacement,
So the local phase symmetry of the field operators is precisely equivalent to the well known symmetry of the classical electromagnetic field.
Maxwell’s Equations