How do you interpret quadratic terms in the gauge field in a Lagrangian?

[QuantumSkippy]

Consider a one dimensional gauge theory where the field has mass. The term,

m^{2}A^{\mu}A_{\mu}

is the conventional mass term. What if you find terms in your Unified Field Theory lagrangian of the form

M_{\mu\nu}A^{\mu}A^{\nu} ?

In this case M_{\mu\nu} is constant.

When it is not the case that

M_{\mu\nu}

is of the form

m^{2}g_{\mu\nu} ,

are these to be interpreted as self-interaction terms, or self-interaction terms somehow related to mass for the gauge field, or as bona fide mass terms?

[chrispb]

As far as my understanding goes, it's the mass eigenstates that would propagate. This is what we (I, at least) believe happens with neutrinos (with a spinor instead of a vector field, of course). You'd diagonalize your M matrix and find that the mass eigenstates are not identical to flavor eigenstates, but mass eigenstates are ones that diagonalize the Hamiltonian and therefore evolve under an e^(-iHt). Writing down the mass eigenstate at t=0, time-evolving it, then rewriting it in the flavor basis allows you to see the now-popular neutrino oscillations. What IS peculiar, as far as I know, is the fact that the neutrino masses are so small.

[chrispb]

I should also point out that in QED, a mass term like that breaks local gauge invariance and is therefore generally disallowed.