Quantum Field Theory in Curved Spacetime: Quantized Fields and Gravity (Cambridge University Press, Cambridge, England, 2009)

The validity of the adiabatic renormalization procedure has been recently brought into question in [15, 16], but then counter-arguments have been given in [17]. Some of the results in [15, 16] are, however, in contrast with basic principles of renormalization in curved spacetime. For example, the introduction of a infrared cut-off proposed in [15, 16] to eliminate subhorizon modes for which adiabatic expansion is not mathematically accurate, gives the physically unacceptable result of having a non-conserved renormalized stress tensor (see, e.g., [18])

As a check of the above results, we can calculate the transverse part of the renormalized energy-momentum tensor for the Proca field and then take the limit m → 0. We straightforwardly obtain that it is twice the renormalized energy-momentum tensor for a massless conformally coupled scalar field, as it should be [20]

A delta-function spectrum of the form P(k) ∝ k δ(k), is the only type of spectrum that gives a true scaleindependent magnetic field. A scale-invariant spectrum P(k) ∝ constant usually used in the literature to define a scale-independent field is both infrared and ultraviolet divergent. Introducing an infrared and an ultraviolet cut-off, kmin and kmax respectively, one straightforwardly finds that the associated field is not constant but scales logarithmically as ln(1/λ) in the range 2π/kmax λ 2π/kmin

The fact that the physical magnetic VEV is (scale-) and time-independent, should not come as a surprise. Indeed, the same is true for the renormalized electromagnetic energy density (see below). Moreover, this result is also confirmed in the ζ-function renormalization approach. To see this, we observe that 0|B(x)2 |0 ∼ ∞ 0 dkk3 /a4 ∼ H4 ∞ 0 dxx3 ∼ H4 ∞ n=1 n3, where we introduced the dimensionless variables x = -kη, and roughly approximated the integral by a sum. The divergent sum can now be cured by analytical continuation [19] by replacing it with ζ(-3) = 1/120, where ζ(x) is the Riemann ζ-function. This gives the renormalized and finite result 0|B(x)2 |0 phys ∼ H4

We expect that the results obtained in de Sitter inflation hold, with minor modifications, also in more realistic models of inflation. This because while the Hubble parameter sets the scale of energy during inflation, giving 0|B(x)2 |0 phys ∼ H4, the mass of the field sets the scale of correlation. For the photon case, then, the renormalized vacuum magnetic fluctuations are allowed to be correlated on arbitrarily large distances