Calculating loop correlators, Pauli-Villars regularization is introduced to avoid divergence. It is to cut off the high frequency(loop-momentum) contribution.

Thus a question naturally arises. Why is the high frequency contribution unphysical?

In Schwartz QFT book explaining the Casimir force, he said that since the high frequency photons does not interact with the walls we can cut off its contribution.(If possible, plz explain the reason)

However, in the case of free theory the regularization is also used. So I'm confused. What is the reason we can discard the contribution of the high momentum?


2 Answers 2


Naive formulations of QFT are often ill-defined to begin with, and this shows up in calculations as UV divergences. This problem would go away if we started with a theory that is mathematically well-defined. In many cases we can do this by replacing space with a discrete lattice (which eliminates UV divergences), but that's obviously artificial. In some cases we can do it without such contrivances, but a general understanding of how to do it properly is still somewhere in the future.

Pauli-Villars is an approach to fixing this within the context of perturbation theory. It is a perturbative regulator. What does it mean physically? Nothing, really. On the contrary, like the non-perturbative lattice regulator, it's just an artificial way of protecting our calculations from what we don't know (or don't want to mess with) about the extreme UV physics.

Here's an excerpt from pages 866-867 in DeWitt (2003), The Global Approach to Quantum Field Theory (two volumes, Oxford University Press):

In [the Pauli-Villars scheme] one replaces Feynman propagators by differences between Feynman propagators corresponding to fields having different masses. ... It does not correspond to any `vacuum' expectation value and cannot be obtained, with any boundary conditions, from the functional integral for the model. Therefore the Pauli-Villars scheme can in no sense be regarded as a 'physical' modification of a quantum field theory, even a modification that allows states of arbitrarily negative energy.

Regarding the Casimir effect, maybe what Schwartz meant is that the magnitude of the effect is not sensitive to the precise value of the UV cutoff, whether implemented using Pauli-Villars or a lattice. The Casimir effect, at least the simplest version of it, can be modeled using boundary conditions on a free-field model instead of as material plates in an interacting model. In the free-field approach, we can do the calculations exactly with a suitable regulator (like a lattice) and confirm that the force has a finite limit as the UV cutoff is sent to infinity. In this sense, the high-frequency modes don't affect the result. I don't have that book on hand, but I'm guessing that this is probably what Schwartz meant.

The Casimir effect can be calculated using any regulator, and can probably even be calculated without using a regulator in the free-field case. The calculation can also be done using a lattice regulator, and it has a finite limit as the lattice spacing becomes much smaller than the distance between the "plates" (modeled artificially as boundary conditions). Which regulator we use shouldn't matter, because we end up removing the cutoff anyway.

Different regulators may have different advantages in different contexts. Pauli-Villars is nice for preserving gauge invariance, if the symmetries allow mass terms. This is a matter of mathematical convenience, though, not a matter of physical significance.

  • $\begingroup$ Thank you. I have one more question. Is the reason why UV region is excluded related with the Landau pole Region where QED breaks down? In cutting ignorance, does "ignorance" mean the regions where the theories do not work well? QCD in IR, QED in UV $\endgroup$
    – Summal
    Jun 3, 2019 at 8:42
  • 1
    $\begingroup$ @Summal The extreme UV region is excluded because that's currently the only way we know how to define many QFTs (like QED and QCD), and excluding the extreme UV is okay because no real experiment has infinite resolution anyway. QCD works well in the IR (perturbation theory doesn't, but QCD itself does), and it almost certainly has a well-defined limit as the UV cutoff is removed (but this hasn't been proven). QED might not, but this is unclear, because the Landau pole is a perturbation-theory prediction that might not really be valid in the extreme UV. $\endgroup$ Jun 4, 2019 at 0:37

I will probably write a detailed answer later but for now just google RENORMALIZATION IN ELECTRIC FIELD OF CHARGED WIRE you will see here also you will see same kind of integral divergent and pauli type regulations actually gives the answer you expect from gasses law. However for conceptual reason I prefer dimensional regularisation as it has direct physical interpretation in renormalization group (see SHANKAR CONDE MATTER FIELD THEORY for details connection) Also you can physically cook up two dimensional electric field formula from there. I can not remember where I saw it my Prof told me. The idea feynmann had was probably to assume the mass of photon not zero but some dependency on energy (just like in Debyes theory you assume kis dependent on energy) Then after calculating the hanging term is the inverse of photon mass which goes to infinity ideally but can be putted after calculations. sorry for terse answer first answer in physics stack Exchange Hope it helps.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.