Just a simple question for lattice QCD experts, is continuum limit of lattice field theory a relativistic quantum field theory?

Because i heard that lattice QCD is done in imaginary time, producing a ground-state (lowest-energy state) in infinite temporal extend limit. And that two-point correlation functions computed in imaginary time can be analytically continued to real-time. But for massive particles (such as quarks) to be relativistic, their kinetic energies should be equal to or greater than $mc^2$.


1 Answer 1


This isn't really specific to lattice theory - doing stuff in "imaginary time" means you're doing Euclidean QFT. This isn't so much "relativistic" or "non-relativistic" as it is just a formal computation - no one thinks this Euclidean theory directly describes any physical setting.

However, analytic continuation of the results to "real time" then gives you quantities for a relativistic Minkowskian theory. It is not at all obvious that analytically continuing the results of the a priori unphysical Euclidean theory should have anything to do with what the physical Minkowskian theory does, but this is the miracle of Wick rotation - it turns out that this actually works (although in many cases you need to think a bit carefully about what Euclidean theory corresponds to what Minkowskian theory, see e.g. this question and its answers), and the formal statement of this is called the Osterwalder-Schrader theorem.

  • $\begingroup$ Well, my question was more about continuum limit of lattice field theory. Is only a data extrapolated in continuum limit gives a relativistic QFT after analytic continuation? Does it mean that a lattice with any finite lattice spacing is non-relativistic? $\endgroup$
    – Peter
    Sep 13, 2022 at 16:50
  • $\begingroup$ @Peter I don't understand the distinction you're making here - the continuum limit of a Euclidean lattice field theory is presumably a Euclidean QFT, no? Wick rotation doesn't care how you produced your Euclidean QFT. $\endgroup$
    – ACuriousMind
    Sep 13, 2022 at 16:52
  • $\begingroup$ @Peter there’s two issues that are separate, the continuum limit and wick rotation. At finite lattice spacing, your Lorentz/Euclidean group of symmetries is broken by the lattice and you get discretisation artifacts. And usually people use relativistic to mean the real-time theory, but this is separate $\endgroup$ Sep 13, 2022 at 16:53
  • $\begingroup$ Thanks, now it's clear. $\endgroup$
    – Peter
    Sep 13, 2022 at 17:04

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.