Title: A critical look into Luttinger's theorem

Abstract: Breakdown of Landau's hypothesis of adiabatic continuation from non-interacting to fully interacting electrons is commonly believed to bring about a violation of Luttinger's theorem. Here, we elucidate what may go wrong in the proof of Luttinger's theorem, which does not concern the non-existence of Luttinger-Ward functional but rather its analytic properties. The analysis suggests a simple way to correct Luttinger's expression of the electron number in models where perturbation theory breaks down close to a half-filled Mott insulator. We show that the modified formula yields the right electron number in well-established examples where Luttinger's theorem is violated. Moreover, in periodic systems it implies that, when only a Luttinger surface exists without Fermi pockets, the electron density gets stuck to half-filling as in the nearby Mott insulator, and thus the state is incompressible. In the hypothetical circumstance that the single-particle density of states has a pseudo gap instead of a hard one, that result envisages the noteworthy existence of incompressible states with gapless fermionic excitations whose `Fermi` surface is just the Luttinger one.