In their recently published paper1l Hugenholtz and Pines have developed a new method to evaluate the ground state energy of a system of interacting bosons, the merit of which consists in the elimination of the zero-momentum state and accordingly the applicability of the linked cluster expansion. Their theory, however, does not give correct results because of the undue treatment of the zero-momentum state; this will be proved in a simple example. Up to third order in coupling parameter of the ground state energy we compare the result of the usual Rayleigh-S~hrodinger perturbation method with that obtained according to Hugenholtz-Pines ' theory and find that the latter does not agree with the former.
We examine in detail the microscopic theories of Beliaev and Hugenholtz and Pines as applied to an interacting system of bosons in its ground state and demonstrate that these theories are strictly applicable to a zero density gas only. After providing a re-normalization of the time dependent perturbation theory of a many fermion system in its ground state, we reformulate the corresponding boson theory without ab initio approximations concerning the k=0 mode.The resulting boson perturbation theory has a diagrammatic structure which is topologically identical with the corresponding fermion theory and provided the implicit assumptions concerning convergence are valid, we conclude that a finite fraction of the particles in an arbitrary interacting Bose gas at T=ok will not condense into the k=0 mode.