Renormalization of the 2-point function of the Hubbard model at half-filling (Q2484038)

It is proved, that the correlation functions of the two-dimensional Hubbard model at half-filling at finite temperature are analytic in the coupling constant \(\lambda\) for \(|\lambda| \leq c/|\log T|^2\) by treating the renormalization of two-particle subgraphs, that was missing in the previous paper: \textit{V. Rivasseau} [J. Stat. Phys. 106, 693--722 (2002; Zbl 1001.82046)]. First, the tools for a multiscale analysis, namely: momentum slices, sectors and their conservations laws introduced in above paper are presented together with additional tools. In order to generalize the results of the first paper to the full model, it is proposed to renormalize the two-point functions of the theory instead of suppressing them. This is accomplished by a second order Taylor expansion of the two-point function with given cutoff. The authors perform a multislice expansion and get a ``Gallavotti-Nicolo'' or clustering tree structure. The central point of the paper is a constructive two-particle irreducible analysis of the self-energy, starting from consideration of the connected amputated two-point Schwinger function. The self-energy is defined through its Feinman graph expansion. First, it is formulated the main bound on the connected amputated and one-particle irreducible 2-point function or self-energy. But for the problem solution, it is not sufficient and therefore, a two-particle irreducible expansion is derived inside each 2-point contribution. Then, the line form of Menger's theorem leads to the explicit construction of three line-disjoint paths for every self-energy contribution, in a way compatible with constructive bounds. A second use of Menger's theorem in the vertex form allows to improve the power couting for two-point subgraphs exploiting the particle-hole symmetry of the theory at half-filling and leads to analyticity result. The upper bounds on the self-energy, obtained from the analysis, strongly suggest, that the second momentum derivative of the self-energy is not uniformly bounded in the region \(|\lambda| \leq c/|\log T|^2\). A rigorous proof of this statement follows from a rigorous lower bound of the same type than these upper bounds, but for the smallest non-trivial self-energy graph. All these bounds prove, that the model is not a Fermi liquid in the sense of Salmhofer's criterion.