It is difficult to judge the power of high-temperature expansion to decide
the nature of T = 0 ordering, but we have to keep an open mind. Rather than trying to guess the eventual outcome, let us look back, and draw conclusions from our solid results. The Hartree-Fock criterion (8.13) would have made us believe that the Nagaoka phase covers most of the ground state phase diagram, extending over all 0 < n < 1, and from U N 4t upwards. Taking correlations into account has drastically reduced the area which may be still permitted for a Nagaoka state, changing its character from a supposedly robust, intermediate-coupling phenomenon to a marginally surviving strong-coupling regime. The square lattice Hubbard model may be ferromagnetic somewhere, but it certainly is not a model noted for its ferromagnetism.
8.4.3 Lattice Structure Dependence of Ferromagnetism
in a Single-Band Model Extending the studies to other lattices in D = 2, 3, and 00, we learn that the shape and extent of the area in the T = 0 phase diagram, which is covered by the fully spin-polarized state, depends very much on the lattice structure [152]. Though the Nagaoka phase cannot be equated with the ferromagnetic phase (there are reasons to worry about systems with partially polarized ground states), still we can claim that systems with a large Nagaoka region are genuinely inclined towards ferromagnetism, while for systems with a small or vanishing Nagaoka phase, the very appearance of ferromagnetic order is in doubt. Though we know that the SKA trial state does not perform so well as states with intersite correlations do, its simplicity allows to make a comparative study of a number of lattices23. At the level of the SKA states, the stability of the Nagaoka state hinges upon the band density of states. To see this, let us discuss the limiting forms of AE(n,U = 00) for n + 0, and for n + 1. First, let us assume that the DOS is finite at both band edges: let it be Pb at the bottom of the band Q,, and pt at the top of the band q (the assumption works in D = 2). Near half-filling, we collect the terms of order (1 - n). Replacing 23Usingthe resolvent method, very similar conclusions can be reached for the most general Basile-Elser states [155].
Negative Mass and Negative Refractive Index in Atom Nuclei - Nuclear Wave Equation - Gravitational and Inertial Control: Part 2: Gravitational and Inertial Control, #2