partite lattices at half-filling but in any other case, we should suspect
that the system prefers a different ordering pattern. This is quite clear in D = 1 where we see that q = 2kF is the best choice. For higher di- mensions, the situation is not so clear-cut and in any case, the resulting states are much more difficult to handle. We refer to the literature for a sample of complicated mean field phase diagrams [179, 3531. Sufficiently far from half-filling, even the ferromagnetic (q=O) so- lution becomes preferable to the antiferromagnetic solution. The sim- plest mean-field diagrams (whose general appearance we sketched in Fig. 4.11) show only these two kinds of ordering. For the simple cu- bic lattice, the corresponding phase diagram was worked out by Penn [324],showing a central region of antiferromagnetism flanked by two domains of ferromagnetism (Fig. 4.11). This gave the hope that itin- erant ferromagnetism can be understood as the typical intermediate-U, intermediate-filling behaviour of the Hubbard model. Note that at large U ,the ferromagnetic domain begins to outflank the AF region, reach- ing the n = 1 axis at U + 00. We have already remarked that the energy scale sustaining AF or SDW behaviour must be suppressed at large enough U. What is the status of the ordered states predicted by Hartee-Fock theories? We understand that such theories are only approximate, and that they usually overestimate the ordering tendencies because they force the system into the straightjacket of long range order, thus even what would be really short range order effects, are compelled to appear in the disguise of long range order. But is it possible that the very prediction of a state with SDW long range order is false? We discuss first ordering at T = 0. It is understood that the pre- diction of an antiferromagnetic insulator for (perfectly nested) cubic lattices at half-filling is a solid result for dimensions d 2 2 . The state- ment is not exactly proven in a mathematical sense, but it is supported by such a variety of techniques that there is no room for reasonable doubt. In fact, the corresponding result is “almost true” even in D = 1: the ground state is insulating, with a gap whose U-dependence is not unlike the SDW result. T h e that there is no long range order but it is
- replaced by antiferromagnetic quasi-long-range order; the correlations fall off algebraically (S&SA) (-l)m-n/lm-n\, and the staggered sus-
(Mathematical Modelling - Theory and Applications 10) C. Rocşoreanu, A. Georgescu, N. Giurgiţeanu (Auth.) - The FitzHugh-Nagumo Model - Bifurcation and Dynamics-Springer Netherlands (2000)