This document discusses how taking the limit of infinite dimensions simplifies the Hubbard model but does not fully solve it. Specifically:
1) In the limit of infinite dimensions, contributions from products of non-interacting propagators become negligible unless the site indices coincide, collapsing the model into an effective single-site problem.
2) This effective single-site problem can be understood as an Anderson impurity model with a fluctuating effective field, which in principle requires an unlimited number of self-consistency conditions to solve fully.
3) While certain quantities may be solved numerically exactly, the effective bandwidth near the Mott transition remains an open debate due to the finite number of self-consistency conditions that can
This document discusses how taking the limit of infinite dimensions simplifies the Hubbard model but does not fully solve it. Specifically:
1) In the limit of infinite dimensions, contributions from products of non-interacting propagators become negligible unless the site indices coincide, collapsing the model into an effective single-site problem.
2) This effective single-site problem can be understood as an Anderson impurity model with a fluctuating effective field, which in principle requires an unlimited number of self-consistency conditions to solve fully.
3) While certain quantities may be solved numerically exactly, the effective bandwidth near the Mott transition remains an open debate due to the finite number of self-consistency conditions that can
This document discusses how taking the limit of infinite dimensions simplifies the Hubbard model but does not fully solve it. Specifically:
1) In the limit of infinite dimensions, contributions from products of non-interacting propagators become negligible unless the site indices coincide, collapsing the model into an effective single-site problem.
2) This effective single-site problem can be understood as an Anderson impurity model with a fluctuating effective field, which in principle requires an unlimited number of self-consistency conditions to solve fully.
3) While certain quantities may be solved numerically exactly, the effective bandwidth near the Mott transition remains an open debate due to the finite number of self-consistency conditions that can
scales accordingly [295, 2701. Since sums over products of GQ,oap-
pear in consecutive orders of perturbation theory, it is easy to ac- cept that such contributions become negligible in the limit D + 00, unless the site indices coincide. The resulting real-space collapse of perturbation-theoretical contributions is the underlying reason why the infinite-dimensional Hubbard model poses a simpler problem than its finite-dimensional counterparts [272]. It is intuitively clear that the simplification arises because the lattice model becomes equivalent to an effective single-site model. It remains, however, a difficult problem to solve. The effective on-site model in question can be understood as an Anderson impurity model, where the impurity is subject to a fluctuating effective field. Owing to the complicated frequency-dependence of the effective field one should, in principle, impose an unlimited number of self-consistency conditions. All methods of solution amount to keeping a finite (and not necessarily large) number of conditions. For certain quantities the results may be regarded as numerically exact, while for small energy scales (like the effective bandwidth which is expected to vanish near the Mott transition) the character of the solution is the subject of continuing debate. The reason why the Hubbard model remains non-trivial even in the limit of infinite dimensionality, can be traced to the on-site character of the interaction. For models with intersite interactions between localized degrees of freedom27, the limit D + 00 brings great simplification be- cause fluctuations of order 1 / a become negligible, and thus the usual mean field approximation becomes exact. In contrast, here we are deal- ing with itinerant degrees of freedom, and the purely on-site interaction UfijtfijiLjl apparently does not care about the number of neighbours. For this reason, the dynamics of the Hubbard model remains complicated even in D = 00. Still, there is a subtler sense, embodied in the effective impurity model, in which the limit D + 00 permits a simpler, mean-
"The spin models of magnetic ordering are an important example. See Ch. 6.