This document discusses the Gutzwiller variational method for evaluating expectation values in quantum many-body systems. It explains that within this method, a small cluster of sites is examined and subjected to an effective field from the surrounding environment. The behavior of this cluster is meant to be characteristic of the whole lattice. Next, it describes using a cluster method to evaluate expectation values by focusing on a two-site cluster immersed in an effective medium, in order to simplify calculations from considering the entire lattice.
This document discusses the Gutzwiller variational method for evaluating expectation values in quantum many-body systems. It explains that within this method, a small cluster of sites is examined and subjected to an effective field from the surrounding environment. The behavior of this cluster is meant to be characteristic of the whole lattice. Next, it describes using a cluster method to evaluate expectation values by focusing on a two-site cluster immersed in an effective medium, in order to simplify calculations from considering the entire lattice.
This document discusses the Gutzwiller variational method for evaluating expectation values in quantum many-body systems. It explains that within this method, a small cluster of sites is examined and subjected to an effective field from the surrounding environment. The behavior of this cluster is meant to be characteristic of the whole lattice. Next, it describes using a cluster method to evaluate expectation values by focusing on a two-site cluster immersed in an effective medium, in order to simplify calculations from considering the entire lattice.
classical statistical mechanics, sums over the exponentially large num-
ber of states of a macroscopic system tend to be dominated by their largest term (see Appendix C) and the overwhelming majority of con- figurations looks very much like an average configuration. It is these “typical” configurations which dominate the behaviour of the system. The same argument applies also to the case of finite U ,where in a typi- cal configuration (illustrated in Fig. 9.l.b), the concentration of doubly occupied sites is n d ( q 0 ) where qo is the optimum value of 7. All this is yet very non-formal and vague but we can already see how the pos- sibility of developing a mean-field-like language for local correlations in an itinerant quantum system emerges. One should be able to carve out a small cluster from the inside of the system, and subject it to an effective field from the environment. It can be achieved that the aver- age behaviour of the cluster will be characteristic of the whole lattice. The “mean field” is, however, not like the fields familiar from models of ordering in statistical physics, but rather the typical environment in a fermionic reservoir. Among other things, fermions must be allowed to travel to and fro between the cluster and its environment, and this must be done phase-coherently so that a sharp Fermi surface survives. Next, we describe a cluster method for the evaluation of expectation values, which puts the above idea into practice. The cluster method is one of the possible ways to arrive at the results of the Gutzwiller approximation; for a different approach, cf. Appendix C. In order not to burden the main text with too many technicalities, we will be satisfied with the most handwaving argument”. In this simplest form, our treatment is applicable to the half-filled case only. Let us observe that 3t consists of an on-site term, and a nearest- neighbour pair term, so to evaluate E(q), it is sufficient to know the average state of a pair of sites i and j. The essence of our effective medium treatment is to replace the true expectation values in (9.17) with values calculated with a clus- ter method. The simplification is obvious: the two-site cluster has only 4 x 4 = 16 states. However, we certainly do not mean to replace the lattice with an isolated pair of sites! The pair of sites is immersed in ”It can be regarded as a simplified version of the argument presented in the review [431] by Vollhardt.