Spatial random permutations and infinite cycles (Q2517970)

This article is devoted to spatial random permutations of \(N\) elements, where the permutations are weighted according to the point locations. First a model for the spatial random permutations in a bounded domain is introduced, and an approach is shown how to suppress permutations with large jumps. This is achieved by assigning a ``one-body energy'' of the form \(\Sigma_i\xi(x_i-x_{\pi(i)})\) to a given permutation \(\pi\) on a finite set \(x\). The one-body potential \(\xi\) is nonnegative, spherically symmetric, and typically monotonically increasing, although there are also more general cases allowed. In addition, ``many-body potentials'' are introduced, which depend on several jumps, as well as a weight for the points \(x=x_i,...,x_N\). To consider thermodynamic limits of systems, then, infinite volume measures for permutations are constructed only, with a fixed x chosen according to some point process. Thus, a natural criterion for the existence of the infinite volume limit is given. This criterion is trivially fulfilled if the interaction prohibits jumps greater than a certain finite distance. Another option for taking the thermodynamic limit is to focus on the existence of the limiting distribution of one special random variable at \(N\rightarrow\infty\). Motivated by the relevance to the Bose-Einstein condensation, the authors chose the probability of the existence of long cycles as random variable. They give a criterion (that corresponds to low density or high temperature) for the absence of infinite cycles. Infinite cycles are usually present at high densities above a critical one. The authors establish the occurrence of infinite, macroscopic cycles for the case where only the one-body potential is present. They average over the point configuration x in a suitable way. Thus, they find also a simple, exact formula for the critical density. It turns out to be nothing else than the critical density of the ideal Bose gas, first computed by Einstein in 1925. Next, the relation between models of spatial random permutations and the Feynman-Kac representation of the Bose gas is studied. The authors are particularly interested in the effect of interactions on the Bose-Einstein condensation. They expect that weakly interacting bosons can be exactly described by a model of spatial permutations with two-body interactions. They show, that the largest terms of their interaction model contributing to the interactions between permutation jumps are due to cycles of length 2. Retaining these contributions only, they are able to explore some effects of interactions on Bose-Einstein condensation. In particular, they compute the critical temperature exactly. It turns out to be higher than the non-interacting one and to deviate linearly in the scattering length of the interaction potential, which is in qualitative agreement with other, up-to-now, findings.