LOCALIZATION AND DECAY OF CORRELATIONS FOR A PINNED LATTICE…一个压制网格的本地化和相关衰变.pdfVIP

LOCALIZATION AND DECAY OF CORRELATIONS FOR A PINNED LATTICE FREE FIELD IN DIMENSION TWO Erwin Bolthausen1 and David Brydges2 Universit¨at Z¨urich and University of Virginia AMS subject classifications: 60K35, 82B41. 1 A survey on models and questions There is a natural class of generalizations of the standard random walks to higher dimensional “time”, which are mainly considered in mathematical physics. To motivate them, we consider first a standard real valued random walk X = 0, X , . . . , X . For simplicity, we assume that the distribution 0 1 n of the increments has a symmetric density f which is bounded and posi- tive everywhere. Therefore, we can write f (x) = exp(−φ(x)), where φ is bounded from below, and symmetric. The joint density of (X , . . . , X ) is 1 n then (x , . . . , x ) → exp(− n φ(x − x )) where we put x = 0. For the 1 n i=1 i i−1 0 higher dimensional versions introduced below, it is usually more natural to look at random walks which are tied down at the endpoint, i.e. conditioned on Xn+1 = 0. As we have assumed the increments to have a density, there is no problem to define that properly: The conditioned random walk has just the n-dimensional density n+1 1 (1) exp − φ(x − x ) i i−1 Zn i=1 where we now set x0 = xn+1 = 0, and where Zn is the appropriate norming n+1 Z = · · · exp − φ(x − x ) dx . . . dx n i i−1 1 n