Markov Superposition Expansion for the Entropy and Correlation Functions in Two and Three Dimensions

Phil Attard

Ian Wark Research Institute, University of South Australia, Mawson Lakes SA 5095 Australia.

in 'Statistical Physics on the Eve of the Twenty-First Century', M.T. Batchelor and L.T. Wille (eds), (World Scientific, Singapore, 1999).

Abstract
The role that entropy plays in information theory is described, together with a pedagogic derivation of Gibb's formula that expresses it in terms of the probability of states. It is then shown how the entropy of a spin lattice system may be expanded in terms of successively higher order spin correlation functions. This expansion is based on a Markov approach and it has previously proven successful in one dimensional spin-lattice systems, which have possible applications to time series and signal processing. Here the procedure is generalized to higher dimensions, where it may be applicable to image analysis and tomography. Tests for the two-dimensional Ising model show that two terms of the expansion suffice for accurate result at both high and low temperatures. The corresponding Markov superposition approximation for the higher order correlation functions in terms of elemental adjacent correlation functions (i.e. between neighbor vertices, bonds, and squares), is given in both two- and three-dimensions.

Back to publications or to Phil Attard's home page.