Multicanonical distribution and the origin of power laws

Abstract

A multicanonical formalism is applied to the problem of statistical equilibrium in a complex system with a hierarchy of dynamical structures. At the small scales the system is in quasi-equilibrium and follows a Maxwell-Boltzmann distribution with a slowly fluctuating temperature. The probability distribution for the temperature is determined using Bayesian analysis and it is then used to average the Maxwell-Boltzmann distribution. The resulting energy distribution law is written in terms of generalized hypergeometric functions, which display power-law tails.