Michael Kiessling, Rutgers University

Abstract

 Relative entropy of a probability measure, relative to a given a priori measure, is a well-defined and extremely useful notion in probability, statistics, and statistical mechanics. A question by Alice Chang has prompted me to introduce the notion of an `entropy of signed and complex measures relative to a signed a priori measure.' In this talk I will explain Alice Chang's question, and why entropy of signed and complex measures relative to a signed measure may hold the answer. I show computer-generated experimental evidence for my proposal.