On Lee's Markovian Entropy-Maximization Model for Population Distribution

Abstract

The conclusions drawn by Lee for his first case, in a paper describing a Markovian entropy-maximizing model of population distribution, are found to be incorrect. It is, however, shown that his model for this case always leads to a global maximum value of entropy, a result which cannot be guaranteed for his third case. This result is then generalized to a more universal measure of entropy. Markovian measures of directed divergence are defined and used to get minimum discrimination distributions.