The identification of potentially optimal strategies in LPI decision making

This note is concerned with a version of the classical choice under uncertainty problem where there exists linear partial information (LPI) on the probabilities of the states of nature, i.e., the probabilities are not known precisely, but are subject to linear constraints reflecting their likelihood of occurrence. A previous paper examined how pairwise statistical dominance can be established in the presence of LPI. In the present paper more powerful results are presented, exploiting the concept of potential optimality.