Probabilistic game semantics
Vincent Danos, Russ Harmer · ACM Transactions on Computational Logic · 2002
A category of HO/N-style games and probabilistic strategies is developed where the possible choices of a strategy are quantified so as to give a measure of the likelihood of seeing a given play. A two-sided die is shown to be universal in this category, in the sense that any strategy breaks down into a composition between some deterministic strategy and that die. The interpretative power of the category is then demonstrated by delineating a Cartesian closed subcategory that provides a fully abstract model of a probabilistic extension of Idealized Algol.