作者: David H. Wolpert
DOI:
关键词:
摘要: A long-running difficulty with conventional game theory has been how to modify it accommodate the bounded rationality characterizing all real-world players. recurring issue in statistical physics is best approximate joint probability distributions decoupled (and therefore far more tractable) distributions. It recently shown that same information theoretic mathematical structure, known as Probability Collectives (PC) underlies both issues. This relationship between and allows techniques insights &om one field be applied other. In particular, PC provides a formal model-independent definition of degree player equilibria. pair papers extends previous work on by introducing new computational approaches effectively find equilibria common-interest (team) games.