Regret matching with finite memory (Q367483)

The article deals with regret matching processes when a bounded recall is considered. The value for the regret is obtained out of the average payoff reached within the finite memory used. The article contains a very pertinent literature review, proper theoretical description of regret matching with finite memory, followed by the definitions for sets closed under some-or-better replies (CUSOBR), product sets that are closed under some-or-better replies (PCUSOBR) and so-called \(\omega\)-sets, which are also accompanied by illustrative example games. Results for weakly acyclic games and random sampling ones are presented. All the propositions presented in the article have a dedicated section with proofs. Despite the examples within it, the article is overall a highly theoretical one.