Mollifier Representation in Non-Constant-Sum Games

An Experimental Test

Bibliographic Data

ID6284658
AuthorsH Andrew Michener, Greg B Macheel, Charles G Depies, Chris A Bowen (University of Wisconsin–Madison)
Year1986
Volume30
Issue2
Pages361-382
Publication date1986-06-01
Peer ReviewedYes
Open AccessYes
TypeARTICLE
VenueJournal of Conflict Resolution (JOURNAL)
Journal identifiersISSN: 0022-0027 • E-ISSN: 1552-8766
PublisherSAGE Publications Inc (PUBLISHER)
DOI10.1177/0022002786030002007
OpenAlexW2101937085
LanguageEN
Citations received2
References cited29

This article reports an experimental test that juxtaposes the von Neumann-Morgenstern characteristic function v(S) against the homomollifier function h(S) proposed by Charnes et al. (1978). The test was conducted in the context of 5-person cooperative sidepayment non-constant-sum games with nonempty core. Experimental results show that payoff predictions by various solution concepts (the Shapley value, the nucleolus, the 2-center) computed from the homomollifier are more accurate than predictions by the same solutions computed from the characteristic function. Supplementary analyses of data show that the payoff function x(S) is more closely approximated by the homomollifier h(S) than by the characteristic function v(S). These findings are interpreted as indicating that the homomollifier is more useful than the characteristic function for purposes of predicting payoffs in non-constant-sum games

Characteristic function (probability theory · Combinatorics · Constant (computer programming · Constant function · Context (archaeology · Core (optical fiber · Function (biology · Game theory · Mathematical analysis · Mathematical economics · Probability density function · Representation (politics · Shapley Value · Statistics · Stochastic game · Computer Science · Experimental Behavioral Economics Studies · Game Theory and Applications · Game Theory and Voting Systems · Mathematics

Tools

Open DOISci-Hub