Co-evolution of Optimal Agents for the Alternating Offers Bargaining Game
Bargaining, as an instance of sequential games, is a widely studied problem in game theory, experimental and computational economics. We consider the problem of evolving computational agents with optimal (Subgame Perfect Equilibrium) strategies for the Alternating Offers Bargaining Game. Previous work co-evolving agents for this problem has argued that it is not possible to achieve optimal agents at the end of the co-evolutionary process due to the myopic properties of the evolutionary agents. Emphasising the notion of a co-evolutionary solution concept, we show that this conclusion is mis-leading and present a co-evolutionary algorithm that evolves optimal strategies for the bargaining game with one round. We conclude by explaining why, using previous evaluation procedures and strategy representations, the algorithm is not able to converge to optimal strategies for games with more rounds.
Unable to display preview. Download preview PDF.
- 1.Beyer, H.: The theory of evolution strategies. Springer, New York (2001)Google Scholar
- 2.van Bragt, D.D.B., Gerding, E.H., La Poutré, J.A.: Equilibrium selection in alternating-offers bargaining models - the evolutionary computing approach. The Electronic Journal of Evolutionary Modeling and Economic Dynamics (2002)Google Scholar
- 4.Darwen, P.J.: Co-evolutionary learning by automatic modularisation with speciation. Ph.D. thesis, University of New South Wales (1996)Google Scholar
- 5.Ficici, S.G.: Solution concepts in coevolutionary algorithms. Ph.D. thesis, Brandeis University (2004)Google Scholar
- 6.Gerding, E., van Bragt, D.D.B., La Poutré, J.A.: Multi-issue negotiation processes by evolutionary simulation: validation and social extensions. Tech. Rep. SEN-R0024, CWI, Amsterdam, The Netherlands (2000)Google Scholar
- 8.Jin, N.: Constraint-based co-evolutionary genetic programming for bargaining problems. Ph.D. thesis, University of Essex (2007)Google Scholar