Title: The constraints of the valuation distribution for solving a class of games by using a best response algorithm

Authors: Gangshu Cai, Jie Zhong, Peter R. Wurman

Addresses: Management Information Systems and Decision Sciences, Texas A&M International University, Laredo, TX 78041, USA. ' Operations Research and Computer Science, North Carolina State University, Raleigh, NC 27695-7535, USA. ' Operations Research and Computer Science, North Carolina State University, Raleigh, NC 27695-7535, USA

Abstract: Infinite games with incomplete information are common in practice. First-price, sealed-bid auctions are a prototypical example. To solve this kind of infinite game, a heuristic approach is to discretise the strategy spaces and enumerate to approximate the equilibrium strategies. However, an approximate algorithm might not be guaranteed to converge. This paper discusses the utilisation of a best response algorithm in solving infinite games with incomplete information. We show the constraints of the valuation distributions define the necessary conditions of the convergence of the best response algorithm for several classes of infinite games, including auctions. A salient feature of the necessary convergence conditions lies in that they can be employed to compute the exact Nash equilibria without discretising the strategy space if the best response algorithm converges.

Keywords: auctions; infinite games; e-commerce; decision support systems; DSS; best response; necessary conditions; valuation distribution; electronic commerce; value chain management.

DOI: 10.1504/IJMTM.2007.013756

International Journal of Manufacturing Technology and Management, 2007 Vol.12 No.4, pp.327 - 341

Published online: 25 May 2007 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article