We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cond-mat

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Condensed Matter > Disordered Systems and Neural Networks

Title: Self-Averaging of Nash Equilibria in Two-Player Games

Abstract: We use techniques from the statistical mechanics of disordered systems to analyse the typical properties of Nash equilibria of bimatrix games with large random payoff matrices. We find two distinct regimes depending on the correlation $\kappa$ between the payoff matrices of either player. For $\kappa < 0$ we find that the statistical properties of Nash equilibria are self-averaging, i.e. for a randomly chosen Nash equilibrium values such as the payoff to either player are realized with probability 1 for any typical realization of the payoff matrices. We derive analytical expressions for these quantities and find good agreement with numerical simulations. For $\kappa \geq 0$ the scaling ansatz of the statistical mechanics approach breaks down since only a sub-extensive number of strategies are played with non-zero probability at a Nash equilibrium. This picture is complemented by the calculation of the properties of Nash equilibria in pure strategies.
Comments: 6 pages, for additional information see this http URL
Subjects: Disordered Systems and Neural Networks (cond-mat.dis-nn); Statistical Mechanics (cond-mat.stat-mech); Adaptation and Self-Organizing Systems (nlin.AO)
Cite as: arXiv:cond-mat/9905076 [cond-mat.dis-nn]
  (or arXiv:cond-mat/9905076v1 [cond-mat.dis-nn] for this version)

Submission history

From: Johannes Berg [view email]
[v1] Thu, 6 May 1999 13:42:40 GMT (21kb)
[v2] Wed, 21 Jul 1999 15:36:12 GMT (19kb)

Link back to: arXiv, form interface, contact.