Nash equilibria of Cauchy-random zero-sum and coordination matrix games (Q2501068): Difference between revisions

The author considers two types of two-person finite random games in the form \((A,-A)\) and \((A,A)\), called \textit{Cauchy-random zero-sum games} or \textit{Cauchy-coordination games}, respectively, where \(A\) is an \((m\times n)\)-matrix with entries chosen independently with respect to the Cauchy distribution. For such games, he analyses the following problem: For a natural \(k\) and an interval \((a, b)\) (finite or infinite), what is the probability of the existence of a unique Nash equilibrium \((x,y)\) with suport size of \(x\) and \(y\) equal to \(k\), and with payoff \((\lambda, -\lambda)\) or \((\lambda, \lambda)\) (in case of a zero-sum game or coordination game, respectively) with \(a\leq \lambda \leq b\). Two theorems are the main results of the paper where the author finds explicit formulae on that probability in both cases. In the second part of the paper he derives the asymptotic behavior associated to those theorems in the limit of large matrices of a given shape. The achieved results complete the recent ones of McLennan and Berg on Gausian random bimatrix games.