Evolutionary Game Theory on Measure Spaces: Asymptotic Behavior of Solutions

TL;DR

This paper extends evolutionary game theory models to measure spaces, analyzing long-term behavior and convergence of solutions, including cases with multiple strategies and perturbations, establishing stability and asymptotic properties.

Contribution

It introduces a framework for analyzing the asymptotic behavior of solutions in measure spaces, including convergence to Dirac measures and stability of equilibria.

Findings

Solutions in measure spaces are asymptotically closed.

Pure replicator dynamics converge to a Dirac measure at the fittest strategy.

Existence of globally stable equilibria under small perturbations.

Abstract

In [12] we formulated an evolutionary game theory model as a dynamical system on the state space of finite signed Borel measures under the weak* topology. The focus of this paper is to extend the analysis to include the long-time behavior of solutions to the model. In particular, we show that M(Q), the finite signed Borel measures are asymptotically closed. This means that if the initial condition is a finite signed Borel measure and if the asymptotic limit of the model solution exists, then it will be a measure (note that function spaces such as L1(Q) and C(Q) do not have this property). We also establish permanence results for the full replicator mutator model. Then, we study the asymptotic analysis in the case where there is more than one strategy of a given fitness (a continuum of strategies of a given fitness), a case that often arises in applications. To study this case our mathematical structure must include the ability to demonstrate the convergence of the model solution to a measure supported on a continuum of strategies. For this purpose, we demonstrate how to perform completions of the space of measures and how to use these completions to formulate weak (generalized) asymptotic limits. In particular, we show that for the pure replicator dynamics the (weak) solution of the dynamical system converges to a Dirac measure centered at the fittest strategy class; thus this Dirac measure is a globally attractive equilibrium point which is termed a continuously stable strategy (CSS). It is also shown that in the discrete case of the pure replicator dynamics and even for small perturbation of the pure replicator dynamics (i.e., selection with small mutation) there exists a globally asymptotically stable equilibrium.