Computation and analysis of evolutionary game dynamics
Is Version Of
Biological processes are usually defined based on the principles of replication, mutation, competition, adaption, and evolution. In evolutionary game theory, such a process is modeled as a so-called evolutionary game, which not only provides an alternative interpretation of dynamical equilibrium in terms of the game nature of the process, but also bridges the stability of the biological process with the Nash equilibrium of the evolutionary game. Computationally, the evolutionary game models are described in terms of inverse and direct games, which are estimating the payoff matrix from data and computing the Nash equilibrium of a given payoff matrix respectively. We discuss the necessary and sufficient conditions for the Nash equilibrium states, and derive the methods for both inverse and direct games in this thesis. The inverse game is solved by a non-parametric smoothing method and penalized least squares method, while different schemes for the computation of the direct game are proposed including a specialized Snow-Shapley algorithm, a specialized Lemke-Howson algorithm, and an algorithm based on the solution of a complementarity problem on a simplex. Computation for the sparsest and densest Nash equilibria is investigated. We develop a new algorithm called dual method with better performance than the traditional Snow-Shapley method on the sparse and dense Nash equilibrium searching. Computational results are presented based on examples. The package incorporating all the schemes, the Toolbox of Evolution Dynamics Analysis (TEDA), is described.