Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.755570
Title: An empirical study on computation of exact and approximate equilibria
Author: Igwe, Tobenna
ISNI:       0000 0004 7428 5635
Awarding Body: University of Liverpool
Current Institution: University of Liverpool
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
The computation of Nash equilibria is one of the central topics in game theory, which has received much attention from a theoretical point of view. Studies have shown that the problem of finding a Nash equilibrium is PPAD-complete, which implies that we are unlikely to find a polynomial-time algorithm for this problem. Naturally, this has led to a line of work studying the complexity of finding approximate Nash equilibria. This thesis examines the computation of such approximate Nash equilibria within several classes of games from an empirical perspective. In this thesis, we address the computation of approximate Nash equilibria in bimatrix and polymatrix games. For both of these game classes, we provide a library of implementations of algorithms for the computation of exact and approximate Nash equilibria, as well as a suite of game generators which were used as a base for our empirical analysis of the algorithms. We investigate the trade-off between quality of approximation produced by the algorithms and the expected runtime. We provide some insight into the inner workings of the state-of-the-art algorithm for computing ε-Nash equilibria, presenting worst-case examples found for our provided suite of game generators. We then show lower bounds on these algorithms. In the case of polymatrix games, we generate this lower bound from a real-world application of game theory. For bimatrix games, we provide a robust means of generating lower bounds for approximation algorithms with the use of genetic algorithms.
Supervisor: Savani, Rahul Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.755570  DOI:
Share: