Automating approximation analysis for Nash equilibria algorithms in two-player games

Computing polynomial-time approximate Nash equilibria (NE) is a fundamental problem in algorithmic game theory, with deep connections to the complexity class TFNP. Recent advances in approximate NE algorithms have become increasingly sophisticated, making the verification of their approximation guar...

Full description

Saved in:

Bibliographic Details
Published in:	Information and computation Vol. 307; p. 105362
Main Authors:	Deng, Xiaotie, Li, Dongchen, Li, Hanyu
Format:	Journal Article
Language:	English
Published:	Elsevier Inc 01.11.2025
Subjects:	Algorithmic game theory Approximation algorithms Automated algorithm analysis Nash equilibrium Verification Nash equilibrium Approximation algorithms Automated algorithm analysis Verification Algorithmic game theory
ISSN:	0890-5401
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	Computing polynomial-time approximate Nash equilibria (NE) is a fundamental problem in algorithmic game theory, with deep connections to the complexity class TFNP. Recent advances in approximate NE algorithms have become increasingly sophisticated, making the verification of their approximation guarantees both complex and error-prone. We present the first automated method for analyzing approximation bounds of algorithms for two-player normal-form games. Given any algorithm that computes approximate NE, our approach automatically derives tight approximation bounds using constraint programming techniques. We demonstrate the effectiveness of our method by applying it to all known algorithms in the literature, reproducing their manually-proven approximation bounds within seconds and without human intervention. Our results provide both a powerful verification tool and new insights into the structure of approximate equilibrium computation.
ISSN:	0890-5401
DOI:	10.1016/j.ic.2025.105362