Approximating N-Player Nash Equilibrium through Gradient Descent

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Dongge, Yan, Xiang, Dou, Zehao, Huang, Wenhan, Yang, Yaodong, Deng, Xiaotie
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913637103304704
author Wang, Dongge
Yan, Xiang
Dou, Zehao
Huang, Wenhan
Yang, Yaodong
Deng, Xiaotie
author_facet Wang, Dongge
Yan, Xiang
Dou, Zehao
Huang, Wenhan
Yang, Yaodong
Deng, Xiaotie
contents Decoding how rational agents should behave in shared systems remains a critical challenge within theoretical computer science, artificial intelligence and economics studies. Central to this challenge is the task of computing the solution concept of games, which is Nash equilibrium (NE). Although computing NE in even two-player cases are known to be PPAD-hard, approximation solutions are of intensive interest in the machine learning domain. In this paper, we present a gradient-based approach to obtain approximate NE in N-player general-sum games. Specifically, we define a distance measure to an NE based on pure strategy best response, thereby computing an NE can be effectively transformed into finding the global minimum of this distance function through gradient descent. We prove that the proposed procedure converges to NE with rate $O(1/T)$ ($T$ is the number of iterations) when the utility function is convex. Experimental results suggest our method outperforms Tsaknakis-Spirakis algorithm, fictitious play and regret matching on various types of N-player normal-form games in GAMUT. In addition, our method demonstrates robust performance with increasing number of players and number of actions.
format Preprint
id arxiv_https___arxiv_org_abs_2501_03001
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Approximating N-Player Nash Equilibrium through Gradient Descent
Wang, Dongge
Yan, Xiang
Dou, Zehao
Huang, Wenhan
Yang, Yaodong
Deng, Xiaotie
Computer Science and Game Theory
Decoding how rational agents should behave in shared systems remains a critical challenge within theoretical computer science, artificial intelligence and economics studies. Central to this challenge is the task of computing the solution concept of games, which is Nash equilibrium (NE). Although computing NE in even two-player cases are known to be PPAD-hard, approximation solutions are of intensive interest in the machine learning domain. In this paper, we present a gradient-based approach to obtain approximate NE in N-player general-sum games. Specifically, we define a distance measure to an NE based on pure strategy best response, thereby computing an NE can be effectively transformed into finding the global minimum of this distance function through gradient descent. We prove that the proposed procedure converges to NE with rate $O(1/T)$ ($T$ is the number of iterations) when the utility function is convex. Experimental results suggest our method outperforms Tsaknakis-Spirakis algorithm, fictitious play and regret matching on various types of N-player normal-form games in GAMUT. In addition, our method demonstrates robust performance with increasing number of players and number of actions.
title Approximating N-Player Nash Equilibrium through Gradient Descent
topic Computer Science and Game Theory
url https://arxiv.org/abs/2501.03001