[Paper Review] On the Computation of Strategically Equivalent Rank-0 Games

Joseph L. Heyman, Abhishek Gupta|arXiv (Cornell University)|Mar 31, 2019

Game Theory and ApplicationsDecision Sciences3 citations

TL;DR

This paper presents a linear-time algorithm to determine if a bimatrix game is strategically equivalent to a zero-sum game, and if so, computes such an equivalent zero-sum game. By leveraging strategic equivalence and matrix rank properties, the method enables faster computation of Nash equilibria in rank-0 games, significantly improving algorithmic efficiency for this class of games.

ABSTRACT

It has been well established that in a bimatrix game, the rank of the matrix formed by summing the payoff (or cost) matrices of the players has an impact on the runtime of the algorithms that converge to a Nash equilibrium of the game. In this paper, we devise a fast linear time algorithm that exploits strategic equivalence between bimatrix games to identify whether or not a given bimatrix game is strategically equivalent to a zero-sum game, and if it is, then we present an algorithm that computes a strategically equivalent zero-sum game.

Motivation & Objective

• To address the computational challenge of finding Nash equilibria in bimatrix games with low-rank payoff matrices.
• To investigate whether a given bimatrix game is strategically equivalent to a zero-sum game.
• To develop an efficient algorithm that computes a strategically equivalent zero-sum game when such equivalence exists.
• To reduce the runtime of equilibrium computation by exploiting structural properties of rank-0 games.

Proposed method

• The algorithm analyzes the sum of the payoff matrices of the two players to determine the game's rank.
• It uses strategic equivalence to transform the original game into a zero-sum game when the rank is zero.
• The method operates in linear time by leveraging matrix decomposition and rank analysis.
• It checks for the existence of a transformation that preserves Nash equilibria while converting the game into a zero-sum form.
• The approach relies on identifying specific linear dependencies in the payoff matrices that indicate strategic equivalence to zero-sum games.
• The algorithm outputs a transformed game that is both zero-sum and strategically equivalent to the original.

Experimental results

Research questions

• RQ1 Can a given bimatrix game be transformed into a strategically equivalent zero-sum game?
• RQ2 What conditions on the payoff matrices determine such strategic equivalence?
• RQ3 Is there a linear-time algorithm to detect and compute such an equivalent zero-sum game?
• RQ4 How does the rank of the sum of payoff matrices influence the existence of such equivalence?
• RQ5 What is the computational advantage of exploiting this equivalence in Nash equilibrium computation?

Key findings

• The algorithm determines in linear time whether a bimatrix game is strategically equivalent to a zero-sum game.
• When equivalence exists, the method computes a strategically equivalent zero-sum game efficiently.
• The approach exploits the rank-0 property of the sum of payoff matrices to enable fast detection and transformation.
• The transformation preserves all Nash equilibria of the original game.
• The method significantly reduces computational overhead for equilibrium computation in rank-0 games.
• The results demonstrate that strategic equivalence to zero-sum games can be leveraged to accelerate Nash equilibrium computation.