Equilibrium Selection in Multi-Agent Policy Gradients via Opponent-Aware Basin Entry

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Shcherbinin, Yevhen, Redina, Arina, Kalpin, Maxim, Kochetov, Vlad
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918509151256576
author Shcherbinin, Yevhen
Redina, Arina
Kalpin, Maxim
Kochetov, Vlad
author_facet Shcherbinin, Yevhen
Redina, Arina
Kalpin, Maxim
Kochetov, Vlad
contents Multi-agent policy-gradient methods have been shown to converge locally near stable Nash equilibria. Local convergence, however, does not determine which equilibrium is reached. We study this question through basin-entry probability with respect to a target set of equilibria selected by an external criterion, such as payoff dominance. For finite-unroll Meta-MAPG, we show that the update decomposes into ordinary policy gradient plus own-learning and peer-learning corrections, with controlled sampling noise and finite-unroll bias. We identify the peer-learning correction as the main equilibrium-selection mechanism: under a local alignment condition, the probability of entering the certified attraction region of the target stable-Nash set increases, relative to ordinary policy gradient. Because persistent correction may shift zero-update points of the original game, annealing the correction after entering the basin recovers ordinary policy-gradient dynamics and inherits local stable-Nash convergence guarantees. Experiments in Stag Hunt, iterated Prisoner's Dilemma, and preliminary neural-policy coordination environments support this basin-entry view, showing increased entry into cooperative basins under peer-aware updates.
format Preprint
id arxiv_https___arxiv_org_abs_2605_18078
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Equilibrium Selection in Multi-Agent Policy Gradients via Opponent-Aware Basin Entry
Shcherbinin, Yevhen
Redina, Arina
Kalpin, Maxim
Kochetov, Vlad
Machine Learning
Multi-agent policy-gradient methods have been shown to converge locally near stable Nash equilibria. Local convergence, however, does not determine which equilibrium is reached. We study this question through basin-entry probability with respect to a target set of equilibria selected by an external criterion, such as payoff dominance. For finite-unroll Meta-MAPG, we show that the update decomposes into ordinary policy gradient plus own-learning and peer-learning corrections, with controlled sampling noise and finite-unroll bias. We identify the peer-learning correction as the main equilibrium-selection mechanism: under a local alignment condition, the probability of entering the certified attraction region of the target stable-Nash set increases, relative to ordinary policy gradient. Because persistent correction may shift zero-update points of the original game, annealing the correction after entering the basin recovers ordinary policy-gradient dynamics and inherits local stable-Nash convergence guarantees. Experiments in Stag Hunt, iterated Prisoner's Dilemma, and preliminary neural-policy coordination environments support this basin-entry view, showing increased entry into cooperative basins under peer-aware updates.
title Equilibrium Selection in Multi-Agent Policy Gradients via Opponent-Aware Basin Entry
topic Machine Learning
url https://arxiv.org/abs/2605.18078