Last-Iterate Convergence in Adaptive Regret Minimization for Approximate Extensive-Form Perfect Equilibrium

Hang Ren, Xiaozhen Sun, Tianzi Ma, Jiajia Zhang, Xuan Wang · 2025

Read Last-Iterate Convergence in Adaptive Regret Minimization for Approximate Extensive-Form Perfect Equilibrium by Hang Ren, Xiaozhen Sun, Tianzi Ma, Jiajia Zhang, Xuan Wang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-07699.txt.

Description

In imperfect-information games, the Nash equilibrium can play badly on branches that rational play never reaches. Hang Ren and four co-authors target a refinement, extensive-form perfect equilibrium, which adds small controlled trembles so every branch gets sensible play. Older solvers average their strategies, which costs heavily, and fix the tremble size, which trades accuracy against speed.

Their algorithm adapts regret minimization so the final strategy itself converges, no averaging needed. A new per-information-set measure steers the tremble size dynamically. Theory confirms convergence to the refined equilibrium, and experiments beat leading solvers on both plain and refined equilibrium tasks. The paper appears in 2025.

Catalog record

CREATORHang Ren, Xiaozhen Sun, Tianzi Ma, Jiajia Zhang, Xuan Wang
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE69 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-07699

Browse the full library · About the Tower of Babel