Semicoarse Correlated Equilibria and LP-Based Guarantees for Gradient Dynamics in Normal-Form Games
Mete Şeref Ahunbay, Martin Bichler · 2025
Read Semicoarse Correlated Equilibria and LP-Based Guarantees for Gradient Dynamics in Normal-Form Games by Mete Şeref Ahunbay, Martin Bichler (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-2502-20466.txt.
Introduction
Ahunbay and Bichler refined what can be said about learning dynamics in games in twenty twenty-five. Gradient ascent, a standard learning rule, is known to avoid external regret. Empirically it often lands on Nash equilibrium even in games where weaker equilibrium notions allow many other outcomes.
The authors prove gradient ascent satisfies a stronger regret notion, leading to a new solution concept they call semicoarse correlated equilibria. In discretized Bertrand price competition with at least two lowest-cost firms, Nash equilibria are the only such outcomes under mild profit conditions. In first-price auctions, finer bid grids for low bids likewise force convergence toward Nash outcomes. Their proofs use linear programming duality, giving a simpler route than earlier epoch-based arguments. The work sharpens predictions for gradient-based learning in markets.
Catalog record
| CREATOR | Mete Şeref Ahunbay, Martin Bichler |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 169 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-20466