On Tractable $\Phi$-Equilibria in Non-Concave Games

Authors: Yang Cai, Constantinos Daskalakis, Haipeng Luo, Chen-Yu Wei, Weiqiang Zheng

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical In this paper, we initiate the study of tractable Φ-equilibria in nonconcave games and examine several natural families of strategy modifications. We show that when Φ is finite, there exists an efficient uncoupled learning algorithm that approximates the corresponding Φ-equilibria. Additionally, we explore cases where Φ is infinite but consists of local modifications, showing that Online Gradient Descent can efficiently approximate Φ-equilibria in non-trivial regimes.
Researcher Affiliation Academia Yang Cai Yale University yang.cai@yale.edu Constantinos Daskalakis MIT CSAIL costis@csail.mit.edu Haipeng Luo University of Southern California haipengl@usc.edu Chen-Yu Wei University of Virginia chenyu.wei@virginia.edu Weiqiang Zheng Yale University weiqiang.zheng@yale.edu
Pseudocode Yes Algorithm 1: Φ-regret minimization for non-concave reward via sampling; Algorithm 2: SAMPLESTRATEGY; Algorithm 3: Conv(Φ)-regret minimization for Lipschitz smooth non-concave rewards
Open Source Code No The paper does not contain any explicit statement about providing open-source code for the methodology described, nor does it provide a link to a code repository.
Open Datasets No This paper does not contain experimental results and therefore does not use a dataset.
Dataset Splits No This paper does not contain experimental results and therefore does not define dataset splits.
Hardware Specification No This paper does not contain experimental results and therefore does not provide hardware specifications.
Software Dependencies No This paper does not contain experimental results and therefore does not list specific software dependencies with version numbers.
Experiment Setup No This paper does not contain experimental results and therefore does not provide details on experimental setup or hyperparameters.