Quantum Speedups for Zero-Sum Games via Improved Dynamic Gibbs Sampling

Authors: Adam Bouland, Yosheb M Getachew, Yujia Jin, Aaron Sidford, Kevin Tian

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical We give a quantum algorithm for computing an ϵ-approximate Nash equilibrium of a zero-sum game in an m n payoff matrix with bounded entries. Given a standard quantum oracle for accessing the payoff matrix our algorithm runs in time e O( m + n ϵ 2.5 +ϵ 3) and outputs a classical representation of the ϵ-approximate Nash equilibrium. This improves upon the best prior quantum runtime of e O( m + n ϵ 3) obtained by (van Apeldoorn & Gilyén, 2019) and the classical e O((m + n) ϵ 2) runtime due to (Grigoriadis & Khachiyan, 1995) whenever ϵ = Ω((m + n) 1). We obtain this result by designing new quantum data structures for efficiently sampling from a slowly-changing Gibbs distribution.
Researcher Affiliation Collaboration 1Stanford University, Stanford, CA, USA. 2Microsoft Research, Redmond, WA, USA.
Pseudocode Yes Algorithm 1: Matrix Game Solver(δ, η, T)
Open Source Code No The paper does not contain any statement or link indicating that the code for the described methodology is open-source or publicly available.
Open Datasets No The paper is theoretical and focuses on algorithm design and proofs; it does not describe empirical evaluation on datasets or mention the use of publicly available datasets for training or evaluation.
Dataset Splits No As a theoretical paper focused on algorithm design and complexity, it does not discuss training/test/validation dataset splits.
Hardware Specification No The paper is theoretical and focuses on algorithm design and complexity analysis, not on practical implementation details or experimental results. Therefore, it does not specify any hardware used.
Software Dependencies No The paper is theoretical and does not describe an implementation or empirical experiments that would require specific software dependencies with version numbers.
Experiment Setup No The paper is theoretical and does not describe empirical experiments. Therefore, no experimental setup details, such as hyperparameters or training settings, are provided.