Small Nash Equilibrium Certificates in Very Large Games

Authors: Brian Zhang, Tuomas Sandholm

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments showed that many large or even infinite games have small certificates, allowing us to find equilibria while exploring a vanishingly small portion of the game. ... 7 Experiments We conducted experiments using the algorithm in Section 6 on the following common zero-sum benchmark games. ... Results of experiments can be found in Table 1.
Researcher Affiliation Collaboration Brian Hu Zhang Computer Science Department Carnegie Mellon University bhzhang@cs.cmu.edu Tuomas Sandholm Computer Science Department, CMU Strategic Machine, Inc. Strategy Robot, Inc. Optimized Markets, Inc. sandholm@cs.cmu.edu
Pseudocode Yes Algorithm 6.7 Finding a certificate in a two-player zero-sum game ... Algorithm 6.9 CORRECT(I): Correcting a strategy in the case of infinite reward
Open Source Code No The paper does not provide any statement or link indicating that the source code for the described methodology is publicly available.
Open Datasets Yes (1) A zero-sum variant of the search game [4]. ... (2) k-rank Goofspiel. ... (3) k-rank limit Leduc poker. ... [4] refers to Branislav Bošanský and Jiří Čermák. Sequence-form algorithm for computing Stackelberg equilibria in extensive-form games. In Twenty-Ninth AAAI Conference on Artificial Intelligence, 2015.
Dataset Splits No The paper discusses game theory and equilibrium finding rather than machine learning models with train/validation/test dataset splits. It does not provide specific details on how data was split for training, validation, or testing.
Hardware Specification No The paper states 'For the LP solver, we used Gurobi v9.0.0 [15],' but it does not specify any hardware components (e.g., CPU, GPU models, or memory) used for running the experiments.
Software Dependencies Yes For the LP solver, we used Gurobi v9.0.0 [15].
Experiment Setup No The paper states, 'We computed 0-certificates in all cases. For the LP solver, we used Gurobi v9.0.0 [15],' and describes the games used. However, it does not provide specific hyperparameter values, detailed training configurations, or other system-level settings for reproducibility beyond the choice of solver.