Double Oracle Algorithm for Computing Equilibria in Continuous Games
Authors: Lukáš Adam, Rostislav Horčík, Tomáš Kasl, Tomáš Kroupa5070-5077
AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our numerical experiments show that it outperforms fictitious play on several examples of games appearing in the literature. In particular, we provide a detailed analysis of experiments with a version of the continuous Colonel Blotto game. |
| Researcher Affiliation | Academia | Luk aˇs Adam, Rostislav Horˇc ık, Tom aˇs Kasl, Tom aˇs Kroupa Artificial Intelligence Center, Faculty of Electrical Engineering, Czech Technical University in Prague {adamluk3, xhorcik, kasltoma, tomas.kroupa}@fel.cvut.cz |
| Pseudocode | Yes | Algorithm 1 Double Oracle Algorithm |
| Open Source Code | Yes | Our experiments codes are available at https://github.com/sadda/Double Oracle. |
| Open Datasets | No | The paper describes the mathematical definitions of the games used (e.g., polynomial game G1, Townsend function, Colonel Blotto game) which serve as the basis for experiments. However, it does not provide concrete access information (links, DOIs, repositories) for these as downloadable datasets in the typical sense. |
| Dataset Splits | No | The paper does not specify training, validation, or test splits for datasets as it defines continuous games and runs algorithms on these game definitions, rather than using traditional datasets that are split for machine learning tasks. |
| Hardware Specification | Yes | All computations were performed on a laptop with Intel Core i5 CPU and 8GB RAM and no GPU was involved. |
| Software Dependencies | No | The examples were implemented in Python with solvers scipy.optimize and mip. No specific version numbers for Python or the solvers are provided. |
| Experiment Setup | Yes | The best responses were computed by selecting the best point of a uniform discretization for the one-dimensional problems and by using a mixed-integer linear programming reformulation for the Colonel Blotto games. Randomness is present only in the initialization of one-dimensional examples when a random pair of pure strategies is found. We observed that the choice of initial strategy sets X1 and Y1 is crucial. Indeed, setting X1 = Y1 = {(1, 0, 0), (0, 1, 0), (0, 0, 1)} provides much faster convergence than starting from a random point. For the numerical results we consider three battlefields (n = 3) with equal weights (a = (1, 1, 1)). In both cases we put c = 1/16. |