For Learning in Symmetric Teams, Local Optima are Global Nash Equilibria

Code to reproduce the simulated experiments in the paper "For Learning in Symmetric Teams, Local Optima are Global Nash Equilibria"

To install:

pip install .

To launch the experiments:

The following command will run the simulated experiments in the paper:

python sim.py --gamut RandomGame --players_min 2 --players_max 5 --actions_min 2 --actions_max 5 --trials 100 --subtrials 10 --pivot --to_latex

The above command will draw RandomGames. To draw CoordinationGames or CollaborationGames, modify the --gamut flag accordingly.

Citation

To cite this work, you can use the following BibTeX entry:

@inproceedings{emmons2022learning,
    title={For Learning in Symmetric Teams, Local Optima are Global Nash Equilibria},
    author={Scott Emmons and Caspar Oesterheld and Andrew Critch and Vincent Conitzer and Stuart Russell},
    booktitle={International Conference on Machine Learning},
    year={2022},
    url={https://arxiv.org/abs/2207.03470}
}