Bounded Rationality Equilibrium Learning in Mean Field Games

Accompanying code for the paper "Bounded Rationality Equilibrium Learning in Mean Field Games" by Y. Eich, C. Fabian, K. Cui, and H. Koeppl. Proceedings of the AAAI Conference on Artificial Intelligence, 39, 2025.

Install Python packages in venv

python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Run example

Run a example, i.e. solving random problem with GFP for different equilibria. First change directory to experiments/random

python exp1_GFP.py

Plot example exploitabilities

Plot the results. For generated figures, see the figures folder. (The "figures" folder needs to be created first)

python plot1_GFP.py

Run custom experiments

Options can be found in args_parser.

python main_fp.py --game=random --variant=QRE_fp  --temperature=0.1

These figures from the paper are created by the following scripts

Figure 2 -> experiments/SIS/exp1_GFP.py -> experiments/SIS/plot1_GFP.py

Figure 3 -> experiments/RPS/exp4_simplex.py -> experiments/RPS/plot4_simplex.py

Figure 4 -> experiments/random/exp4_RH_QRE.py -> experiments/random/plot4_RH_QRE.py

Figure 5 -> experiments/random/exp2_GFPI.py -> experiments/random/plot2_GFPI.py

Figure 6 -> experiments/SIS/exp2_GFPI.py -> experiments/SIS/plot2_GFPI.py

Figure 7 -> experiments/RPS/exp2_GFPI.py -> experiments/SIS/plot2_GFPI.py

Figure 8 -> experiments/random/exp1_GFP.py -> experiments/random/plot1_GFP.py

Figure 9 -> experiments/RPS/exp1_GFP.py -> experiments/RPS/plot1_GFP.py

Figure 10 -> experiments/RPS/exp5_RH_seq_vs_parallel.py -> experiments/RPS/plot5_RH_seq_vs_parallel.py