This is the code for the numerical examples presented in the paper Zero-Sum Games between Mean-Field Teams: A Commen Information and Reachability based Analysis.
- To install,
cd
into the root directory and typepip install -r requirements.txt
- Known dependencies: Python (3.8), Numpy (1.24.2), Scipy (1.10.1), Matplotlib (3.7.1).
- Game environment is located in
simple_example/examples
as the classSimpleExample1
. - To solve and plot the value functions, run
simple_example_1.py
- Options available
SOLVE_COR
: solve the coordinator game via discretization if set true, else load from pkl filePLOT_VALUE
: plot the value functions if truePLOT_RSet
: visualize the reachable set if set true
- Game environment is located in
simple_example/examples
as the classSimpleExample2
. - To solve and plot the value functions, run
simple_example_2.py
- Options available
SOLVE_COR
: solve the coordinator game via discretization if set true, else load from pkl filePLOT_VALUE
: plot the value functions if truePLOT_RSet
: visualize the reachable set if set trueSOLVE_DEVIATE
: solve the optimal performance in finite-population case if truePLOT_DEVIATE
: plot the trend of performance gained from deviation vs. number of agents