This repository contains an RSPD and ISPD implementation for the course "Controversies in Game Theory 2021".
We recommend setting up a Python 3 environment in conda on windows and pip on linux/mac. All dependencies of the project are listed in the requirements.txt
file.
To install via conda on windows use
conda env create --name cgt -f requirements.txt
conda activate cgt
To install on linux/mac with pip use
python3 -m venv .venv
source .venv/bin/activate
python3 -m pip install -r requirements.txt
The simulator can be used in one of two ways:
- In a web dashboard that allows quick adjustments of strategies, hyperparameters, and live updates on all statistics and the board state.
- A console-based application that enables a user to run multiple thousands of experiments without supervision and receive aggregated results.
In order to start the dashboard, simply execute the following from the root directory of the project
python3 main_dash.py
afterward open a browser at 127.0.0.0.8050
or the address that is presented in your console. On the webpage you can dynamically setup experiments and get live results. There is currently no switch to run the ISPD in the web dashboard.
You can specify one or multiple experiments in a file called experiments.csv
. We included several example setups in the report_experiments
folder. After specification you can execute all experiments by running
python3 run_experiments.py
which will automatically put the results in the data
folder with labelling.
Experiments in experiments.csv can handle two special cases:
- sets of the form
a:b:c
-> one experiment for each of the values - ranges of the form
a;b;c
-> one experiment for each of the values in range(a,b,c) where a is the start b is the end and c is the step size