A repository for the project "Evolution of cooperation among individuals with limited payoff memory," a collaboration with @chilbe3 and Alex McAvoy, at the Max Planck Research Group Dynamics of Social Behavior.
Preprint: https://arxiv.org/abs/2311.02365.
The code for simulating the evolutionary process described in the paper has been
implemented in Matlab. The code used for the results can be found in the folder
matlab.
As Matlab is not an open-source language, we have also implemented the process in Python. Note that the Python code is slower.
To use the source code (Python or Matlab), you will first need to clone the repository locally to your computer. You just need to run the following command in the terminal:
$ git clone git@github.com:Nikoleta-v3/evolution-of-cooperation-with-limited-updating-payoff-memory.gitThe simulations have been run on the supercomputer. We acknowledge and are thankful to the Scientific Compute Cluster at GWDG, the joint data center of the Max Planck Society for the Advancement of Science (MPG), and the University of Göttingen.
For the results, the evolutionary process when the updating considers the two last rounds with two interactions and the results for higher mutation have the longest running times. We recommend running them on a remote computer.
The data generated from running the processes have been archived. However, if you want to reproduce the work, you can use the Matlab scripts (see README.md).
The scripts output a .csv files with the residents at different time steps of
the evolutionary process. You need to run the script clean_data.py, which can
be found in the folder scripts, before running the analysis.
The analysis and figures have been created using Jupyter Notebooks. Before running the analysis, you have to download the simulation data. They are archived in Zenodo: link here.
The notebooks for the figures can be found in the folder nbs.
The analysis only uses basic Python packages that should be installed with your Python. However, we have created an environment file. To install it, navigate to the project using the terminal and run the command:
$ conda env create -f environment.ymlThis installs all the dependencies on a conda environment. You can activate
the environment by running:
$ conda activate stochastic-payoffsThe environment can also appear in your Jupyter Notebook by running the following command:
$ python -m ipykernel install --user --name stochastic-payoffs --display-name stochastic-payoffsThe paper and supplementary information are written in LaTeX. All the necessary
files to compile the documents can be found in the folder paper.
The repository is under an MIT license.
All contributions are welcome, whether they are in the form of code, feedback, or reporting issues.
As a reminder, the Python code for the evolutionary process is rather slow, so we recommend using the Matlab code. We are working on creating a more efficient Python code for such processes.
To install the Python package, navigate to the repository using the terminal. Once you are there, run the following command:
$ python setup.py installThe Python code for the project has been developed using test-driven development.
To run the test suite and the test suite's coverage, run the following command (while the environment is activated):
$ pytest --cov=src tests/