MATLAB^® and Python packages for local resource-allocation games

Authors: Rahul Chandan, Dario Paccagnan and Jason Marden

The MATLAB^® and Python packages available here have been developed as companion code for our papers on local resource-allocation games (e.g., Optimal mechanisms for distributed resource-allocation). These packages are freely available, and are covered by the GNU Standard License, Version 3 (GPLv3). You are encouraged to download, develop and redistribute these packages so long as you abide by the conditions of the license agreement (e.g., do not commercialize any software developed from these packages).

We are by no means experienced programmers, and we welcome any and all feedback concerning these packages. Though we are very interested in maintaining and improving this code, we cannot guarantee that your feedback will be addressed in a timely manner. Nevertheless, please feel free to contact us by email at rchandan at ucsb dot edu.

Please note that the MATLAB^® package currently uses the built-in linprog function, and the Python package uses SciPy's linprog function. Since both packages use built-in simplex solvers, they are fairly slow. Updated versions of both packages that allow the user to specify the solver are coming soon. The packages are:

• MATLAB^®: [GitHub]
• Python 3 (with SciPy 1.3.0 or higher): [GitHub]

We also provide scripts that run the examples from their corresponding papers, and illustrate how the above packages are to be used. The examples provided currently cover the following papers:

• D Paccagnan, R Chandan, JR Marden (2020) Incentivizing efficient use of shared infrastructure: Optimal tolls in congestion games. Transactions on Economics and Computation. Submitted.
• R Chandan, D Paccagnan, JR Marden (2020) When Smoothness is Not Enough: Toward Exact Quanification and Optimization of the Price of Anarchy. Operations Research. Submitted.