Socially Optimal Non-discriminatory Restrictions for Continuous-Action Games

Authors: Michael Oesterle, Guni Sharon

AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experimental results with Braess Paradox and the Cournot Game show that this method leads to an optimized social utility of the Nash Equilibria, even when the assumptions are not guaranteed to hold.
Researcher Affiliation Academia Michael Oesterle1, Guni Sharon2 1Institute for Enterprise Systems, University of Mannheim, Germany 2Department of Computer Science & Engineering, Texas A&M University
Pseudocode Yes Algorithm 1: Socially Optimal Action-Space Restrictor (SOAR)
Open Source Code Yes The notebook is publicly available at https://github.com/michoest/aaai-2023.
Open Datasets No The paper defines parameterized versions of the Cournot Game and Braess Paradox rather than utilizing external, publicly available datasets. No specific links, DOIs, or citations to public datasets are provided.
Dataset Splits No The paper conducts experiments on defined game models (Cournot Game, Braess Paradox) rather than using traditional datasets with explicit training, validation, and test splits.
Hardware Specification No The paper states that experiments were executed in a Google Colaboratory notebook but does not provide specific hardware details such as GPU or CPU models.
Software Dependencies No The paper mentions that experiments were conducted in a Google Colaboratory notebook and provides a link to a GitHub repository, but it does not specify software dependencies with version numbers (e.g., Python version, library versions).
Experiment Setup Yes The values of ϵ were empirically chosen to provide a good balance between run-time and accuracy, but the results are actually reasonably insensitive to this choice: Varying ϵ by a factor of 5 caused the MESU to change by less than 1% in both games. Cournot Game Figure 3 shows the results of SOAR for λ {10, 11, ..., 200} with ϵ = 0.1. Braess Paradox Figure 4 shows the results of SOAR for b [4, 18] in steps of 0.1 with ϵ = 0.001.