Computing Nash Equilibria in Potential Games with Private Uncoupled Constraints

Authors: Nikolas Patris, Stelios Stavroulakis, Fivos Kalogiannis, Rose Zhang, Ioannis Panageas

AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we empirically validate our theoretical results using constrained congestion games as a testing ground.
Researcher Affiliation Academia 1University of California, Irvine, 2Archimedes Research Unit
Pseudocode Yes Algorithm 1: IGDλ: Independent Gradient Descent on ϕ( ) = maxλ L( , λ) of the regularized Lagrangian L
Open Source Code Yes 1The code is available at the Git Hub repository: https://github.com/steliostavroulakis/constrained-potential-games
Open Datasets No The paper describes a simulation environment for 'constrained congestion games' with a 'rooted directed acyclic graph (DAG)' and 'five players' with defined 'congestion functions'. This describes the experimental setup and parameters, not a pre-existing or released dataset.
Dataset Splits No The paper describes metrics of convergence for evaluating the algorithm but does not mention specific training, validation, or test dataset splits, nor does it describe a cross-validation setup.
Hardware Specification No The paper describes its experimental setup in Section 5 but does not specify any particular hardware, such as GPU or CPU models, or cloud computing resources used for the experiments.
Software Dependencies No The paper does not provide specific software dependencies or their version numbers, such as programming language versions, library versions, or specific solver versions.
Experiment Setup Yes Our experimental setup involves a rooted directed acyclic graph (DAG)... The graph comprises four paths connecting a source node s and a target node t... In addition, there are five players... The congestion experienced on each path is influenced by the number of players selecting that particular route... Let the learning rate η be 1/β, where β = c/µ and c = 4((n Amax)2 + (Λmaxγ)2) is constant.