First-Order Algorithms for Nonlinear Generalized Nash Equilibrium Problems
Authors: Michael I. Jordan, Tianyi Lin, Manolis Zampetakis
JMLR 2023 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results demonstrate the efficiency of our algorithms in practice. We conduct some preliminary numerical experiments and highlight the practical performance of our algorithms. We summarize the results in Table 1 |
| Researcher Affiliation | Academia | Michael I. Jordan EMAIL Department of Electrical Engineering and Computer Science and Department of Statistics University of California, Berkeley Berkeley, CA 94720-1776, USA Tianyi Lin darren EMAIL Department of Electrical Engineering and Computer Science University of California, Berkeley Berkeley, CA 94720-1776, USA Manolis Zampetakis EMAIL Department of Electrical Engineering and Computer Science University of California, Berkeley Berkeley, CA 94720-1776, USA |
| Pseudocode | Yes | Algorithm 1 Accelerated Mirror-Prox Quadratic Penalty Method (AMPQP) ... Algorithm 2 Accelerated Mirror-Prox Augmented Lagrangian Method (AMPAL) |
| Open Source Code | No | The paper does not provide a direct link to a code repository or an explicit statement about releasing the source code for the methodology described. It mentions that "All of the algorithms were implemented with MATLAB R2020b" but this refers to the tools used, not their own code release. |
| Open Datasets | Yes | We conduct experiments using several datasets from Facchinei and Kanzow (2009). |
| Dataset Splits | No | The paper describes experimental parameters and stopping criteria but does not provide specific details on how datasets were split into training, validation, or test sets. |
| Hardware Specification | Yes | All of the algorithms were implemented with MATLAB R2020b on a Mac Book Pro with an Intel Core i9 2.4GHz (8 cores and 16 threads) and 16GB memory. |
| Software Dependencies | Yes | All of the algorithms were implemented with MATLAB R2020b |
| Experiment Setup | Yes | We implement our Algorithm 1 and 2 with player-specific parameters: uν max = 106 and βν 0 = ρν 0 = 1 for each ν N, where the AMP algorithm is used for solving a general VI. ... The associated parameters are chosen according to the size of the problem: γ = 4 if n < 100 and γ = 2 otherwise. ... Also, we initialize the multipliers (λ0, µ0) using the same nonnegative least-squares approach as in Kanzow and Steck (2016) and also set the same stopping criterion but with the tolerance 10 4. The maximum iteration number is set as 50 and the maximum penalty parameter is set as 1012. ... The maximum iteration number is set as 2000 and we stop each inner loop when either the iteration number exceeds 2000 or F(x)+ G(x) is below 10 6. |