Synchronization in Learning in Periodic Zero-Sum Games Triggers Divergence from Nash Equilibrium
Authors: Yuma Fujimoto, Kaito Ariu, Kenshi Abe
AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Furthermore, our experiments observe this behavior even if these assumptions are removed. |
| Researcher Affiliation | Collaboration | Yuma Fujimoto1,2,3 , Kaito Ariu1, Kenshi Abe1,4 1Cyber Agent 2University of Tokyo 3Soken University 4University of Electro-Communications |
| Pseudocode | No | The paper describes mathematical equations and derivations for the learning dynamics (e.g., Eqs. (1) and (2)) but does not present them in a structured pseudocode or algorithm block format. |
| Open Source Code | Yes | Code https://github.com/Cyber Agent AILab/ periodic games synchronization |
| Open Datasets | No | The paper uses simulated game environments such as 'periodic matching pennies' and various 'matrix games' (2x2, 3x3, 6x6) for its experiments. It does not refer to or provide access information for any external, pre-existing public datasets. |
| Dataset Splits | No | The paper simulates learning dynamics in defined game environments rather than using pre-existing datasets with explicit training/test/validation splits. |
| Hardware Specification | No | The paper describes the numerical method used for simulation (Runge-Kutta fourth-order method) but does not provide any specific details about the hardware (e.g., GPU models, CPU types, memory) on which these simulations were executed. |
| Software Dependencies | No | The paper mentions using the 'Runge-Kutta fourth-order method' for simulations, which is a mathematical algorithm. However, it does not specify any particular software libraries, programming languages, or other dependencies with version numbers used for its implementation. |
| Experiment Setup | Yes | To simulate the learning dynamics, we use the Runge-Kutta fourth-order method with a step size of 1/40. The initial strategies are set to the Nash equilibrium of the time-average game. The numerical method is the same as Fig. 1 but the step size is 1/(4 103). |