pengyang7881187/sicmdp-new

Implementation of the environments and algorithms in Semi-Infinitely Constrained Markov Decision Processes and Efficient Reinforcement Learning

SICMDP

Liangyu Zhang, Yang Peng, Wenhao Yang, Zhihua Zhang. Semi-Infinitely Constrained Markov Decision Processes and Efficient Reinforcement Learning

The repository includes the implementation of the environments and algorithms (SI-CPO and SI-CPPO) in the paper, see https://github.com/pengyang7881187/SICMDP for the implemntation of SI-CRL algorithm.

The tabular environment discharge of sewage and SI-CPO algorithm are included in tabular_envs, and other directories include the code for the ship route planning environment and SI-CPPO algorithm based on RLlib.

Requirements

• Gurobipy
• Gymnasium: pip install gymnasium
• Pytorch: pip install torch==1.13.1+cu117 --extra-index-url https://download.pytorch.org/whl/cu117
• RLlib: pip install "ray[rllib]"