ChemGymRL is a collection of reinforcement learning environments in the open AI standard which are designed for experimentation and exploration with reinforcement learning in the context of chemical discovery. These experiments are virtual variants of experiments and processes that would otherwise be performed in real-world chemistry labs and in industry.
ChemGymRL simulates real-world chemistry experiments by allocating agents to benches to perform said experiments as dictated by a lab manager who has operational control over all agents in the environment. The environment supports the training of Reinforcement Learning agents by associating positive and negative rewards based on the procedure and outcomes of actions taken by the agents.
ChemGymRL has as a goal to create an environment which would allow scientists and researchers to simulate chemistry labs with the use of reinforcement learning agents. For people interested in experimenting with material design to drug discovery, this is the right software for you!
The environment is fully compatible with modern reinforcement learning frameworks such as Keras-RL and is currently under active development.
ChemGymRL is proudly hosted on github.
ChemGymRL has built-in documentation web pages. Find detailed information, instructions and tutorials at https://chemgymrl.readthedocs.io/en/latest/