search

zach401 / acnportal

Research tools for the Adaptive Charging Network
BSD 3-Clause "New" or "Revised" License
75 stars 32 forks source link

Open ai prototyping #76

Closed sunash closed 4 years ago

sunash commented 4 years ago

Adds a step function to a simulator, with which a simulation may be stepped forward one iteration using an input schedule. This function is used by the gym-acnportal package to realize the agent-controlled agent-environment loop required for training reinforcement learning agents.

The Simulator object now accepts None as the input scheduler. This is useful if the Simulator object is not driving the flow of the simulation (e.g. if an external agent is stepping the simulator) in which case a scheduling algorithm is not needed.

Includes minor stylistic changes and some additional tests.

review-notebook-app[bot] commented 4 years ago

Check out this pull request on  ReviewNB

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB