Giters

ADGEfficiency / energy-py

Reinforcement learning for energy systems

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Backwards induction

ADGEfficiency opened this issue · comments

commented

Backwards induction is a dynamic programming model - it uses an environment model (i.e. a model that predicts state transitions and rewards) to solve MDPs.

Backwards induction can be implemented for any energy_py environment by manipulating the time series structure of the energy_py envs. This will allow any energy_py env to be solved by backwards induction.

I have drafted out an implementation of backwards induction and start to think about how to manipulate the env structure.

Next step is for me to upload these!