Code to reproduce examples in NeurIPS submission: Robust exploration in linear quadratic reinforcement learning.
Run:
multistep_rrl_empirical.m ~ to reproduce numerical simulations in which the method is applied to the simulated system
multistep_rrl_theoretical.m _ to reproduce numerical simulations in which the method is applied, and costs are calculated based on theoretical worst-case system behavior.
Dependencies:
Yalmip: https://yalmip.github.io/
Mosek: https://www.mosek.com/