Provably Efficient Model-Free Algorithms for Non-stationary CMDPs

We study model-free reinforcement learning (RL) algorithms in episodic non-stationary constrained Markov decision processes (CMDPs), in which an agent aims to maximize the expected cumulative reward subject to a cumulative constraint on the expected utility (cost). In the non-stationary environment, the reward, utility functions, and the transition kernels can vary arbitrarily over time as long as the cumulative variations do not exceed certain variation budgets. We propose the first model-free, simulator-free RL algorithms with sublinear regret and zero constraint violation for non-stationary CMDPs in both tabular and linear function approximation settings with provable performance guarantee. Our results on regret bound and constraint violation for tabular case match the corresponding best results for stationary CMDPs when the total budget is known. Furthermore, we provide a general framework for dealing with the well-known challenges when analyzing non-stationary CMDPs without requiring prior knowledge of the variation budget and apply the approach for both tabular and linear approximation settings.