Value function based reinforcement learning (RL) algorithms, for example, -learning, learn optimal policies from datasets of actions, rewards, and state transitions. However, when the underlying state transition dynamics are stochastic and evolve on a high-dimensional space, generating independent and identically distributed (IID) data samples for creating these datasets poses a significant challenge due to the intractability of the associated normalizing integral. In these scenarios, Hamiltonian Monte Carlo (HMC) sampling offers a computationally tractable way to generate data for training RL algorithms. In this paper, we introduce a framework, called \textit{Hamiltonian -Learning}, that demonstrates, both theoretically and empirically, that values can be learned from a dataset generated by HMC samples of actions, rewards, and state transitions. Furthermore, to exploit the underlying low-rank structure of the function, Hamiltonian -Learning uses a matrix completion algorithm for reconstructing the updated function from value updates over a much smaller subset of state-action pairs. Thus, by providing an efficient way to apply -learning in stochastic, high-dimensional settings, the proposed approach broadens the scope of RL algorithms for real-world applications.
View on arXiv