1 problem

Let $\mathsf{H}$ denote an upper bound on the span of the optimal bias function in an average-reward Markov decision process. In the online setting, regret is measured over the int…