Optimal and instance-dependent guarantees for Markovian linear stochastic approximation

Open in new window