The complexity of solving MAB (multi-armed bandit) using Markov decision theory increases exponentially with the number of bandit processes. Instead of solving the n-dimensional MDP with the state-sp