We consider the Unimodal Multi-Armed Bandit problem where the goal is to find the optimal price under an unknown unimodal reward function, with an additional "markdown" constraint that requires that the price exploration is non-increasing. This markdown optimization problem faithfully models a...