We reformulate a deterministic optimization problem as a filtering problem, where the goal is to compute the conditional distribution of the unobserved state given the observation history. We prove ...