Reduced order estimation for non-Markov processes

Reduced-order smoothers are developed for linear stochastic boundary-value problems. The design of the smoothers is shown to involve solving linear two-point boundary-value problems.