General backwards Markov models
DEMETRIOS G. LAINIOTIS · IEEE Transactions on Automatic Control · 1976
In this note general backward Markovian models for second. order stochastic processes are derived by simple backward differentiation of the "partitioned" algorithms. These backward models are equivalent to the forward process models in the sense that when solved in the backward direction they yield the same state covariance as the forward model. The general backward models are of theoretical interest as well as of computational importance in several applications.