The Monte-Carlo method for filtering with discrete-time observations (Q5945633)

The paper deals with discrete-time filtering problems. Three different ways of occurring noisy observations are considered, but in all cases there is no explicit form for the transition semigroup of the Markov process. The performance is measured in terms of how many ''single'' random variables of fixed dimension are necessary to simulate in order to achieve a given error in the approximation. Thus, a Monte-Carlo filtering algorithm is developed and studied. Some preliminaries with necessary proofs and explanations are provided. Also, some numerical results are presented to illustrate the efficiency of the developed algorithm. The paper can be of interest for all specialists and experts involved in the Monte-Carlo method applications in filtering and estimation problems.