Second order processes with restrictions in the index set and applications (Q1092580)

Let \(X(.)=(X(t)\), \(t\in T)\) be a second order random process with index set T. There are many situations where this process may only be observed for t in a subset S of T. Some examples are: discretization of continuous time series; censored data; missing data; aggregation and grouping etc. The idea is that even if interest is concentrated on (X(t), \(t\in S)\), it may be advantageous to embed it into a larger process (X(t), \(t\in T\supseteq S)\). It is also of interest to evaluate the loss of information caused by the restriction in the index set. The author discusses the effect of the restrictions and presents applications to the Wiener and tied-down Wiener processes. These results are also applied to the computation of generalized inverses of non- negative definite matrices. It is given as well an application to asymptotic statistical inference, presenting a heuristic derivation of the asymptotic Fisher information matrix in the censored case.