Markovian representation and prediction of stochastic processes with time-varying rational spectrum (Q1180381)

This paper is concerned with the spectral factorization problem and the prediction problem for a class of stochastic processes \(\{X(t)\}\) with time-dependent rational spectrum. The prediction problem is formulated as an integral equation of the first kind, \(\int^ T_ 0 K(t,\tau)V(\tau)d\tau=f(t)\), \(0\leq\tau \leq T\) for the weighting function \(V(t)\), with \(f(t)\), a known function defined on \(0\leq t\leq T\) and \(K(\tau,t)\), the covariance function of \(\{X(t)\}\). First, state-space representation problem for the processes is considered. The representation relies on the reproducing kernel Hilbert space (RKHS) techniques. Based on this representation, the prediction problem is solved using the concept of integral operators. It should be noted that the solution is obtained under a slowly-varying condition on the coefficients of the process.