Adaptive covariance estimation of locally stationary processes (Q1807064): Difference between revisions

Let \(X(t)\) be a real-valued zero-mean process with covariance \(R(t,s)= EX(t)X(s)\). The covariance operator is defined for any \(f\in L^2(R)\) by \[ Tf(t)= \int^\infty_{-\infty} R(t,s)f(s)ds. \] It is shown that the covariance operator of a locally stationary process has approximate eigenvectors that are local cosine functions. The problem of estimation of covariance operators with a ``best'' basis search is treated. Fast numerical algorithms and their application to examples of locally stationary processes are described.