DETERMINISM TEST AND NOISE ESTIMATE FOR A COMPLEX TIME-SERIES

We address the issue of recognizing determinism in a time series. Spec ifically, we employ the method of singular-value decomposition (SVD) t o derive the eigenvalue spectra of the trajectory matrices constructed from a number of scalar time series, mainly white noise and chaotic s ignals, where a very large embedding dimension is used. The results su ggest that the SVD eigenvalue spectrum can be employed as a measure of determinism and an estimate for the strength of a noise contained in the time series can be deduced.