Detecting and analysing nonstationarity in a time series with nonlinear cross-predictions

摘要

We propose an informal test for stationarity in a time series which checksfor the compatibility of nonlinear approximations to the dynamics made indifferent segments of the sequence. The segments are compared directly, ratherthan via statistical parameters. The approach provides detailed informationabout episodes with similar dynamics during the measurement period. Thusphysically relevant changes in the dynamics can be followed.