Slow and fast dynamics in coupled systems: A time series analysis view

We study the dynamics of systems with different timescales, when access only to the slow variables is allowed. We use the concept of finite size Lyapunov exponent (FSLE) and consider both the cases when the equations of motion for the slow components are known, and the situation when a scalar time series of one of the slow variables has been measured. A discussion on the effects of parametrizing the fast dynamics is given. We show that, although the computation of the largest Lyapunov exponent can be practically infeasible in complex dynamical systems, the computation of the FSLE allows to extract information on the characteristic time and on the predictability of the large-scale, slow-time dynamics even with moderate statistics and unresolved small scales. Copyright (C) 1998 Elsevier Science B.V.