“…One of the main challenges in nonlinear time series analysis are the time-consuming algorithms. Although a number of improvements have been proposed to efficiently reconstruct high-dimensional state spaces (see Hegger et al, 1999 for an overview), estimate dimensions (Grassberger, 1990;Toledo et al, 1997;Lai and Lerner, 1998;Widman et al, 1998;Sprott and Rowlands, 2001), Lyapunov exponents (von Bremen et al, 1997;Oiwa and Fiedler-Ferrara, 1998a,b), and entropies (Corana and Rolando, 1995), along with efficient techniques to search for neighbors in high-dimensional state spaces (Schreiber, 1995;Merkwirth et al, 2000), real time applications are nevertheless limited by the number of time series M (i.e. the number of recording channels) and by the number of data points N in each time series.…”