Differential phase space reconstructed for chaotic time series

A new numerical differential filter is built to estimate the numerical differential for a chaotic time series and then a differential phase space for the chaotic time series is reconstructed. Correlation dimensions, Lyapunov exponents and forecasting are discussed for the chaotic time series on the reconstructed differential phase space and on the delay phase space, respectively. Comparison results show that the numerical results on the differential phase space are better than that on the delay phase space.