Abstract

Experimental chaotic signals inevitably contain a certain amount of noise. Low pass filtering is known to alter the signal such that subsequent dimension analysis yields false results. We show that a certain acausal filter reduces the noise but avoids the filter-induced error. A discussion reveals that the Kaplan-Yorke equation, which relates attractor dimension to Lyapunov exponents, implicitly relies on the forward direction of time.

© 1991 Optical Society of America