The authors propose a modification of an earlier scheme of theirs to estimate the maximal Lyapunov characteristic exponent (MLCE) of a dynamical system. This scheme, which is similar to the bootstrap method, only involves the use of available data and its interpolation in the time domain, whereby new time series are generated, each having a larger size relative to the original time series. The MLCE is then estimated for each new time series, thus obtaining a distribution for the MLCE of the original time series.
The algorithm is discussed in some detail, but the strength of this paper is in the application of the methodology to several well-known dynamical systems (including the Lorenz and Rossler attractors) for which information about the MLCE is available by other means. The results for the clean and noisy data are detailed, demonstrating reasonable agreement with the true values.