In different fields of science and engineering, a model of a given underlying dynamical system can be obtained by means of measurement data records called time series. This model becomes very important to understand the original system behaviour and to predict the future values of that system. From the model, parameters such as the prediction horizon can be computed to obtain the point where the prediction becomes useless. In this work, a new parallel kd-tree based approach for computing the prediction horizon is presented. The parallel approach uses the maximal Lyapunov exponent, which is computed by Wolf’s method, as an estimator of the prediction horizon.

No abstract

No abstract

No abstract