Highly scalable algorithm for computation of recurrence quantitative analysis

Recurrence plot analysis is a well-established method to analyse time series in numerous areas of research. However, it has exponential computational and spatial complexity. As the main result of this paper, a technique for the computation of recurrence quantitative analysis (RQA) is outlined. This method significantly reduces spatial complexity of computation by computing RQA directly from the time series, optimizing memory accesses and reducing computational time. Additionally, parallel implementation of this technique is tested on the Salomon cluster and is proved to be extremely fast and scalable. This means that recurrence quantitative analysis may be applied to longer time series or in applications with the need of real-time analysis.