AccRQA library: accelerating recurrence quantification analysis

Abstract Recurrence quantification analysis is a powerful tool for identifying and quantifying patterns in dynamical systems, widely used across many disciplines. The rapid growth of data in these fields demands more efficient techniques for analysis. We present AccRQA, a high-performance library available in Python, R, and C/C++, which utilizes novel, scalable, and portable parallel algorithms. AccRQA is parallelized using OpenMP and can leverage NVIDIA GPUs when available, providing portability across computational platforms (CPUs, GPUs) and user environments (PC, HPC), thus offering flexibility between exploration and systematic mapping of a vast parameter space. AccRQA supports long time series and efficient computations for different embedding dimensions m and delay $$\tau$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mi>τ</mml:mi> </mml:math> with minimal memory requirements. We also present performance benchmarks demonstrating an average 10 $$\times$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mo>×</mml:mo> </mml:math> speed-up and at least a 6 $$\times$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mo>×</mml:mo> </mml:math> speed-up compared to state-of-the-art RQA packages on both CPUs and GPUs.