Algorithm Optimization and Parallelization of PRESTO Single Pulse Search

Since the discovery of some objects without obvious periodicity, such as fast radio bursts and rotating radio temporary sources, the single pulse search approach has attracted extensive attention of researchers. Meanwhile, the radio telescope equipment is becoming more and more perfect, which brings higher resolution and wider observation space. Therefore, the amount of data generated by pulsar observation increases dramatically, and it is urgent to speed up the data processing of pulsars. The single pulse search in PRESTO (PulsaR Exploration and Search TOolkit) is introduced, the detrending algorithm in single pulse search is optimized by Cython programming, and the parallelization of single pulse search on CPU (Central Processing Units) is realized by Ray Framework. The experimental results show that the parallelization of the optimized single pulse search algorithm can significantly improve the performance of the search program and shorten the data processing time. At the same time, the parallel strategy is only based on CPU, so it can realize high-performance data processing in pure CPU environment without modifying the code.