Architecture, implementation and parallelization of the software to search for periodic gravitational wave signals

Abstract

The parallelization, design and scalability of the PolGrawAllSky code to search for periodic gravitational waves from rotating neutron stars is discussed. The code is based on an efficient implementation of the F -statistic using the Fast Fourier Transform algorithm. To perform an analysis of data from the advanced LIGO and Virgo gravitational wave detectors' network, which will start operating in 2015, hundreds of millions of CPU hours will be required -the code utilizing the potential of massively parallel supercomputers is therefore mandatory. We have parallelized the code using the Message Passing Interface standard, implemented a mechanism for combining the searches at different sky-positions and frequency bands into one extremely scalable program. The parallel I/O interface is used to escape bottlenecks, when writing the generated data into file system. This allowed to develop a highly scalable computation code, which would enable the data analysis at large scales on acceptable time scales. Benchmarking of the code on a Cray XE6 system was performed to show efficiency of our parallelization concept and to demonstrate scaling up to 50 thousand cores in parallel.

Search Methods for Continuous Gravitational-Wave Signals from Unknown Sources in the Advanced-Detector Era

“…This particular implementation of the F -statistic uses its own template bank setup in order to optimize the number of fast Fourier transform (FFT) computations [103,104], which normally takes a significant part of the overall computing cost of the search. Further improvements at parameter-estimation level are optimization algorithms to resolve the characteristic frequency multi-modality of the F -statistic [105] and the inclusion of machinelearning algorithms to filter out non-astrophysical candidates [106].…”

Search methods for continuous gravitational-wave signals from unknown sources in the advanced-detector era

“…This particular implementation of the F -statistic uses its own template bank setup in order to optimize the number of fast Fourier transform (FFT) computations [103,104], which normally takes a significant part of the overall computing cost of the search. Further improvements at parameter-estimation level are optimization algorithms to resolve the characteristic frequency multi-modality of the F -statistic [105] and the inclusion of machine-learning algorithms to filter out non-astrophysical candidates [106].…”

On a Dynamic Scheduling Algorithm for Massively Parallel Computations of Atomic Isotope

“…However, the scalability of the approach is limited due to statistics I/O bottleneck, and the prediction is not possible on the whole range of parameters. This technique remains useful for the simulations where the task size dependency on the input parameters can be approximated with monotonous function, such as the software to search for periodic gravitational wave signals [9], where we successfully applied the grouping approach. In case of nuclear clusters study, where the computation of each set of parameters must be done only once, this approach must be advanced.…”