Exploring the use of Random Projections for Gravitational Wave Data Analysis

Abstract

The rst observation run of advanced LIGO returned a surprising 3 detections of coalescing Binary Black Hole (BBH) systems. Having shown much promise as strong Gravitational Wave candidates lying within the aLIGO detector sensitivity, e orts are now being directed at ne-tuning our searches to detect more Compact Binary Coalescence (CBC) sources, especially ones involving Neutron Stars. A combination of algorithmic and software speedup strategies have been explored for achieving a low-latency detection of signals from these systems, to generate timely alerts for Electromagnetic Follow-up observations. In this study, we investigate another mathematical technique, called Random Projection,which guarantees the preservation of information in high-dimensional data structures under projection to a lower dimension following the Johnson-Lindenstrauss lemma. We explore the applicability of Random Projections for reducing Gravitational Wave templates in order to speed up the computation of matched filtering in the time domain.