Bulletin of the American Physical Society
APS April Meeting 2021
Volume 66, Number 5
Saturday–Tuesday, April 17–20, 2021; Virtual; Time Zone: Central Daylight Time, USA
Session Q17: Gravitational Wave Data Analysis Methods ILive
|
Hide Abstracts |
Sponsoring Units: DGRAV Chair: Derek Davis, Caltech |
Monday, April 19, 2021 10:45AM - 10:57AM Live |
Q17.00001: Can a computer learn if LIGO and Virgo will observe gravitational waves? Davide Gerosa, Geraint Pratten, Alberto Vecchio We present a novel machine-learning approach to estimate selection effects in gravitational-wave observations. Using techniques similar to those commonly employed in image classification and pattern recognition, we train a series of neural-network classifiers to predict the LIGO/Virgo detectability of gravitational-wave signals from compact-binary mergers. We include the effect of spin precession, higher-order modes, and multiple detectors and show that their omission, as it is common in large population studies, tends to overestimate the inferred merger rate in selected regions of the parameter space. Although here we train our classifiers using a simple signal-to-noise ratio threshold, our approach is ready to be used in conjunction with full pipeline injections, thus paving the way toward including actual distributions of astrophysical and noise triggers into gravitational-wave population analyses. See arxiv:2007.06585. [Preview Abstract] |
Monday, April 19, 2021 10:57AM - 11:09AM Live |
Q17.00002: Searches for Compact Binary Coalescence Events using Neural Networks in LIGO/Virgo Second Observation Period Alexis Menendez-Vazquez, Machiel Kolstein, Mario Martinez, Lluïsa-Maria Mir We present results on the search for the coalescence of compact binary mergers using convolutional neural networks and the LIGO/Virgo data, corresponding to the O2 observation period. Two-dimensional images in time and frequency are used as input, and two sets of neural networks are trained separately for low mass (0.2 - 2.0 solar masses)and high mass (25 - 100 solar masses) compact binary coalescence events. We explored neural networks trained with input information from a single or a pair of interferometers, indicating that the use of information from pairs leads to an improved performance. A scan over the full O2 data set using the convolutional neural networks for detection demonstrates that the performance is compatible with that from canonical pipelines using matched filtering techniques. No additional events with significant signal-to-noise ratio are found in the O2 data. [Preview Abstract] |
Monday, April 19, 2021 11:09AM - 11:21AM Live |
Q17.00003: Deep learning for clustering of continuous gravitational wave candidates in broad searches Banafsheh Beheshtipour, Maria Alessandra Papa Searching continuous gravitational waves from unseen objects is computationally expensive and relies on hierarchies of follow-up stages for candidates above a given significance threshold. Clustering is a powerful technique which simplifies these follow-ups and reduces the computational cost by bundling together nearby candidates in a single follow-up. We present deep learning networks that automate the clustering procedure. We implemented two networks, one can identify clusters due to large signals, and one can detect clusters due to much fainter signals. These two networks are complementary and using them in cascade achieves an excellent detection efficiency across a wide range of signal strengths. Also, this method shows a false alarm rate comparable/lower than that of clustering methods currently in use. [Preview Abstract] |
Monday, April 19, 2021 11:21AM - 11:33AM Live |
Q17.00004: Enhancing detection of gravitational waves with machine learning Tanmaya Mishra The Coherent WaveBurst (cWB) search algorithm identifies gravitational wave (GW) signals in the LIGO-Virgo data by looking for excess power events in the time-frequency domain. In order to efficiently and robustly separate signal events from noise events, we propose a machine learning algorithm to improve the cWB detection efficiency for binary black hole (BBH) mergers. A decision tree machine learning algorithm based on XGBoost is incorporated into the cWB framework. We test the enhanced cWB search on the O1-O2 data of LIGO-Virgo and successfully recover all the BBH events previously detected by cWB. We demonstrate an improvement of $\sim 25\%$ of the detection efficiency on a simulation set of stellar-mass BBHs, and an improvement of $\sim 15\%$ for the intermediate mass black hole mergers with total mass above 100 $M_{\odot}$. We demonstrate that the enhanced cWB search also has increased sensitivity to the eccentric binary mergers even when trained only on circular binary waveforms. [Preview Abstract] |
Monday, April 19, 2021 11:33AM - 11:45AM Live |
Q17.00005: Enhancing gravitational-wave burst detection confidence in expanded detector networks with the BayesWave pipeline Yi Shuen Christine Lee, Margaret Millhouse, Andrew Melatos The global gravitational-wave detector network achieves higher detection rates, better parameter estimates, and more accurate sky localisation, as the number of detectors, $\mathcal{I}$ increases. This talk quantifies network performance as a function of $\mathcal{I}$ for \textit{BayesWave}, a source-agnostic, wavelet-based, Bayesian algorithm which distinguishes between true astrophysical signals and instrumental glitches. Detection confidence is quantified using the signal-to-glitch Bayes factor, $\mathcal{B}_{\mathcal{S},\mathcal{G}}$. An analytic scaling is derived for $\mathcal{B}_{\mathcal{S},\mathcal{G}}$ versus $\mathcal{I}$, the number of wavelets, and the network signal-to-noise ratio, SNR$_\text{net}$, which is confirmed empirically via injections into detector noise of networks comprising two, three and four interferometers. We also show how larger detector networks impact the quality of the waveform reconstruction, and the sky localisation of the signal. [Preview Abstract] |
Monday, April 19, 2021 11:45AM - 11:57AM Live |
Q17.00006: A novel signal consistency check for gravitational waves Ryan Magee Gravitational wave detection pipelines have successfully identified dozens of candidate GWs originating from the merger of binary black holes, binary neutron stars, and neutron star black hole binaries. Despite the success of these pipelines, noise transients — or glitches — can occasionally mimic characteristics of astrophysical signals. We describe a new signal consistency check in the GstLAL-based matched filter pipeline that measures the response of the entire template bank to each GW candidate. [Preview Abstract] |
Monday, April 19, 2021 11:57AM - 12:09PM Live |
Q17.00007: Detection and waveform reconstruction of gravitational-wave signals with coherent Wave Burst and Wavelet Detection Filter methods. Yanyan Zheng, Alberto Less, Filip Morawski, Marco Cavaglia, Elena Cuoco The coherent-Wave Burst (cWB) and the Wavelet Detection Filter (WDF) pipelines are wavelet-based software tools designed for the detection and reconstruction of unmodelled, transient gravitational-wave signals for a network of gravitational-wave detectors. cWB is one of the main pipelines in use by the LIGO Scientific and Virgo collaborations to search for unmodelled signals. WDF is an event trigger generator that is designed to incorporate machine learning. We compare the detection efficiency of the two pipelines by injecting binary black hole signals in Advanced detector recolored noise. We find that the efficiency of the pipelines is comparable for signals with signal-to-noise ratio above \textasciitilde 10. We perform a statistical comparison of the signal waveforms that are reconstructed by the pipelines. Both cWB and WDF provide comparable reconstructed waveforms that are consistent with the injected signals. [Preview Abstract] |
Monday, April 19, 2021 12:09PM - 12:21PM Live |
Q17.00008: Simulation-based inference for compact binaries Stephen Green, Jonathan Gair Over the past five years, LIGO and Virgo have published 50 detections of gravitational waves from compact binary coalescences. To infer the system parameters, iterative sampling algorithms such as MCMC are typically used with Bayes' theorem to obtain posterior samples---by repeatedly generating waveforms and comparing to measured strain data. In this talk, I will describe instead the use of simulation-based inference with deep neural networks to learn non-iterative surrogate models for the posterior, which can be used to perform accurate inference in seconds. We use normalizing flows to represent the full 15-dimensional posterior distribution for binary black holes, and we demonstrate inference on real data. These approaches therefore represent a path forward for fast multimessenger alerts and a means to address the growing rate of detections. I will conclude by discussing prospects for treating non-Gaussian detector noise using these methods. [Preview Abstract] |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700