Bulletin of the American Physical Society
APS April Meeting 2022
Volume 67, Number 6
Saturday–Tuesday, April 9–12, 2022; New York
Session E17: Machine Learning Techniques for Gravitational Wave DetectionRecordings Available
|
Hide Abstracts |
Sponsoring Units: DGRAV GDS Chair: Clifford Johnson, University of Southern California Room: 16th Floor Sky Lobby |
Saturday, April 9, 2022 3:45PM - 3:57PM |
E17.00001: Search for binary black hole mergers in the third observing run of Advanced LIGO-Virgo using coherent WaveBurst enhanced with Machine Learning Tanmaya Mishra, Brendan D O'Brien, Marek Szczepanczyk, Gabriele Vedovato, Shubhagata Bhaumik, Gayathri Vivekananthaswamy, Giovanni Prodi, Francesco Salemi, Edoardo Milotti, Imre Bartos, Sergey G Klimenko Coherent WaveBurst (cWB) is a search algorithm that identifies generic gravitational wave (GW) signals by looking for excess power events in the time-frequency domain with minimal assumptions on the signal model. We use a Machine Learning (ML) method to improve the search sensitivity of cWB to binary black hole (BBH) mergers by automating the signal-noise classification of excess power events reconstructed by cWB. In this work, the ML-enhanced cWB search is used to detect BBH signals in the third observing run of Advanced LIGO-Virgo. We detect, with higher significance, all the GW events previously reported by the standard cWB search in the GW Transient Catalogs. We also detect marginal candidate events not listed in the GW Transient Catalogs and estimate their source frame masses. For simulated events found with a false alarm rate of less than 1 per year, we present the improvement in the detection efficiency of approximately 20% for both the stellar-mass and intermediate-mass black hole binary mergers. We demonstrate the robustness of the ML-enhanced search for detection of generic BBH signals by reporting increased sensitivity to the spin precessing and eccentric BBH events. Furthermore, we compare the performance of the ML-enhanced cWB search with different detector networks. |
Saturday, April 9, 2022 3:57PM - 4:09PM |
E17.00002: Rapid gravitational wave source properties inference using supervised learning Shaon Ghosh Compact binary coalescences of at least one neutron star component is of special interest in the field of transient astronomy. These objects are strong emitters of well-modeled gravitational waves and simultaneously progenitors of short GRBs and kilonova. Rapid identification of the properties of these events that can inform us about the probability of having an electromagnetic counterpart (EM-Bright) is of great importance to the transient astronomy community. In this presentation I will talk about the latest updates in this work. I will show how we use supervised learning to improve the estimates of the EM-Bright probabilities in realtime. I will also present how we incorporate information from earlier events (for example inferences on the equation of state) to calculate more astrophysically accurate probabilities of EM-Bright. |
Saturday, April 9, 2022 4:09PM - 4:21PM |
E17.00003: Gravitational-wave burst searches enhanced with Machine Learning Marek Szczepanczyk The wealth of gravitational-wave (GW) discoveries by LIGO-Virgo-KAGRA provides new insight into the Universe. All of them come from compact binary systems, and a non-binary GW source is still awaiting discovery. Core-collapse supernovae are prime examples but also cosmic strings, pulsar glitches, and many others. The LIGO-Virgo-KAGRA were performing many searches for these generic sources, and coherent WaveBurst (cWB) is one of the algorithms used to identify short-duration transients. It was the only pipeline providing public alerts during LIGO-Virgo-KAGRA third observing run for unmodeled transients. The sensitivity of cWB was recently enhanced with Machine Learning. In my presentation, I will show that the new cWB version is more sensitive to a wide range of GW morphologies compared to the standard version. I will present results from the generic search with the cWB-enhanced algorithm. |
Saturday, April 9, 2022 4:21PM - 4:33PM |
E17.00004: Wavescan: multiresolution regression of gravitational-wave data. Sergey G Klimenko Identification of a transient gravitational-wave signal embedded into non-stationary noise requires the analysis of time-dependent spectral components in the resulting time series. This talk presents a regression method where a stack of wavelets with different windows spanning a wide range of resolutions, is used to scan power at each time-frequency location. Such a wavelet scan (or ``wavescan'') extends the conventional multi-resolution analysis to capture the local variations of power due to the temporal and spectral leakage. To achieve the high-resolution localization, a wavelet, least affected by the leakage, is selected from the stack at each time-frequency location. The presented method is used to obtain the high-resolution time-frequency distribution of the signal power, extract signals from noise in the wavelet domain, and reconstruct the corresponding time-domain waveforms. To demonstrate the performance of the method for detection of GW signals, the multiresolution regression is applied to the analysis of the gravitational wave data from the LIGO detectors. |
Saturday, April 9, 2022 4:33PM - 4:45PM |
E17.00005: Learning orbital dynamics of binary black hole systems from gravitational wave measurements Scott E Field, Brendan Keith, Akshay Khadse We introduce a gravitational waveform inversion strategy that discovers mechanical models of binary black hole (BBH) systems. We show that only a single time series of (possibly noisy) waveform data is necessary to construct the equations of motion for a BBH system. Starting with a class of universal differential equations parameterized by feed-forward neural networks, our strategy involves the construction of a space of plausible mechanical models and a physics-informed constrained optimization within that space to minimize the waveform error. We apply our method to various BBH systems including extreme and comparable mass ratio systems in eccentric and non-eccentric orbits. We show the resulting differential equations apply to time durations longer than the training interval, and relativistic effects, such as perihelion precession, radiation reaction, and orbital plunge, are automatically accounted for. The methods outlined here provide a new, data-driven approach to studying the dynamics of binary black hole systems. |
Saturday, April 9, 2022 4:45PM - 4:57PM |
E17.00006: Real-time localization of gravitational waves from compact binary coalescences using deep learning Chayan Chatterjee, Linqing Wen, Damon Beveridge Accurate, real-time source localization of gravitational wave (GW) events is important for electromagnetic follow-up observations of short gamma ray bursts, which follow binary neutron star mergers. Current parameter estimation methods like Markov Chain Monte Carlo (MCMC) and Nested sampling used by the LIGO-Virgo-Kagra collaboration, however, can take anywhere between a few hours to several days to infer the GW source parameter posteriors. Faster, approximately Bayesian methods like Bayestar (Singer and Price, 2016) can localize GWs in around 1 sec, but is less accurate. In this work, we introduce a deep learning model using the Normalizing Flows technique to estimate the sky direction posteriors of all compact binary coalescence (CBC) sources - binary black hole, binary neutron star, and neutron star - black hole mergers at orders of magnitude faster speed of a few milliseconds, at comparable accuracy to existing observed and published results. This is the only deep learning-based model, at the time of writing, that can achieve accurate source localization of all kinds of CBC sources at milli-second latency. We demonstrate the performance of this model on simulated two and three detector CBC signals injected in stationary, Gaussian noise and coloured by advanced LIGO power spectral density. We also discuss the application of this model for rapid estimation of other source parameter posteriors which are important for electromagnetic follow-up, like the masses and distance of the binary system. |
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