Improved deep learning techniques in gravitational-wave data analysis
Peking University（北京大学）；Kavli Institute for Astronomy and Astrophysics, Peking University（北京大学科维理天文与天体物理研究所）；National Astronomical Observatories, Chinese Academy of Sciences（中国科学院国家天文台）；Beijing Normal University（北京师范大学）
Heming Xia, Lijing Shao, Junjie Zhao, Zhoujian Cao
In recent years, convolutional neural network (CNN) and other deep learning models have been gradually introduced into the area of gravitational-wave (GW) data processing. Compared with the traditional matched-filtering techniques, CNN has significant advantages in efficiency in GW signal detection tasks. In addition, matched-filtering techniques are based on the template bank of the existing theoretical waveform, which makes it difficult to find GW signals beyond theoretical expectation. In this paper, based on the task of GW detection of binary black holes, we introduce the optimization techniques of deep learning, such as batch normalization and dropout, to CNN models. Detailed studies of model performance are carried out. Through this study, we recommend to use batch normalization and dropout techniques in CNN models in GW signal detection tasks. Furthermore, we investigate the generalization ability of CNN models on different parameter ranges of GW signals. We point out that CNN models are robust to the variation of the parameter range of the GW waveform. This is a major advantage of deep learning models over matched-filtering techniques.
I present a numerical fit to the peak harmonic gravitational wave frequency emitted by an eccentric binary system in the post-Newtonian approximation. This fit significantly improves upon a previous commonly-used fit in population synthesis studies, in particular for eccentricities $\lesssim 0.8$.
Ajou University（亚洲大学）；University at Buffalo, State University of New York（纽约州立大学布法罗分校）
The gravitational wave detection problem is challenging because the noise is typically overwhelming. Convolutional neural networks (CNNs) have been successfully applied, but require a large training set and the accuracy suffers significantly in the case of low SNR. We propose an improved method that employs a feature extraction step using persistent homology. The resulting method is more resilient to noise, more capable of detecting signals with varied signatures and requires less training. This is a powerful improvement as the detection problem can be computationally intense and is concerned …
In the era of second generation ground-based gravitational wave detectors, short gamma-ray bursts (GRBs) will be among the most promising astrophysical events for joint electromagnetic and gravitational wave observation. A targeted search for gravitational wave compact binary merger signals in coincidence with short GRBs was developed and used to analyze data from the first generation LIGO and Virgo instruments. In this paper, we present improvements to this search that enhance our ability to detect gravitational wave counterparts to short GRBs. Specifically, we introduce an improved method fo…
Low mass X-ray binaries are prime targets for continuous gravitational wave searches by ground-based interferometers. Results are presented from a search for five low-mass X-ray binaries whose spin frequencies and orbital elements are measured accurately from X-ray pulsations: HETE J1900.1-2455, IGR J00291+5934, SAX J1808.4-3658, XTE J0929-314, and XTE J1814-338. Data are analysed from Observing Run 2 of the Advanced Laser Interferometer Gravitational-wave Observatory (LIGO). The search algorithm uses a hidden Markov model to track spin wandering, the $\mathcal{J}$-statistic maximum likelihood…
Recently several studies have pointed out that gravitational-wave detectors are sensitive to ultralight vector dark matter and can improve the current best constraints given by the Equivalence Principle tests. While a gravitational-wave detector is a highly precise measuring tool of the length difference of its arms, its sensitivity is limited because the displacements of its test mass mirrors caused by vector dark matter are almost common. In this Letter we point out that the sensitivity is significantly improved if the effect of finite light-traveling time in the detector's arms is taken int…
The recent detection of a neutron star binary through gravitational waves, GW170817, has offered another source of information about the properties of cold supranuclear matter. Information from the signal emitted before the neutron stars merged has been used to study the equation of state of these bodies, however, any complementary information included in the signal emitted after the merger has been lost in the detector noise. In this paper we investigate the prospects of studying GW170817-like post-merger signals with future gravitational-wave detectors. We first compute the expected properti…
We give an improved estimate of the detectability of gravitational waves from magnetically confined mountains on accreting neutron stars. The improved estimate includes the following effects for the first time: three-dimensional hydromagnetic ("fast") relaxation, three-dimensional resistive ("slow") relaxation, realistic accreted masses $M_a \la 2 \times 10^{-3} M_\odot$, (where the mountain is grown ab initio by injection), and verification of the curvature rescaling transformation employed in previous work. Typically, a mountain does not relax appreciably over the lifetime of a low-mass X-ra…
Max Planck Institute for Gravitational Physics (Albert Einstein Institute)（马克斯·普朗克引力物理研究所（阿尔伯特·爱因斯坦研究所））；Leibniz Universität Hannover（莱布尼茨汉诺威大学）；University of Wisconsin Milwaukee（威斯康星大学密尔沃基分校）
Broad searches for continuous gravitational wave signals rely on hierarchies of follow-up stages for candidates above a given significance threshold. An important step to simplify these follow-ups and reduce the computational cost is to bundle together in a single follow-up nearby candidates. This step is called clustering and we investigate carrying it out with a deep learning network. In our first paper [1], we implemented a deep learning clustering network capable of correctly identifying clusters due to large signals. In this paper, a network is implemented that can detect clusters due to …
Searches for continuous gravitational waves from \textit{unknown} Galactic neutron stars provide limits on the shapes of neutron stars. A rotating neutron star will produce gravitational waves if asymmetric deformations exist in its structure that are characterized by the star's ellipticity. In this study, we use a simple model of the spatial and spin distribution of Galactic neutron stars to estimate the total number of neutron stars probed, using gravitational waves, to a given upper limit on the ellipticity. This may help optimize future searches with improved sensitivity. The improved sens…
The pre-merger (early-warning) gravitational-wave (GW) detection and localization of a compact binary merger would enable astronomers to capture potential electromagnetic (EM) emissions around the time of the merger, thus shedding light on the complex physics of the merger. While early detection and sky localization are of primary importance to the multimessenger follow-up of the event, improved estimates of luminosity distance and orbital inclination could also provide insights on the observability of the EM emission. In this work, we demonstrate that the inclusion of higher modes of gravitat…
NCSA（美国国家超级计算应用中心）；University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）
The recent Nobel-prize-winning detections of gravitational waves from merging black holes and the subsequent detection of the collision of two neutron stars in coincidence with electromagnetic observations have inaugurated a new era of multimessenger astrophysics. To enhance the scope of this emergent science, we proposed the use of deep convolutional neural networks for the detection and characterization of gravitational wave signals in real-time. This method, Deep Filtering, was initially demonstrated using simulated LIGO noise. In this article, we present the extension of Deep Filtering usi…
University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；NCSA（美国国家超级计算应用中心）
The recent Nobel-prize-winning detections of gravitational waves from merging black holes and the subsequent detection of the collision of two neutron stars in coincidence with electromagnetic observations have inaugurated a new era of multimessenger astrophysics. To enhance the scope of this emergent field of science, we pioneered the use of deep learning with convolutional neural networks, that take time-series inputs, for rapid detection and characterization of gravitational wave signals. This approach, Deep Filtering, was initially demonstrated using simulated LIGO noise. In this article, …
The next generation of ground-based gravitational-wave detectors are likely to observe gravitational waves from the coalescences of compact-objects binaries. We describe the state of the art for predictions of the rate of compact-binary coalescences and report on initial efforts to develop a framework for converting gravitational-wave observations into improved constraints on astrophysical parameters.
NCSA, University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校国家超级计算应用中心）；Department of Statistics, University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校统计系）；Department of Astronomy, University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校天文系）；Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校电气与计算机工程系）
Gravitational wave astronomy is a rapidly growing field of modern astrophysics, with observations being made frequently by the LIGO detectors. Gravitational wave signals are often extremely weak and the data from the detectors, such as LIGO, is contaminated with non-Gaussian and non-stationary noise, often containing transient disturbances which can obscure real signals. Traditional denoising methods, such as principal component analysis and dictionary learning, are not optimal for dealing with this non-Gaussian noise, especially for low signal-to-noise ratio gravitational wave signals. Furthe…
We discuss the astrophysical science case for a decihertz gravitational-wave mission. We focus on unique opportunities for scientific discovery in this frequency range, including probes of type IA supernova progenitors, mergers in the presence of third bodies, intermediate mass black holes, seeds of massive black holes, improved sky localization, and tracking the population of merging compact binaries.
Next-generation observations will revolutionize our understanding of binary black holes and will detect new sources, such as intermediate-mass black holes. Primary science goals include: Discover binary black holes throughout the observable Universe; Reveal the fundamental properties of black holes; Uncover the seeds of supermassive black holes.
The IceCube DeepCore is a dense infill array of the IceCube Neutrino Observatory at the South Pole. While IceCube is best suited for detecting neutrinos with energies of several 100 GeV and above, DeepCore allows to probe neutrinos with lower energies. We focus on a sample of neutrinos with energies above approximately 10 GeV, which was originally optimised for oscillation experiments. Recently, it has been adapted to enable searches for transient sources of astrophysical neutrinos in the sky. In particular, this low-energy dataset can be used to conduct follow-up searches of gravitational wav…
We present an improved search for binary compact-object mergers using a network of ground-based gravitational-wave detectors. We model a volumetric, isotropic source population and incorporate the resulting distribution over signal amplitude, time delay, and coalescence phase into the ranking of candidate events. We describe an improved modeling of the background distribution, and demonstrate incorporating a prior model of the binary mass distribution in the ranking of candidate events. We find a $\sim 10\%$ and $\sim 20\%$ increase in detection volume for simulated binary neutron star and neu…
University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；NCSA（美国国家超级计算应用中心）
Gravitational wave astronomy has set in motion a scientific revolution. To further enhance the science reach of this emergent field, there is a pressing need to increase the depth and speed of the gravitational wave algorithms that have enabled these groundbreaking discoveries. To contribute to this effort, we introduce Deep Filtering, a new highly scalable method for end-to-end time-series signal processing, based on a system of two deep convolutional neural networks, which we designed for classification and regression to rapidly detect and estimate parameters of signals in highly noisy time-…
A crucial component to maximizing the science gain from the multi-messenger follow-up of gravitational-wave (GW) signals from compact binary mergers is the prompt discovery of the electromagnetic counterpart. Ideally, the GW detection and localization must be reported early enough to allow for telescopes to slew to the location of the GW-event before the onset of the counterpart. However, the time available for early warning is limited by the short duration spent by the dominant ($\ell = m = 2$) mode within the detector's frequency band. Nevertheless, we show that, including higher modes - whi…
A gravitational-wave (GW) early-warning of a compact-binary coalescence event, with a sufficiently tight localisation skymap, would allow telescopes to point in the direction of the potential electromagnetic counterpart before its onset. This will enable astronomers to extract valuable information of the complex astrophysical phenomena triggered around the time of the merger. Use of higher-modes of gravitational radiation, in addition to the dominant mode typically used in templated real-time searches, was recently shown to produce significant improvements in early-warning times and skyarea lo…
In recent years, improvements in Deep Learning (DL) techniques towards Gravitational Wave (GW) astronomy have led to a significant rise in the development of various classification algorithms that have been successfully employed to extract GWs of binary blackhole merger events from noisy time-series data. However, the success of these models is constrained by the length of time-sample and the class of GW source: binary blackhole and neutron star binaries to some extent. In this work, we intended to advance the boundaries of DL techniques using Convolutional Neural Networks, to go beyond binary…
We combine hierarchical Bayesian modeling with a flow-based deep generative network, in order to demonstrate that one can efficiently constraint numerical gravitational wave (GW) population models at a previously intractable complexity. Existing techniques for comparing data to simulation,such as discrete model selection and Gaussian process regression, can only be applied efficiently to moderate-dimension data. This limits the number of observable (e.g. chirp mass, spins.) and hyper-parameters (e.g. common envelope efficiency) one can use in a population inference. In this study, we train a n…
Similar to light, gravitational waves (GWs) can be lensed. Such lensing phenomena can magnify the waves, create multiple images observable as repeated events, and superpose several waveforms together, inducing potentially discernible patterns on the waves. In particular, when the lens is small, $\lesssim 10^5 M_\odot$, it can produce lensed images with time delays shorter than the typical gravitational-wave signal length that conspire together to form ``beating patterns''. We present a proof-of-principle study utilizing deep learning for identification of such a lensing signature. We bring the…
The catalog of gravitational-wave events is growing, and so are our hopes of constraining the underlying astrophysics of stellar-mass black-hole mergers by inferring the distributions of, e.g., masses and spins. While conventional analyses parametrize this population with simple phenomenological models, we propose an emulation-based approach that can compare astrophysical simulations against gravitational-wave data. We combine state-of-the-art deep-learning techniques with hierarchical Bayesian inference and exploit our approach to constrain the properties of repeated black-hole mergers from t…
Self-gravity computation by multipole expansion is a common approach in problems such as core-collapse and Type Ia supernovae, where single large condensations of mass must be treated. The standard formulation of multipole self-gravity suffers from two significant sources of error, which we correct in the formulation presented in this article. The first source of error is due to the numerical approximation that effectively places grid cell mass at the central point of the cell, then computes the gravitational potential at that point, resulting in a convergence failure of the multipole expansio…
Neutron stars in low mass X-ray binaries are hypothesised to emit continuous gravitational waves that may be detectable by ground-based observatories. The torque balance model predicts that a higher accretion rate produces larger-amplitude gravitational waves, hence low mass X-ray binaries with high X-ray flux are promising targets for gravitational wave searches. The detection of X-ray pulsations would identify the spin frequency of these neutron stars, and thereby improve the sensitivity of continuous gravitational-wave searches by reducing the volume of the search parameter space. We perfor…
Gravitational waves are ripples in the space time fabric when high energy events such as black hole mergers or neutron star collisions take place. The first Gravitational Wave (GW) detection (GW150914) was made by the Laser Interferometer Gravitational-wave Observatory (LIGO) and Virgo Collaboration on September 14, 2015. Furthermore, the proof of the existence of GWs had countless implications from Stellar Evolution to General Relativity. Gravitational waves detection requires multiple filters and the filtered data has to be studied intensively to come to conclusions on whether the data is a …
We demonstrate the application of a convolutional neural network to the gravitational wave signals from core collapse supernovae. Using simulated time series of gravitational wave detectors, we show that based on the explosion mechanisms, a convolutional neural network can be used to detect and classify the gravitational wave signals buried in noise. For the waveforms used in the training of the convolutional neural network, our results suggest that a network of advanced LIGO, advanced VIRGO and KAGRA, or a network of LIGO A+, advanced VIRGO and KAGRA is likely to detect a magnetorotational co…
Modeling the nanohertz gravitational wave background with the t-process: Improved frequentist analysis of gravitational-wave power
Università degli Studi di Milano-Bicocca（米兰比可卡大学）；INFN, Sezione di Milano-Bicocca（意大利国家核物理研究所米兰比可卡分部）；INAF – Osservatorio Astronomico di Cagliari（意大利国家天体物理研究所卡利亚里天文台）；INAF – Osservatorio Astronomico di Brera（意大利国家天体物理研究所布雷拉天文台）；Vanderbilt University（范德堡大学）
