Gravitational-wave surrogate models powered by artificial neural networks

Inferring the properties of black holes and neutron stars is a key science goal of gravitational-wave (GW) astronomy. To extract as much information as possible from GW observations we must develop methods to reduce the cost of Bayesian inference. In this paper, we use artificial neural networks (ANNs) and the parallelisation power of graphics processing units (GPUs) to improve the surrogate modelling method, which can produce accelerated versions of existing models. As a first application of our method, ANN-Sur, we build a time-domain surrogate model of the spin-aligned binary black hole (BBH) waveform model SEOBNRv4. We achieve median mismatches of 2e-5 and mismatches no worse than 2e-3. For a typical BBH waveform generated from 12 Hz with a total mass of $60 M_\odot$ the original SEOBNRv4 model takes 1812 ms. Existing bespoke code optimisations (SEOBNRv4opt) reduced this to 91.6 ms and the interpolation based, frequency-domain surrogate SEOBNRv4ROM can generate this waveform in 6.9 ms. Our ANN-Sur model, when run on a CPU takes 2.7 ms and just 0.4 ms when run on a GPU. ANN-Sur can also generate large batches of waveforms simultaneously. We find that batches of up to 10^4 waveforms can be evaluated on a GPU in just 163 ms, corresponding to a time per waveform of 0.016 ms. This method is a promising way to utilise the parallelisation power of GPUs to drastically increase the computational efficiency of Bayesian parameter estimation.

[1]  Scott E. Field,et al.  Surrogate model of hybridized numerical relativity binary black hole waveforms , 2018, Physical Review D.

[2]  Flanagan,et al.  Gravitational waves from merging compact binaries: How accurately can one extract the binary's parameters from the inspiral waveform? , 1994, Physical review. D, Particles and fields.

[3]  Daniel Williams,et al.  Precessing numerical relativity waveform surrogate model for binary black holes: A Gaussian process regression approach , 2019 .

[4]  Michael Purrer,et al.  Frequency-domain gravitational waves from nonprecessing black-hole binaries. I. New numerical waveforms and anatomy of the signal , 2015, 1508.07250.

[5]  Chad R. Galley,et al.  Reduced-Order Modeling with Artificial Neurons for Gravitational-Wave Inference. , 2019, Physical review letters.

[6]  Brookhaven National Laboratory,et al.  Accelerating parameter inference with graphics processing units , 2019, Physical Review D.

[7]  G. Pratten,et al.  Multipolar effective one body model for nonspinning black hole binaries , 2019, Physical Review D.

[8]  Jian-Xun Wang,et al.  Non-intrusive model reduction of large-scale, nonlinear dynamical systems using deep learning , 2019, Physica D: Nonlinear Phenomena.

[9]  Omer San,et al.  An artificial neural network framework for reduced order modeling of transient flows , 2018, Commun. Nonlinear Sci. Numer. Simul..

[10]  I. Mandel,et al.  Rapidly evaluating the compact-binary likelihood function via interpolation , 2013, 1305.3798.

[11]  Michael Purrer,et al.  Fast and accurate inference on gravitational waves from precessing compact binaries , 2016, 1604.08253.

[12]  I. Mandel,et al.  Towards Rapid Parameter Estimation on Gravitational Waves from Compact Binaries using Interpolated Waveforms , 2012, 1211.1254.

[13]  Jan S. Hesthaven,et al.  Fast prediction and evaluation of gravitational waveforms using surrogate models , 2013, ArXiv.

[14]  Drew Keppel,et al.  Interpolating compact binary waveforms using the singular value decomposition , 2011, 1108.5618.

[15]  Parameswaran Ajith,et al.  Accurate inspiral-merger-ringdown gravitational waveforms for nonspinning black-hole binaries including the effect of subdominant modes , 2017, 1708.03501.

[16]  Scott E. Field,et al.  A Surrogate model of gravitational waveforms from numerical relativity simulations of precessing binary black hole mergers , 2017, 1701.00550.

[17]  W. Marsden.  I and J , 2012 .

[18]  L. Haegel,et al.  Predicting the properties of black-hole merger remnants with deep neural networks , 2019, Classical and Quantum Gravity.

[19]  M. Colleoni,et al.  Multimode frequency-domain model for the gravitational wave signal from nonprecessing black-hole binaries , 2020, 2001.10914.

[20]  Michael Boyle,et al.  Improved effective-one-body model of spinning, nonprecessing binary black holes for the era of gravitational-wave astrophysics with advanced detectors , 2016, 1611.03703.

[21]  R. Poggiani.  Multi-messenger Observations of a Binary Neutron Star Merger , 2019, Proceedings of Frontier Research in Astrophysics – III — PoS(FRAPWS2018).

[22]  D. Barta,et al.  Fast prediction and evaluation of eccentric inspirals using reduced-order models , 2018, Physical Review D.

[23]  ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , 2019 .

[24]  R. Cotesta,et al.  Frequency-domain reduced-order model of aligned-spin effective-one-body waveforms with higher-order modes , 2020, 2003.12079.

[25]  Darina Dvinskikh,et al.  On Primal and Dual Approaches for Distributed Stochastic Convex Optimization over Networks , 2019, 2019 IEEE 58th Conference on Decision and Control (CDC).

[26]  N. Nguyen,et al.  A general multipurpose interpolation procedure: the magic points , 2008 .

[27]  Lawrence E. Kidder,et al.  Aligned-spin neutron-star–black-hole waveform model based on the effective-one-body approach and numerical-relativity simulations , 2020, Physical Review D.

[28]  Tsuyoshi Murata,et al.  {m , 1934, ACML.

[29]  Frank Ohme,et al.  Phenomenological model for the gravitational-wave signal from precessing binary black holes with two-spin effects , 2018, Physical Review D.

[30]  E. Porter,et al.  Bayesian inference for binary neutron star inspirals using a Hamiltonian Monte Carlo algorithm , 2018, Physical Review D.

[31]  Yoshinta Setyawati,et al.  Regression methods in waveform modeling: a comparative study , 2019, Classical and Quantum Gravity.

[32]  C. Pankow,et al.  Novel scheme for rapid parallel parameter estimation of gravitational waves from compact binary coalescences , 2015, 1502.04370.

[33]  S. McWilliams,et al.  Optimizing spinning time-domain gravitational waveforms for advanced LIGO data analysis , 2016, 1601.03393.

[34]  Walter Hoffmann,et al.  Iterative algorithms for Gram-Schmidt orthogonalization , 1989, Computing.

[35]  C. Mishra,et al.  Including mode mixing in a higher-multipole model for gravitational waveforms from nonspinning black-hole binaries , 2019, Physical Review D.

[36]  Y. Bouffanais,et al.  Gravitational-wave detection rates for compact binaries formed in isolation: LIGO/Virgo O3 and beyond , 2019, Physical Review D.

[37]  C. García-Quirós,et al.  Accelerating the evaluation of inspiral–merger–ringdown waveforms with adapted grids , 2020, Classical and Quantum Gravity.

[38]  Michael Purrer,et al.  Frequency-domain gravitational waves from nonprecessing black-hole binaries. II. A phenomenological model for the advanced detector era , 2015, 1508.07253.

[39]  Michele Vallisneri,et al.  Learning Bayesian Posteriors with Neural Networks for Gravitational-Wave Inference. , 2020, Physical review letters.

[40]  Frank Ohme,et al.  Including higher order multipoles in gravitational-wave models for precessing binary black holes , 2020 .

[41]  Thibault Damour,et al.  Time-domain effective-one-body gravitational waveforms for coalescing compact binaries with nonprecessing spins, tides, and self-spin effects , 2018, Physical Review D.

[42]  C. Messenger,et al.  Deep-learning continuous gravitational waves , 2019, Physical Review D.

[43]  S. Roy,et al.  Effectual template banks for upcoming compact binary searches in Advanced-LIGO and Virgo data , 2017, Physical Review D.

[44]  Andrea Taracchini,et al.  Enriching the symphony of gravitational waves from binary black holes by tuning higher harmonics , 2018, Physical Review D.

[45]  Shirley Ho,et al.  Gravitational-wave population inference with deep flow-based generative network , 2020, Physical Review D.

[46]  S. Privitera,et al.  Searching for Gravitational Waves from Compact Binaries with Precessing Spins , 2016, 1603.02444.

[47]  A Hamiltonian Monte–Carlo method for Bayesian inference of supermassive black hole binaries , 2014, 1311.7539.

[48]  Shang-hai chiao tʿung ta hsüeh.  Communications on pure and applied analysis , 2002 .

[50]  Zachariah B. Etienne,et al.  Improving performance of SEOBNRv3 by ∼300× , 2018, Classical and Quantum Gravity.

[51]  Cambridge,et al.  Gravitational wave parameter estimation with compressed likelihood evaluations , 2013, 1304.0462.

[52]  E. Clothiaux,et al.  Neural Networks and Their Applications , 1994 .

[53]  C. Broeck,et al.  Matter imprints in waveform models for neutron star binaries: Tidal and self-spin effects , 2018, Physical Review D.

[54]  Lawrence E. Kidder,et al.  Multipolar effective-one-body waveforms for precessing binary black holes: Construction and validation , 2020, Physical Review D.

[55]  B. A. Boom,et al.  Prospects for observing and localizing gravitational-wave transients with Advanced LIGO, Advanced Virgo and KAGRA , 2013, Living Reviews in Relativity.

[56]  Vincent Delmas,et al.  Non-intrusive reduced-order modeling using uncertainty-aware Deep Neural Networks and Proper Orthogonal Decomposition: Application to flood modeling , 2021, J. Comput. Phys..

[57]  Drew Keppel,et al.  Interpolation in waveform space: enhancing the accuracy of gravitational waveform families using numerical relativity , 2013 .

[59]  N. Nguyen,et al.  An ‘empirical interpolation’ method: application to efficient reduced-basis discretization of partial differential equations , 2004 .

[60]  Aaas News,et al.  Book Reviews , 1893, Buffalo Medical and Surgical Journal.

[61]  F. S. Guzmán,et al.  Time series analysis of gravitational wave signals using neural networks , 2015 .

[62]  Rory Smith,et al.  Accelerated gravitational wave parameter estimation with reduced order modeling. , 2014, Physical review letters.

[63]  F. Ohme,et al.  Enhancing gravitational waveform models through dynamic calibration , 2018, Physical Review D.

[64]  R. Stephenson.  A and V , 1962, The British journal of ophthalmology.

[65]  Marlin B. Schäfer,et al.  Detection of gravitational-wave signals from binary neutron star mergers using machine learning , 2020, Physical Review D.

[66]  P. K. Panda,et al.  GW190814: Gravitational Waves from the Coalescence of a 23 M$_\odot$ Black Hole with a 2.6 M$_\odot$ Compact Object , 2020, 2006.12611.

[67]  Scott E. Field,et al.  Surrogate models for precessing binary black hole simulations with unequal masses , 2019, Physical Review Research.

[68]  A. Samajdar,et al.  Improving the NRTidal model for binary neutron star systems , 2019, Physical Review D.

[70]  Kurunathan Ratnavelu,et al.  FRONTIERS IN PHYSICS , 2009 .

[71]  Frank Ohme,et al.  First Higher-Multipole Model of Gravitational Waves from Spinning and Coalescing Black-Hole Binaries. , 2017, Physical review letters.

[72]  Maria Alessandra Papa,et al.  Deep learning for clustering of continuous gravitational wave candidates II: identification of low-SNR candidates , 2020, ArXiv.

[73]  Matthew West,et al.  The PyCBC search for gravitational waves from compact binary coalescence , 2015, 1508.02357.

[74]  D. Keitel,et al.  Faster search for long gravitational-wave transients: GPU implementation of the transient F-statistic , 2018, Classical and Quantum Gravity.

[76]  G. Pratten,et al.  Gravitational-wave selection effects using neural-network classifiers , 2020, Physical Review D.

[77]  S. Duane,et al.  Hybrid Monte Carlo , 1987 .

[78]  G. Pratten,et al.  Multipolar effective one body waveform model for spin-aligned black hole binaries , 2020, 2001.09082.

[79]  Michael Pürrer,et al.  Frequency-domain reduced order models for gravitational waves from aligned-spin compact binaries , 2014 .

[80]  Baijiong Lin,et al.  Binary neutron stars gravitational wave detection based on wavelet packet analysis and convolutional neural networks , 2019, Frontiers of Physics.

[81]  E. Thrane,et al.  Parallelized inference for gravitational-wave astronomy , 2019, Physical Review D.

[82]  Jan S. Hesthaven,et al.  Non-intrusive reduced order modeling of nonlinear problems using neural networks , 2018, J. Comput. Phys..

[83]  Michael Boyle,et al.  Numerical relativity waveform surrogate model for generically precessing binary black hole mergers , 2017, 1705.07089.

[84]  I. Mandel,et al.  Accelerating gravitational wave parameter estimation with multi-band template interpolation , 2017, 1703.02062.

[85]  A. Nagar,et al.  Efficient effective one body time-domain gravitational waveforms , 2018, Physical Review D.