2019/03/05 by Hongyu Shen, E. A. Huerta, Eamonn O'Shea +4
Computer Science · Mathematics · Physics and Astronomy · #Algorithm #Artificial intelligence #Astrophysical Phenomena and Observations #Astrophysics #Bayesian probability #Binary number #Computer science #Consistency (knowledge bases) #Estimator #Gamma-ray bursts and supernovae #Gaussian #Gravitational wave #Mathematics #Omega #Physics #Posterior probability #Pulsars and Gravitational Waves Research #Quantum mechanics #Statistical physics #Statistics #acm:68T01 #acm:68T35 #acm:83C35 #acm:83C57 #astro-ph.HE #cs.AI #cs.LG #gr-qc #msc:68T01 #msc:68T35 #msc:83C35 #msc:83C57 #stat.ML
paper · pdf · doi:10.1088/2632-2153/ac3843
published as Machine Learning: Science and Technology, Volume 3, Number 1, Year 2022 · v4: 13 pages, 6 figures, First application of Neural Networks for gravitational wave parameter posterior estimation across multiple events with single training
openalex publication_date 2019/03/05 · openalex created_date 2021/11/22 · arxiv created 2021/12/19 · arxiv updated 2021/12/21 · openalex updated_date 2026/08/06
We introduce deep learning models to estimate the masses of the binary components of black hole mergers, (m1,m2), and three astrophysical properties of the post-merger compact remnant, namely, the final spin, af, and the frequency and damping time of the ringdown oscillations of the fundamental ℓ=m=2 bar mode, (ωR, ωI). Our neural networks combine a modified WaveNet architecture with contrastive learning and normalizing flow. We validate these models against a Gaussian conjugate prior family whose posterior distribution is described by a closed analytical expression. Upon confirming that our models produce statistically consistent results, we used them to estimate the astrophysical parameters (m1,m2, af, ωR, ωI) of five binary black holes: GW150914, GW170104, GW170814, GW190521 and GW190630. We use PyCBC Inference to directly compare traditional Bayesian methodologies for parameter estimation with our deep-learning-based posterior distributions. Our results show that our neural network models predict posterior distributions that encode physical correlations, and that our data-driven median results and 90% confidence intervals are similar to those produced with gravitational wave Bayesian analyses. This methodology requires a single V100 NVIDIA GPU to produce median values and posterior distributions within two milliseconds for each event. This neural network, and a tutorial for its use, are available at the Data and Learning Hub for Science.