TemplateGeNN: Neural Networks used to accelerate Gravitational Wave Template Bank Generation

2025/02/21 by Green, Susanna, Lundgren, Andrew
#FOS: Physical sciences #General Relativity and Quantum Cosmology (gr-qc) #Instrumentation and Methods for Astrophysics (astro-ph.IM)

paper · doi:10.48550/arxiv.2502.15337

Abstract

We introduce TemplateGeNN, a fast stochastic template bank generation algorithm which uses Graphical Processing Units (GPUs) and a LearningMatch model (Siamese neural network). TemplateGeNN generated a binary black hole template bank (chirp mass varied from 5 M_\odot ≤ M_c ≤ 20M_\odot, symmetric mass ratio varied from 0.1 ≤ η≤ 0.24999, and equal aligned spin varied from -0.99 ≤ χ_1,2≤ 0.99) of 31,640 templates in ∼ 1 day on a single A100 GPU. To test the sensitivity of this template bank we injected 7746 binary black hole templates into LIGO Gaussian noise. This template bank recovered 98% of the injections with a fitting factor greater than 0.97. For lower mass regions (black hole mass region between 5 M_\odot ≤ m_{1, 2} ≤ 25 M_\odot), 99% of 9469 injections were recovered with a fitting factor greater than 0.97. LearningMatch and TemplateGeNN are a machine-learning pipeline that can be used to accelerate template bank generation for future gravitational-wave data analysis.

Related

• LearningMatch: Siamese Neural Network Learns the Match Manifold
• Gravitational-wave template banks for novel compact binaries
• Generating optimal Gravitational-Wave template banks with metric-preserving autoencoders
• Fast, accurate, and differentiable: a neural-network surrogate for NRSur7dq4 precessing binary black hole waveforms
• Efficient Stochastic Template Bank Using Inner Product Inequalities