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

TL;DR

TemplateGeNN is a GPU-accelerated, neural network-based algorithm that rapidly generates gravitational wave template banks, achieving high sensitivity and efficiency in black hole binary detection simulations.

Contribution

It introduces a novel GPU-based stochastic template bank generation method using a Siamese neural network, significantly reducing computation time for gravitational wave data analysis.

Findings

Generated a large template bank in about 1 day on a single GPU.

Achieved 98% recovery rate with high fitting factors in simulated data.

Demonstrated the pipeline's effectiveness for future gravitational wave searches.

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 $5 M_{\odot} \leq \mathcal{M}_{c} \leq 20M_{\odot}$, symmetric mass ratio varied from $0.1 \leq \eta \leq 0.24999$, and equal aligned spin varied from $-0.99 \leq \chi_{1,2}\leq 0.99$) of 31,640 templates in $\sim 1$ 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 $5 M_{\odot} \leq m_{1, 2} \leq 25 M_{\odot}$), 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.