Abstract:Matched-filter based gravitational-wave search pipelines identify candidate events within seconds of their arrival on Earth, offering a chance to guide electromagnetic follow-up and observe multi-messenger events. Understanding the detectors' response to an astrophysical transient across the searched signal manifold is paramount to inferring the parameters of the progenitor and deciding which candidates warrant telescope time. We describe a framework that uses artificial neural networks to interpolate gravitational waves and, equivalently, the signal-to noise ratio (SNR) across sufficiently local patches of the signal manifold. Our machine-learning based model generates a single waveform in 6 milliseconds on a CPU and 0.4 milliseconds on a GPU. When using a GPU to generate batches of waveforms simultaneously, we find that we can produce $10^4$ waveforms in $\lesssim 1$ ms. This is achieved while remaining faithful, on average, to 1 part in $10^4$ (1 part in $10^5$) for binary black hole (binary neutron star) waveforms. The model we present is designed to directly utilize intermediate detection pipeline outputs in the hopes of facilitating a better real-time understanding of gravitational-wave candidates.

What problem does this paper attempt to address?

### What problems does this paper attempt to solve? This paper aims to solve several key problems in the low - latency analysis of gravitational waves (GW), specifically including: 1. **Rapid generation of gravitational - wave waveforms and signal - to - noise ratio (SNR)**: - The paper proposes a neural - network - based interpolation method for rapidly generating gravitational - wave waveforms and their corresponding SNR in the local area of the signal manifold. This method can generate a single waveform in only 6 milliseconds on a CPU and only 0.4 milliseconds on a GPU, and can generate a large number of waveforms simultaneously (for example, generating 10,000 waveforms on a GPU takes less than 1 millisecond). This greatly improves the speed of waveform generation, making real - time analysis possible. 2. **Improving the accuracy and speed of low - latency parameter estimation**: - Although current low - latency parameter estimation pipelines (such as GstLAL) can quickly identify candidate events, their accuracy is limited. By using neural - network interpolation techniques, the parameter estimation process can be accelerated while maintaining high fidelity. This helps to more accurately understand the nature of gravitational - wave candidate events and provides better guidance for subsequent electromagnetic follow - up observations. 3. **Coping with the impact of changes in detector noise characteristics**: - Changes in the power spectral density (PSD) of the detector may lead to errors in waveform reconstruction. The paper explores how to minimize these errors by regularly updating and retraining the neural - network model to ensure the accuracy of the model under different PSD conditions. 4. **Optimizing the low - latency data processing flow**: - By integrating the neural - network interpolation technique into the existing low - latency search pipeline, gravitational - wave data can be processed and analyzed more efficiently. For example, after a candidate event is detected, the interpolation technique can be used to quickly estimate the SNR at any parameter point, thereby reducing the amount of computation required for Bayesian inference and improving the efficiency of the entire analysis flow. In summary, by introducing neural - network interpolation techniques, this paper aims to accelerate the generation of gravitational - wave waveforms and SNR, improve the accuracy and speed of low - latency parameter estimation, and optimize the entire low - latency data analysis flow to better support the development of multi - messenger astronomy.

A neural network-based gravitational wave interpolant with applications to low-latency analyses

A machine-learning pipeline for real-time detection of gravitational waves from compact binary coalescences

Detection and Parameter Estimation of Gravitational Waves from Binary Neutron-Star Mergers in Real LIGO Data using Deep Learning

Deep Neural Networks to Enable Real-time Multimessenger Astrophysics

Neural network method to search for long transient gravitational waves

Improving Early Detection of Gravitational Waves from Binary Neutron Stars Using CNNs and FPGAs

Real-Time Detection of Unmodelled Gravitational-Wave Transients Using Convolutional Neural Networks

Real-Time Detection of Gravitational Waves from Binary Neutron Stars using Artificial Neural Networks

Real-time gravitational-wave inference for binary neutron stars using machine learning

Neural network time-series classifiers for gravitational-wave searches in single-detector periods

What to do when things get crowded? Scalable joint analysis of overlapping gravitational wave signals

Generalized approach to matched filtering using neural networks

Mitigating the impact of noise transients in gravitational-wave searches using reduced basis timeseries and convolutional neural networks

A Binary Neutron Star Merger Search Pipeline Powered by Deep Learning

Rapid localization of gravitational wave sources from compact binary coalescences using deep learning

Random Convolutional Kernels for Space-Detector Based Gravitational Wave Signals

Gravitational wave search by time-scale-recursive denoising and matched filtering

Enabling multi-messenger astronomy with continuous gravitational waves: early warning and sky localization of binary neutron stars in Einstein Telescope

Toward real-time detection of unmodeled gravitational wave transients using convolutional neural networks

Enabling multimessenger astronomy with continuous gravitational waves: Early warning and sky localization of binary neutron stars in the Einstein Telescope