arXiv subjects

Lucy M. Thomas

Publications and source records attributed to Lucy M. Thomas.

9 recordsLinked to original sources

arXiv2026

Fast, accurate, and differentiable: a neural-network surrogate for NRSur7dq4 precessing binary black hole waveforms

We present a neural network surrogate model that emulates the NRSur7dq4 gravitational waveform model for precessing binary black hole mergers. The surrogate decomposes the waveform into constituent quantities and trains an independent multilayer perceptron (MLP) for each. We validate the surrogate against NRSur7dq4 on 10,000 waveforms spanning its full parameter space ($1 \leq q \leq 4$, $|χ_{A,B}| \leq 0.8$). For representative total masses between 60 and 300 $M_\odot$, median sky-averaged frequency-domain mismatches range from $8.0 \times 10^{-5}$ to $1.7 \times 10^{-4}$, with 95th percentiles below $10^{-3}$. On an NVIDIA L40S GPU the JAX surrogate evaluates a single waveform in about 1 ms end-to-end, roughly 10 times faster than the LALSimulation C implementation of NRSur7dq4, and sustains about 140 times the LALSimulation throughput at batch size 64, making it well suited for both low-latency parameter-estimation samplers and large-scale waveform generation. The full NRSur7dq4 NN waveform-to-likelihood pipeline is implemented in JAX and is differentiable. This is the first neural-network surrogate of a precessing numerical-relativity waveform model to combine validated NR-faithful accuracy with a fully differentiable, GPU-accelerated inference pipeline, enabling gradient-based inference approaches via automatic differentiation including Fisher information matrices, GPU-accelerated nested sampling, gradient-based MCMC and importance sampling.

gr-qc↗

arXiv2026

labrador: A domain-optimized machine-learning tool for gravitational wave inference

Fast and reliable inference of gravitational-wave source parameters is crucial for analyzing large catalogs that are reaching the size of hundreds of detections, and for identifying short-lived electromagnetic counterparts. Neural posterior estimation has emerged as a powerful inference method, where the model is trained on simulated gravitational-wave data at considerable computational cost, but thereafter enables extremely fast and inexpensive inference at test time. Here, we extend this approach by incorporating domain-specific physical insights and methods in the model architecture. These include compressing the data by heterodyning against a reference waveform chosen via approximate likelihood maximization, removing parameter degeneracies through tailored coordinate systems, and eliminating known multimodalities by folding the parameter space. As a result, the network is approximately equivariant to changes in the source parameters, and achieves a reduced training cost and improved model interpretability. Our implementation, called labrador, can be trained end-to-end on a 1-day timescale on $\sim 10^2$ CPU cores and a V100 GPU, achieving a median importance-sampling efficiency of 1% on quadrupolar, aligned-spin signals in a broad mass range (chirp mass $\mathcal{M} \in 1\text{-}50\,\mathrm{M}_\odot$, mass ratio $q > 0.1$). labrador is the first neural inference code to achieve extensive coverage of long-duration signals with secondary masses $m_2 < 10\,\mathrm{M}_\odot$, rendered possible by its equivariance property. Among our novel contributions is a numerically stable procedure that enables neural posterior estimation when the simulation and inference priors differ.

gr-qc↗

arXiv2026

Revisiting the Coprecessing Frame in the Presence of Orbital Eccentricity

Accurate inclusion of both spin precession and orbital eccentricity effects in gravitational waveform models represents a key hurdle in our ability to fully characterize the properties of compact binaries. Virtually all efforts to model precession rely on a coprecessing frame transformation, a time-dependent spatial rotation that tracks the dominant emission direction and simplifies the waveform morphology. We assess the utility of the coprecessing frame transformation to separate out the effect of the precession of the orbital plane from the waveform in the presence of non-negligible orbital eccentricity. We rely on 20 numerical relativity simulations, which include the complete physical effects of spin precession and eccentricity in the strong-field, and compare waveforms in both the inertial and coprecessing frames. Comparing against the eccentric, spin-aligned model SEOBNRv5EHM, we find that while the waveform mismatches decrease in the coprecessing frame, they remain above the level required for accurate waveform modeling, $\sim$ 0.01 or higher for large inclinations. Further improvements, e.g., modeling mode asymmetries as already pursued for quasicircular binaries, will likely prove essential. We also find that by removing the dominant amplitude and phase modulations from the waveform, the coprecessing frame facilitates surrogate modeling, achieving lower errors at a fixed number of basis elements compared to the inertial frame. Our results demonstrate both the utility and the limitations of the coprecessing frame as a cornerstone in waveform modeling for eccentric and precessing binaries.

gr-qc↗

arXiv2025

Optimizing Neural Network Surrogate Models: Application to Black Hole Merger Remnants

Surrogate models of numerical relativity simulations of merging black holes provide the most accurate tools for gravitational-wave data analysis. Neural network-based surrogates promise evaluation speedups, but their accuracy relies on (often obscure) tuning of settings such as the network architecture, hyperparameters, and the size of the training dataset. We propose a systematic optimization strategy that formalizes setting choices and motivates the amount of training data required. We apply this strategy on NRSur7dq4Remnant, an existing surrogate model for the properties of the remnant of generically precessing binary black hole mergers and construct a neural network version, which we label NRSur7dq4Remnant_NN. The systematic optimization strategy results in a new surrogate model with comparable accuracy, and provides insights into the meaning and role of the various network settings and hyperparameters as well as the structure of the physical process. Moreover, NRSur7dq4Remnant_NN results in evaluation speedups of up to $8$ times on a single CPU and a further improvement of $2,000$ times when evaluated in batches on a GPU. To determine the training set size, we propose an iterative enrichment strategy that efficiently samples the parameter space using much smaller training sets than naive sampling. NRSur7dq4Remnant_NN requires $O(10^4)$ training data, so neural network-based surrogates are ideal for speeding-up models that support such large training datasets, but at the moment cannot directly be applied to numerical relativity catalogs that are $O(10^3)$ in size. The optimization strategy is available through the gwbonsai package.

gr-qc↗

arXiv2022

Accelerating multimodal gravitational waveforms from precessing compact binaries with artificial neural networks

Gravitational waves from the coalescences of black hole and neutron stars afford us the unique opportunity to determine the sources' properties, such as their masses and spins, with unprecedented accuracy. To do so, however, theoretical models of the emitted signal that are i) extremely accurate and ii) computationally highly efficient are necessary. The inclusion of more detailed physics such as higher-order multipoles and relativistic spin-induced orbital precession increases the complexity and hence also computational cost of waveform models, which presents a severe bottleneck to the parameter inference problem. A popular method to generate waveforms more efficiently is to build a fast surrogate model of a slower one. In this paper, we show that traditional surrogate modelling methods combined with artificial neural networks can be used to build a computationally highly efficient while still accurate emulation of multipolar time-domain waveform models of precessing binary black holes. We apply this method to the state-of-the-art waveform model SEOBNRv4PHM and find significant computational improvements: On a traditional CPU, the typical generation of a single waveform using our neural network surrogate SEOBNN_v4PHM_4dq2 takes 18ms for a binary black hole with a total mass of $44 M_{\odot}$ when generated from 20Hz. In comparison to SEOBNRv4PHM itself, this amounts to an improvement in computational efficiency by two orders of magnitude. Utilising additional GPU acceleration, we find that this speed-up can be increased further with the generation of batches of waveforms simultaneously. Even without additional GPU acceleration, this dramatic decrease in waveform generation cost can reduce the inference timescale from weeks to hours.

gr-qc↗

arXiv2022

The last three years: multiband gravitational-wave observations of stellar-mass binary black holes

Understanding the formation and evolution of the stellar-mass binary black holes discovered by LIGO and Virgo is a challenge that spans many areas of astrophysics, from stellar evolution, dynamics and accretion disks, to possible exotic early universe processes. Over the final years of their lives, stellar-mass binaries radiate gravitational waves that are first observable by space-based detectors (such as LISA) and then ground-based instruments (such as LIGO, Virgo and the next generation observatories Cosmic Explorer and the Einstein Telescope). Using state-of-the-art waveform models and parameter-estimation pipelines for both ground- and space-based observations, we show that (the expected handful of) these multiband observations will allow at least percent-level measurements of all 17 parameters that describe the binary, the possible identification of a likely host galaxy, and the forewarning of the merger days in advance allowing telescopes at multiple wavelengths to search for any electromagnetic signature associated to it. Multiband sources will therefore be a gold mine for astrophysics, but we also show that they could be less useful as laboratories for fundamental tests of general relativity than has been previously suggested.

astro-ph.HE↗

arXiv2021

New effective precession spin for modeling multimodal gravitational waveforms in the strong-field regime

Accurately modelling the complete gravitational-wave signal from precessing binary black holes through the late inspiral, merger and ringdown remains a challenging problem. The lack of analytic solutions for the precession dynamics of generic double-spin systems, and the high dimensionality of the problem, obfuscate the incorporation of strong-field spin-precession information into semi-analytic waveform models used in gravitational-wave data analysis. Previously, an effective precession spin, $χ_p$, was introduced to reduce the number of spin degrees of freedom. Here, we show that $χ_p$ alone does not accurately reproduce higher-order multipolar modes, in particular the ones that carry strong imprints due to precession such as the $(2,1)$-mode. To improve the higher-mode content, and in particular to facilitate an accurate incorporation of precession effects in the strong-field regime into waveform models, we introduce a new dimensional reduction through an effective precession spin vector, $\vecχ_\perp$, which takes into account precessing spin information from both black holes. We show that this adapted effective precession spin (i) mimics the precession dynamics of the fully precessing configuration remarkably well, (ii) captures the signature features of precession in higher-order modes, and (iii) reproduces the final state of the remnant black hole with high accuracy for the overwhelming majority of configurations. We demonstrate the efficacy of this two-dimensional precession spin in the strong-field regime, paving the path for meaningful calibration of the precessing sector of semi-analytic waveform models with a faithful representation of higher-order modes through merger and the remnant black hole spin.

gr-qc↗

arXiv2021

A generalized precession parameter $χ_\mathrm{p}$ to interpret gravitational-wave data

Originally designed for waveform approximants, the effective precession parameter $χ_\mathrm{p}$ is the most commonly used quantity to characterize spin-precession effects in gravitational-wave observations of black-hole binary coalescences. We point out that the current definition of $χ_\mathrm{p}$ retains some, but not all, variations taking place on the precession timescale. We rectify this inconsistency and propose more general definitions that either fully consider or fully average those oscillations. Our generalized parameter $χ_\mathrm{p}\in[0,2]$ presents an exclusive region $χ_\mathrm{p}>1$ that can only be populated by binaries with two precessing spins. We apply our prescriptions to current LIGO/Virgo events and find that posterior distributions of $χ_\mathrm{p}$ tend to show longer tails at larger values. This appears to be a generic feature, implying that (i) current $χ_\mathrm{p}$ measurement errors might be underestimated, but also that (ii) evidence for spin precession in current data might be stronger than previously inferred. Among the gravitational-wave events released to date, that which shows the most striking behavior is GW190521.

gr-qc↗

arXiv2020

Measuring precession in asymmetric compact binaries

Gravitational-wave observations of merging compact binaries hold the key to precision measurements of the objects' masses and spins. General-relativistic precession, caused by spins misaligned with the orbital angular momentum, is considered a crucial tracer for determining the binary's formation history and environment, and it also improves mass estimates -- its measurement is therefore of particular interest with wide-ranging implications. Precession leaves a characteristic signature in the emitted gravitational-wave signal that is even more pronounced in binaries with highly unequal masses. The recent observations of GW190412 and GW190814 have confirmed the existence of such asymmetric compact binaries. Here, we perform a systematic study to assess the confidence in measuring precession in gravitational-wave observations of high mass ratio binaries and, our ability to measure the mass of the lighter companion in neutron star -- black hole type systems. Using Bayesian model selection, we show that precession can be decisively identified for low-mass binaries with mass ratios as low as $1:3$ and mildly precessing spins with magnitudes $\lesssim 0.4$, even in the presence of systematic waveform errors.

gr-qc↗