arXiv subjects

Antonio Onofre

Publications and source records attributed to Antonio Onofre.

12 recordsLinked to original sources

arXiv2026

DANSur_HM: Modularly incorporating higher modes in a deep learning based gravitational-wave surrogate

Numerical relativity (NR) simulations provide the most faithful representation of the gravitational waves (GWs) emitted by binary black hole (BBH) systems during merger. In the context of GW astronomy, tasks such as parameter estimation can require vast numbers of waveform evaluations per second across the entire parameter space. Since performing full NR simulations for each evaluation is not computationally feasible, interpolating methods for existing NR waveforms, known as surrogate models, have been developed with marked success. In this paper, we build on our previous work to introduce methods to train a fast surrogate model based on neural networks in order to generate BBH merger waveforms, including the fundamental (2,2) mode, as well as the (3,3), (2,1), (4,4), (3,2), (4,3) and (5,5) higher-order modes. Applying a pretraining step on approximant data before fine-tuning on NR data allows us to smooth out the parameter space, and making use of the parallelization ability of GPUs to project the NR waveforms in the inclination-phase $(ι, ϕ)$ sphere during training allows the fitting of all the explored modes simultaneously. The developed surrogate model, \texttt{DANSur\_HM}, achieves average mismatches of the order of $10^{-4}$, with the worst mismatch at $2.5\times10^{-3}$, and achieves throughput above $6\times10^5$ waveforms/second on an NVIDIA V100 GPU. Parameter estimation tests confirm the usefulness of the inclusion of higher modes.

gr-qc↗

arXiv2026

Automatic classification pipeline for glitches in the Virgo detector

Glitches frequently contaminate data in gravitational-wave detectors, complicating the observation and analysis of astrophysical signals. This work introduces VIGILant, an automatic pipeline for classification and visualization of glitches in the Virgo detector. Using a curated dataset of Virgo O3b glitches, two machine learning approaches are evaluated: tree-based models (Decision Tree, Random Forest and XGBoost) using structured Omicron parameters, and Convolutional Neural Networks (ResNet) trained on spectrogram images. While tree-based models offer higher interpretability and fast training, the ResNet34 model achieved superior performance, reaching a F1 score of 0.9772 and accuracy of 0.9833 in the testing set, with inference times of tens of milliseconds per glitch. The pipeline has been deployed for daily operation at the Virgo site since observing run O4c, providing the Virgo collaboration with an interactive dashboard to monitor glitch populations and detector behavior. This allows to identify low-confidence predictions, highlighting glitches requiring further attention.

gr-qc↗

arXiv2026

Searches for Extra Higgs Bosons using $t\bar{t}+$Higgs{$(\to b\bar b)$} Events within 2HDMs: Direct versus Indirect Probes

We study the possibility of establishing the production of additional Higgs states in the process $gg,q\bar q\to t\bar t Φ$, where $Φ$ = $H_{2,3}$, with $H_2$ being CP-even and $H_3$ being CP-odd, at the Large Hadron Collider (LHC), by solely exploiting the kinematic features of the reconstructed $t\bar t$ system. We adopt as reference theoretical framework a generic CP-Conserving 2-Higgs Doublet Model (2HDM), which also accommodate a Standard Model (SM)-like Higgs state $H_1$. We show that the masses $m_{H_{2,3}}$ exhibit clear correlations with the $t\bar{t}$ system properties and could, in principle, be extracted from these. Moreover, the CP properties of the $H_{2,3}$ states can be determined, even when both states are produced simultaneously. We then compare the results produced using this method with those obtained from a full kinematic reconstruction of the $H_{2,3}$ decays in the most studied $b\bar b$ channel (we take $m_{H_{2,3}}< 2m_t$), thus proving the superiority of the approach here proposed. This paves the way to both the discovery and characterization of additional Higgs states produced {\sl inclusively} in association with top-antitop quark pairs, thereby dispensing of the complications intrinsic to the {\sl exclusive} reconstruction of such states from their decay products. We test this by establishing the sensitivity of our approach in the case of a Composite 2HDM (C2HDM), describing the Higges as pseudo-Nambu Goldstone Bosons (pNGBs) and naturally predicting Higgs mass spectra in the range of sensitivity of the described analysis.

hep-ph↗

arXiv2026

An autoencoder-based surrogate waveform model for quasi-circular binary-black-hole mergers

The generation of accurate waveforms from binary black hole (BBH) mergers is a major effort in Gravitational-Wave Astronomy. In recent years, machine-learning-based surrogate models for BBH waveforms have been proposed. Those offer the potential to dramatically accelerate waveform generation while maintaining accuracy competitive with that of traditional waveform approximants. In this work, we investigate the viability of autoencoders as generative models for gravitational-wave signals from quasi-circular BBH mergers. We introduce AESur3dq8, a novel surrogate waveform model based on autoencoders that enables the rapid and accurate construction of large template banks, producing millions of waveforms in under a second using modest computational resources. The model is trained on the numerical-relativity-informed surrogate NRHybSur3dq8 and subsequently fine-tuned using the SXS catalog of BBH simulations. We demonstrate that waveforms generated by AESur3dq8 achieve mismatches of order $10^{-4}$ with respect to Numerical Relativity waveforms, and that parameter estimation performed with these templates yields results fully consistent with those reported by the LIGO-Virgo-KAGRA Collaboration for observed gravitational-wave events.

astro-ph.IM↗

arXiv2025

A Deep Learning Powered Numerical Relativity Surrogate for Binary Black Hole Waveforms

Gravitational-wave approximants are essential for gravitational-wave astronomy, allowing the coverage binary black hole parameter space for inference or match filtering without costly numerical relativity (NR) simulations, but generally trading some accuracy for computational efficiency. To reduce this trade-off, NR surrogate models can be constructed using interpolation within NR waveform space. We present a 2-stage training approach for neural network-based NR surrogate models. Initially trained on approximant-generated waveforms and then fine-tuned with NR data, these dual-stage artificial neural surrogate (\texttt{DANSur}) models offer rapid and competitively accurate waveform generation, generating millions in under 20ms on a GPU while keeping mean mismatches with NR around $10^{-4}$. Implemented in the \textsc{bilby} framework, we show they can be used for parameter estimation tasks.

gr-qc↗

arXiv2024

Deep-Learning Classification and Parameter Inference of Rotational Core-Collapse Supernovae

We test deep-learning (DL) techniques for the analysis of rotational core-collapse supernovae (CCSN) gravitational-wave (GW) signals by performing classification and parameter inference of the maximum (peak) frequency and the GW strain amplitude ($Δh$) multiplied by the luminosity distance ($D$) attained at core bounce, respectively, $(f_{peak})$ and $(D \cdot Δh)$. Our datasets are built from a catalog of numerically generated CCSN waveforms assembled by Richers et al. 2017. Those waveforms are injected into noise from the Advanced Laser Interferometer Gravitational Wave Observatory and Advanced Virgo detectors corresponding to the O2 and O3a observing runs. For a network signal-to-noise ratio (SNR) above 5, our classification network using time series detects Galactic CCSN GW signals buried in detector noise with a false positive rate of 0.10% and a 98% accuracy, being able to detect all signals with SNR>10. The inference of $f_{peak}$ is more accurate than for $D \cdot Δh $, particularly for our datasets with the shortest time window (0.25 s) and for a minimum SNR=15. From the calibration plots of predicted versus true values of the two parameters, the standard deviation ($σ$) and the slope deviation with respect to the ideal value are computed. We find $σ_{D \cdot Δh} = 52.6$ cm and $σ_{f_{peak}} = 18.3$ Hz, with respective slope deviations of 11.6% and 8.3%. Our best model is also tested on waveforms from a recent CCSN catalog built by Mitra et al. 2023, different from the one used for the training. For these new waveforms, the true values of the two parameters are mostly within the $1σ$ band around the network's predicted values. Our results show that DL techniques hold promise to infer physical parameters of Galactic rotational CCSN events.

astro-ph.HE↗

arXiv2023

Comparison of neural network architectures for feature extraction from binary black hole merger waveforms

We evaluate several neural-network architectures, both convolutional and recurrent, for gravitational-wave time-series feature extraction by performing point parameter estimation on noisy waveforms from binary-black-hole mergers. We build datasets of 100,000 elements for each of four different waveform models (or approximants) in order to test how approximant choice affects feature extraction. Our choices include \texttt{SEOBNRv4P} and \texttt{IMRPhenomPv3}, which contain only the dominant quadrupole emission mode, alongside \texttt{IMRPhenomPv3HM} and \texttt{NRHybSur3dq8}, which also account for high-order modes. Each dataset element is injected into detector noise corresponding to the third observing run of the LIGO-Virgo-KAGRA (LVK) collaboration. We identify the Temporal Convolutional Network (TCN) architecture as the overall best performer in terms of training and validation losses and absence of overfitting to data. Comparison of results between datasets shows that the choice of waveform approximant for the creation of a dataset conditions the feature extraction ability of a trained network. Hence, care should be taken when building a dataset for the training of neural networks, as certain approximants may result in better network convergence of evaluation metrics. However, this performance does not necessarily translate to data which is more faithful to numerical relativity simulations. We also apply this network on actual signals from LVK runs, finding that its feature-extracting performance can be effective on real data.

gr-qc↗

arXiv2023

Convolutional Neural Networks for the classification of glitches in gravitational-wave data streams

We investigate the use of Convolutional Neural Networks (including the modern ConvNeXt network family) to classify transient noise signals (i.e.~glitches) and gravitational waves in data from the Advanced LIGO detectors. First, we use models with a supervised learning approach, both trained from scratch using the Gravity Spy dataset and employing transfer learning by fine-tuning pre-trained models in this dataset. Second, we also explore a self-supervised approach, pre-training models with automatically generated pseudo-labels. Our findings are very close to existing results for the same dataset, reaching values for the F1 score of 97.18% (94.15%) for the best supervised (self-supervised) model. We further test the models using actual gravitational-wave signals from LIGO-Virgo's O3 run. Although trained using data from previous runs (O1 and O2), the models show good performance, in particular when using transfer learning. We find that transfer learning improves the scores without the need for any training on real signals apart from the less than 50 chirp examples from hardware injections present in the Gravity Spy dataset. This motivates the use of transfer learning not only for glitch classification but also for signal classification.

gr-qc↗

arXiv2022

Machine-Learning Love: classifying the equation of state of neutron stars with Transformers

The use of the Audio Spectrogram Transformer (AST) model for gravitational-wave data analysis is investigated. The AST machine-learning model is a convolution-free classifier that captures long-range global dependencies through a purely attention-based mechanism. In this paper a model is applied to a simulated dataset of inspiral gravitational wave signals from binary neutron star coalescences, built from five distinct, cold equations of state (EOS) of nuclear matter. From the analysis of the mass dependence of the tidal deformability parameter for each EOS class it is shown that the AST model achieves a promising performance in correctly classifying the EOS purely from the gravitational wave signals, especially when the component masses of the binary system are in the range $[1,1.5]M_{\odot}$. Furthermore, the generalization ability of the model is investigated by using gravitational-wave signals from a new EOS not used during the training of the model, achieving fairly satisfactory results. Overall, the results, obtained using the simplified setup of noise-free waveforms, show that the AST model, once trained, might allow for the instantaneous inference of the cold nuclear matter EOS directly from the inspiral gravitational-wave signals produced in binary neutron star coalescences.

astro-ph.IM↗

arXiv2022

Identification of Binary Neutron Star Mergers in Gravitational-Wave Data Using YOLO One-Shot Object Detection

We demonstrate the application of the YOLOv5 model, a general purpose convolution-based single-shot object detection model, in the task of detecting binary neutron star (BNS) coalescence events from gravitational-wave data of current generation interferometer detectors. We also present a thorough explanation of the synthetic data generation and preparation tasks based on approximant waveform models used for the model training, validation and testing steps. Using this approach, we achieve mean average precision ($\text{mAP}_{[0.50]}$) values of 0.945 for a single class validation dataset and as high as 0.978 for test datasets. Moreover, the trained model is successful in identifying the GW170817 event in the LIGO H1 detector data. The identification of this event is also possible for the LIGO L1 detector data with an additional pre-processing step, without the need of removing the large glitch in the final stages of the inspiral. The detection of the GW190425 event is less successful, which attests to performance degradation with the signal-to-noise ratio. Our study indicates that the YOLOv5 model is an interesting approach for first-stage detection alarm pipelines and, when integrated in more complex pipelines, for real-time inference of physical source parameters.

astro-ph.IM↗

arXiv2020

Exploring gravitational-wave detection and parameter inference using Deep Learning methods

We explore machine learning methods to detect gravitational waves (GW) from binary black hole (BBH) mergers using deep learning (DL) algorithms. The DL networks are trained with gravitational waveforms obtained from BBH mergers with component masses randomly sampled in the range from 5 to 100 solar masses and luminosity distances from 100 Mpc to, at least, 2000 Mpc. The GW signal waveforms are injected in public data from the O2 run of the Advanced LIGO and Advanced Virgo detectors, in time windows that do not coincide with those of known detected signals. We demonstrate that DL algorithms, trained with GW signal waveforms at distances of 2000 Mpc, still show high accuracy when detecting closer signals, within the ranges considered in our analysis. Moreover, by combining the results of the three-detector network in a unique RGB image, the single detector performance is improved by as much as 70%. Furthermore, we train a regression network to perform parameter inference on BBH spectrogram data and apply this network to the events from the the GWTC-1 and GWTC-2 catalogs. Without significant optimization of our algorithms we obtain results that are mostly consistent with published results by the LIGO-Virgo Collaboration. In particular, our predictions for the chirp mass are compatible (up to 3$σ$) with the official values for 90% of events

gr-qc↗

arXiv2017

Pseudoscalar couplings in $t \bar{t} H$ production at NLO+NLL accuracy

We study the production of a Higgs boson in association to a top-antitop pair at the Large Hadron Collider. We show how precise predictions for the differential distributions with respect to the transverse momentum of the Higgs boson, to the invariant mass of the top-antitop-Higgs system and to the invariant mass of the top-antitop pair can provide useful information on the possible presence of a pseudoscalar component in the coupling of the top quark with the Higgs boson. We evaluate the production of a top-antitop pair and a Higgs boson to next-to-leading order in fixed order perturbation theory and we carry out the resummation of soft emission corrections to next-to-leading-logarithmic accuracy for the LHC operating at a center of mass energy of $13$ TeV. We discuss how the shape of these distributions can be employed experimentally, making a physics case for the kinematic reconstruction of dilepton channels.

hep-ph↗