arXiv subjects

Gabriele Demasi

Publications and source records attributed to Gabriele Demasi.

4 recordsLinked to original sources

arXiv2026

The Sequential Monte Carlo goes NUTS: Boosting Gravitational-Wave Inference

Sequential Monte Carlo (SMC) methods have recently been applied to gravitational-wave inference as a powerful alternative to standard sampling techniques, such as Nested Sampling. At the same time, gradient-based Markov Chain Monte Carlo algorithms, most notably the No-U-Turn Sampler (NUTS), provide an efficient way to explore high-dimensional parameter spaces. In this work we present SHARPy, a Bayesian inference framework that combines the parallelism and evidence-estimation capabilities of SMC with the state-of-the-art sampling performance of NUTS. Moreover, SHARPy exploits the local geometric structure of the posterior to further improve efficiency. Built on JAX, a high-performance computing framework that enables automatic differentiation and hardware acceleration, SHARPy performs gravitational-wave inference on binary black-hole events in around ten minutes, yielding posterior samples and Bayesian evidence estimates that are consistent with those obtained through Nested Sampling. This work sets a new milestone in Gravitational-Wave inference with likelihood-based methods and paves the way for model comparison tasks to be accomplished in minutes.

gr-qc↗

arXiv2025

Exploration of features in the black hole mass spectrum inspired by non-parametric analyses of gravitational wave observations

Current gravitational-wave data reveal structures in the mass function of binary compact objects. Properly modelling and deciphering such structures is the ultimate goal of gravitational-wave population analysis: in this context, non-parametric models are a powerful tool to infer the distribution of black holes from gravitational waves without committing to any specific functional form. Here, we aim to quantitatively corroborate the findings of non-parametric methods with parametrised models incorporating the features found in such analyses. We propose two modifications of the currently favoured PowerLaw+Peak model, inspired by non-parametric studies, and use them to analyse the third Gravitational Wave Transient Catalogue. Our analysis marginally supports the existence of two distinct, differently redshift-evolving subpopulations in the black hole primary mass function, and suggests that, to date, we are still unable to robustly assess the shape of the mass ratio distribution for symmetric ($q>0.7$) binaries.

astro-ph.HE↗

arXiv2024

Enhancing gravitational-wave host localization with SKYFAST: rapid volume and inclination angle reconstruction

The scientific impact of GW170817 strongly supports the need for an efficient electromagnetic follow-up campaign to gravitational-wave event candidates. The success of such campaigns depends critically on a fast and accurate localization of the source. In this paper, we present SKYFAST, a new pipeline for rapid localization of gravitational-wave event hosts. SKYFAST runs alongside a full parameter estimation (PE) algorithm, from which posterior samples are taken. It uses these samples to reconstruct an analytical posterior for the sky position, luminosity distance, and inclination angle using a Dirichlet Process Gaussian Mixture Model, a Bayesian non-parametric method. This approach allows us to provide an accurate localization of the event using only a fraction of the total samples produced by the full PE analysis. Depending on the PE algorithm employed, this can lead to significant time savings, which is crucial for identifying the electromagnetic counterpart. Additionally, in a few minutes, SKYFAST generates a ranked list of the most probable galaxy hosts from a galaxy catalog of choice. This list includes information on the inclination angle posterior conditioned to the position of each candidate host, which is useful for assessing the detectability of gamma-ray burst structured jet emissions.

astro-ph.HE↗

arXiv2024

Hierarchical inference of evidence using posterior samples

The Bayesian evidence, crucial ingredient for model selection, is arguably the most important quantity in Bayesian data analysis: at the same time, however, it is also one of the most difficult to compute. In this paper we present a hierarchical method that leverages on a multivariate normalised approximant for the posterior probability density to infer the evidence for a model in a hierarchical fashion using a set of posterior samples drawn using an arbitrary sampling scheme.

stat.ME↗