Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2609.38122

Fast and Flow-rious: Gravitational Wave Background Bayesian Model Comparison using Normalizing Flows and Nested Sampling

David C. Wright·Aaron D. Johnson·Jeffrey S. Hazboun·William G. Lamb
View original source ↗Read source document ↗Export metadata ↓

Abstract

In 2023, pulsar timing array (PTA) collaborations around the world announced evidence for a nanohertz gravitational-wave background (GWB). Beyond increasing the detection significance, the next major goal is to characterize the GWB and identify its source(s). The typical model-comparison metric used in PTA analyses is the Bayes factor (BF). However, BF computation methods used in the PTA literature, such as product-space sampling and thermodynamic integration, are computationally expensive or inefficient if the models compared are dissimilar or non-nested. In this work, we introduce a PTA model-comparison method using a combination of normalizing flows, nested sampling, and a PTA likelihood that is marginalized over all non-GWB parameters. The flows are trained using a newly-developed Python package coppuccino, which learns the data's copula: the dependence structure after transforming to uniform marginals. We validate the method on both analytic targets and simulated PTA data, recovering the analytic evidence within error bars and achieving an area-under-the-curve (AUC) score of 0.996 when treating the BFs as binary classifiers of the injected models. The result is a method to perform Bayesian model comparison of arbitrary GWB spectral models in minutes on consumer hardware.

Explore related subjects

gr-qcastro-ph.HEastro-ph.IM

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

David C. Wright, Aaron D. Johnson, Jeffrey S. Hazboun, William G. Lamb. 2026-09-29. Fast and Flow-rious: Gravitational Wave Background Bayesian Model Comparison using Normalizing Flows and Nested Sampling. https://arxiv.org/abs/2609.38122

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

The Depletion of Collisionless Dark Matter Spikes

Dense concentrations of dark matter surrounding black holes provide a compelling opportunity to probe the nature of dark matter. In the classic Gondolo--Silk model, the adiabatic growth of a massive black hole in a dark matter cusp produces a steep density spike ($ρ\propto r^{-7/3}$), potentially inducing measurable gravitational-wave dephasings in intermediate and extreme mass-ratio inspirals (IMRIs/EMRIs). We challenge this paradigm by considering a collisionless dark matter spike embedded in a realistic nuclear star cluster. Using Fokker--Planck models of isotropic nuclear clusters, we show that mass segregation in a multi-mass stellar cusp accelerates relaxation relative to single-mass models, thereby driving the dark matter to the lower density $r^{-3/2}$ Bahcall--Wolf profile within 1 Gyr. In the inner regions, where the Fokker--Planck description breaks down, we model strong triple interactions between dark matter particles and EMRIs using post-Newtonian 3-body simulations. We show that EMRIs eject dark matter particles via gravitational slingshots, depleting the inner spike over a few Gyr. Because EMRI number densities are too low to drive two-body relaxation, and replenishment by DM self-relaxation is negligible, this depletion is irreversible. While the extent of EMRI-induced DM depletion depends on the EMRI rate and mass, we find reductions in densities by several orders of magnitude. As a result, the dark-matter-induced dephasings for EMRIs may fall below the LISA detectability threshold for massive black holes at $z = 3$ (2.14 Gyr) with masses $\lesssim 10^{5}\,M_\odot$ (for a low $\mathcal{O}(10) \, \mathrm{Gyr}^{-1}$ EMRI rate), extending to $\lesssim 10^6\,M_\odot$ for more realistic rates of $\mathcal{O}(100 - 300)\,$Gyr$^{-1}$. Our findings substantially reduce the parameter space over which massive black holes can host detectable collisionless dark matter spikes.

gr-qc↗
arXiv2026

Unruh-DeWitt Detector Response in Toroidal Spacetime

The global topology of spacetime, though invisible to local curvature measurements, leaves signatures on the correlation functions of quantum fields. We study these signatures using an Unruh-DeWitt particle detector operating in four-dimensional Minkowski spacetime with two spatial directions periodically identified, yielding a spatial topology $\mathbb{R}\times T^2$. We compute detector transition rates for three trajectories: uniform inertial motion, uniform proper acceleration directed along one of the compact axes, and uniform proper acceleration along the non-compact axis. Our results show how a local quantum measurement can reveal features of the large-scale spatial topology.

gr-qc↗
arXiv2026

Holographic Dark Energy with Hubble Radius as an Infrared Cutoff in Einstein-Cartan Gravity

In this work, we investigate non-interacting holographic dark energy (HDE) with the Hubble radius as the infrared cutoff in Einstein-Cartan gravity. We derive the Einstein-Cartan equations from the action principle and obtain Friedmann-like equations by introducing a torsion scalar. Considering a Weyssenhoff spin fluid, we determine the scaling behavior of the torsion scalar as $Φ\sim a^{-3}$ without introducing an ad hoc ansatz, resolving the ansatz problem of previous torsion scalar scenarios. In the absence of interactions between dark matter and dark energy, the torsion scalar shifts the equation of state for holographic dark energy toward negative values from the dust-like value obtained in HDE without torsion, making cosmic acceleration possible. In particular, the resulting equation of state can approach $ω_X \simeq -1$ and cross the phantom divide within the weak torsion regime $|Φ/H| < 1$. The model predicts a dynamical equation of state in which cosmic acceleration gradually weakens, potentially consistent with recent DESI observations. In spacetimes with torsion, the cosmic distance duality relation between the luminosity distance $d_L$ and the angular diameter distance $d_A$ is modified as $d_L = d_A (1+z)^2 (1+η)$. In the presence of the torsion scalar, we show that the standard relation between redshift and the scale factor is preserved, while the deviation parameter arising from torsion effects is determined as $η\sim \int_{t_S}^{t_O} dt a^{-3}$, where $t_S$ and $t_O$ denote the emission time at the source and the observation time at the observer, respectively. Overall, our results support the feasibility of the model and provide a theoretical framework for preparing likelihood analyses.

gr-qc↗