Gravitational-wave constraints on causal nonlocal kernels: Ringdown bounds and spectral density limits from GWTC-3 Preprint: UBA-GW-2026-01
Christian Balfagón · 2026
Read Gravitational-wave constraints on causal nonlocal kernels: Ringdown bounds and spectral density limits from GWTC-3 Preprint: UBA-GW-2026-01 by Christian Balfagón (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-16026.txt.
Introduction
Balfagon bounds exotic gravity tweaks with black hole ringdown data. Causal nonlocal gravity models carry kernels with positive spectral weight. Two wave channels test such models against GWTC-3 data. A Bayesian ringdown study covers 17 binary black hole events. Universal mode shifts must stay under 5 percent at 90 percent trust. Dispersion bounds plus GW170817 speed limits map the kernel space. Broad infrared-heavy spectral forms fall to the combined cuts. Tabletop sub-millimetre tests may probe the causal scale sooner.
Catalog record
| CREATOR | Christian Balfagón |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 29459 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.16026
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Gravitational-wave constraints on causal nonlocal kernels: Ringdown bounds and spectral density limits from GWTC-3 Preprint: UBA-GW-2026-01 Christian Balfagón Abstract We establish the first observational bounds on causal nonlocal extensions of gravity characterised by retarded Stieltjes-type kernels with positive spectral density ρ ( μ ) ≥ 0 \rho(\mu)\geq 0 , using two complementary gravitational-wave channels. From a Bayesian ringdown analysis of 17 binary black hole events in the LIGO–Virgo GWTC-3 catalogue, we set the observational ceiling on universal fractional quasi-normal mode deformations at | ε Ω | < 0.05 |\varepsilon_{\Omega}|<0.05 (90% C.L.), with a cumulative log Bayes factor ln B = − 0.46 ± 0.77 \ln B=-0.46\pm 0.77 . By mapping published GWTC-3 modified dispersion relation bounds and the GW170817 propagation speed measurement onto the Stieltjes spectral parameter space ( μ char , M 0 ) (\mu_{\rm char},M_{0}) , we exclude a broad class of infrared-extended spectral densities with μ ≲ 10 − 6 m − 2 \mu\lesssim 10^{-6}\,{\rm m}^{-2} , thereby ruling out non-trivial regions of the nonlocal kernel parameter space for the first time. The theoretically motivated fiducial range μ char ∼ M ∗ 2 ∼ 10 8 \mu_{\rm char}\sim M_{*}^{2}\sim 10^{8} – 10 10 m − 2 10^{10}\,{\rm m}^{-2} satisfies all current bounds, and we show that sub-millimetre gravity experiments — which already operate at the predicted causal scale ℓ ∗ ∼ 10 − 4 \ell_{*}\sim 10^{-4} m — offer the most promising path to a direct test. These results define quantitative benchmarks against which future observations across the gravitational-wave, short-range, and cosmological sectors can be compared. I Introduction Nonlocal extensions of general relativity (GR) have been proposed as a route to ultraviolet completion, singularity resolution, and causal–informational closure of the gravitational sector 40 ; 23 ; 37 ; 27 ; 20 . A mathematically tractable class of such extensions employs retarded kernels constructed as positive Stieltjes superpositions of massive Klein–Gordon propagators 15 ; 17 , \mathcal{K}^{-1}=\int_{0}^{\infty}\rho(\mu)\,(-\Box_{g}+\mu)^{-1}_{R}\,d\mu,\qquad\rho(\mu)\geq 0. (1) Positivity of the spectral density \rho(\mu) ensures unitarity, causal propagation, and the absence of ghost-like excitations at the linear level 18 ; 24 ; 35 . Depending on the support and shape of \rho , such kernels predict two classes of gravitational-wave (GW) observables: (i) perturbative shifts to the quasi-normal mode (QNM) spectrum of black holes 34 ; 21 ; 29 , scaling as |\delta\omega/\omega|\sim(\ell_{*}/r_{H})^{2} where \ell_{*}=M_{*}^{-1} is the causal correlation length; and (ii) frequency-dependent dispersion during cosmological propagation, governed by the Stieltjes transfer function m(\omega^{2}) . In this work we constrain both channels using data from the LIGO–Virgo–KAGRA (LVK) third observing run (GWTC-3) 9 ; 3 . Section II presents a model-agnostic Bayesian ringdown analysis of 17 binary black hole (BBH) events, constraining the universal deformation parameter \varepsilon_{\Omega} . Section III maps published LVK modified dispersion relation (MDR) bounds and the GW170817 speed constraint onto the Stieltjes spectral parameter space. Section IV discusses sensitivity projections for future detectors and identifies short-range gravity experiments as the most promising near-term probe. Section V summarises the results. Notation. —We denote the causal nonlocal scale by M_{*} (units of {\rm m}^{-1} ) and the corresponding correlation length by \ell_{*}\equiv M_{*}^{-1} (units of m). The spectral density \rho(\mu) is a function of the spectral mass parameter \mu (units of {\rm m}^{-2} ), with total weight M_{0}=\int_{0}^{\infty}\rho(\mu)\,d\mu and inverse moments \mu_{\rm eff}^{-n}=M_{0}^{-1}\int\rho\,\mu^{-n}\,d\mu . For a spectral density concentrated around a single scale we write \mu_{\rm char} ; in this limit \mu_{\rm eff}\approx\mu_{\rm char}\approx M_{*}^{2} . The fiducial parameter window is M_{*}\in[10^{4},10^{5}]\,{\rm m}^{-1} , equivalently \ell_{*}\in[10^{-5},10^{-4}] m and \mu_{\rm char}\sim M_{*}^{2}\in[10^{8},10^{10}]\,{\rm m}^{-2} . II Ringdown analysis II.1 Method We model the ringdown signal in each detector as a damped sinusoid with QNM frequency and damping time computed from the remnant mass M_{f} and spin a using the Berti et al. phenomenological fits 22 ; 21 , converted to SI units. A single universal parameter \varepsilon_{\Omega} deforms both the frequency and damping time by the common factor (1+\varepsilon_{\Omega}) , preserving the quality factor. The Kerr hypothesis corresponds to \varepsilon_{\Omega}=0 . The per-detector amplitude is analytically marginalised under a Gaussian prior A_{i}\sim\mathcal{N}(0,\sigma_{A}^{2}) , yielding a marginalised log-likelihood free of the biases inherent in profile-likelihood methods 46 . The noise variance \sigma^{2} is estimated from the data using a robust median-absolute-deviation estimator. Templates are whitened in the frequency domain using the same PSD applied to the data, ensuring spectral consistency. Nested sampling 44 is performed with dynesty 45 via Bilby 13 , with n_{\rm live}=1000 and \Delta\ln Z_{\rm tol}=0.1 . Our single-mode template does not capture subdominant overtones or higher multipoles; however, within the Stieltjes framework the deformation is universal across the QNM spectrum, so unmodelled modes do not bias the Bayes factor comparison 32 ; 26 ; 31 . II.2 Results Table 1 lists the individual-event results for 17 GWTC-3 BBH events 9 ; 6 ; 7 spanning remnant masses 25 – 150\,M_{\odot} . All log Bayes factors are consistent with zero within 2\sigma , with both positive and negative values represented. The most precise measurement is GW190412 ( \ln B=+0.000\pm 0.020 ). Table 1: Individual ringdown results for 17 GWTC-3 BBH events. Event M_{f}\,[M_{\odot}] \ln B \sigma_{\ln B} GW150914 68 +0.18 0.18 GW170104 50 -0.06 0.17 GW170814 53 -0.04 0.19 GW190412 38 +0.00 0.02 GW190521 150 -0.35 0.19 GW190814 26 -0.11 0.20 GW191105 35 +0.02 0.21 GW191109 79 -0.09 0.19 GW191113 30 -0.13 0.18 GW191222 42 +0.07 0.19 GW200129 38 +0.15 0.19 GW200208 39 -0.09 0.18 GW200219 39 +0.14 0.20 GW200220 40 -0.21 0.18 GW200224 43 +0.35 0.21 GW200225 35 -0.25 0.19 GW200311 38 -0.03 0.20 Stacked — \mathbf{-0.46} \mathbf{0.77} The cumulative log Bayes factor, computed under the assumption of statistically independent events with a shared deformation parameter, is \begin{split}\ln B_{\rm stack}&=\sum_{k=1}^{N}\ln B_{k}=-0.46,\\ \sigma_{\rm stack}&=\sqrt{\sum_{k=1}^{N}\sigma_{\ln B,k}^{2}}=0.77.\end{split} (2) where N=17 . This stacking is exact when: (i) the events are independent realisations of noise plus signal, which holds for well-separated BBH mergers observed in disjoint data segments; and (ii) the deformation parameter \varepsilon_{\Omega} is universal (identical across events), as predicted by the single-scale Stieltjes framework. A leave-one-out (jackknife) analysis confirms stability (Fig. 4 ): the largest single-event contribution to the stacked Bayes factor is |\Delta\ln B|=0.35 (GW190521 and GW200224), and removing any individual event shifts the cumulative result by less than 0.5\sigma_{\rm stack} . We verified prior robustness by repeating the analysis with a narrower prior \varepsilon_{\Omega}\in[-0.02,0.02] ; the resulting Bayes factors shift by less than 0.05 per event, confirming that the results are not driven by prior volume effects. At 90% credibility, the population-level posterior on \varepsilon_{\Omega} obtained by multiplying the individual-event posteriors yields the constraint (Fig. 3 ) \varepsilon_{\Omega}\in[-0.047,\;+0.029]\quad(90\%\;{\rm C.L.}), (3) i.e. |\varepsilon_{\Omega}|<0.05 . This establishes a quantitative observational ceiling on universal fractional QNM deformations at the percent level, applicable to any modified-gravity theory predicting a single-parameter shift of the (2,2,0) mode 38 ; 25 . For comparison, the LVK GWTC-3 analysis 10 parametrises QNM deviations via independent fractional shifts \delta f_{220} and \delta\tau_{220} to the frequency and damping time, obtaining combined 90% bounds of \delta f_{220}\in[-0.05,+0.05] and \delta\tau_{220}\in[-0.2,+0.1] from hierarchical combination of \sim 10 events 30 . Our parametrisation differs in that \varepsilon_{\Omega} deforms both quantities by a common factor: \delta f_{220}=\delta\tau_{220}=\varepsilon_{\Omega} , preserving the quality factor Q_{220}=\pi f_{220}\tau_{220} .