Constraining cosmological parameters from very-high-energy $\gamma$-ray attenuation in blazar spectra: Bayesian inference and systematic uncertainties
Very-high-energy $\gamma$ rays from active galactic nuclei are attenuated through pair production on the extragalactic background light (EBL), providing a probe of cosmic expansion. We develop a Bayesian inference analysis framework that uses an observationally driven, redshift-dependent EBL model and anchors intrinsic source spectra to the Fermi-LAT 3FHL catalogue while marginalising spectral-shape departures and dataset normalisations. The method is validated on realistic mock observations and applied to 241 archival H.E.S.S., MAGIC and VERITAS spectra of 50 AGN compiled in STeVECat. Because the leading opacity depends on the ratio of the EBL normalisation to the reduced Hubble constant, the gamma-ray data primarily constrain this ratio rather than $H_0$ alone. We measure $r=1.723^{+0.096}_{-0.095}$. For a fixed EBL scale, this gives $H_0=58.3^{+3.4}_{-3.0},\mathrm{km,s^{-1},Mpc^{-1}}$; allowing an 18\% EBL-scale uncertainty gives $H_0=60.5^{+10.8}_{-10.0},\mathrm{km,s^{-1},Mpc^{-1}}$. Matched injections recover the input opacity without appreciable bias, and the tested intrinsic-spectrum and EBL-shape perturbations are subdominant except for large infrared-background mismodelling. The predominantly low-redshift sample leaves $\Omega\mathrm{M}$ weakly constrained. The measurement is therefore limited by the EBL normalisation rather than statistics; reducing its uncertainty to a few per cent would make gamma-ray opacity a competitive, distance-ladder-independent probe of $H_0$.