Gamma-ray Bursts as Cosmological Probes Calibrated with Strongly Lensed Quasars and Neural Networks

Abstract

Gamma-ray bursts (GRBs) extend distance measurements to redshifts beyond those routinely probed by Type Ia supernovae, but their use in cosmology is limited by calibration circularity and possible redshift-dependent effects. We investigate strongly lensed quasars as external calibrators of the Amati and Combo GRB relations. We reconstruct the luminosity-distance relation using an artificial neural network trained on TDCOSMO strong-lensing distances, with the absolute scale inherited from the lensing-only calibration. To avoid extrapolation, only GRBs within the lens-supported redshift interval are used for calibration. We then infer cosmological parameters through a hierarchical Bayesian analysis of flat $Λ$CDM, with $w_0w_a$CDM examined as a consistency test. For flat $Λ$CDM, Amati yields $H_0=78.26^{+17.87}_{-16.30}\,\mathrm{km\,s^{-1}\,Mpc^{-1}}$ and $Ω_m=0.670^{+0.223}_{-0.276}$, while Combo yields $H_0=78.71^{+13.83}_{-11.29}\,\mathrm{km\,s^{-1}\,Mpc^{-1}}$ and $Ω_m=0.635^{+0.235}_{-0.242}$. Amati is relatively insensitive to the phenomenological redshift term, whereas Combo shows a stronger dependence. Injection--recovery tests indicate that an unmodelled positive Combo trend can bias $H_0$ and $Ω_m$ upward. Strong-lens calibration therefore offers a route to extending the GRB Hubble diagram, but the inferred $H_0$ is not an independent GRB measurement: it retains the TDCOSMO distance scale. The precision is limited by GRB scatter, redshift-dependent systematics, and lens-modelling uncertainties.