GRB Hubble Diagram Calibrated on TDCOSMO Lenses Returns H0 of 78.3 km/s/Mpc but Inherits Lens Distance Scale

The study trains an artificial neural network on TDCOSMO time-delay distances to reconstruct the luminosity-distance relation, then restricts GRB calibration to the overlapping redshift interval to eliminate extrapolation. Hierarchical Bayesian fitting of the calibrated Amati and Combo relations follows, with injection-recovery tests exposing upward bias in H0 and Omega_m when an unmodelled positive redshift trend is present in the Combo relation. This approach directly addresses the circularity problem that has historically limited GRB cosmology.

Results remain tied to the TDCOSMO absolute distance scale rather than providing an independent GRB-derived H0; the reported uncertainties are dominated by intrinsic GRB scatter, possible redshift-dependent evolution, and residual lens-model systematics. The higher central H0 value aligns numerically with local measurements yet cannot be read as independent confirmation because the calibration ladder originates in the same strong-lensing data.

The analysis connects to ongoing Hubble-tension efforts by demonstrating that strong-lens calibrators can extend the distance ladder beyond Type Ia supernovae while quantifying how unmodelled systematics propagate into cosmological parameters. Future LSST and Roman Space Telescope lenses will shrink the lens-modelling contribution, but GRB scatter will still require sample sizes several times larger than current catalogs.

Next steps include joint analyses with existing BAO and CMB datasets to test whether the apparent redshift trend in Combo persists and to assess consistency across w0waCDM extensions.