“…In this work, we present the first χEFT results for the energy-dependent np ↔ dγ cross section for a range of energies from threshold to MeV scale, that encompasses the BBN regime. We extend the Bayesian procedure for estimating the χEFT truncation error, which was earlier applied to various other observablesnucleon-mass expansion [2,5], nucleon-nucleon elastic scattering [4,7,8,13,19], nucleon-deuteron scattering [16,18], nuclear-matter equation of state [42,43], pion-photoproduction on the nucleon [21,25], and properties of light nuclei [20,22]-to an electromagnetic reaction cross section. We adopt the error model developed by Melendez et al [13] that used machine learning with Gaussian processes (GPs) calibrated by physics-based hyperparameters to determine the convergence of EFT predictions that may be correlated across independent variables (e.g., energy).…”