SN 2021hpr and its two siblings in the Cepheid calibrator galaxy NGC 3147: A hierarchical BayeSN analysis of a Type Ia supernova trio, and a Hubble constant constraint

To improve Type Ia supernova (SN Ia) standardisability, the consistency of distance estimates to siblings -- SNe in the same host galaxy -- should be investigated. We present Young Supernova Experiment Pan-STARRS-1 $grizy$ photometry of SN 2021hpr, the third spectroscopically confirmed SN Ia in the high-stellar-mass Cepheid-calibrator galaxy NGC 3147. We analyse NGC 3147's trio of SN Ia siblings: SNe 1997bq, 2008fv and 2021hpr, using a new version of the BayeSN model of SN Ia spectral-energy distributions, retrained simultaneously using optical-NIR $BgVrizYJH$ (0.35--1.8 $\mu$m) data. The distance estimates to each sibling are consistent, with a sample standard deviation $\lesssim$0.01 mag, much smaller than the total intrinsic scatter in the training sample: $\sigma_0\approx0.09$ mag. Fitting normal SN Ia siblings in three additional galaxies, we estimate a $\approx$90% probability that the siblings' intrinsic scatter is smaller than $\sigma_0$. We build a new hierarchical model that fits light curves of siblings in a single galaxy simultaneously; this yields more precise estimates of the common distance and the dust parameters. Fitting the trio for a common dust law shape yields $R_V=2.69\pm0.52$. Our work motivates future hierarchical modelling of more siblings, to tightly constrain their intrinsic scatter, and better understand SN-host correlations. Finally, we estimate the Hubble constant, using a Cepheid distance to NGC 3147, the siblings trio, and 109 Hubble flow ($0.01 < z_{\rm{CMB}} < 0.08$) SNe Ia; marginalising over the siblings' and population's intrinsic scatters, and the peculiar velocity dispersion, yields $H_0=77.9\pm6.5 \text{ km s}^{-1}\text{Mpc}^{-1}$.