2015MNRAS.451.2390H

Type Ia supernovae (SNe Ia) are used as `standard candles' for cosmological distance scales. To fit their light-curve shape-absolute luminosity relation, one needs to assume an intrinsic colour and a likelihood of host galaxy extinction or a convolution of these, a colour distribution prior. The host galaxy extinction prior is typically assumed to be an exponential drop-off for the current supernova programmes (P(A_V) ∝e^-A_V/τ_0). We explore the validity of this prior using the distribution of extinction values inferred when two galaxies accidentally overlap (an occulting galaxy pair). We correct the supernova luminosity distances from the SDSS-III supernova projects (SDSS-SN) by matching the host galaxies to one of three templates from occulting galaxy pairs based on the host galaxy mass and the AV-bias-prior-scale (τ0) relation from Jha et al. We find that introducing an AV prior that depends on host mass results in lowered luminosity distances for the SDSS-SN on average but it does not reduce the scatter in individual measurements. This points, in our view, to the need for many more occulting galaxy templates to match to SN Ia host galaxies to rule out this possible source of scatter in the SN Ia distance measurements. We match occulting galaxy templates based on both mass and projected radius and we find that one should match by stellar mass first with radius as a secondary consideration. We discuss the caveats of the current approach: the lack of enough radial coverage, the small sample of priors (occulting pairs with HST data), the effect of gravitationally interacting as well as occulting pairs, and whether an exponential distribution is appropriate. Our aim is to convince the reader that a library of occulting galaxy pairs observed with HST will provide sufficient priors to improve (optical) SN Ia measurements to the next required accuracy in cosmology.