Like vs like: Strategy and improvements in supernova cosmology systematics

Abstract

Control of systematic uncertainties in the use of type Ia supernovae as standardized distance indicators can be achieved through contrasting subsets of observationally characterized, like supernovae. Essentially, like supernovae at different redshifts reveal the cosmology, and differing supernovae at the same redshift reveal systematics, including evolution not already corrected for by the standardization. Here we examine the strategy for use of empirically defined subsets to minimize the cosmological parameter risk, the quadratic sum of the parameter uncertainty and systematic bias. We investigate the optimal recognition of subsets within the sample and discuss some issues of observational requirements on accurately measuring subset properties. Neglecting like vs like comparison (i.e. creating only a single Hubble diagram) can cause cosmological constraints on dark energy to be biased by 1{sigma} or degraded by a factor of 1.6 for a total drift of 0.02 mag. Recognition of subsets at the 0.016 mag level (relative differences) erases bias and reduces the degradation to 2%.