No evidence for bulk velocity from type Ia supernovae

We revisit the effect of peculiar velocities on low-redshift type Ia supernovae. Velocities introduce an additional guaranteed source of correlations between supernova magnitudes that should be considered in all analyses of nearby supernova samples but has largely been neglected in the past. Applying a likelihood analysis to the latest compilation of nearby supernovae, we find no evidence for the presence of these correlations, although, given the significant noise, the data is also consistent with the correlations predicted for the standard $\Lambda$CDM model. We then consider the dipolar component of the velocity correlations -- the frequently studied "bulk velocity" -- and explicitly demonstrate that including the velocity correlations in the data covariance matrix is crucial for drawing correct and unambiguous conclusions about the bulk flow. In particular, current supernova data is consistent with no excess bulk flow on top of what is expected in $\Lambda$CDM and effectively captured by the covariance. We further clarify the nature of the apparent bulk flow that is inferred when the velocity covariance is ignored. We show that a significant fraction of this quantity is expected to be noise bias due to uncertainties in supernova magnitudes and not any physical peculiar motion.