Rotation Curve Decompositions with Gaussian Processes: Taking into Account Data Correlations Leads to Unbiased Results

Abstract

Correlations between velocity measurements in disk galaxy rotation curves are usually neglected when fitting dynamical models. Here I show how data correlations can be taken into account in rotation curve decompositions using Gaussian Processes. I find that marginalizing over correlation parameters proves critical to obtain unbiased estimates of the luminous and dark matter distributions in galaxies.