Rotation curve decompositions with Gaussian Processes: taking into account data correlations leads to unbiased results

TL;DR

This paper demonstrates that accounting for data correlations in galaxy rotation curves using Gaussian Processes is essential for unbiased estimates of dark and luminous matter distributions.

Contribution

It introduces a method to incorporate data correlations via Gaussian Processes in rotation curve analysis, improving the accuracy of mass distribution estimates.

Findings

Ignoring data correlations leads to biased results.

Using Gaussian Processes reduces bias in mass estimates.

Accounting for correlations is critical for reliable galaxy modeling.

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.