BAYES-LOSVD: a bayesian framework for non-parametric extraction of the line-of-sight velocity distribution of galaxies

TL;DR

BAYES-LOSVD is a Bayesian, non-parametric method for extracting galaxy line-of-sight velocity distributions, improving flexibility and robustness over traditional parametric techniques, and is applicable to various IFU data sets.

Contribution

It introduces a Bayesian framework with PCA-based regularization for non-parametric LOSVD extraction, outperforming parametric methods and applicable to multiple galaxy surveys.

Findings

Successfully models diverse LOSVD shapes

Overcomes parametric method limitations

Validated on real galaxy data

Abstract

We introduce BAYES-LOSVD, a novel implementation of the non-parametric extraction of line-of-sight velocity distributions (LOSVDs) in galaxies. We employ bayesian inference to obtain robust LOSVDs and associated uncertainties. Our method relies on principal component analysis to reduce the dimensionality of the base of templates required for the extraction and thus increase the performance of the code. In addition, we implement several options to regularise the output solutions. Our tests, conducted on mock spectra, confirm the ability of our approach to model a wide range of LOSVD shapes, overcoming limitations of the most widely used parametric methods (e.g. Gauss-Hermite expansion). We present examples of LOSVD extractions for real galaxies with known peculiar LOSVD shapes, i.e. NGC4371, IC0719 and NGC4550, using MUSE and SAURON integral-field unit (IFU) data. Our implementation can also handle data from other popular IFU surveys (e.g. ATLAS3D, CALIFA, MaNGA, SAMI). Details of the code and relevant documentation are freely available to the community in the dedicated repositories.