Global Properties of M31's Stellar Halo from the SPLASH Survey: III. Measuring the Stellar Velocity Dispersion Profile

TL;DR

This study measures the velocity dispersion profile of M31's stellar halo using over 5000 stars, revealing a mild decrease with radius, and employs a comprehensive Gaussian Mixture Model to account for structural components and debris features.

Contribution

It provides the first detailed velocity dispersion profile of M31's halo across a wide radial range using an advanced mixture modeling approach.

Findings

Velocity dispersion decreases mildly from 108 km/s to 80-90 km/s with radius.

The dispersion profile follows a power-law with a slope of -0.12±0.05.

The model incorporates all major structural components and debris features.

Abstract

We present the velocity dispersion of red giant branch (RGB) stars in M31's halo, derived by modeling the line of sight velocity distribution of over 5000 stars in 50 fields spread throughout M31's stellar halo. The dataset was obtained as part of the SPLASH (Spectroscopic and Photometric Landscape of Andromeda's Stellar Halo) Survey, and covers projected radii of 9 to 175 kpc from M31's center. All major structural components along the line of sight in both the Milky Way (MW) and M31 are incorporated in a Gaussian Mixture Model, including all previously identified M31 tidal debris features in the observed fields. The probability an individual star is a constituent of M31 or the MW, based on a set of empirical photometric and spectroscopic diagnostics, is included as a prior probability in the mixture model. The velocity dispersion of stars in M31's halo is found to decrease only mildly with projected radius, from 108 km/s in the innermost radial bin (8.2 to 14.1 kpc) to $\sim 80$ to 90 km/s at projected radii of $\sim 40$ to 130 kpc, and can be parameterized with a power-law of slope $-0.12\pm 0.05$. The quoted uncertainty on the power-law slope reflects only the precision of the method, although other sources of uncertainty we consider contribute negligibly to the overall error budget.