The lives and deaths of faint satellite galaxies around M31

Abstract

We present predictions for proper motions, infall times and times of first pericentric passage for 39 of M31's satellite galaxies. We estimate these by sampling satellite orbits from cosmological N-body simulations matched on mass, distance and velocity along the line of sight, in addition to properties of the host system. Our predictions are probabilistic based on repeated sampling from the uncertainty distributions of all quantities involved. We use these constraints on the satellites' orbital histories in conjunction with their published star formation histories to investigate the dominant environmental mechanisms for quenching satellites of M31-like hosts. Around half of the satellites appear to have quenched before their first pericentric passage around M31. Only the most massive satellites (with stellar masses > 10^8 M_sun) are able to maintain star formation for up to billions of years after infall. The majority of faint satellites, with stellar masses < 10^8 M_sun , were likely quenched before entering the M31 system. We compare our results for M31 against predictions for the Milky Way's satellites from the literature; M31's has a more active recent accretion history with more recently quenched satellites than the Milky Way.

Figures (6)

• Figure 1: The dependence of first pericentre time on our chosen parameter space within the simulation. 3D radial distance from the host is on the horizontal axis (normalized by $r_\mathrm{vir}$), line of sight velocity relative to the host is placed on the vertical axis (normalized by $\sigma_\mathrm{3D}$), and satellite stellar mass is indicated by the 3 panels of low, middling, and high masses, as labelled. The space is coloured by the median lookback time to the first pericentre passage, ranging from ancient times in white/yellow to future times (negative values) in black/purple. We show contours of the number of simulated satellites per bin at 3, 30, and 300. The white points show the parameters (most probable values) for each observed satellite of M31.
• Figure 2: Illustration of the method for predicting orbital parameters for examples Andromeda XVIII (top) and Andromeda XIX (bottom). The left panels show the dependence in the simulation of pericentre times on phase space as in Fig. \ref{['fig:allsim']} but restricted to halo masses within a factor of 2 of the observed satellite. The middle panels show the probability distributions of the observed distance and line of sight velocity of the satellites, with a logarithmic colour scale. These provide a weighting to select the simulated haloes shown in the right panels. In these panels, we show comoving distance from the host for 100 randomly selected (with weighting) simulated satellites, with the lower histograms depicting the resultant spread of first pericentre times. The green and blue paths correspond to the later and earlier peaks in the histogram, respectively. For reference, the black dashed line in this panel shows the median virial radius of the hosts at each timestep.
• Figure 3: The time we predict to be taken from the first crossing of M31's virial radius to the first pericentric passage for each observed satellite, plotted against the time of its first pericentric passage. This interval is calculated as a probability distribution in its own right, not from our median times for the two events. The black dotted line shows the evolution of the host haloes' median dynamical timescale. The Triangulum galaxy lies outside of the plot area near $(-1,4)\,\mathrm{Gyr}$.
• Figure 4: From top to bottom, the blue points show the medians of lookback time to first pericentre passage, lookback time to quenching, and their difference (the quenching timescale), each plotted against the satellites' stellar mass. Galaxies that have not yet quenched (as determined by a morphological classification of dwarf irregular or spiral) are shown by triangles. In the quenching time panel, error bars are used to indicate the $16^\mathrm{th}$ to $84^\mathrm{th}$ percentiles; the top and bottom panels uncertainty is visualised as a summed probability density, with darker regions having a higher probability of hosting an M31 satellite. This is obtained from the values' probability distributions and smoothed using Gaussian kernels across a scale of $0.1\,\mathrm{dex}$ for the logarithmic stellar mass and $0.5\,\mathrm{Gyr}$ for the lookback time ($\sim2$ per cent of each axis). The dotted line in the bottom panel indicates simultaneous pericentre and quenching times.
• Figure 5: The proper motions in the plane of the sky relative to M31 for the 9 satellites with observed proper motions available. Filled points show these observed values, while open points show our predictions from Table \ref{['tab:predict']}.