Young stellar populations in early-type galaxies in the Sloan Digital Sky Survey

Abstract

We use a purely data‐driven rectified factor analysis to identify early‐type galaxies with recent star formation in Data Release 4 of the Sloan Digital Sky Survey Spectroscopic Catalogue. We compare the spectra and environment of these galaxies with those of ‘normal’ early‐type galaxies, and a sample of independently selected E+A galaxies. We calculate the projected local galaxy surface density from the nearest five and 10 neighbours (Σ5 and Σ10) for each galaxy in our sample, and find that the dependence on projected local density, of the properties of E+A galaxies, is not significantly different from that of early‐type galaxies with young stellar populations, dropping off rapidly towards denser environments, and flattening off at densities ≲0.1–0.3 Mpc−2. The dearth of E+A galaxies in dense environments confirms that E+A galaxies are most likely the products of galaxy–galaxy merging or interactions, rather than star‐forming galaxies whose star formation has been quenched by processes unique to dense environments, such as ram‐pressure stripping or galaxy harassment. We see a tentative peak in the number of E+A galaxies at Σ10∼ 0.1–0.3 Mpc−2, which may represent the local galaxy density at which the rate of galaxy–galaxy merging or interaction rate peaks. Analysis of the spectra of our early‐type galaxies with young stellar populations suggests that they have a stellar component dominated by F stars, ∼1–4 Gyr old, together with a mature, metal‐rich population characteristic of ‘typical’ early‐type galaxies. The young stars represent ≳10 per cent of the stellar mass in these galaxies. This, together with the similarity of the environments in which this ‘E+F’ population and the E+A galaxy sample are found, suggests that E+F galaxies used to be E+A galaxies, but have evolved by a further ∼ one to a few Gyr. Our rectified factor analysis is sensitive enough to identify this hidden population, which allows us to study the global and intrinsic properties of early‐type galaxies created in major mergers or interactions, and compare them with those early‐types which have had the bulk of their stars in place since a much earlier epoch.

Computational Intelligence in Astronomy – A Win-Win Situation

“…We run a rolling window of size 3 over the series f, g, h), starting with simulation sets (f, g, h) at stages (0,1,2) and ending with the triplet (f, g, h) containing simulation set for stages (20,21,22). For each triplet of consecutive simulation sets (f, g, h), we estimated 3 models, one on 90% of data from f , one on 90% of data from g, and one on 90% of data from h. We also compared with Gaussian Mixture Model initialized randomly from the data (GMM) or initialized with K-means (GMM k ).…”

“…As an illustration, we present in figure 2 two sets of 3-dimensional projections of the simulation data for the triplet (f, g, h) = (20,21,22). The first projection is onto the spatial coordinates, the second is onto the leading 3 eigenvectors found by the Principal Component Analysis of the merged f, g and h sets.…”

“…Artificial neural networks [19,1] or support vector machines [36] have been popular in studying the nature of galaxy evolution from photometric images. In recent years, with the availability of optical spectra of > 10 6 galaxies, the study of individual stellar populations within galaxies have been helped by incorporating independent component analysis and other data-driven techniques [21,22].…”

Young stellar populations in early-type galaxies in the SDSS

“…Full details of the selection can be found in Nolan, Raychaudhury & Kabán (2007), and Nolan, Benson & Raychaudhury (2007, in preparation).…”

“…In Nolan et al (2007), we find that the spectra of ETGs is well-described by two underlying components, so we set K = 2. For a complete description of the rectified factor analysis, see Nolan et al (2007). Figure 1 shows the two components recovered from the analysis of the SDSS ETG spectra.…”