We study the link between observed ultraviolet luminosity, stellar mass, and dust attenuation within rest-frame UV-selected samples at z~ 4, 3, and 1.5. We measure by stacking at 250, 350, and 500 um in the Herschel/SPIRE images from the HerMES progr
am the average infrared luminosity as a function of stellar mass and UV luminosity. We find that dust attenuation is mostly correlated with stellar mass. There is also a secondary dependence with UV luminosity: at a given UV luminosity, dust attenuation increases with stellar mass, while at a given stellar mass it decreases with UV luminosity. We provide new empirical recipes to correct for dust attenuation given the observed UV luminosity and the stellar mass. Our results also enable us to put new constraints on the average relation between star formation rate and stellar mass at z~ 4, 3, and 1.5. The star formation rate-stellar mass relations are well described by power laws (SFR~ M^0.7), with the amplitudes being similar at z~4 and z~3, and decreasing by a factor of 4 at z~1.5 at a given stellar mass. We further investigate the evolution with redshift of the specific star formation rate. Our results are in the upper range of previous measurements, in particular at z~3, and are consistent with a plateau at 3<z<4. Current model predictions (either analytic, semi-analytic or hydrodynamic) are inconsistent with these values, as they yield lower predictions than the observations in the redshift range we explore. We use these results to discuss the star formation histories of galaxies in the framework of the Main Sequence of star-forming galaxies. Our results suggest that galaxies at high redshift (2.5<z<4) stay around 1 Gyr on the Main Sequence. With decreasing redshift, this time increases such that z=1 Main Sequence galaxies with 10^8<M_*/Msun<10^10 stay on the Main Sequence until z=0.
We study the far-infrared (IR) and sub-millimeter properties of a sample of ultraviolet (UV) selected galaxies at zsim1.5. Using stacking at 250, 350 and 500 um from Herschel Space Observatory SPIRE imaging of the COSMOS field obtained within the Her
MES key program, we derive the mean IR luminosity as a function of both UV luminosity and slope of the UV continuum beta. The IR to UV luminosity ratio is roughly constant over most of the UV luminosity range we explore. We also find that the IR to UV luminosity ratio is correlated with beta. We observe a correlation that underestimates the correlation derived from low-redshift starburst galaxies, but is in good agreement with the correlation derived from local normal star-forming galaxies. Using these results we reconstruct the IR luminosity function of our UV-selected sample. This luminosity function recovers the IR luminosity functions measured from IR selected samples at the faintest luminosities (Lir ~ 10^{11} L_sun), but might underestimate them at the bright-end (Lir > 5.10^{11} L_sun). For galaxies with 10^{11}<Lir/L_sun<10^{13}, the IR luminosity function of a UV selection recovers (given the differences in IR-based estimates) 52-65 to 89-112 per cent of the star-formation rate density derived from an IR selection. The cosmic star-formation rate density derived from this IR luminosity function is 61-76 to 100-133 per cent of the density derived from IR selections at the same epoch. Assuming the latest Herschel results and conservative stacking measurements, we use a toy model to fully reproduce the far IR luminosity function from our UV selection at zsim 1.5. This suggests that a sample around 4 magnitudes deeper (i.e. reaching u sim 30 mag) and a large dispersion of the IR to UV luminosity ratio are required.
We investigate the quality of associations of astronomical sources from multi-wavelength observations using simulated detections that are realistic in terms of their astrometric accuracy, small-scale clustering properties and selection functions. We
present a general method to build such mock catalogs for studying associations, and compare the statistics of cross-identifications based on angular separation and Bayesian probability criteria. In particular, we focus on the highly relevant problem of cross-correlating the ultraviolet Galaxy Evolution Explorer (GALEX) and optical Sloan Digital Sky Survey (SDSS) surveys. Using refined simulations of the relevant catalogs, we find that the probability thresholds yield lower contamination of false associations, and are more efficient than angular separation. Our study presents a set of recommended criteria to construct reliable cross-match catalogs between SDSS and GALEX with minimal artifacts.
We measure the projected spatial correlation function w_p(r_p) from a large sample combining GALEX ultraviolet imaging with the SDSS spectroscopic sample. We study the dependence of the clustering strength for samples selected on (NUV - r)_abs color,
specific star formation rate (SSFR), and stellar mass. We find that there is a smooth transition in the clustering of galaxies as a function of this color from weak clustering among blue galaxies to stronger clustering for red galaxies. The clustering of galaxies within the green valley has an intermediate strength, and is consistent with that expected from galaxy groups. The results are robust to the correction for dust extinction. The comparison with simple analytical modeling suggests that the halo occupation number increases with older star formation epochs. When splitting according to SSFR, we find that the SSFR is a more sensitive tracer of environment than stellar mass.
We present the first measurements of the angular correlation function of galaxies selected in the far (1530 A) and near (2310 A) Ultraviolet from the GALEX survey fields overlapping SDSS DR5 in low galactic extinction regions. The area used covers 12
0 sqdeg (GALEX - MIS) down to magnitude AB = 22, yielding a total of 100,000 galaxies. The mean correlation length is ~ 3.7 pm 0.6 Mpc and no significant trend is seen for this value as a function of the limiting apparent magnitude or between the GALEX bands. This estimate is close to that found from samples of blue galaxies in the local universe selected in the visible, and similar to that derived at z ~ 3 for LBGs with similar rest frame selection criteria. This result supports models that predict anti-biasing of star forming galaxies at low redshift, and brings an additional clue to the downsizing of star formation at z<1.
We analyze the clustering properties of ultraviolet selected galaxies by using GALEX-SDSS data at z<0.6 and CFHTLS deep u imaging at z=1. These datasets provide a unique basis at z< 1 which can be directly compared with high redshift samples built wi
th similar selection criteria. We discuss the dependence of the correlation function parameters (r0, delta) on the ultraviolet luminosity as well as the linear bias evolution. We find that the bias parameter shows a gradual decline from high (b > 2) to low redshift (b ~ 0.79^{+0.1}_{-0.08}). When accounting for the fraction of the star formation activity enclosed in the different samples, our results suggest that the bulk of star formation migrated from high mass dark matter halos at z>2 (10^12 < M_min < 10^13 M_sun, located in high density regions), to less massive halos at low redshift (M_min < 10^12 M_sun, located in low density regions). This result extends the ``downsizing picture (shift of the star formation activity from high stellar mass systems at high z to low stellar mass at low z) to the dark matter distribution.
