arXiv subjects

S. Charlot

Publications and source records attributed to S. Charlot.

At least 145 records · Page 8Linked to original sources

arXiv2007

VVDS-SWIRE: Clustering evolution from a spectroscopic sample of galaxies with redshift 0.2<z<2.1 selected from Spitzer IRAC 3.6 micron and 4.5 micron photometry

By combining the VVDS with the SWIRE data, we have built the currently largest spectroscopic sample of galaxies selected in the rest-frame near-infrared. These allow us to investigate, for the first time using spectroscopic redshifts, the clustering evolution of galaxies selected from their rest-frame near-infrared luminosity in the redshift range 0.2<z<2.1. Therefore we use the projected two-point correlation function wp(rp) to study the three dimensional clustering properties of galaxies detected at 3.6 and 4.5 microns with IRAC and for which we have spectroscopic redshifts from first epoch VVDS. We find that in the flux limited samples at 3.6 and 4.5 microns, the clustering length does not change from redshift ~2 to the present. In addition, we find that the clustering lengths are systematically higher when galaxy samples are selected from increasingly redder wavelengths. These values are larger than the typical values found for I-band selected galaxies in the same redshift range, but we find that the difference in clustering length between I-band and 3.6/4.5 microns selected samples is decreasing with increasing redshift to become comparable at z~1.5. We interpret this as evidence that galaxies with older stellar populations and galaxies actively forming stars reside in comparably overdense environments at epochs earlier than z~1.5. The increasing difference in clustering length observed between rest-frame UV-optical and infrared selected samples could then be an indication that star formation is gradually shifting to lower density regions as cosmic time increases.

astro-ph↗

arXiv2007

The cosmic star formation rate evolution from z=5 to z=0 from the VIMOS VLT Deep Survey

We present the evolution of the comoving SFR density in the redshift range 0 < z < 5 using the first epoch data release of the VVDS, that is 11564 spectra selected at I_AB=24 over 2200 arcmin^2 in two fields of view, the VVDS-0226-04 and the VVDS-CDFS-0332-27, and the cosmological parameters (Omega_M, Omega_L, h)=(0.3, 0.7, 0.7). We study the multi-wavelength non dust-corrected luminosity densities at 0 < z < 2 from the rest-frame FUV to the optical passbands, and the rest-frame 1500A luminosity functions and densities at 2.7 < z < 5. They evolve from z=1.2 to 0.05 according to (1+z)^{x} with x = 2.05, 1.94, 1.92, 1.14, 0.73, 0.42, 0.30 in the FUV-1500, NUV-2800, U-3600, B-4400, V-5500, R-6500, and I-7900 passbands, respectively. From z = 1.2 to 0.2 the B-band density for the irregular-like galaxies decreases markedly by a factor 3.5 while it increases by a factor 1.7 for the elliptical-like galaxies. We identify several SFR periods; from z = 5 to 3.4 the FUV-band density increases by at most 0.5dex, from z=3.4 to 1.2 it decreases by 0.08dex, from z=1.2 to 0.05 it declines steadily by 0.6dex. For the most luminous M_AB(1500) < -21 galaxies the FUV-band density drops by 2dex from z = 3.9 to 1.2, and for the intermediate -21 < M_AB(1500) < -20 galaxies it drops by 2dex from z = 0.2 to 0. Comparing with dust corrected surveys, at 0.4 < z < 2 the FUV seems obscured by a constant factor of ~1.8-2 mag, while at z < 0.5 it seems progressively less obscured by up to ~0.9-1 mag when the dust-deficient early-type population is increasingly dominating the B-band density. The VVDS results agree with a downsizing picture where the most luminous sources cease to efficiently produce new stars 12 Gyrs ago (at z~4), while intermediate luminosity sources keep producing stars until 2.5 Gyrs ago (at z~0.2).(abridged)

astro-ph↗

arXiv2007

The VVDS type-1 AGN sample: The faint end of the luminosity function

In a previous paper (Gavignaud et al. 2006), we presented the type-1 Active Galactic Nuclei (AGN) sample obtained from the first epoch data of the VIMOS-VLT Deep Survey (VVDS). The sample consists of 130 faint, broad-line AGN with redshift up to z=5 and 17.5< I <24.0, selected on the basis of their spectra. In this paper we present the measurement of the Optical Luminosity Function up to z=3.6 derived from this sample, we compare our results with previous results from brighter samples both at low and at high redshift. Our data, more than one magnitude fainter than previous optical surveys, allow us to constrain the faint part of the luminosity function up to high redshift. By combining our faint VVDS sample with the large sample of bright AGN extracted from the SDSS DR3 (Richards et al., 2006b) and testing a number of different evolutionary models, we find that the model which better represents the combined luminosity functions, over a wide range of redshift and luminosity, is a luminosity dependent density evolution (LDDE) model, similar to those derived from the major X-surveys. Such a parameterization allows the redshift of the AGN space density peak to change as a function of luminosity and explains the excess of faint AGN that we find at 1.0< z <1.5. On the basis of this model we find, for the first time from the analysis of optically selected samples, that the peak of the AGN space density shifts significantly towards lower redshift going to lower luminosity objects. This result, already found in a number of X-ray selected samples of AGN, is consistent with a scenario of "AGN cosmic downsizing", in which the density of more luminous AGN, possibly associated to more massive black holes, peaks earlier in the history of the Universe, than that of low luminosity ones.

astro-ph↗

arXiv2007

The VIMOS-VLT Deep Survey. Color bimodality and the mix of galaxy populations up to z~2

In this paper we discuss the mix of star-forming and passive galaxies up to z~2, based on the first epoch VIMOS-VLT Deep Survey (VVDS) data.In agreement with previous works we find that the galaxy rest-frame color distribution follows a bimodal distribution at z<=1, and we establish that this bimodality holds up to z~2. The details of the rest-frame color distribution depend however on redshift and on galaxy luminosity: faint galaxies are bluer than the luminous ones over the whole redshift range of our data, and galaxies become bluer as redshift increases. This blueing trend does not depend, to a first approximation, on galaxy luminosity. Using our spectroscopic dataset we can also separate galaxies based on a star-formation, estimated combining the equivalent width of the [OII] emission line and the strength of the Dn(4000) break. The comparison between this spectral classification and the rest-frame colors shows that about 35-40 % of the red objects are star forming galaxies. Hence we conclude that the red sequence cannot be used to effectively isolate a sample of purely passively evolving objects within a cosmological survey. We also find that the color-magnitude relations derived for the color and for the spectroscopically selected early-type galaxies have remarkably similar properties, with the contaminating star-forming galaxies within the red sequence objects introducing no significant offset in the rest frame colors. Therefore the average color of the red objects does not appear to be a very sensitive indicator for measuring the evolution of the early-type galaxy population.Finally, we use the multi-band VVDS photometric data and SED fitting to derive multi-color galaxy types, which have a slightly higher efficiency than rest-frame color in isolating the passive, non star-forming galaxies within the VVDS sample.

astro-ph↗

arXiv2006

The VIRMOS-VLT Deep Survey: the last 10 billion years of evolution of galaxy clustering

We discuss the evolution of clustering of galaxies in the Universe from the present epoch back to z ~ 2, using the first-epoch data from the VIMOS-VLT Deep Survey (VVDS). We present the evolution of the projected two-point correlation function of galaxies for the global galaxy population, as well as its dependence on galaxy intrinsic luminosities and spectral types. While we do not find strong variations of the correlation function parameters with redshift for the global galaxy population, the clustering of objects with different intrinsic luminosities evolved significantly during last 8-10 billion years. Our findings indicate that bright galaxies in the past traced higher density peaks than they do now and that the shape of the correlation function of most luminous galaxies is different from observed for their local counterparts, which is a supporting evidence of a non-trivial evolution of the galaxy vs. dark matter bias.

astro-ph↗

arXiv2006

The VIMOS VLT Deep Survey:The ultraviolet galaxy luminosity function and luminosity density at 3<z<4

We study the LF of the high-z galaxy population with 3<z<4 using a purely I-band magnitude-selected spectroscopic sample obtained in the framework of the VVDS. We determine the LF from the VVDS, taking care to add as few assumptions and as simple corrections as possible, and compare our results with those obtained from photometric studies, based on Lyman-break selections or photo-z measurements. We find that in the range 3<z<4, the VVDS LF is parameterized by phi*=1.24+-0.50 10-3 mag-1 Mpc-3 and M*=-21.49+-0.19, assuming a slope alpha=-1.4 consistent with most previous studies. While phi* is comparable to previously found values, M* is significantly brighter by about 0.5 mag at least. Using the conservative slope -1.4, we find a LD at 1700A rho(M<-18.5)=2.4 10 19 W Mpc-3 and rho Tot=3.1 10 19 W Mpc-3, comparable to that estimated in other studies. The unexpectedly large number of very bright galaxies found in the VVDS indicates that the color-selection and photo-z techniques that are generally used to build high-z galaxy samples may be affected by a significant fraction of color-measurement failures or by incomplete modelling of the mix of stellar emission, AGN contribution, dust absorption and intergalactic extinction assumed to identify high-z galaxies, making pure magnitude selection better able to trace the full population. Because of the difficulty to identify all low-luminosity galaxies in a spectroscopic survey, the LD could still be significantly underestimated. We also find that the relative contribution of the most luminous galaxies compared to the fainter ones is at least twice as large in the VVDS compared to former estimates. Therefore, the VVDS paints a quite different picture of the role of the most actively star-forming galaxies in the history of star formation.

astro-ph↗

arXiv2006

Witnessing the build-up of the colour-density relation

We investigate the redshift and luminosity evolution of the galaxy colour-density relation using the data from the First Epoch VIMOS-VLT Deep Survey (VVDS) on scales of R=5 h^(-1)Mpc up to redshift z ~ 1.5. While at lower redshift we confirm the existence of a steep colour-density relation, with the fraction of the reddest(/bluest) galaxies of the same luminosity increasing(/decreasing) as a function of density, this trend progressively disappears in the highest redshift bins investigated.Our results suggest the existence of an epoch (more remote for brighter galaxies) characterized by the absence of the colour-density relation on the R=5 h^(-1)Mpc scales investigated. The rest frame u*-g' colour-magnitude diagram shows a bimodal pattern in both low and high density environments up to redshift z ~ 1.5. We find that the bimodal distribution is not universal but strongly depends upon environment. Both the colour-density and colour-magnitude-density relations, on the R=5 h^(-1)Mpc scales, appear to be a transient, cumulative product of genetic and environmental factors that have been operating over at least a period of 9 Gyr. These findings support an evolutionary scenario in which star formation/gas depletion processes are accelerated in more luminous objects and in high density environments: star formation activity is progressively shifting with cosmic time towards lower luminosity galaxies (downsizing), and out of high density environments.

astro-ph↗

arXiv2006

NIR Follow-Up of the VVDS 02hr Field

We present a new K-band survey covering 623 arcmin$^2$ in the VVDS 0226-0430 deep field down to a limiting magnitude K$_{\rm{Vega}}$ $\leq$ 20.5. We use the spectroscopic sample extracted from this new K-band catalogue to assess the effectiveness of optical-near infrared color selections in identifying extreme classes of objects at high redshift.

astro-ph↗

arXiv2006

Properties of the galaxy population in hydrodynamical simulations of clusters

We present a study of the galaxy population predicted by hydrodynamical simulations for a set of 19 galaxy clusters based on the GADGET-2 Tree+SPH code. These simulations include gas cooling, star formation, a detailed treatment of stellar evolution and chemical enrichment, as well as SN energy feedback in the form of galactic winds. We compute the spectro-photometric properties of the simulated galaxies. All simulations have been performed for two choices of the stellar initial mass function: a standard Salpeter IMF, and a top-heavier IMF. Several of the observational properties of the galaxy population in nearby clusters are reproduced fairly well by simulations. A Salpeter IMF is successful in accounting for the slope and the normalization of the color-magnitude relation for the bulk of the galaxy population. Simulated clusters have a relation between mass and optical luminosity which generally agrees with observations, both in normalization and slope. We find that galaxies are generally bluer, younger and more star forming in the cluster outskirts, thus reproducing the observational trends. However, simulated clusters have a total number of galaxies which is significantly smaller than the observed one, falling short by about a factor 2-3. Finally, the brightest cluster galaxies are always predicted to be too massive and too blue, when compared to observations, due to gas overcooling in the core cluster regions, even in the presence of a rather efficient SN feedback.

astro-ph↗

arXiv2006

The VIMOS VLT Deep Survey: The build-up of the colour-density relation

We investigate the redshift and luminosity evolution of the galaxy colour-density relation using the data from the First Epoch VIMOS-VLT Deep Survey (VVDS). The size (6582 galaxies), depth (I_AB<=24) and redshift sampling rate of the survey enable us to reconstruct the 3D galaxy environment on relatively local scales (R=5 Mpc) up to z~1.5. Particular attention has been devoted to calibrate a density reconstruction scheme, which factors out survey selection effects and reproduces in an unbiased way the underlying `real' galaxy environment. While at lower redshift we confirm the existence of a steep colour-density relation, with the fraction of the reddest(/bluest) galaxies of the same luminosity increasing(/decreasing) as a function of density, this trend progressively disappears in the highest redshift bins investigated. The rest frame u*-g' colour-magnitude diagram shows a bimodal pattern in both low and high density environments up to z~1.5. We find that the bimodal distribution is not universal but strongly depends upon environment: at lower redshifts the colour-magnitude diagrams in low and high density regions are significantly different while the progressive weakening of the colour-density relation causes the two bimodal distributions to nearly mirror each other in the highest redshift bin investigated. Both the colour-density and the colour-magnitude-density relations appear to be a transient, cumulative product of genetic and environmental factors operating over at least a period of 9 Gyr. These findings support an evolutionary scenario in which star formation/gas depletion processes are accelerated in more luminous objects and in high density environments: star formation activity is shifting with cosmic time towards lower luminosity (downsizing), and out of high density environments.

astro-ph↗

arXiv2006

The Rest-frame Optical Colors of 99,000 SDSS Galaxies

We synthesize the rest-frame Stroemgren colors using SDSS spectra for 99,088 galaxies selected from Data Release 1. This narrow-band ~200 AA photometric system (uz, vz, bz, yz), first designed for the determination of effective temperature, metallicity and gravity of stars, measures the continuum spectral slope of galaxies in the rest-frame 3200-5800 AA wavelength range. Galaxies form a remarkably narrow locus (~0.03 mag) in the resulting color-color diagram. The Bruzual & Charlot population synthesis models suggest that the position of a galaxy along this locus is controlled by a degenerate combination of metallicity and age of the dominant stellar population. Galaxy distribution along the locus is bimodal, with the local minimum corresponding to an ~1 Gyr old single stellar population. The position perpendicular to the locus is independent of metallicity and age, and reflects the galaxy's dust content, as implied by both the models and the statistics of IRAS detections. A comparison of this locus with the galaxy locus in the H_delta-D_n(4000) diagram, utilized by Kauffmann et al. (2003) to estimate stellar masses, reveals a tight correlation, although the two analyzed spectral ranges barely overlap. Overall, the galaxy spectral energy distribution in the entire UV to near-IR range can be described as a single-parameter family with an accuracy of 0.1 mag, or better. This nearly one-dimensional distribution of galaxies in the multi-dimensional space of measured parameters strongly supports the conclusion of Yip et al. (2004), based on a principal component analysis, that SDSS galaxy spectra can be described by a small number of eigenspectra. Apparently, the contributions of stellar populations that dominate the optical emission from galaxies are combined in a simple and well-defined way.

astro-ph↗

arXiv2006

Ages and metallicities of early-type galaxies in the SDSS: new insight into the physical origin of the colour-magnitude and the Mg2-sigmaV relations

We exploit recent constraints on the ages and metallicities of early-type galaxies in the Sloan Digital Sky Survey (SDSS) to gain new insight into the physical origin of two fundamental relations obeyed by these galaxies: the colour-magnitude and the Mg2-sigmaV relations. Our sample consists of 26,003 galaxies selected from the SDSS DR2 on the basis of their concentrated light profiles, for which we have previously derived median-likelihood estimates of stellar metallicity, light-weighted age and stellar mass. Our analysis provides the most unambiguous demonstration to date of the fact that both relations are primarily sequences in stellar mass and that total stellar metallicity, alpha-elements-to-iron abundance ratio and light-weighted age all increase with mass along the two relations. For high-mass ellipticals, the dispersion in age is small and consistent with the error. At the low-mass end, there is a tail towards younger ages, which dominates the scatter in colour and index strength at fixed mass. A small, but detectable, intrinsic scatter in the mass-metallicity relation also contributes to the scatter in the two observational scaling relations, even at high masses. Our results suggest that the chemical composition of an early-type galaxy is more tightly related to its dynamical mass (including stars and dark matter) than to its stellar mass. The ratio between stellar mass and dynamical mass appears to decrease from the least massive to the most massive galaxies in our sample.

astro-ph↗

arXiv2006

The VIMOS VLT Deep Survey: the faint type-1 AGN sample

We present the type-1 active galactic nuclei (AGN) sample extracted from the VIMOS VLT Deep Survey first observations of 21000 spectra in 1.75 square degree. This sample, which is purely magnitude limited, free of morphological or color selection biases, contains 130 broad line AGN (BLAGN) spectra with redshift up to 5. Our data are divided into a wide (Iab < 22.5) and a deep (Iab < 24) subsample containing 56 and 74 objects respectively. Because of its depth and selection criteria, this sample is uniquely suited to study the population of faint type-1 AGN. Our measured surface density (~ 472 +- 48 BLAGN per square degree with Iab < 24) is significantly higher than that of any other optically selected sample of BLAGN with spectroscopic confirmation. By applying a morphological and color analysis to our AGN sample we find that: (1)~23% of the AGN brighter than Iab=22.5 are classified as extended; this percentage increases to ~42% for those with z < 1.6; (2) a non-negligible fraction of our BLAGN are lying close to the color space area occupied by stars in u*-g' versus g'-r' color-color diagram. This leads us to the conclusion that classical optical ultraviolet preselection technique, if employed at such deep magnitudes (Iab=22.5) in conjuction with a preselection of point-like sources, can miss miss up to ~35% of the AGN population. Finally, we present a composite spectrum of our sample of objects. While the continuum shape is very similar to that of the SDSS composite at short wavelengths, it is much redder than it at lambda > 3000 A. We interpret this as due to significant contamination from emission of the host galaxies, as expected from the faint absolute magnitudes sampled by our survey.

astro-ph↗

arXiv2006

The star formation history of luminous infrared galaxies

We constrain the past star formation histories of a sample of 25 distant (z~0.7) luminous infrared galaxies (LIRGs) detected with the mid infrared cameras ISOCAM and MIPS onboard the ISO and Spitzer satellites. We use high resolution VLT-FORS2 spectroscopy in addition to a comprehensive library of 200,000 model optical spectra to derive Bayesian likelihood estimates of the star formation histories of these galaxies based on the analysis of Balmer absorption lines and the 4000 A break. The locus of distant LIRGs in the diagram defined by Hdelta_A and D4000 is roughly comparable to that of local LIRGs observed with IRAS, suggesting that no trend for an evolution is detected between the local and distant LIRGs. We obtain similar results when using either the H8 or the H$δ_A$ Balmer absorption-line indices in combination with D4000. We compute a birthrate parameter (b=SFR/ ) of 4\pm1, confirming that the distant LIRGs are currently experiencing a major phase of star formation. The most likely duration of the bursts is 0.10 Gyr, during which the LIRGs produce 5-10 % of their current stellar mass. No evidence is found for successive starbursts on the scale of a few times $10^7$ yr, such as those predicted by some numerical simulations of major mergers. However, the large number density of those galaxies suggest that they could experience between two and four LIRG phases until the present epoch.

astro-ph↗

arXiv2006

Dissecting Galaxy Colors with GALEX, SDSS, and Spitzer

We combine data from SDSS and the GALEX and Spitzer observatories to create a sample of galaxies observed homogeneously from the UV to the Far-IR. This sample, consisting of ~460 galaxies observed spectroscopically by SDSS provides a multiwavelength (0.15-24 micron) view of obscured and unobscured star formation in nearby (z<0.3) galaxies with SFRs ranging from 0.01 to 100 M_solar/yr. We calculate a robust dust measure from the infrared to UV ratio (IRX) and explore the influence of star formation history (SFH) on the dust-UV color relation (i.e. the IRX-beta relation). We find that the UV colors of galaxies are only weakly dependent on their SFH as measured by the 4000A break. However, we find that the contributions of dust and SFH are distinguishable when colors at widely separated wavelengths (e.g. 0.23-3.6 micron) are introduced. We show this explicitly by recasting the IRX-beta relation as a more general IRX-SFH-color relation, which we examine in different projections. We also determine simple fits to this relation.

astro-ph↗

arXiv2006

The VIMOS VLT Deep Survey - Evolution of the luminosity functions by galaxy type up to z=1.5 from first epoch data

From the first epoch observations of the VVDS up to z=1.5 we have derived luminosity functions (LF) of different spectral type galaxies. The VVDS data, covering ~70% of the life of the Universe, allow for the first time to study from the same sample and with good statistical accuracy the evolution of the LFs by galaxy type in several rest frame bands from a purely magnitude selected sample. The magnitude limit of the VVDS allows the determination of the faint end slope of the LF with unprecedented accuracy. Galaxies have been classified in four spectral classes, using their colours and redshift, and LFs have been derived in the U, B, V, R and I rest frame bands from z=0.05 to z=1.5. We find a significant steepening of the LF going from early to late types. The M* parameter is significantly fainter for late type galaxies and this difference increases in the redder bands. Within each of the galaxy spectral types we find a brightening of M* with increasing redshift, ranging from =< 0.5 mag for early type galaxies to ~1 mag for the latest type galaxies, while the slope of the LF of each spectral type is consistent with being constant with redshift. The LF of early type galaxies is consistent with passive evolution up to z~1.1, while the number of bright early type galaxies has decreased by ~40% from z~0.3 to z~1.1. We also find a strong evolution in the normalization of the LF of latest type galaxies, with an increase of more than a factor 2 from z~0.3 to z~1.3: the density of bright late type galaxies in the same redshift range increases of a factor ~6.6. These results indicate a strong type-dependent evolution and identifies the latest spectral types as responsible for most of the evolution of the UV-optical luminosity function out to z=1.5.

astro-ph↗

arXiv2006

Galaxy luminosity function per morphological type up to z=1.2

We have computed the evolution of the rest-frame B-band luminosity function (LF) for bulge and disk-dominated galaxies since z=1.2. We use a sample of 605 spectroscopic redshifts with I_{AB}<24 in the Chandra Deep Field South from the VIMOS-VLT Deep Survey, 3555 galaxies with photometric redshifts from the COMBO-17 multi-color data, coupled with multi-color HST/ACS images from the Great Observatories Origin Deep Survey. We split the sample in bulge- and disk-dominated populations on the basis of asymmetry and concentration parameters measured in the rest-frame B-band. We find that at z=0.4-0.8, the LF slope is significantly steeper for the disk-dominated population (α=-1.19 \pm 0.07) compared to the bulge-dominated population (α=-0.53 \pm 0.13). The LF of the bulge-dominated population is composed of two distinct populations separated in rest-frame color: 68% of red (B-I)_{AB}>0.9 and bright galaxies showing a strongly decreasing LF slope α=+0.55 \pm 0.21, and 32% of blue (B-I)_{AB}<0.9 and more compact galaxies which populate the LF faint-end. We observe that red bulge-dominated galaxies are already well in place at z~1, but the volume density of this population is increasing by a factor 2.7 between z~1 and z~0.6. It may be related to the building-up of massive elliptical galaxies in the hierarchical scenario. In addition, we observe that the blue bulge-dominated population is dimming by 0.7 magnitude between z~1 and z~0.6. Galaxies in this faint and more compact population could possibly be the progenitors of the local dwarf spheroidal galaxies.

astro-ph↗

arXiv2006

Accurate photometric redshifts for the CFHT Legacy Survey calibrated using the VIMOS VLT Deep Survey

We present photometric redshifts for an uniquely large and deep sample of 522286 objects with i'_{AB}<25 in the Canada-France Legacy Survey ``Deep Survey'' fields, which cover a total effective area of 3.2 deg^2. We use 3241 spectroscopic redshifts with 0<z<5 from the VIMOS VLT Deep Survey as a calibration to derive these photometric redshifts. We devise a robust calibration method which removes systematic trends in the photometric redshifts and significantly reduces the fraction of catastrophic errors. We use our unique spectroscopic sample to present a detailed assessment of the robustness of the photometric redshift sample. For a sample selected at i'_{AB}<24, we reach a redshift accuracy of σ_{Δz/(1+z)}=0.037 with η=3.7% of catastrophic error. The reliability of our photometric redshifts is lower for fainter objects: we find σ_{Δz/(1+z)}=0.029, 0.043 and η=1.7%, 5.4% for samples selected at i'_{AB}=17.5-22.5 and 22.5-24 respectively. We find that the photometric redshifts of starburst galaxies in our sample are less reliable: although these galaxies represent only 18% of the spectroscopic sample they are responsible for 54% of the catastrophic errors. We find an excellent agreement between the photometric and the VVDS spectroscopic redshift distributions at i'_{AB}<24. Finally, we compare the redshift distributions of i' selected galaxies on the four CFHTLS deep fields, showing that cosmic variance is already present on fields of 0.8 deg^2.

astro-ph↗