Aims. We present and release photometric redshifts for a uniquely large and deep sample of 522286 objects with i AB ≤ 25 in the Canada-France Hawaii Telescope Legacy Survey (CFHTLS) "Deep Survey" fields D1, D2, D3, and D4, which cover a total effective area of 3.2 deg 2 . Methods. We use 3241 spectroscopic redshifts with 0 ≤ z ≤ 5 from the VIMOS VLT Deep Survey (VVDS) as a calibration and training set to derive these photometric redshifts. Using the "Le Phare" photometric redshift code, we developed a robust calibration method based on an iterative zero-point refinement combined with a template optimisation procedure and the application of a Bayesian approach. This method removes systematic trends in the photometric redshifts and significantly reduces the fraction of catastrophic errors (by a factor of 2), a significant improvement over traditional methods. We use our unique spectroscopic sample to present a detailed assessment of the robustness of the photometric redshift sample. Results. For a sample selected at i AB ≤ 24, we reach a redshift accuracy of σ ∆z/(1+z) = 0.029 with η = 3.8% of catastrophic errors (η is defined strictly as those objects with |∆z|/(1 + z) > 0.15). The reliability of our photometric redshifts decreases for faint objects: we find σ ∆z/(1+z) = 0.025, 0.034 and η = 1.9%, 5.5% for samples selected at i AB = 17.5-22.5 and 22.5-24 respectively. We find that the photometric redshifts of starburst galaxies are less reliable: although these galaxies represent only 22% of the spectroscopic sample, they are responsible for 50% of the catastrophic errors. An analysis as a function of redshift demonstrates that our photometric redshifts work best in the redshift range 0.2 ≤ z ≤ 1.5. 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 still present on fields of 0.7-0.9 deg 2 . These photometric redshifts are made publicly available at http://terapix.iap.fr (complete ascii catalogues) and http://cencos.oamp.fr/cencos/CFHTLS/ (searchable database interface).Key words. galaxies: distances and redshifts -galaxies: photometry -methods: data analysis Article published by EDP Sciences and available at http://www.edpsciences.org/aa or http://dx

We introduce a novel technique to construct spatially resolved maps of stellar mass surface density in galaxies based on optical and near‐infrared (NIR) imaging. We use optical/NIR colour(s) to infer effective stellar mass‐to‐light ratios (M/L) at each pixel, which are then multiplied by the surface brightness to obtain the local surface stellar mass density. We build look‐up tables to express M/L as a function of colour(s) by marginalizing over a Monte Carlo library of 50 000 stellar population synthesis (SPS) models by Charlot & Bruzual, which include a revised prescription for the thermally pulsing asymptotic giant branch (TP‐AGB) stellar evolutionary phase. Moreover, we incorporate a wide range of possible dust extinction parameters. In order to extract reliable flux and colour information at any position in the galaxy, we perform a median adaptive smoothing of the images that preserves the highest possible spatial resolution. As the most practical and robust, and hence fiducial method, we express the M/L in the H band as a function of (g−i) and (i−H). Stellar mass maps computed in this way have a typical accuracy of 30 per cent or less at any given pixel, determined from the scatter in the models. We compare maps obtained with our fiducial method with those derived using other combinations of bandpasses: (i) mass maps based on the M/L in NIR bands require one optical and one optical‐NIR colour to avoid significant biases as a function of the local physical properties of a galaxy; (ii) maps based on M/L in i band as a function of (g−i) only are generally in excellent agreement with our best optical‐NIR set, except for extremely star‐forming and dust extincted regions. We further compute stellar mass maps using a model library identical to the previous one except for being based on older SPS models, which assume shorter lived TP‐AGB stars. The M/L in the NIR inferred using these old models may be up to 2.5 times larger than the new ones, but this varies strongly as a function of colours and is maximal for the bluest colours. Finally, we compare total stellar mass estimates obtained by integrating resolved mass maps with those obtained with unresolved photometry. In galaxies with evident dust lanes, unresolved estimates may miss up to 40 per cent of the total stellar mass because dusty regions are strongly under‐represented in the luminous fluxes.

We compile a new sample of 115 nearby, non‐Seyfert galaxies spanning a wide range of star formation activities, from starburst to nearly dormant, based on ultraviolet observations with various satellites. We combine these observations with infrared observations to study the relation between ratio of total far‐infrared to ultraviolet luminosity and ultraviolet spectral slope (the IRX–UV relation). We show that, at fixed ultraviolet spectral slope, quiescent star‐forming galaxies in our sample have systematically lower ratio of total far‐infrared to ultraviolet luminosity than starburst galaxies. The strengths of spectral indices sensitive to star formation history, such as the 4000‐Å spectral discontinuity and the Hα emission equivalent width, correlate well with distance from the mean relation for starburst galaxies in the IRX–UV diagram, while there is little or no correlation between the dust‐sensitive Hα/Hβ ratio and this distance. This is strong observational evidence that the star formation history is relevant to the second parameter affecting the IRX–UV relation. We show that these results can be understood in the framework of the simple model of Charlot & Fall for the transfer of starlight through the interstellar medium in galaxies. We confirm that, for starburst galaxies, the tight IRX–UV relation can be understood most simply as a sequence in overall dust content. In addition, we find that the broadening of the relation for quiescent star‐forming galaxies can be understood most simply as a sequence in the ratio of present to past‐averaged star formation rate. We use a library of Monte Carlo realizations of galaxies with different star formation histories and dust contents to quantify the accuracy to which the ultraviolet attenuation AFUV of a galaxy can be estimated from either the ratio of far‐infrared to ultraviolet luminosity or the ultraviolet spectral slope. We provide simple formulae for estimating AFUV as a function of either of these observational quantities and show that the accuracy of these estimates can be significantly improved if some constraints are available on the ratio of present to past‐averaged star formation rate.