arXiv subjects

M. Brescia

Publications and source records attributed to M. Brescia.

At least 325 records · Page 18Linked to original sources

arXiv2018

Photometric redshifts for the Kilo-Degree Survey. Machine-learning analysis with artificial neural networks

We present a machine-learning photometric redshift analysis of the Kilo-Degree Survey Data Release 3, using two neural-network based techniques: ANNz2 and MLPQNA. Despite limited coverage of spectroscopic training sets, these ML codes provide photo-zs of quality comparable to, if not better than, those from the BPZ code, at least up to zphot<0.9 and r<23.5. At the bright end of r<20, where very complete spectroscopic data overlapping with KiDS are available, the performance of the ML photo-zs clearly surpasses that of BPZ, currently the primary photo-z method for KiDS. Using the Galaxy And Mass Assembly (GAMA) spectroscopic survey as calibration, we furthermore study how photo-zs improve for bright sources when photometric parameters additional to magnitudes are included in the photo-z derivation, as well as when VIKING and WISE infrared bands are added. While the fiducial four-band ugri setup gives a photo-z bias $δz=-2e-4$ and scatter $σ_z<0.022$ at mean z = 0.23, combining magnitudes, colours, and galaxy sizes reduces the scatter by ~7% and the bias by an order of magnitude. Once the ugri and IR magnitudes are joined into 12-band photometry spanning up to 12 $μ$, the scatter decreases by more than 10% over the fiducial case. Finally, using the 12 bands together with optical colours and linear sizes gives $δz<4e-5$ and $σ_z<0.019$. This paper also serves as a reference for two public photo-z catalogues accompanying KiDS DR3, both obtained using the ANNz2 code. The first one, of general purpose, includes all the 39 million KiDS sources with four-band ugri measurements in DR3. The second dataset, optimized for low-redshift studies such as galaxy-galaxy lensing, is limited to r<20, and provides photo-zs of much better quality than in the full-depth case thanks to incorporating optical magnitudes, colours, and sizes in the GAMA-calibrated photo-z derivation.

astro-ph.CO↗

arXiv2017

Euclid: Superluminous supernovae in the Deep Survey

In the last decade, astronomers have found a new type of supernova called `superluminous supernovae' (SLSNe) due to their high peak luminosity and long light-curves. These hydrogen-free explosions (SLSNe-I) can be seen to z~4 and therefore, offer the possibility of probing the distant Universe. We aim to investigate the possibility of detecting SLSNe-I using ESA's Euclid satellite, scheduled for launch in 2020. In particular, we study the Euclid Deep Survey (EDS) which will provide a unique combination of area, depth and cadence over the mission. We estimated the redshift distribution of Euclid SLSNe-I using the latest information on their rates and spectral energy distribution, as well as known Euclid instrument and survey parameters, including the cadence and depth of the EDS. We also applied a standardization method to the peak magnitudes to create a simulated Hubble diagram to explore possible cosmological constraints. We show that Euclid should detect approximately 140 high-quality SLSNe-I to z ~ 3.5 over the first five years of the mission (with an additional 70 if we lower our photometric classification criteria). This sample could revolutionize the study of SLSNe-I at z>1 and open up their use as probes of star-formation rates, galaxy populations, the interstellar and intergalactic medium. In addition, a sample of such SLSNe-I could improve constraints on a time-dependent dark energy equation-of-state, namely w(a), when combined with local SLSNe-I and the expected SN Ia sample from the Dark Energy Survey. We show that Euclid will observe hundreds of SLSNe-I for free. These luminous transients will be in the Euclid data-stream and we should prepare now to identify them as they offer a new probe of the high-redshift Universe for both astrophysics and cosmology.

astro-ph.CO↗

arXiv2017

Properties of Hi-GAL clumps in the inner Galaxy]{The Hi-GAL compact source catalogue. I. The physical properties of the clumps in the inner Galaxy ($-71.0^{\circ}< \ell < 67.0^{\circ}$)

Hi-GAL is a large-scale survey of the Galactic plane, performed with Herschel in five infrared continuum bands between 70 and 500 $μ$m. We present a band-merged catalogue of spatially matched sources and their properties derived from fits to the spectral energy distributions (SEDs) and heliocentric distances, based on the photometric catalogs presented in Molinari et al. (2016a), covering the portion of Galactic plane $-71.0^{\circ}< \ell < 67.0^{\circ}$. The band-merged catalogue contains 100922 sources with a regular SED, 24584 of which show a 70 $μ$m counterpart and are thus considered proto-stellar, while the remainder are considered starless. Thanks to this huge number of sources, we are able to carry out a preliminary analysis of early stages of star formation, identifying the conditions that characterise different evolutionary phases on a statistically significant basis. We calculate surface densities to investigate the gravitational stability of clumps and their potential to form massive stars. We also explore evolutionary status metrics such as the dust temperature, luminosity and bolometric temperature, finding that these are higher in proto-stellar sources compared to pre-stellar ones. The surface density of sources follows an increasing trend as they evolve from pre-stellar to proto-stellar, but then it is found to decrease again in the majority of the most evolved clumps. Finally, we study the physical parameters of sources with respect to Galactic longitude and the association with spiral arms, finding only minor or no differences between the average evolutionary status of sources in the fourth and first Galactic quadrants, or between "on-arm" and "inter-arm" positions.

astro-ph.GA↗

arXiv2017

The third data release of the Kilo-Degree Survey and associated data products

The Kilo-Degree Survey (KiDS) is an ongoing optical wide-field imaging survey with the OmegaCAM camera at the VLT Survey Telescope. It aims to image 1500 square degrees in four filters (ugri). The core science driver is mapping the large-scale matter distribution in the Universe, using weak lensing shear and photometric redshift measurements. Further science cases include galaxy evolution, Milky Way structure, detection of high-redshift clusters, and finding rare sources such as strong lenses and quasars. Here we present the third public data release (DR3) and several associated data products, adding further area, homogenized photometric calibration, photometric redshifts and weak lensing shear measurements to the first two releases. A dedicated pipeline embedded in the Astro-WISE information system is used for the production of the main release. Modifications with respect to earlier releases are described in detail. Photometric redshifts have been derived using both Bayesian template fitting, and machine-learning techniques. For the weak lensing measurements, optimized procedures based on the THELI data reduction and lensfit shear measurement packages are used. In DR3 stacked ugri images, weight maps, masks, and source lists for 292 new survey tiles (~300 sq.deg) are made available. The multi-band catalogue, including homogenized photometry and photometric redshifts, covers the combined DR1, DR2 and DR3 footprint of 440 survey tiles (447 sq.deg). Limiting magnitudes are typically 24.3, 25.1, 24.9, 23.8 (5 sigma in a 2 arcsec aperture) in ugri, respectively, and the typical r-band PSF size is less than 0.7 arcsec. The photometric homogenization scheme ensures accurate colors and an absolute calibration stable to ~2% for gri and ~3% in u. Separately released are a weak lensing shear catalogue and photometric redshifts based on two different machine-learning techniques.

astro-ph.GA↗

arXiv2017

The VOICE Survey : VST Optical Imaging of the CDFS and ES1 Fields

We present the VST Optical Imaging of the CDFS and ES1 Fields (VOICE) Survey, a VST INAF Guaranteed Time program designed to provide optical coverage of two 4 deg$^2$ cosmic windows in the Southern hemisphere. VOICE provides the first, multi-band deep optical imaging of these sky regions, thus complementing and enhancing the rich legacy of longer-wavelength surveys with VISTA, Spitzer, Herschel and ATCA available in these areas and paving the way for upcoming observations with facilities such as the LSST, MeerKAT and the SKA. VOICE exploits VST's OmegaCAM optical imaging capabilities and completes the reduction of WFI data available within the ES1 fields as part of the ESO-Spitzer Imaging Extragalactic Survey (ESIS) program providing $ugri$ and $uBVR$ coverage of 4 and 4 deg$^2$ areas within the CDFS and ES1 field respectively. We present the survey's science rationale and observing strategy, the data reduction and multi-wavelength data fusion pipeline. Survey data products and their future updates will be released at http://www.mattiavaccari.net/voice/ and on CDS/VizieR.

astro-ph.GA↗

arXiv2016

Towards a census of super-compact massive galaxies in the Kilo Degree Survey

The abundance of compact, massive, early-type galaxies (ETGs) provides important constraints to galaxy formation scenarios. Thanks to the area covered, depth, excellent spatial resolution and seeing, the ESO Public optical Kilo Degree Survey (KiDS), carried out with the VLT Survey Telescope (VST), offers a unique opportunity to conduct a complete census of the most compact galaxies in the Universe. This paper presents a first census of such systems from the first 156 square degrees of KiDS. Our analysis relies on g-, r-, and i-band effective radii ($R_{\rm e}$), derived by fitting galaxy images with PSF-convolved Sérsic models, high-quality photometric redshifts, $z_{\rm phot}$, estimated from machine learning techniques, and stellar masses, $M_{\rm \star}$, calculated from KiDS aperture photometry. After massiveness ($M_{\rm \star} > 8 \times 10^{10}\, \rm M_{\odot}$) and compactness ($R_{\rm e} < 1.5 \, \rm kpc$ in g-, r- and i-bands) criteria are applied, a visual inspection of the candidates plus near-infrared photometry from VIKING-DR1 are used to refine our sample. The final catalog, to be spectroscopically confirmed, consists of 92 systems in the redshift range $z \sim 0.2-0.7$. This sample, which we expect to increase by a factor of ten over the total survey area, represents the first attempt to select massive super-compact ETGs (MSCGs) in KiDS. We investigate the impact of redshift systematics in the selection, finding that this seems to be a major source of contamination in our sample. A preliminary analysis shows that MSCGs exhibit negative internal colour gradients, consistent with a passive evolution of these systems. We find that the number density of MSCGs is only mildly consistent with predictions from simulations at $z>0.2$, while no such system is found at $z < 0.2$.

astro-ph.GA↗

arXiv2015

Luminosity functions in the CLASH-VLT cluster MACS J1206.2-0847: the importance of tidal interactions

We present the optical luminosity functions (LFs) of galaxies for the CLASH-VLT cluster MACS J1206.2-0847 at z=0.439, based on HST and SUBARU data, including ~600 spectroscopically confirmed member galaxies. The LFs on the wide SUBARU FoV are well described by a single Schechter function down to M~M*+3, whereas this fit is poor for HST data, due to a faint-end upturn visible down M~M*+7, suggesting a bimodal behaviour. We also investigate the effect of local environment by deriving the LFs in four different regions, according to the distance from the centre, finding an increase in the faint-end slope going from the core to the outer rings. Our results confirm and extend our previous findings on the analysis of mass functions, which showed that the galaxies with stellar mass below 10^10.5, M_sun have been significantly affected by tidal interaction effects, thus contributing to the intra cluster light.

astro-ph.GA↗

arXiv2015

CLASH-VLT: Environment-driven evolution of galaxies in the z=0.209 cluster Abell 209

The analysis of galaxy properties and the relations among them and the environment, can be used to investigate the physical processes driving galaxy evolution. We study the cluster A209 by using the CLASH-VLT spectroscopic data combined with Subaru photometry, yielding to 1916 cluster members down to a stellar mass of 10^{8.6} Msun. We determine: i) the stellar mass function of star-forming and passive galaxies; ii) the intra-cluster light and its properties; iii) the orbits of low- and high-mass passive galaxies; and iv) the mass-size relation of ETGs. The stellar mass function of the star-forming galaxies does not depend on the environment, while the slope found for passive galaxies becomes flatter in the densest region. The color distribution of the intra-cluster light is consistent with the color of passive members. The analysis of the dynamical orbits shows that low-mass passive galaxies have tangential orbits, avoiding small pericenters around the BCG. The mass-size relation of low-mass passive ETGs is flatter than that of high mass galaxies, and its slope is consistent with that of field star-forming galaxies. Low-mass galaxies are also more compact within the scale radius of 0.65 Mpc. The ratio between stellar and number density profiles shows a mass segregation in the center. The comparative analysis of the stellar and total density profiles indicates that this effect is due to dynamical friction. Our results are consistent with a scenario in which the "environmental quenching" of low-mass galaxies is due to mechanisms such as harassment out to R200, starvation and ram-pressure stripping at smaller radii, as supported by the analysis of the mass function, of the dynamical orbits and of the mass-size relation of passive early-types in different regions. Our analyses support the idea that the intra-cluster light is formed through the tidal disruption of subgiant galaxies.

astro-ph.GA↗

arXiv2015

Shapley Supercluster Survey: Construction of the Photometric Catalogues and i-band Data Release

The Shapley Supercluster Survey is a multi-wavelength survey covering an area of ~23 deg^2 (~260 Mpc^2 at z=0.048) around the supercluster core, including nine Abell and two poor clusters, having redshifts in the range 0.045-0.050. The survey aims to investigate the role of the cluster-scale mass assembly on the evolution of galaxies, mapping the effects of the environment from the cores of the clusters to their outskirts and along the filaments. The optical (ugri) imaging acquired with OmegaCAM on the VLT Survey Telescope is essential to achieve the project goals providing accurate multi-band photometry for the galaxy population down to m*+6. We describe the methodology adopted to construct the optical catalogues and to separate extended and point-like sources. The catalogues reach average 5sigma limiting magnitudes within a 3\arcsec diameter aperture of ugri=[24.4,24.6,24.1,23.3] and are 93% complete down to ugri=[23.8,23.8,23.5,22.0] mag, corresponding to ~m*_r+8.5. The data are highly uniform in terms of observing conditions and all acquired with seeing less than 1.1 arcsec full width at half-maximum. The median seeing in r-band is 0.6 arcsec, corresponding to 0.56 kpc h^{-1}_{70} at z=0.048. While the observations in the u, g and r bands are still ongoing, the i-band observations have been completed, and we present the i-band catalogue over the whole survey area. The latter is released and it will be regularly updated, through the use of the Virtual Observatory tools. This includes 734,319 sources down to i=22.0 mag and it is the first optical homogeneous catalogue at such a depth, covering the central region of the Shapley supercluster.

astro-ph.GA↗

arXiv2015

Galaxy evolution within the Kilo-Degree Survey

The ESO Public Kilo-Degree Survey (KiDS) is an optical wide-field imaging survey carried out with the VLT Survey Telescope and the OmegaCAM camera. KiDS will scan 1500 square degrees in four optical filters (u, g, r, i). Designed to be a weak lensing survey, it is ideal for galaxy evolution studies, thanks to the high spatial resolution of VST, the good seeing and the photometric depth. The surface photometry have provided with structural parameters (e.g. size and Sérsic index), aperture and total magnitudes have been used to derive photometric redshifts from Machine learning methods and stellar masses/luminositites from stellar population synthesis. Our project aimed at investigating the evolution of the colour and structural properties of galaxies with mass and environment up to redshift $z \sim 0.5$ and more, to put constraints on galaxy evolution processes, as galaxy mergers.

astro-ph.GA↗

arXiv2015

CLASH-VLT: Substructure in the galaxy cluster MACS J1206.2-0847 from kinematics of galaxy populations

In the effort to understand the link between the structure of galaxy clusters and their galaxy populations, we focus on MACSJ1206.2-0847 at z~0.44 and probe its substructure in the projected phase space through the spectrophotometric properties of a large number of galaxies from the CLASH-VLT survey. Our analysis is mainly based on an extensive spectroscopic dataset of 445 member galaxies, mostly acquired with VIMOS@VLT as part of our ESO Large Programme, sampling the cluster out to a radius ~2R200 (4 Mpc). We classify 412 galaxies as passive, with strong Hdelta absorption (red and blue galaxies, and with emission lines from weak to very strong. A number of tests for substructure detection are applied to analyze the galaxy distribution in the velocity space, in 2D space, and in 3D projected phase-space. Studied in its entirety, the cluster appears as a large-scale relaxed system with a few secondary, minor overdensities in 2D distribution. We detect no velocity gradients or evidence of deviations in local mean velocities. The main feature is the WNW-ESE elongation. The analysis of galaxy populations per spectral class highlights a more complex scenario. The passive galaxies and red strong Hdelta galaxies trace the cluster center and the WNW-ESE elongated structure. The red strong Hdelta galaxies also mark a secondary, dense peak ~2 Mpc at ESE. The emission line galaxies cluster in several loose structures, mostly outside R200. The observational scenario agrees with MACS J1206.2-0847 having WNW-ESE as the direction of the main cluster accretion, traced by passive galaxies and red strong Hdelta galaxies. The red strong Hdelta galaxies, interpreted as poststarburst galaxies, date a likely important event 1-2 Gyr before the epoch of observation. The emission line galaxies trace a secondary, ongoing infall where groups are accreted along several directions.

astro-ph.CO↗

arXiv2014

Extending the supernova Hubble diagram to z~1.5 with the Euclid space mission

We forecast dark energy constraints that could be obtained from a new large sample of Type Ia supernovae where those at high redshift are acquired with the Euclid space mission. We simulate a three-prong SN survey: a z<0.35 nearby sample (8000 SNe), a 0.2<z<0.95 intermediate sample (8800 SNe), and a 0.75<z<1.55 high-z sample (1700 SNe). The nearby and intermediate surveys are assumed to be conducted from the ground, while the high-z is a joint ground- and space-based survey. This latter survey, the "Dark Energy Supernova Infra-Red Experiment" (DESIRE), is designed to fit within 6 months of Euclid observing time, with a dedicated observing program. We simulate the SN events as they would be observed in rolling-search mode by the various instruments, and derive the quality of expected cosmological constraints. We account for known systematic uncertainties, in particular calibration uncertainties including their contribution through the training of the supernova model used to fit the supernovae light curves. Using conservative assumptions and a 1-D geometric Planck prior, we find that the ensemble of surveys would yield competitive constraints: a constant equation of state parameter can be constrained to sigma(w)=0.022, and a Dark Energy Task Force figure of merit of 203 is found for a two-parameter equation of state. Our simulations thus indicate that Euclid can bring a significant contribution to a purely geometrical cosmology constraint by extending a high-quality SN Hubble diagram to z~1.5. We also present other science topics enabled by the DESIRE Euclid observations

astro-ph.CO↗

arXiv2014

Data Driven Discovery in Astrophysics

We review some aspects of the current state of data-intensive astronomy, its methods, and some outstanding data analysis challenges. Astronomy is at the forefront of "big data" science, with exponentially growing data volumes and data rates, and an ever-increasing complexity, now entering the Petascale regime. Telescopes and observatories from both ground and space, covering a full range of wavelengths, feed the data via processing pipelines into dedicated archives, where they can be accessed for scientific analysis. Most of the large archives are connected through the Virtual Observatory framework, that provides interoperability standards and services, and effectively constitutes a global data grid of astronomy. Making discoveries in this overabundance of data requires applications of novel, machine learning tools. We describe some of the recent examples of such applications.

astro-ph.IM↗

arXiv2014

CLASH-VLT: The stellar mass function and stellar mass density profile of the z=0.44 cluster of galaxies MACS J1206.2-0847

Context. The study of the galaxy stellar mass function (SMF) in relation to the galaxy environment and the stellar mass density profile, rho(r), is a powerful tool to constrain models of galaxy evolution. Aims. We determine the SMF of the z=0.44 cluster of galaxies MACS J1206.2-0847 separately for passive and star-forming (SF) galaxies, in different regions of the cluster, from the center out to approximately 2 virial radii. We also determine rho(r) to compare it to the number density and total mass density profiles. Methods. We use the dataset from the CLASH-VLT survey. Stellar masses are obtained by SED fitting on 5-band photometric data obtained at the Subaru telescope. We identify 1363 cluster members down to a stellar mass of 10^9.5 Msolar. Results. The whole cluster SMF is well fitted by a double Schechter function. The SMFs of cluster SF and passive galaxies are statistically different. The SMF of the SF cluster galaxies does not depend on the environment. The SMF of the passive population has a significantly smaller slope (in absolute value) in the innermost (<0.50 Mpc), highest density cluster region, than in more external, lower density regions. The number ratio of giant/subgiant galaxies is maximum in this innermost region and minimum in the adjacent region, but then gently increases again toward the cluster outskirts. This is also reflected in a decreasing radial trend of the average stellar mass per cluster galaxy. On the other hand, the stellar mass fraction, i.e., the ratio of stellar to total cluster mass, does not show any significant radial trend. Conclusions. Our results appear consistent with a scenario in which SF galaxies evolve into passive galaxies due to density-dependent environmental processes, and eventually get destroyed very near the cluster center to become part of a diffuse intracluster medium.

astro-ph.GA↗

arXiv2014

A catalogue of photometric redshifts for the SDSS-DR9 galaxies

Accurate photometric redshifts for large samples of galaxies are among the main products of modern multiband digital surveys. Over the last decade, the Sloan Digital Sky Survey (SDSS) has become a sort of benchmark against which to test the various methods. We present an application of a new method to the estimation of photometric redshifts for the galaxies in the SDSS Data Release 9 (SDSS-DR9). Photometric redshifts for more than 143 million galaxies were produced and made available at the URL: http://dame.dsf.unina.it/catalog/DR9PHOTOZ/. The MLPQNA (Multi Layer Perceptron with Quasi Newton Algorithm) model provided within the framework of the DAMEWARE (DAta Mining and Exploration Web Application REsource) is an interpolative method derived from machine learning models. The obtained redshifts have an overall uncertainty of sigma=0.023 with a very small average bias of about 3x10^-5, and a fraction of catastrophic outliers of about 5%. This result is slightly better than what was already available in the literature, particularly in terms of the smaller fraction of catastrophic outliers.

astro-ph.IM↗

arXiv2014

Intra Cluster Light properties in the CLASH-VLT cluster MACS J1206.2-0847

We aim at constraining the assembly history of clusters by studying the intra cluster light (ICL) properties, estimating its contribution to the fraction of baryons in stars, f*, and understanding possible systematics/bias using different ICL detection techniques. We developed an automated method, GALtoICL, based on the software GALAPAGOS to obtain a refined version of typical BCG+ICL maps. We applied this method to our test case MACS J1206.2-0847, a massive cluster located at z=0.44, that is part of the CLASH sample. Using deep multi-band SUBARU images, we extracted the surface brightness (SB) profile of the BCG+ICL and we studied the ICL morphology, color, and contribution to f* out to R500. We repeated the same analysis using a different definition of the ICL, SBlimit method, i.e. a SB cut-off level, to compare the results. The most peculiar feature of the ICL in MACS1206 is its asymmetric radial distribution, with an excess in the SE direction and extending towards the 2nd brightest cluster galaxy which is a Post Starburst galaxy. This suggests an interaction between the BCG and this galaxy that dates back to t <= 1.5 Gyr. The BCG+ICL stellar content is 8% of M_(*,500) and the (de-) projected baryon fraction in stars is f*=0.0177 (0.0116), in excellent agreement with recent results. The SBlimit method provides systematically higher ICL fractions and this effect is larger at lower SB limits. This is due to the light from the outer envelopes of member galaxies that contaminate the ICL. Though more time consuming, the GALtoICL method provides safer ICL detections that are almost free of this contamination. This is one of the few ICL study at redshift z > 0.3. At completion, the CLASH/VLT program will allow us to extend this analysis to a statistically significant cluster sample spanning a wide redshift range: 0.2<z<0.6.

astro-ph.CO↗

arXiv2013

CLASH-VLT: The mass, velocity-anisotropy, and pseudo-phase-space density profiles of the z=0.44 galaxy cluster MACS 1206.2-0847

We use an unprecedented data-set of about 600 redshifts for cluster members, obtained as part of a VLT/VIMOS large programme, to constrain the mass profile of the z=0.44 cluster MACS J1206.2-0847 over the radial range 0-5 Mpc (0-2.5 virial radii) using the MAMPOSSt and Caustic methods. We then add external constraints from our previous gravitational lensing analysis. We invert the Jeans equation to obtain the velocity-anisotropy profiles of cluster members. With the mass-density and velocity-anisotropy profiles we then obtain the first determination of a cluster pseudo-phase-space density profile. The kinematics and lensing determinations of the cluster mass profile are in excellent agreement. This is very well fitted by a NFW model with mass M200=(1.4 +- 0.2) 10^15 Msun and concentration c200=6 +- 1, only slightly higher than theoretical expectations. Other mass profile models also provide acceptable fits to our data, of (slightly) lower (Burkert, Hernquist, and Softened Isothermal Sphere) or comparable (Einasto) quality than NFW. The velocity anisotropy profiles of the passive and star-forming cluster members are similar, close to isotropic near the center and increasingly radial outside. Passive cluster members follow extremely well the theoretical expectations for the pseudo-phase-space density profile and the relation between the slope of the mass-density profile and the velocity anisotropy. Star-forming cluster members show marginal deviations from theoretical expectations. This is the most accurate determination of a cluster mass profile out to a radius of 5 Mpc, and the only determination of the velocity-anisotropy and pseudo-phase-space density profiles of both passive and star-forming galaxies for an individual cluster [abridged]

astro-ph.CO↗

arXiv2013

Photometric redshifts for Quasars in multi band Surveys

MLPQNA stands for Multi Layer Perceptron with Quasi Newton Algorithm and it is a machine learning method which can be used to cope with regression and classification problems on complex and massive data sets. In this paper we give the formal description of the method and present the results of its application to the evaluation of photometric redshifts for quasars. The data set used for the experiment was obtained by merging four different surveys (SDSS, GALEX, UKIDSS and WISE), thus covering a wide range of wavelengths from the UV to the mid-infrared. The method is able i) to achieve a very high accuracy; ii) to drastically reduce the number of outliers and catastrophic objects; iii) to discriminate among parameters (or features) on the basis of their significance, so that the number of features used for training and analysis can be optimized in order to reduce both the computational demands and the effects of degeneracy. The best experiment, which makes use of a selected combination of parameters drawn from the four surveys, leads, in terms of DeltaZnorm (i.e. (zspec-zphot)/(1+zspec)), to an average of DeltaZnorm = 0.004, a standard deviation sigma = 0.069 and a Median Absolute Deviation MAD = 0.02 over the whole redshift range (i.e. zspec <= 3.6), defined by the 4-survey cross-matched spectroscopic sample. The fraction of catastrophic outliers, i.e. of objects with photo-z deviating more than 2sigma from the spectroscopic value is < 3%, leading to a sigma = 0.035 after their removal, over the same redshift range. The method is made available to the community through the DAMEWARE web application.

astro-ph.IM↗