J. A. Peacock

The University of Edinburgh, Edinburgh, Scotland, United Kingdom

Are you J. A. Peacock?

Claim your profile

Publications (245)991 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: We investigate the dependence of the galaxy luminosity function on geometric environment within the Galaxy And Mass Assembly (GAMA) survey. The tidal tensor prescription, based on the Hessian of the pseudo-gravitational potential, is used to classify the cosmic web and define the geometric environments: for a given smoothing scale, we classify every position of the surveyed region, $0.04<{z}<0.26$, as either a void, a sheet, a filament or a knot. We consider how to choose appropriate thresholds in the eigenvalues of the Hessian in order to partition the galaxies approximately evenly between environments. We find a significant variation in the luminosity function of galaxies between different geometric environments; the normalisation, characterised by $\phi^{*}$ in a Schechter function fit, increases by an order of magnitude from voids to knots. The turnover magnitude, characterised by $M^*$, brightens by approximately $0.5$ mag from voids to knots. However, we show that the observed modulation can be entirely attributed to the indirect local-density dependence. We therefore find no evidence of a direct influence of the cosmic web on the galaxy luminosity function.
    12/2014;
  • http://dx.doi.org/10.1051/0004-6361/201014586. 09/2014;
  • [Show abstract] [Hide abstract]
    ABSTRACT: We report results obtained during the characterization of a commercial front-illuminated progressive scan interline transfer CCD camera. We demonstrate that the unmodified camera operates successfully in temperature and pressure conditions (-40C, 4mBar) representative of a high altitude balloon mission. We further demonstrate that the centroid of a well-sampled star can be determined to better than 2% of a pixel, even though the CCD is equipped with a microlens array. This device has been selected for use in a closed-loop star-guiding and tip-tilt correction system in the BIT-STABLE balloon mission.
    08/2014;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Our view of the low-redshift Cosmic Web has been revolutionized by galaxy redshift surveys such as 6dFGS, SDSS and 2MRS. However, the trade-off between depth and angular coverage limits a systematic three-dimensional account of the entire sky beyond the Local Volume (z<0.05). In order to reliably map the Universe to cosmologically significant depths over the full celestial sphere, one must draw on multiwavelength datasets and state-of-the-art photometric redshift techniques. We have undertaken a dedicated program of cross-matching the largest photometric all-sky surveys -- 2MASS, WISE and SuperCOSMOS -- to obtain accurate redshift estimates of millions of galaxies. The first outcome of these efforts -- the 2MASS Photometric Redshift catalog (2MPZ, Bilicki et al. 2014a) -- has been publicly released and includes almost 1 million galaxies with a mean redshift of z=0.08. Here we summarize how this catalog was constructed and how using the WISE mid-infrared sample together with SuperCOSMOS optical data allows us to push to redshift shells of z~0.2--0.3 on unprecedented angular scales. Our catalogs, with ~20 million sources in total, provide access to cosmological volumes crucial for studies of local galaxy flows (clustering dipole, bulk flow) and cross-correlations with the cosmic microwave background such as the integrated Sachs-Wolfe effect or lensing studies.
    08/2014;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We present statistically significant detections at 850um of the Lyman Break Galaxy (LBG) population at z=3, 4, and 5 using data from the Submillimetre Common User Bolometer Array 2 (SCUBA-2) Cosmology Legacy Survey (S2CLS) in the United Kingdom Infrared Deep Sky Survey Ultra Deep Survey (UKIDSS-UDS) field. We employ a stacking technique to probe beneath the survey limit to measure the average 850um flux density of LBGs at z=3, 4, and 5 with typical ultraviolet luminosities of L(1700A)~10^29 erg/s/Hz. We measure 850um flux densities of (0.25 +/- 0.03, (0.41 +/- 0.06), and (0.88 +/- 0.23) mJy respectively, and find that they contribute at most 20 per cent to the cosmic far-infrared background at 850um. Fitting an appropriate range of spectral energy distributions to the z=3, 4, and 5 LBG stacked 24-850um fluxes, we derive infrared (IR) luminosities of L(8-1000um)~3.2, 5.5, and 11.0x10^11 Lsun (corresponding to star formation rates of ~50-200 Msun/yr) respectively. We find that the evolution in the IR luminosity density of LBGs is broadly consistent with model predictions for the expected contribution of luminous IR galaxy (LIRG) to ultraluminous IR galaxy (ULIRG) type systems at these epochs. We also see a strong positive correlation between stellar mass and IR luminosity. Our data are consistent with the main sequence of star formation showing little or no evolution from z=3 to 5. We have also confirmed that, for a fixed mass, the reddest LBGs (UV slope Beta -> 0) are indeed redder due to dust extinction, with SFR(IR)/SFR(UV) increasing by approximately an order of magnitude over -2<Beta<0 such that SFR(IR)/SFR(UV)~20 for the reddest LBGs. Furthermore, the most massive LBGs also tend to have higher obscured-to-unobscured ratio, hinting at a variation in the obscuration properties across the mass range.
    07/2014;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The characterisation of cosmic voids gives unique information about the large-scale distribution of galaxies, their evolution and the cosmological model. We identify and characterise cosmic voids in the VIMOS Public Extragalactic Redshift Survey (VIPERS) at redshift 0.55 < z < 0.9. A new void search method is developed based upon the identification of empty spheres that fit between galaxies. The method can be used to characterise the cosmic voids despite the presence of complex survey boundaries and internal gaps. We investigate the impact of systematic observational effects and validate the method against mock catalogues. We measure the void size distribution and the void-galaxy correlation function. We construct a catalogue of voids in VIPERS. The distribution of voids is found to agree well with the distribution of voids found in mock catalogues. The void-galaxy correlation function shows indications of outflow velocity from the voids.
    07/2014;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We use the first release of the VImos Public Extragalactic Redshift Survey of galaxies (VIPERS) of ~50,000 objects to measure the biasing relation between galaxies and mass in the redshift range z=[0.5,1.1]. We estimate the 1-point distribution function [PDF] of VIPERS galaxies from counts in cells and, assuming a model for the mass PDF, we infer their mean bias relation. The reconstruction of the bias relation from PDFs is performed through a novel method that accounts for Poisson noise, redshift distortions, inhomogeneous sky coverage and other selection effects. With this procedure we constrain galaxy bias and its deviations from linearity down to scales as small as 4 Mpc/h and out to z=1.1. We detect small (~3%) but significant deviations from linear bias. The mean biasing function is close to linear in regions above the mean density. The mean slope of the biasing relation is a proxy to the linear bias parameter. It increases both with luminosity, in agreement with results of previous analyses, and with redshift. However, we detect a strong bias evolution only for z>0.9 in agreement with some, but not all, previous studies. We also detected a significant increase of the bias with the scale, from 4 to 8 Mpc/h, now seen for the first time out to z=1. The amplitude of nonlinearity depends on redshift, luminosity and on scales but no clear trend is detected. Thanks to the large cosmic volume probed by VIPERS we find that the mismatch between the previous estimates of bias at z~1 from zCOSMOS and VVDS-Deep galaxy samples is fully accounted for by cosmic variance. The results of our work confirm the importance of going beyond the over-simplistic linear bias hypothesis showing that non-linearities can be accurately measured through the applications of the appropriate statistical tools to existing datasets like VIPERS.
    06/2014;
  • Source
    David Alonso, Elizabeth Eardley, John A. Peacock
    [Show abstract] [Hide abstract]
    ABSTRACT: We investigate the dependence of the mass function of dark-matter haloes on their environment within the cosmic web of large-scale structure. A dependence of the halo mass function on large-scale mean density is a standard element of cosmological theory, allowing mass-dependent biasing to be understood via the peak-background split. On the assumption of a Gaussian density field, this analysis can be extended to ask how the mass function depends on the geometrical environment: clusters, filaments, sheets and voids, as classified via the tidal tensor (the Hessian matrix of the gravitational potential). In linear theory, the problem can be solved exactly, and the result is attractively simple: the conditional mass function has no explicit dependence on the local tidal field, and is a function only of the local density on the filtering scale used to define the tidal tensor. There is nevertheless a strong implicit predicted dependence on geometrical environment, because the local density couples statistically to the derivatives of the potential. We compute the predictions of this model and study the limits of their validity by comparing them to results deduced empirically from N-body simulations. For sufficiently large filtering sizes, the agreement is good; but there are deviations from the Gaussian prediction at high nonlinearities. We discuss how to obtain improved predictions in this regime, using the `effective-universe' approach.
    06/2014;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We present a maximum-likelihood weak lensing analysis of the mass distribution in optically selected spectroscopic Galaxy Groups (G3Cv1) in the Galaxy And Mass Assembly (GAMA) survey, using background Sloan Digital Sky Survey (SDSS) photometric galaxies. The scaling of halo mass, $M_h$, with various group observables is investigated. Our main results are: 1) the measured relations of halo mass with group luminosity, virial volume and central galaxy stellar mass, $M_\star$, agree very well with predictions from mock group catalogues constructed from a GALFORM semi-analytical galaxy formation model implemented in the Millennim $\Lambda$CDM N-body simulation; 2) the measured relations of halo mass with velocity dispersion and projected half-abundance radius show weak tension with mock predictions, hinting at problems in the mock galaxy dynamics and their small scale distribution; 3) the median $M_h|M_\star$ measured from weak lensing depends more sensitively on the dispersion in $M_\star$ at fixed $M_h$ than it does on the median $M_\star|M_h$. Our measurements suggest an intrinsic dispersion of $\sigma_{\log(M_\star)}\sim 0.15$; 4) Comparing our mass estimates with those in the catalogue, we find that the G3Cv1 mass can give biased results when used to select subsets of the group sample. Of the various new halo mass estimators that we calibrate using our weak lensing measurements, group luminosity is the best single-proxy estimator of group mass.
    04/2014;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We explore the evolution of the Colour-Magnitude Relation (CMR) and Luminosity Function (LF) at 0.4<z<1.3 from the VIMOS Public Extragalactic Redshift Survey (VIPERS) using ~45,000 galaxies with precise spectroscopic redshifts down to i'_AB<22.5 over ~10.32 deg^2 in two fields. From z=0.5 to z=1.3 the LF and CMR are well defined for different galaxy populations and M^*_B evolves by ~1.04(1.09)+/-0.06(0.10) mag for the total (red) galaxy sample. We compare different criteria for selecting early-type galaxies (ETGs): (1) fixed cut in rest-frame (U-V) colours, (2) evolving cut in (U-V) colours, (3) rest-frame (NUV-r')-(r'-K) colour selection, and (4) SED classification. Regardless of the method we measure a consistent evolution of the red-sequence (RS). Between 0.4<z<1.3 we find a moderate evolution of the RS intercept of Delta(U-V)=0.28+/-0.14 mag, favouring exponentially declining star formation (SF) histories with SF truncation at 1.7<=z<=2.3. Together with the rise in the ETG number density by 0.64 dex since z=1, this suggests a rapid build-up of massive galaxies (M>10^11 M_sun) and expeditious RS formation over a short period of ~1.5 Gyr starting before z=1. This is supported by the detection of ongoing SF in ETGs at 0.9<z<1.0, in contrast with the quiescent red stellar populations of ETGs at 0.5<z<0.6. There is an increase in the observed CMR scatter with redshift, two times larger than in galaxy clusters and at variance with theoretical models. We discuss possible physical mechanisms that support the observed evolution of the red galaxy population. Our findings point out that massive galaxies have experienced a sharp SF quenching at z~1 with only limited additional merging. In contrast, less-massive galaxies experience a mix of SF truncation and minor mergers which build-up the low- and intermediate-mass end of the CMR.
    01/2014;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: [Abridged] Non-uniform sampling and gaps in sky coverage are common in galaxy redshift surveys but these effects can degrade galaxy counts-in-cells and density estimates. We carry out a comparison of methods that aim to fill the gaps to correct for the systematic effects. Our study is motivated by the analysis of the VIMOS Extragalactic Redshift Survey (VIPERS), a flux-limited survey (i<22.5) based on one-pass observations with VIMOS, with gaps covering 25% of the surveyed area and a mean sampling rate of 35%. Our findings are applicable to other surveys with similar observing strategies. We compare 1) two algorithms based on photometric redshift, that assign redshifts to galaxies based on the spectroscopic redshifts of the nearest neighbours, 2) two Bayesian methods, the Wiener filter and the Poisson-Lognormal filter. Using galaxy mock catalogues we quantify the accuracy of the counts-in-cells measurements on scales of R=5 and 8 Mpc/h after applying each of these methods. We also study how they perform to account for spectroscopic redshift error and inhomogeneous and sparse sampling rate. We find that in VIPERS the errors in counts-in-cells measurements on R<10 Mpc/h scales are dominated by the sparseness of the sample. All methods underpredict by 20-35% the counts at high densities. This systematic bias is of the same order as random errors. No method outperforms the others. Random and systematic errors decrease for larger cells. We show that it is possible to separate the lowest and highest densities on scales of 5 Mpc/h at redshifts 0.5<z<1.1, over a large volume such as in VIPERS survey. This is vital for the characterisation of cosmic variance and rare populations (e.g, brightest galaxies) in environmental studies at these redshifts.
    01/2014;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The Galaxy And Mass Assembly (GAMA) survey furnishes a deep redshift catalog that, when combined with the Wide-field Infrared Explorer ($WISE$), allows us to explore for the first time the mid-infrared properties of $> 110, 000$ galaxies over 120 deg$^2$ to $z\simeq 0.5$. In this paper we detail the procedure for producing the matched GAMA-$WISE$ catalog for the G12 and G15 fields, in particular characterising and measuring resolved sources; the complete catalogs for all three GAMA equatorial fields will be made available through the GAMA public releases. The wealth of multiwavelength photometry and optical spectroscopy allows us to explore empirical relations between optically determined stellar mass (derived from synthetic stellar population models) and 3.4micron and 4.6micron WISE measurements. Similarly dust-corrected Halpha-derived star formation rates can be compared to 12micron and 22micron luminosities to quantify correlations that can be applied to large samples to $z<0.5$. To illustrate the applications of these relations, we use the 12micron star formation prescription to investigate the behavior of specific star formation within the GAMA-WISE sample and underscore the ability of WISE to detect star-forming systems at $z\sim0.5$. Within galaxy groups (determined by a sophisticated friends-of-friends scheme), results suggest that galaxies with a neighbor within 100$\,h^{-1} $kpc have, on average, lower specific star formation rates than typical GAMA galaxies with the same stellar mass.
    01/2014; 782(2).
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Key cosmological applications require the three-dimensional galaxy distribution on the entire celestial sphere. These include measuring the gravitational pull on the Local Group, estimating the large-scale bulk flow and testing the Copernican principle. However, the largest all-sky redshift surveys -- the 2MRS and IRAS PSCz -- have median redshifts of only z=0.03 and sample the very local Universe. There exist all-sky galaxy catalogs reaching much deeper -- SuperCOSMOS in the optical, 2MASS in the near-IR and WISE in the mid-IR -- but these lack complete redshift information. At present, the only rapid way towards larger 3D catalogs covering the whole sky is through photometric redshift techniques. In this paper we present the 2MASS Photometric Redshift catalog (2MPZ) containing 1 million galaxies, constructed by cross-matching 2MASS XSC, WISE and SuperCOSMOS all-sky samples and employing the artificial neural network approach (the ANNz algorithm), trained on such redshift surveys as SDSS, 6dFGS and 2dFGRS. The derived photometric redshifts have errors nearly independent of distance, with an all-sky accuracy of sigma_z=0.015 and a very small percentage of outliers. In this way, we obtain redshift estimates with a typical precision of 12% for all the 2MASS XSC galaxies that lack spectroscopy. In addition, we have made an early effort towards probing the entire 3D sky beyond 2MASS, by pairing up WISE with SuperCOSMOS and training the ANNz on GAMA redshift data reaching currently to z_med~0.2. This has yielded photo-z accuracies comparable to those in the 2MPZ. These all-sky photo-$z$ catalogs, with a median z~0.1 for the 2MPZ, and significantly deeper for future WISE-based samples, will be the largest and most complete of their kind for the foreseeable future.
    The Astrophysical Journal Supplement Series 11/2013; 210(1). · 16.24 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We use a sample of about 22,000 galaxies at $0.65<z<1.2$ from the VIPERS PDR-1 catalogue, to constrain the cosmological model through a measurement of the galaxy {\it clustering ratio} $\eta_{g,R}$. This statistic has favourable properties, being defined as the ratio of two quantities characterizing the smoothed density field in spheres of given radius $R$: the value of its correlation function on a multiple of this scale, $\xi(nR)$, and its variance $\sigma^2(R)$. For sufficiently large values of $R$, this is a universal number, capturing 2-point clustering information independently of the linear bias and linear redshift-space distortions of the specific galaxy tracers. In this paper we discuss in detail how to extend the application of $\eta_{g,R}$ to quasi-linear scales and how to control and remove observational selection effects which are typical of redshift surveys as VIPERS. We verify the accuracy and efficiency of these procedures using mock catalogues that match the survey selection process. These results evidence the robustness of $\eta_{g,R}$ to non-linearities and observational effects, which is related to its very definition as a ratio of quantities that are similarly affected. We measure $\Omega_{m,0}=0.270_{-0.025}^{+0.029}$. In addition to the great precision achieved on our estimation of $\Omega_m$ using VIPERS PDR-1, this result is remarkable because it appears to be in good agreement with a recent estimate $z\simeq 0.3$, obtained applying the same technique to the SDSS-LRG catalogue. It, therefore, suports the robustness of the present analysis. Moreover, the combination of these two measurements at $z\sim 0.3$ and $z\sim 0.9$ provides us with a very precise estimate $\Omega_{m,0}=0.274\pm0.017$ which highlights the great consistency between our estimation and other cosmological probes such as BAOs, CMB and Supernovae.
    10/2013;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We present the first Public Data Release (PDR-1) of the VIMOS Public Extragalactic Survey (VIPERS). It comprises 57 204 spectroscopic measurements together with all additional information necessary for optimal scientific exploitation of the data, in particular the associated photometric measurements and quantification of the photometric and survey completeness. VIPERS is an ESO Large Programme designed to build a spectroscopic sample of ' 100 000 galaxies with iAB < 22.5 and 0.5 < z < 1.5 with high sampling rate (~45%). The survey spectroscopic targets are selected from the CFHTLS-Wide five-band catalogues in the W1 and W4 fields. The final survey will cover a total area of nearly 24 deg2, for a total comoving volume between z = 0.5 and 1.2 of ~4x10^7 h^(-3)Mpc^3 and a median galaxy redshift of z~0.8. The release presented in this paper includes data from virtually the entire W4 field and nearly half of the W1 area, thus representing 64% of the final dataset. We provide a detailed description of sample selection, observations and data reduction procedures; we summarise the global properties of the spectroscopic catalogue and explain the associated data products and their use, and provide all the details for accessing the data through the survey database (http://vipers.inaf.it) where all information can be queried interactively.
    10/2013;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We present the first application of a "multiple-tracer" redshift-space distortion (RSD) analysis to an observational galaxy sample, using data from the Galaxy and Mass Assembly survey (GAMA). Our dataset is an r < 19.8 magnitude-limited sample of 178,579 galaxies covering redshift interval z < 0.5 and area 180 deg^2. We obtain improvements of 10-20% in measurements of the gravitational growth rate compared to a single-tracer analysis, deriving from the correlated sample variance imprinted in the distributions of the overlapping galaxy populations. We present new expressions for the covariances between the auto-power and cross-power spectra of galaxy samples that are valid for a general survey selection function and weighting scheme. We find no evidence for a systematic dependence of the measured growth rate on the galaxy tracer used, justifying the RSD modelling assumptions, and validate our results using mock catalogues from N-body simulations. For multiple tracers selected by galaxy colour, we measure normalized growth rates in two independent redshift bins f*sigma_8(z=0.18) = 0.36 +/- 0.09 and f*sigma_8(z=0.38) = 0.44 +/- 0.06, in agreement with standard GR gravity and other galaxy surveys at similar redshifts.
    Monthly Notices of the Royal Astronomical Society 09/2013; 436(4). · 5.52 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We measure the evolution of the galaxy stellar mass function from z=1.3 to z=0.5 using the first 53,608 redshifts of the ongoing VIMOS Public Extragalactic Survey (VIPERS). We estimate the galaxy stellar mass function at several epochs discussing in detail the amount of cosmic variance affecting our estimate. We find that Poisson noise and cosmic variance of the galaxy mass function in the VIPERS survey are comparable with the statistical uncertainties of large surveys in the local universe. VIPERS data allow us to determine with unprecedented accuracy the high-mass tail of the galaxy stellar mass function, which includes a significant number of galaxies that are usually too rare to detect with any of the past spectroscopic surveys. At the epochs sampled by VIPERS, massive galaxies had already assembled most of their stellar mass. We apply a photometric classification in the (U-V) rest-frame colour to compute the mass function of blue and red galaxies, finding evidence for the evolution of their contribution to the total number density budget: the transition mass above which red galaxies dominate is found to be about 10^10.4 M_sun at z=0.55 and evolves proportionally to (1+z)^3. We are able to trace separately the evolution of the number density of blue and red galaxies with masses above 10^11.4 M_sun, in a mass range barely studied in previous work. We find that for such large masses, red galaxies show a milder evolution with redshift, when compared to objects at lower masses. At the same time, we detect a population of similarly massive blue galaxies, which are no longer detectable below z=0.7. These results show the improved statistical power of VIPERS data, and give initial promising indications of mass-dependent quenching of galaxies at z~1. [Abridged]
    03/2013;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We present in this paper the general real- and redshift-space clustering properties of galaxies as measured in the first data release of the VIPERS survey. VIPERS is a large redshift survey designed to probe the distant Universe and its large-scale structure at 0.5 < z < 1.2. We describe in this analysis the global properties of the sample and discuss the survey completeness and associated corrections. This sample allows us to measure the galaxy clustering with an unprecedented accuracy at these redshifts. From the redshift-space distortions observed in the galaxy clustering pattern we provide a first measurement of the growth rate of structure at z = 0.8: f\sigma_8 = 0.47 +/- 0.08. This is completely consistent with the predictions of standard cosmological models based on Einstein gravity, although this measurement alone does not discriminate between different gravity models.
    03/2013;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We investigate the dependence of galaxy clustering on luminosity and stellar mass in the redshift range 0.5<z<1.1, using the first ~55000 redshifts from the VIMOS Public Extragalactic Redshift Survey (VIPERS). We measured the redshift-space two-point correlation functions (2PCF), and the projected correlation function, in samples covering different ranges of B-band absolute magnitudes and stellar masses. We considered both threshold and binned galaxy samples, with median B-band absolute magnitudes -21.6<MB-5log(h)<-19.5 and median stellar masses 9.8<log(M*[Msun/h^2])<10.7. We assessed the real-space clustering in the data from the projected correlation function, which we model as a power law in the range 0.2<r_p[Mpc/h]<20. Finally, we estimated the galaxy bias as a function of luminosity, stellar mass, and redshift, assuming a flat LCDM model to derive the dark matter 2PCF. We provide the best-fit parameters of the power-law model assumed for the real-space 2PCF -- the correlation length and the slope -- as well as the linear bias parameter, as a function of the B-band absolute magnitude, stellar mass, and redshift. We confirm and provide the tightest constraints on the dependence of clustering on luminosity at 0.5<z<1.1. We prove the complexity of comparing the clustering dependence on stellar mass from samples that are originally flux-limited and discuss the possible origin of the observed discrepancies. Overall, our measurements provide stronger constraints on galaxy formation models, which are now required to match, in addition to local observations, the clustering evolution measured by VIPERS galaxies between z=0.5 and z=1.1 for a broad range of luminosities and stellar masses.
    03/2013;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The aim of this work is to develop a comprehensive method for classifying sources in large sky surveys and we apply the techniques to the VIMOS Public Extragalactic Redshift Survey (VIPERS). Using the optical (u*, g', r', i') and NIR data (z', Ks), we develop a classifier, based on broad-band photometry, for identifying stars, AGNs and galaxies improving the purity of the VIPERS sample. Support Vector Machine (SVM) supervised learning algorithms allow the automatic classification of objects into two or more classes based on a multidimensional parameter space. In this work, we tailored the SVM for classifying stars, AGNs and galaxies, and applied this classification to the VIPERS data. We train the SVM using spectroscopically confirmed sources from the VIPERS and VVDS surveys. We tested two SVM classifiers and concluded that including NIR data can significantly improve the efficiency of the classifier. The self-check of the best optical + NIR classifier has shown a 97% accuracy in the classification of galaxies, 97 for stars, and 95 for AGNs in the 5-dimensional colour space. In the test on VIPERS sources with 99% redshift confidence, the classifier gives an accuracy equal to 94% for galaxies, 93% for stars, and 82% for AGNs. The method was applied to sources with low quality spectra to verify their classification, and thus increasing the security of measurements for almost 4 900 objects. We conclude that the SVM algorithm trained on a carefully selected sample of galaxies, AGNs, and stars outperforms simple colour-colour selection methods, and can be regarded as a very efficient classification method particularly suitable for modern large surveys.
    03/2013;

Publication Stats

15k Citations
991.00 Total Impact Points

Institutions

  • 1988–2014
    • The University of Edinburgh
      • Institute for Astronomy (IfA)
      Edinburgh, Scotland, United Kingdom
  • 2012
    • Durham University
      • Department of Physics
      Durham, England, United Kingdom
  • 2011
    • The Astronomical Observatory of Brera
      Merate, Lombardy, Italy
  • 2009
    • University of St Andrews
      • School of Physics and Astronomy
      Saint Andrews, Scotland, United Kingdom
  • 1986–2009
    • The Royal Observatory, Edinburgh
      Edinburgh, Scotland, United Kingdom
  • 2007
    • Space Telescope Science Institute
      Baltimore, Maryland, United States
  • 2004
    • Queen's University
      Kingston, Ontario, Canada
    • University of Bristol
      Bristol, England, United Kingdom
    • Middle East Technical University
      Engüri, Ankara, Turkey
    • University College London
      • Department of Physics and Astronomy
      London, ENG, United Kingdom
  • 2003
    • University of California, Berkeley
      • Department of Astronomy
      Berkeley, CA, United States
  • 2002
    • Princeton University
      • Department of Astrophysical Sciences
      Princeton, New Jersey, United States
    • University of Cambridge
      • Institute of Astronomy
      Cambridge, ENG, United Kingdom
  • 1999
    • Hebrew University of Jerusalem
      Yerushalayim, Jerusalem District, Israel
  • 1997
    • The University of Western Ontario
      • Department of Physics and Astronomy
      London, Ontario, Canada
  • 1989
    • French National Centre for Scientific Research
      • Institut d'astrophysique spatiale (IAS)
      Paris, Ile-de-France, France