The 2–10 keV unabsorbed luminosity function of AGN from the LSS, CDFS, and COSMOS surveys^⋆,⋆⋆

¹ Institute for Astronomy, Astrophysics, Space Applications and Remote Sensing (IAASARS), National Observatory of Athens, 15236 Penteli, Greece
² INAF–Osservatorio Astronomico di Bologna, via Ranzani 1, 40127 Bologna, Italy
³ Lund Observatory, Department of Astronomy and Theoretical Physics, Lund University, Box 43, 22100 Lund, Sweden
e-mail: piero@astro.lu.se
⁴ Service d’astrophysique, IRFU, CEA Saclay, 91191 Gif-sur-Yvette, France
⁵ Geneva Observatory, University of Geneva, ch. des Maillettes 51, 1290 Versoix, Switzerland
⁶ Max-Planck-Institut für extraterrestrische Physick, 85478 Garching, Germany
⁷ Department of Physics, Yale University, PO Box 208121, New Haven, CT 06520, USA
⁸ Instituto de Física de Cantabria (CSIC-UC), 39005 Santander, Spain
⁹ INAF–IASF Milano, via Bassini 15, 20133 Milano, Italy
¹⁰ ICREA and Institut de Ciències del Cosmos (ICC), Universitat de Barcelona (IEEC-UB), Martí y Franquès 1, 08028 Barcelona, Spain
¹¹ Argelander Institute for Astronomy, Bonn University, 53121 Bonn, Germany
¹² Physics Department, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece
¹³ Università di Bologna, Dipartimento di Fisica e Astronomia, via Berti Pichat 6/2, 40127 Bologna, Italy

Received: 20 July 2015
Accepted: 16 December 2015

Abstract

The XMM-Large scale structure (XMM-LSS), XMM-Cosmological evolution survey (XMM-COSMOS), and XMM-Chandra deep field south (XMM-CDFS) surveys are complementary in terms of sky coverage and depth. Together, they form a clean sample with the least possible variance in instrument effective areas and point spread function. Therefore this is one of the best samples available to determine the 2–10 keV luminosity function of active galactic nuclei (AGN) and their evolution. The samples and the relevant corrections for incompleteness are described. A total of 2887 AGN is used to build the LF in the luminosity interval 10⁴²–10⁴⁶ erg s^-1 and in the redshift interval 0.001–4. A new method to correct for absorption by considering the probability distribution for the column density conditioned on the hardness ratio is presented. The binned luminosity function and its evolution is determined with a variant of the Page-Carrera method, which is improved to include corrections for absorption and to account for the full probability distribution of photometric redshifts. Parametric models, namely a double power law with luminosity and density evolution (LADE) or luminosity-dependent density evolution (LDDE), are explored using Bayesian inference. We introduce the Watanabe-Akaike information criterion (WAIC) to compare the models and estimate their predictive power. Our data are best described by the LADE model, as hinted by the WAIC indicator. We also explore the recently proposed 15-parameter extended LDDE model and find that this extension is not supported by our data. The strength of our method is that it provides unabsorbed, non-parametric estimates, credible intervals for luminosity function parameters, and a model choice based on predictive power for future data.