The Demographics of Broad-Line Quasars in the Mass-Luminosity Plane. I. Testing FWHM-Based Virial Black Hole Masses

Abstract

We jointly constrain the luminosity function (LF) and black hole mass function (BHMF) of broad-line quasars with forward Bayesian modeling in the quasar mass-luminosity plane, based on a homogeneous sample of ~ 58,000 SDSS DR7 quasars at z ~ 0.3-5. We take into account the selection effect of the sample flux limit; more importantly, we deal with the statistical scatter between true BH masses and FWHM-based single-epoch virial mass estimates, as well as potential luminosity-dependent biases of these mass estimates. The LF is tightly constrained in the regime sampled by SDSS, and makes reasonable predictions when extrapolated to ~ 3 magnitudes fainter. Downsizing is seen in the model LF. On the other hand, we find it difficult to constrain the BHMF to within a factor of a few at z>~0.7 (with MgII and CIV-based virial BH masses). This is mainly driven by the unknown luminosity-dependent bias of these mass estimators and its degeneracy with other model parameters, and secondly driven by the fact that SDSS quasars only sample the tip of the active BH population at high redshift. Nevertheless, the most likely models favor a positive luminosity-dependent bias for MgII and possibly for CIV, such that at fixed true BH mass, objects with higher-than-average luminosities have over-estimated FWHM-based virial masses. There is tentative evidence that downsizing also manifests itself in the active BHMF, and the BH mass density in broad-line quasars contributes an insignificant amount to the total BH mass density at all times. Within our model uncertainties, we do not find a strong BH mass dependence of the mean Eddington ratio; but there is evidence that the mean Eddington ratio (at fixed BH mass) increases with redshift.