2012ApJ...760...74H

We use stellar population synthesis modeling to analyze the host-galaxy properties of a sample of 33 UV-selected, narrow-lined active galactic nuclei (AGNs) at z ∼ 2-3. In order to quantify the contribution of AGN emission to host galaxy broadband spectral energy distributions (SEDs), we use the subsample of 11 AGNs with photometric coverage spanning from rest-frame UV through near-IR wavelengths. Modeling the SEDs of these objects with a linear combination of stellar population and AGN templates, we infer the effect of the AGN on derived stellar population parameters. We also estimate the typical bias in derived stellar populations for AGNs lacking rest-frame near-IR wavelength coverage, and develop a method for inferring the true host-galaxy properties. We compare AGN host-galaxy properties to those of a sample of UV-selected, star-forming non-AGNs in the same redshift range, including a subsample carefully matched in stellar mass. Although the AGNs have higher masses and star-formation rates than the full non-active sample, their stellar population properties are consistent with those of the mass-selected sample, suggesting that the presence of an AGN is not connected with the cessation of star formation activity in star-forming galaxies at z ∼ 2-3. We suggest that a correlation between M_BH and galaxy stellar mass is already in place at this epoch. Assuming a roughly constant Eddington ratio for AGNs at all stellar masses, we are unable to detect the AGNs in low-mass galaxies because they are simply too faint.