Stern et al. (2012) presented a study of WISE selection of AGN in the 2 deg 2 COSMOS field, finding that a simple criterion W1-W2≥0.8 provides a highly reliable and complete AGN sample for W2<15.05, where the W1 and W2 passbands are centered at 3.4µm and 4.6µm, respectively. Here we extend this study using the larger 9 deg 2 NOAO Deep Wide-Field Survey Boötes field which also -2has considerably deeper WISE observations than the COSMOS field, and find that this simple color-cut significantly loses reliability at fainter fluxes. We define a modified selection criterion combining the W1−W2 color and the W2 magnitude to provide highly reliable or highly complete AGN samples for fainter WISE sources. In particular, we define a color-magnitude cut that finds 130±4 deg −2 AGN candidates for W2<17.11 with 90% reliability. Using the extensive UV through mid-IR broad-band photometry available in this field, we study the spectral energy distributions of WISE AGN candidates. We find that, as expected, the WISE AGN selection can identify highly obscured AGN, but that it is biased towards objects where the AGN dominates the bolometric luminosity output. We study the distribution of reddening in the AGN sample and discuss a formalism to account for sample incompleteness based on the step-wise maximum-likelihood method of Efstathiou et al. (1988). The resulting dust obscuration distributions depend strongly on AGN luminosity, consistent with the trend expected for a Simpson (2005) receding torus. At L AGN ∼ 3 × 10 44 erg s −1 , 29±7% of AGN are observed as Type 1, while at ∼ 4 × 10 45 erg s −1 the fraction is 64±13%. The distribution of obscuration values suggests that dust in the torus is present as both a diffuse medium and in optically thick clouds.

The WISE mission has unveiled a rare population of high-redshift (z = 1 − 4.6), dusty, hyperluminous galaxies, with infrared luminosities L IR > 10 13 L ⊙ , and sometimes exceeding 10 14 L ⊙ . Previous work has shown that their dust temperatures and overall far-IR spectral energy distributions (SEDs) are significantly hotter than expected to be powered by star-formation. We present here an analysis of the rest-frame optical through mid-IR SEDs for a large sample of these so-called "Hot, Dust-Obscured Galaxies" (Hot DOGs). We find that the SEDs of Hot DOGs are generally well modeled by the combination of a luminous, yet obscured AGN that dominates the rest-frame emission at λ > 1µm and the bolometric luminosity output, and a less luminous host galaxy that is responsible for the bulk of the rest optical/UV emission. Even though the stellar mass of the host galaxies may be as large as 10 11 − 10 12 M ⊙ , the AGN emission, with a range of luminosities comparable to those of the most luminous QSOs known, require that either Hot DOGs have black hole masses significantly in excess of the local relations, or that they radiate significantly above the Eddington limit, at a level at least 10 times more efficiently than z ∼ 2 QSOs. We show that, while rare, the number density of Hot DOGs is comparable to that of equally luminous but unobscured (i.e., Type 1) QSOs. This may be at odds with the trend suggested at lower luminosities for the fraction of obscured AGN to decrease with increasing luminosity. That trend may, instead, reverse at higher luminosities. Alternatively, Hot DOGs may not be the torus-obscured counterparts of the known optically selected, largely unobscured Hyper-Luminous QSOs, and may represent a new component of the galaxy evolution paradigm. Finally, we discuss the environments of Hot DOGs and statistically show that these objects are in regions as dense as those of known high-redshift proto-clusters.

The Wide-field Infrared Survey Explorer (WISE) has completed its all-sky survey in four channels at 3.4 -22 µm, detecting hundreds of millions of objects. We merge the WISE mid-infrared data with optical data from the Sloan Digital Sky Survey (SDSS) and provide a phenomenological characterization of WISE extragalactic sources. WISE is most sensitive at 3.4 µm (W 1) and least sensitive at 22 µm (W 4). The W 1 band probes massive early-type galaxies out to z ∼ > 1. This is more distant than SDSS identified early-type galaxies, consistent with the fact that 28% of 3.4 µm sources have faint or no r-band counterparts (r > 22.2). In contrast, 92 -95% of 12 µm and 22 µm sources have SDSS optical counterparts with r ≤ 22.2. WISE 3.4 µm detects 89.8% of the entire SDSS QSO catalog at arXiv:1209.2065v2 [astro-ph.CO] 19 Dec 2012 SNR W 1 >7σ, but only 18.9% at 22 µm with SNR W 4 >5σ. We show that WISE colors alone are effective in isolating stars (or local early-type galaxies), starforming galaxies and strong AGN/QSOs at z ∼ < 3. We highlight three major applications of WISE colors: (1) Selection of strong AGN/QSOs at z ≤ 3 using W 1 − W 2 > 0.8 and W 2 < 15.2 criteria, producing a better census of this population. The surface density of these strong AGN/QSO candidates is 67.5 ± 0.14 per deg 2 . (2) Selection of dust-obscured, type-2 AGN/QSO candidates. We show that WISE W 1 − W 2 > 0.8, W 2 < 15.2 combined with r − W 2 > 6 (Vega) colors can be used to identify type-2 AGN candidates. The fraction of these type-2 AGN candidates is 1/3rd of all WISE color-selected AGNs.(3) Selection of ultra-luminous infrared galaxies at z ∼ 2 with extremely red colors, r − W 4 > 14 or well-detected 22 µm sources lacking detections in the 3.4 and 4.6 µm bands. The surface density of z ∼ 2 ULIRG candidates selected with r − W 4 > 14 is 0.9 ± 0.07 per deg 2 at SNR W 4 ≥ 5 (the corresponding, lowest flux density of 2.5 mJy), which is consistent with that inferred from smaller area Spitzer surveys. Optical spectroscopy of a small number of these high-redshift ULIRG candidates confirms our selection, and reveals a possible trend that optically fainter or r−W 4 redder candidates are at higher redshifts.