1 Overview

Emission from active galactic nuclei (AGNs) is generally explained in terms of accretion onto a supermassive black hole in the center of the galaxy. It is common belief that all galaxies go through an active phase during their lives, and several empirical laws relate physical properties of the black hole to properties of the host galaxy. These empirical laws constitute evidence for the existence of a strong feedback between galaxy evolution and nuclear black hole formation. As a consequence, in order to understand galaxy evolution, it is crucial that we gain a deeper knowledge of black hole formation and evolution.

The search for AGNs has been based on a wide range of selection techniques over the last decades (e.g., [1, 3, 9, 13]), but none of them is completely bias-free. Provided the availability of multiepoch data, a way to identify AGNs is based on the detection of their variability, which is usually attributed to instabilities in the accretion disk and changes in the accretion rate (e.g., [12]). In the work that we illustrate here, we select AGNs in the COSMOS field on the basis of their optical variability; for details, see [7]. A complementary analysis of optically variable AGNs in the Chandra Deep Field South is presented in the work by Falocco et al. in this same volume.

Our analysis is based on data from the VLT Survey Telescope (VST; see [5]), located at Cerro Paranal Observatory, Chile. The analysis of optical variability in the COSMOS field is based on images from the Supernova Diversity And Rate Evolution (SUDARE) survey (see [2] for details): the dataset consists of multiple 1 × 1 deg2 images of the cosmos field spread over a time baseline of ≈ 5 months.

2 The Sample of AGN Candidates

Following the approach described in [14], we derived from the light curve of each source i an average magnitude \(\overline{\mbox{ mag}}_{i}\) and the corresponding r.m.s. deviation \(\sigma _{i}\); hence, in order to obtain a sample of optically variable sources, we computed the running average of the r.m.s. deviation \(\langle \sigma _{i}^{ltc}\rangle\) and its own r.m.s. deviation r.m.s.\(_{\langle \sigma _{i}^{ltc}\rangle }\) over a 0. 5 mag wide bin, and then defined the variability threshold as \(\sigma _{i}^{ltc} \geq \langle \sigma _{i}^{ltc}\rangle + 3 \times \mbox{ r.m.s.}_{\langle \sigma _{i}^{ltc}\rangle }\). Our analysis is restricted to the objects detected in at least 20 % of the epochs and with r(AB) < 23 mag. We visually inspected each of the optically variable sources and the corresponding light curves; after removing all the candidates whose variability is likely spurious (because of close neighbors, defective pixels, etc.), the final sample consists of 83 sources.

In order to investigate the nature and properties of our AGN candidates, we made use of data from multiwavelength COSMOS catalogs available from other observatories.

Spectroscopy is widely used to identify AGNs, since their spectra are characterized by prominent emission lines and are generally broadband, with a continuum emission of approximately the same energy per decade in the frequency range 1013–1020 Hz [8]. Another reliable indicator of the active nature of a galaxy is its X-ray emission, especially when it is characterized by variability. 76 % of our sources have an X-ray counterpart in the catalogs from Chandra X-ray Observatory [6] and XMM-Newton [4], and can be classified as AGNs on the sole basis of their X-ray properties. All of them have an X-ray luminosity (both soft and hard) \(L_{X} > 10^{42}\) erg s−1: since this is generally assumed to be an upper limit for the X-ray luminosity of non-active galaxies (e.g., [3]), it constitutes evidence for AGN activity.

A widely used diagnostic for AGN identification compares their X-ray and optical luminosities (e.g., [10]). On a diagram like the one shown in Fig. 42.1, AGNs generally place themselves in the region defined by − 1 ≤ X/O ≤ 1 (e.g., [11]); all the sources in our sample of AGN candidates with an X-ray counterpart do lie in that area if we take into account hard X-ray fluxes, while only two of them lie below the defined region if we refer to soft X-ray fluxes. It is worth mentioning that all but one of the VST sources with an X-ray counterpart are also classified as AGNs after their spectra ([4, 6]).

Fig. 42.1
figure 1

Hard (2–10 keV) X-ray vs. optical flux. Larger red dots: AGN candidates with an X-ray counterpart from X-ray catalogs (Chandra or XMM); smaller grey dots: reference population of the X-ray sources in the Chandra catalog having a VST counterpart and r(AB) < 23 mag; dashed line: X/O  = 0; lower and upper solid lines: X/O \(= -1\mbox{ and X/O } = 1\), respectively

Full size image

Another class of diagnostics makes use of colors to select AGNs: on a color vs. color plane, objects define distinct regions depending on their nature, because of their different spectral energy distributions. In Fig. 42.2 we show such a diagram in the optical/NIR: it is apparent that, apart from the stellar sequence and the area defined by galaxies, there is a different population of sources (65 % of the sample) which are quasar-like, i.e., point-like in shape but extended as for their colors; they are also optically variable, and are confirmed AGNs on the basis of their X-ray properties.

Fig. 42.2
figure 2

r-z vs z-k diagram. Larger dots: AGN candidates; smaller dots: reference population of the VST-COSMOS sources for which stellarity index and color information are available; crosses: AGNs confirmed by X-ray properties; triangles: SNe; boxes: ex-novo confirmed quasar-like AGNs. Stellarity index: 0 = extended, 1 = point-like

Full size image

3 Results and Conclusions

We confirm as AGNs 67 of the sources in our sample of candidates, while 12 are classified as SNe (see the work by Botticella et al. in this same volume). The purity of the sample is 81 %, while the completeness with respect to the X-ray confirmed AGNs with a VST counterpart is 15 %. We estimate that, if we had a longer (e.g., 2 year) baseline, the intrinsic variability of our sources would increase up to 50 %, returning a higher completeness. We also show (Fig. 42.3) that the subsample of VST sources with an X-ray counterpart is characterized by an average optical variability which is higher than the average optical variability of the whole VST sample; this means that, although 85 % of the X-Ray selected AGNs lie below our variability threshold, on average they are more variable than the rest of the optical sample, and thus could be detected using a lower threshold/longer baseline.

Fig. 42.3
figure 3

Light curve r.m.s. as a function of magnitude. Black symbols: X-ray emitters that are confirmed AGNs; grey dots: non-variable sources in the VST complete sample. Thin dashed red line and thick dashed blue line: running average of the r.m.s. deviation of the complete sample and of the subsample of X-ray emitters, respectively; solid line: variability threshold

Full size image

The selection techniques based on optical variability allow us to identify AGNs in wide sky areas with a ground-based telescope; this will prove of great importance in the scenario of time-domain astronomy, whose relevance has been growing exponentially in the last years and is certainly going to grow even more in the near future.