0810.1709/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2: \usepackage{lscape}
3: \usepackage{color}
4: 
5: 
6: \shorttitle{Characterization of AGN and their hosts.}
7: \shortauthors{Ramos Almeida et al.}
8: 
9: 
10: \begin{document}
11: 
12: 
13: \title{Characterization of AGN and their hosts in the Extended Groth Strip: a multiwavelength analysis.}
14: 
15: 
16: 
17: \author{C. Ramos Almeida\altaffilmark{1}, J.M. Rodr\'\i guez Espinosa\altaffilmark{1},
18: G. Barro\altaffilmark{2}, J. Gallego\altaffilmark{2}, and P. G. P\'{e}rez-Gonz\'{a}lez\altaffilmark{2,3}}
19: 
20: 
21: \altaffiltext{1}{Instituto de Astrof\'\i sica de Canarias (IAC), 
22:               C/V\'\i a L\'{a}ctea, s/n, E-38205, La Laguna, Tenerife, Spain.
23: cra@iac.es and jre@iac.es}
24: 
25: \altaffiltext{2}{Departamento de Astrof\'\i sica  y Ciencias de la Atm\'{o}sfera, Facultad de Ciencias 
26: F\'\i sicas, Universidad Complutense de Madrid, E-28040 Madrid, Spain.
27: gbc@astrax.fis.ucm.es, jgm@astrax.fis.ucm.es, and pgperez@astrax.fis.ucm.es}
28: 
29: \altaffiltext{3}{Associate Astronomer at Steward Observatory, University of Arizona, Tucson, AZ 85721.}
30: 
31: \begin{abstract}
32: 
33: We have employed a reliable technique of classification of Active Galactic Nuclei (AGN) based on 
34: the fit of well-sampled spectral energy distributions (SEDs) with a complete set of AGN and 
35: starburst galaxy templates. We have compiled ultraviolet, optical, and infrared data
36: for a sample of 116 AGN originally selected for their X-ray and mid-infrared emissions 
37: (96 with single detections and 20 with double optical counterparts).  
38: This is the most complete compilation of multiwavelength data for such a big sample of AGN in the Extended
39: Groth Strip (EGS).
40: Through these SEDs, we are able to obtain highly reliable photometric redshifts and to 
41: distinguish between pure and host-dominated AGN. 
42: For the objects with unique detection we find that they can be separated into 
43: five main groups, namely:
44: {\it Starburst-dominated AGN} (24 \% of the sample), {\it Starburst-contaminated AGN} (7 \%), 
45: {\it Type-1 AGN} (21 \%), {\it Type-2 AGN} (24 \%), and {\it Normal galaxy hosting AGN} (24 \%). 
46: We find these groups concentrated at different redshifts: {\it Type-2 AGN} and 
47: {\it Normal galaxy hosting AGN} are concentrated at low redshifts, whereas {\it Starburst-dominated AGN} 
48: and {\it Type-1 AGN} show a larger span.
49: Correlations between hard/soft X-ray and ultraviolet, optical and infrared luminosities, respectively, 
50: are reported for the first time for such a sample of AGN spanning a wide range of redshifts.
51: For the 20 objects with double detection the percentage of {\it Starburst-dominated AGN} increases up to 48\%.
52: 
53: \end{abstract}
54: 
55: \keywords{galaxies:active - galaxies:nuclei - galaxies:starburst - ultraviolet:galaxies - 
56: infrared:galaxies - X-rays:galaxies}
57: 
58: 
59: \section{Introduction}
60: 
61: The role of AGN in the formation and evolution of galaxies is still not well
62: established. It is not clear whether AGN represent episodic phenomena in the life of galaxies, are random
63: processes (given that the Supermassive Black Hole is already there), or are more fundamental. 
64: Some authors claim that AGN are
65: key in quenching the star formation bursts in their host galaxies \citep{Granato04,Springel05}.
66: It has also been shown that the mass dependence
67: of the peak star-formation epoch appears to mirror the mass dependence of Black Hole (BH) activity, 
68: as recently seen in redshift surveys of both radio-and X-ray-selected active 
69: galactic nuclei \citep{Waddington01,Hasinger03b}.
70: For these reasons, searching for signatures of AGN feedback in 
71: the properties of AGN host galaxies is one of the most promising ways of testing the role of AGN in galaxy evolution. 
72: 
73: One way of finding variations in the AGN population with redshift is to compare their SEDs 
74: defined over a broad wavelength range. The SED of an AGN can reveal the presence of the underlying 
75: central engine, together with the luminosity of the host galaxy, the reddening, and
76: the role of the star formation in the various frequency regimes. SED determination in samples of
77: AGN at different redshifts is an efficient method to search for evolutionary trends.
78: Accuracy in the photometry and a filterset spanning a broad wavelength range are required to characterize correctly  
79: different types of AGN. 
80: 
81: 
82: Multiwavelength surveys are fundamental in the study of active galactic nuclei, since these appear 
83: considerably
84: different depending on the wavelength range of consideration. The hard X-ray selection of AGN using 
85: deep observations is one of the most reliable methods of finding AGN \citep{Mushotzky04}, although a percentage of them  
86: remains undetected using this technique \citep{Peterson06}, specially the most highly obscured objects.
87: For this reason, it is important to characterize AGN at different wavelength ranges, 
88: in order to be capable of identifying them by more than one selection technique, and to distinguish between
89: the different groups of active nuclei, including those that could be contaminated, or even hidden, by starbursts. 
90: Mid-infrared surveys have been very successful in finding X-ray undetected AGN in large numbers, 
91: but in this case it 
92: is crucial to distinguish the AGN from the non-active star-forming galaxies. This can be achieved  
93: using typical mid-infrared colors of AGN \citep{Lacy04,Stern05,Alonso06,Donley08} or by combining mid-infrared 
94: and radio detections \citep{Donley05,Alonso06,Martinez-Sansigre05,Martinez-Sansigre07,Lacy07,Park08}.
95: 
96: 
97: The Extended Groth Strip ($\alpha$ = 14$^{h}$ 17$^{m}$, $\delta$ = 52$^{o}$ 30') enlarges the Hubble Space
98: Telescope Groth-Westphal strip \citep{Groth94} up to 2$^{o}$x15', having the advantage of
99: being a low extinction area in the northern sky, with low galactic and zodiacal infrared emission, and
100: good schedulability by space observatories. For these reasons, there is a huge amount of public data at
101: different wavelength ranges that only require to be compiled and cross-correlated in a consistent way.
102: %The Chandra X-ray observatory, XMM-Newton, and the Spitzer Space Telescope have greatly improved 
103: %the accuracy in the cross-identifications between data from different instruments, since the size of their 
104: %point-spread functions has decreased noticeably with respect to previous space telescopes. 
105: The overall majority of the observational work in the EGS
106: have been coordinated by the AEGIS proyect\footnote{The AEGIS project is a collaborative effort to obtain both 
107: deep imaging covering all major wavebands from X-ray to radio and optical spectroscopy over a large area 
108: of sky. http://aegis.ucolick.org/index.html} \citep{Davis07}. 
109: 
110: 
111: With the huge amount of data available for this region of the sky, we have constructed a robust AGN sample, 
112: detected in the X-rays and in the mid-infrared, intermediate in depth and area in comparison with other surveys
113: \citep{Jannuzi99,Dickinson01,Lonsdale03,Eisenhardt04,Franceschini05}.
114: The photometry has been performed over the publicy available images, in several bands, in order to compile
115: as best-sampled SEDs as possible.
116: %This paper compiles multiwavelength data from X-ray to mid-infrared for 116 AGN in the EGS.
117: The biggest advantage of our AGN sample, compared with other works, is the robustness of the photometry, 
118: performed in a consistent way among the different bands, and its multiwavelength nature: it is the most complete 
119: compilation of data for such a big sample of AGN in the EGS. This allows us to determine accurate photometric redshifts, 
120: and to distinguish clearly between the different groups of AGN.
121: We have used a comprenhensive set of AGN plus starburst templates 
122: from \citet{Polletta07}, to fit the SEDs of the galaxies in the sample, 
123: and to separate them in five different main groups. Through this classification, we can study the properties 
124: of the different types of AGN in this sample.
125: Section 2 describes the sample and the cross-matching of the multiwavelength data, Section 3 explains the  
126: technique of classification of AGN based on the fit of their spectral energy distributions, in Section 4 the 
127: results of this paper are discussed, and finally in Section 5 the main conclusions are summarized.
128: Throughout this paper  we assume an 
129: H$_{0}$=75 km s$^{-1}$ Mpc$^{-1}$ and a $\Lambda$CDM cosmology with $\Omega_{m}$=0.3 and 
130: $\Omega_{\Lambda}$=0.7.
131: 
132: 
133: 
134: \section{Sample and multi-wavelength data}
135: \label{sample}
136: 
137: 
138: 	The sample studied in this work, that comprises 116 AGN candidates, was built from the 
139: previously published X-ray catalogs found in \citet{Waskett03}, \citet{Nandra05}, and \citet{Barmby06}. 
140: These objects were originally selected by \citet{Barmby06}, both in the X-ray (Chandra and XMM-Newton) 
141: and in the mid-infrared (Spitzer). 
142: The X-ray and mid-infrared observations in the EGS are intermediate in depth and area between GOODS
143: \citep{Dickinson01}, the shallower NOAO Deep-Wide Field \citep{Jannuzi99,Eisenhardt04}, and SWIRE
144: \citep{Lonsdale03} surveys. Therefore, this region provides a valuable test of AGN properties at
145: intermediate fluxes. 
146: In addition to this, we have compiled ultraviolet, optical, and near-infrared archival data for these AGN,  
147: in order to increase the definition of their SEDs. 
148: These well-sampled SEDs allow us, first, to classify the objects in different galaxy population
149: groups and calculate their photometric redshifts, and secondly, to study the physical 
150: properties of this representative sample of AGN.
151: 
152: The Chandra data were
153: taken with ACIS-I in 2002 August \citep{Nandra05}, consisting of a 200 ks exposure with a limiting
154: full-band flux (0.5-10 keV) of 3.5x10$^{-16}$ erg cm$^{-2}$ s$^{-1}$. The XMM-Newton data were obtained
155: in 2000 July with a 56 ks exposure and with a limiting 0.5-10 keV flux of 2x10$^{-15}$ 
156: erg cm$^{-2}$ s$^{-1}$ \citep{Waskett03}. \citet{Barmby06} combined both catalogues producing a list of
157: 152 sources within the limits of the Spitzer mid-infrared observations. 
158: 
159: The Spitzer data (IRAC and MIPS) are part of the Infrared Array Camera Deep Survey,
160: taken during 2003 December and 2004 
161: June-July with 2.7 hr exposure per pointing. In the case of the MIPS 
162: data, the observations were done in 2004 January and June with a depth of 1200 s per pointing.
163: The 5$\sigma$ limiting AB magnitudes are 24.0, 24.0, 21.9, and 22.0 for the IRAC bands, and 19.1 in the
164: case of MIPS. \citet{Barmby06} finally selected 138 objects with secure detections in all four IRAC bands,
165: out of the 152 X-ray emitters. 
166: The detection of these objects in both the X-rays and the mid-infrared gives confidence in their 
167: classification as AGN. Besides, we have also checked that the values of the hard X-ray and 24 \micron~fluxes 
168: lie inside the AGN-characteristic region (see Figure 1 of \citet{Alonso04}). 
169: 
170: 
171: 
172: In addition to the previous, we have compiled near and far-ultraviolet
173: images from the GALEX GR2/GR3 data release (3$\sigma$ limiting AB magnitudes = 25 in both Far- and Near-UV filters);  
174: optical images from the CFHT Legacy Survey, T0003 worldwide release (Gwyn et al., in preparation),
175: taken with the MegaCam imager on the 4 m Canada-France Hawaii Telescope \citep{Boulade03} (5$\sigma$ limiting 
176: AB magnitudes = 26.3, 27.0, 26.5, 26.0, and 25.0 in u,g,r,i, and z bands);
177: and J and K$_{S}$ data from the version 3.3 of the Palomar-WIRC K-selected catalog of \citet{Bundy06}, 
178: (5$\sigma$ limiting Vega magnitudes = 23 and 20.6 in the J and K$_{S}$ bands). 
179: 
180: The fluxes employed in this work have been measured in a
181: compilation of publicly available imaging data, which is outlined briefly
182: in \citet{Villar08} and will be described in detail in Barro et al., in preparation   
183: (see also \citet{Perez08b}). Photometry
184: in consistent apertures was measured in all bands with available
185: imaging data following the procedure described in \citet{Perez05,Perez08}.
186: In the near-infrared, no deep J- and K$_{S}$-band imaging data were
187: available and we have used the photometric catalogs published by \citet{Bundy06}.
188: The same happens with the X-ray data, that have been drawn from the catalogs \citep{Waskett03,Nandra05,Barmby06}.
189: 
190: 
191: We have performed the cross-matching of the 138 sources between the X-ray and Spitzer data, adding  
192: ultraviolet, optical, and near-infrared data points,  
193: avoiding the false matches that \citet{Barmby06} expected in their sample. We identify these 
194: objects through their IRAC positions in our merged photometric catalog \citep{Perez05,Perez08}. 
195: The source coordinates
196: on the IRAC 3.6~\micron~images are then cross-correlated with each one of the ultraviolet, optical and
197: near-infrared catalogs using a search radius of 2.5\arcsec~, starting with the deepest images. When a source is
198: identified in one of these images, the \citet{Kron80} elliptical aperture from this reference image is taken and 
199: overlaid onto each of the other bands. The aperture employed is large enough to enclose the PSF in all the 
200: ultraviolet, optical and
201: near-infrared images (the seeing being less than 1.5\arcsec). For IRAC and MIPS, because of their large PSFs,
202: integrated magnitudes measured in small apertures (applying aperture corrections) are employed. 
203: %For the GALEX data, due
204: %to the large FWHM of the PSF (6\arcsec-7\arcsec), the \emph{MAG$_{-}$BEST} magnitude given by \emph{SEXTRACTOR} is chosen.
205: The hard and soft X-ray fluxes are obtained by cross-correlating the IRAC positions with the X-ray catalogs, 
206: using a search radius of 2\arcsec~in this case. 
207: Uncertainties of each measured flux are obtained from the sky pixel-to-pixel variations, 
208: detector readout noise, Poisson noise in the measured fluxes, errors in the World Coordinate 
209: System, and errors in the absolute photometric calibration. 
210: 
211: In some cases,  for a single IRAC source,  there are several 
212: counterparts in the ground-based images within the 2.5\arcsec~search radius. For these objects, the ground-based
213: optical/near-infrared reference image is used to determine the positions of each source separately. 
214: The IRAC images are then deconvolved using the IRAC PSFs.
215: Although the IRAC PSFs have
216: FWHMs of aproximately 2\arcsec, determination of the central position of each IRAC source can be performed more 
217: accurately, and sources can be resolved for separations $\sim$1\arcsec~from each other.   
218: IRAC fluxes are then remeasured by fixing the positions of the objects in each pair, and by scaling the
219: flux of each object in an aperture of 0.9\arcsec.  
220: For a more detailed description of the cross-matching and aperture photometry see \citet{Perez05,Perez08}. 
221: 
222: Out of the 138 sources that comprise the final sample chosen by \citet{Barmby06}, we find 96 sources
223: that have unique detections in all bands, plus other 20 objects with double detection in the ground-based 
224: images. We discard the remaining 22 objects because 21 of them show multiple (more than two) 
225: detections in the optical/near-infrared images, leading to possible source confusion, plus another object 
226: that shows a star-like SED.
227: The analysis of the data will be done first for the 96 objects that are definitely free from contamination 
228: from other sources. Nevertheless, in Section 4.5, we
229: analyse the images and photometric redshifts of those additional 20 objects with double detection.
230: 
231: 
232: 
233: \section{Spectral energy distributions and photometric redshifts of objects with unique detection.}
234: \label{spectral}
235: 
236: In order to classify the 96 spectral energy distributions and to estimate their photometric redshifts, 
237: we combine optical (u,g,r,i,z), near-infrared (J,K), and mid-infrared data 
238: (IRAC 3.6, 4.5, 5.8, 8 \micron~and MIPS 24 \micron) to build well-sampled SEDs that we then fit with 
239: the library of starburst, AGN and galaxy templates 
240: taken from \citet{Polletta07}. We make use of the photometric redshift code
241: HyperZ \citep{Bolzonella00} to perform the fits. This code determines the best photometric redshifts
242: (z$_{phot}$) by minimization of the $\chi^{2}$ derived from a comparison between the photometric 
243: SEDs and the set of template spectra, leaving the redshift as a variable. 
244: The code also takes into account the effects of dust extinction according to the selected reddening law 
245: \citep{Calzetti00}. Choosing a wide range of reddening values seems to be essential to reproduce the SEDs of 
246: high redshift galaxies \citep{Bolzonella00}.
247: According to \citet{Steidel99}, the typical E(B-V) for galaxies up
248: to z~$\sim$~4 is 0.15 mag, thus A$_{V}$~$\sim$~0.6 mag when using the Calzetti's law. 
249: The maximum A$_{V}$ allowed in our calculations is about 2 times this value, thus A$_{V}$ ranges 
250: from 0.0 to 1.2, with a step between them of 0.3. Similar values of A$_{V}$ are typically chosen 
251: in the literature \citep{Bolzonella00,Babbedge04}.
252: 
253: 
254: The chosen set of templates contains 23 SED-types, that we have arranged into the following five 
255: main groups: 
256: {\it Starburst-dominated AGN} (which includes four Starbursts and 
257: Starburst/ULIRGs templates), {\it Starburst-contaminated AGN} (three templates, namely: 
258: Starburst/ULIRG/Seyfert 1, Starburst/Seyfert 2, and Starburst/ULIRG/Seyfert 2), 
259: {\it Type-1 AGN} (three Type-1 QSO templates), {\it Type-2 AGN} 
260: (Type-2 QSO, Torus-QSO, Seyfert 1.8, and Seyfert 2 templates), and finally, {\it Normal galaxy hosting AGN}
261: (nine templates including 2, 5, and 13 Gyr ellipticals plus S0, Sa, Sb, Sc, Sd, Sdm type spirals). 
262: These templates span the range in wavelength between 0.1 and 1000
263: \micron. For a detailed explanation of their synthesis see \citet{Polletta07}. Our main 
264: interest is to classify all of our sources into these five main groups and to determine the 
265: distribution of the sources into each of these groups.
266: Notice that although all of the sources are AGN, the {\it Starburst-dominated AGN} have their SEDs 
267: dominated by the starburst emission from the optical to the mid-infrared; the 
268: {\it Normal galaxy hosting AGN} would be low-luminosity AGN embeded in an otherwise normal galaxy
269: emission; and
270: in the case of the {\it Starburst-contaminated AGN}, the emission of both the starburst 
271: or the AGN dominate depending on the wavelength we are looking at. Indeed, some of these objects 
272: show noticeably the AGN power-law beyond the near-infrared.
273: 
274: 
275: 
276: We fit data from the optical $u$ band up to the MIPS 24 \micron~band. We avoid the use of GALEX data because 
277: few galaxies have these, and because their use introduce big errors in the fits.
278: As explained in \citet{Polletta07}, including
279: mid-infrared data improves considerably the photometric redshift calculations, since some spectral types suffer
280: degeneration that is broken by the non-extinguished longer wavelengths, even if the errors in the magnitudes are 
281: larger in the mid-infrared than in the optical and near-infrared bands. 
282: 
283: Examples of HyperZ fits for each of the employed templates are shown in Figure \ref{fits}. 
284: In the {\it Type-2 AGN} pannel,
285: only three templates are shown because none of the 96 galaxies were fitted with the Torus-QSO template. 
286: The {\it Normal galaxy hosting AGN} pannel contains only one example of elliptical template 
287: (the 2 Gyr elliptical) and one
288: example of spiral (Sb). See Table \ref{photoz} to check the SED types and their corresponding group.
289: Photometric redshifts derived from the fits 
290: are reported in Table \ref{photoz}, together with the $\chi_{\nu}^{2}$ and probabilities given by 
291: HyperZ, the A$_{V}$, and the template used for the fit of each galaxy.
292: In the cases where spectroscopic redshifts are available, these are also given in Table \ref{photoz}.
293:   
294: 
295: \begin{figure}[!h]
296: \centering
297: \includegraphics[width=11cm,angle=90]{f1.eps}
298: \caption{\footnotesize{Examples of SEDs in our sample (dots) fitted with different templates from
299: \citet{Polletta07} for the five main groups considered. The legends in the bottom right of each pannel 
300: refer to the actual template within the groups from \citet{Polletta07}.
301: The data have been scaled for clarity. The X-axis corresponds to observed wavelength.
302: The fifteen galaxies represented here are labelled with the ID from Table \ref{photoz}. The 
303: percentages of objects enclosed in each group are typed in the upper left corner of each pannel.}
304: \label{fits}}
305: \end{figure}
306: 
307: A comparison between the photometric and spectroscopic redshifts for the 39 sources with publicly
308: available $z_{spec}$ from the DEEP data archive \citep{Weiner05} is shown in Figure \ref{z}. Notice that 
309: only 31 out of these 39 galaxies have reliable spectroscopic redshifts (flag = 3 or 4 in the DEEP data archive). 
310: Horizontal error bars 
311: indicating the reliability of the $z_{spec}$ are represented in Figure \ref{z}, together with vertical error bars
312: that indicate the discrepancies between the $z_{spec}$ and $z_{phot}$.
313: The dashed lines represent 20\%
314: agreement in (1+$z$). The fractional error $\Delta$z=${z_{phot}-z_{spec} \over 1+ z_{spec}}$ quantifies the 
315: number of catastrophic outliers, which are those with $\mid\Delta z\mid$ $>$ 0.2. Our measured mean
316: $\Delta$z for the 39 sources with spectroscopic redshifts is 0.05, with a $\sigma_{z}$ = 0.37, and an outlier 
317: fraction of $\sim$18\%, corresponding to seven
318: discordant objects, labelled in Figure \ref{z}. However, if we consider only the 31 objects with reliable 
319: $z_{spec}$ (flags = 3 or 4), $\Delta$z = -0.03, and $\sigma_{z}$ = 0.11, with three outliers (8\%). 
320: These results point to the goodness of our fits, and thus we rather trust our photometric redshifts 
321: better than the spectroscopic ones for the outliers indicated (all of them with $z_{spec}$ with 
322: flags = 1 or 2 in the DEEP database).
323: We nevertheless note a slight underestimation of our photometric redshifts (see Figure \ref{z}) 
324: in comparison with 
325: the spectroscopic ones ($\Delta$z = -0.03). Although this effect is negligible,
326: we are aware of it, and we assume that all the calculated z$_{phot}$'s might be affected by this slight underestimation.
327: 
328: 
329: 
330: %Notice however that all these outliers have 
331: %secondary redshifts values, also given by HyperZ,
332: %which are in good agreement with the spectroscopic ones, i.e., inside the 20\% boundaries in (1+$z$).
333: 
334: Based on the good agreement between spectroscopic and photometric redshifts in this fairly large 
335: subsample of sources (the results shown are better than
336: those typically obtained for AGN samples \citep{Babbedge04,Kitsionas05,Bundy08} and with practically the same 
337: $\sigma_{z}$ and outlier fraction than those reported by \citet{Polletta07}), we can
338: confidently extrapolate the results to the rest of the sample. This, together with the SED classification into the 
339: five groups described above, allows us to perform a reliable 
340: statistical analysis of the different AGN populations. 
341: 
342: \begin{figure}[!h]
343: \centering
344: \includegraphics[width=13cm,height=11cm]{f2.eps}
345: \caption{\footnotesize{Comparison between photometric and spectrospic redshifts for the 39 sources with publicly
346: available $z_{spec}$ from the DEEP data archive \citep{Weiner05}. The solid line corresponds to 
347: z$_{spec}$=z$_{phot}$.
348: Dashed lines represent 20\% boundaries in (1+$z$). Horizontal error bars 
349: indicating the reliability of the $z_{spec}$ (flag=1 for the less reliable and flag=4 for the most reliable) 
350: are represented. Vertical error bars ilustrate the discrepancies between the $z_{spec}$ 
351: and $z_{phot}$. The seven outliers are labelled (4, 6, 38, 52, 80, 98 and 126). 
352: Note that there are only three outliers when only the 31 galaxies with reliable $z_{spec}$ are considered.}
353: \label{z}}
354: \end{figure}
355: 
356: 
357: It is worth to mention that the distribution of object type in the subsample
358: with spectroscopic redshifts is completely different from the total sample. 
359: Considering only the 31 objects with highly reliable z$_{spec}$, there are 4 {\it Starburst-dominated AGN}, 
360: 4 {\it Starburst-contaminated AGN}, 3 {\it Type-1 AGN}, 5 {\it Type-2 AGN}, and 15 {\it Normal galaxy hosting AGN}.
361: Thus, it is very difficult to check the redshift failure rate for the different groups. Only for the 
362: {\it Normal galaxy hosting AGN} we can confirm the success in the redshift determination with this set of
363: templates, since $\sim$50\% of the total number of objects fitted with elliptical or spiral templates have 
364: z$_{spec}$ to compare with. Due to the flat and featureless SED typical of Type-1 QSOs, 
365: the {\it Type-1 AGN} group of templates could produce the less reliable photometric redshifts of the sample
366: \citep{Franceschini05}. 
367: We can not discard then that any subset of templates produces higher redshift failure rates
368: than others, but looking at the distribution of the objects belonging to the different groups of AGN in the various 
369: diagnostic diagrams in the following sections, and at the correlations displayed by them, we are confident 
370: that our SED classification and redshift determination are as good as for the {\it Normal galaxy hosting AGN} 
371: for the rest of the groups. 
372: 
373: 
374: \section{Discussion}
375: \subsection{Classification by SEDs and photometric redshift distribution of the sample.}
376: 
377: 
378: Together with the photometric redshift calculations reported in the previous section, we obtain spectral
379: energy distribution fits, that allow us to distinguish between different types of AGN populations, i.e., whether
380: they are pure AGN, AGN hosted by starburst-dominated galaxies, or AGN in otherwise normal 
381: galaxies.
382: 
383: For the five main groups described before we obtain the following distribution:
384: {\it Starburst-dominated AGN} (24 \% of the sample), {\it Starburst-contaminated AGN} (7 \%),
385: {\it Type-1 AGN} (21 \%), {\it Type-2 AGN} (24 \%), and {\it Normal galaxy hosting AGN} (24 \%).
386: 
387: 
388: 
389: We consider the {\it Type-1 AGN},  {\it Type-2 AGN} and {\it Starburst-contaminated AGN} as 
390: representative groups of AGN-dominated galaxies (since their SEDs are AGN-like at all or almost all 
391: wavelength ranges). The {\it Starburst-dominated AGN} and {\it Normal galaxy hosting AGN} are likewise considered 
392: AGN somehow masked by their host emission. With this simple classification, we find that 52\% of the 
393: sample is AGN-dominated while
394: 48\% is host galaxy-dominated; i.e., half of the objects in the EGS sample of AGN
395: show AGN-like SEDs while the other half show host-dominated SEDs.
396: This is consistent with the finding that between 40\% and 60\% of the Chandra-selected galaxies in the 
397: Hawaii Deep Survey Field SSA13 and in the Chandra Deep Field North (\citet{Barger01} and \citet{Hornschemeier01}, 
398: respectively) have optical spectra with no-signs of nuclear activity.
399: 
400: Also \citet{Barmby06}, based on the IRAC slopes ($\alpha <$ 0 for the red power-law IRAC SEDs, 
401: and $\alpha >$ 0 for the blue ones) divided their 
402: sample in sources where the central engine dominates the IRAC SEDs and stellar-dominated galaxies.
403: They found that 40\% of the sources have red power-law SEDs, another 40\% have blue host-dominated 
404: mid-infrared SEDs, and the remaining 20\% could not be fitted with a power-law. 
405: 
406: 
407: The method employed in this paper constitutes a powerful technique of classification of high redshift 
408: AGN provided we are able to procure well-sampled SEDs. This is important, for instance, for 
409: multi-band deep surveys of galaxies for which spectroscopic data will be necessarily scarce. 
410: Having SEDs over the largest wavelength range as possible is mandatory to identify the entire 
411: AGN population \citep{Dye08}. Otherwise, depending on the observed wavelength, the galaxies could be
412: missclasified. This is crucial, for example, for our {\it Starburst-contaminated AGN}, that in the optical
413: range look like starburst galaxies, and towards redder wavelengths appear as Type-1 or Type-2 AGN.
414: \citet{Dye08} finds also that the results of the SED fitting show little difference between two
415: filtersets that span the same wavelength range, despite the number of filters used. Nevertheless, from our work, 
416: we find that including a large number of filters can reveal 
417: details in the SED shape that help the code choosing between different templates. This is crucial 
418: to distinguish among the different templates of a given group, for which little differences in the SED
419: determine the type of object, or its age \citep{Polletta07}.
420: 
421: 
422: 
423: We use now the classification of the galaxies obtained to investigate the properties of the different 
424: AGN groups. The distribution of redshifts for all the 96 objects with unique detection in our sample is shown 
425: in the top-left pannel of Figure \ref{hyperz}. 58\% of the sample have z$<$1, with the rest of the sources distributed in a
426: decreasing tail up to z=3. This is expected for X-ray selected samples with similar or even deeper flux
427: limits \citep{Hasinger03,Barger05}. Figure \ref{hyperz} also shows histograms for the photometric redshift 
428: distributions of the {\it Starburst-dominated AGN}, {\it Starburst-contaminated AGN}, {\it Type-1 AGN}, 
429: {\it Type-2 AGN} , and the {\it Normal galaxy hosting AGN} groups. 
430: 
431: \begin{figure}[!h]
432: \centering
433: \includegraphics[width=14cm]{f3.eps}
434: \caption{\footnotesize{Distribution of photometric redshifts for the 96 objects with unique detection in our sample and 
435: for all the five main groups considered in this work. 
436: Mean photometric redshift for each group and corresponding standard deviation, together with the 
437: median redshift and the number of objects are labelled in each pannel.}
438: \label{hyperz}}
439: \end{figure}
440: 
441: 
442: 
443: {\it Type-2 AGN} and {\it Normal galaxy hosting AGN} are concentrated at lower redshifts, whereas the 
444: {\it Starburst-dominated AGN} show a high concentration around z$_{phot}$=1.3. The {\it Type-1 AGN} group has 
445: the largest spread in redshift, its mean value being z$_{phot}$=1.24$\pm$0.95. 
446: The {\it Starburst-contaminated AGN} group contains only seven objects, six of them within the redshift range [0.6,1.2], 
447: the other having a z$_{phot}$=2.45
448: 
449: This indicates that the {\it Starburst-dominated AGN} constitute the high-redshift population of AGN
450: masked by powerful host emission, whilst the {\it Normal galaxy hosting AGN} group represents the low-redshift 
451: population of low-luminosity AGN also masked by their host galaxies. Previous studies suggest that most
452: low luminosity AGN are found in massive, mostly spheroidal galaxies \citep{Dunlop03,Kauffmann03,Grogin05,Pierce07}.
453: Something similar happens with the AGN-dominated group: 
454: the {\it Type-1 AGN} span a large redshift distribution, the {\it Starburst-contaminated AGN} 
455: are located at intermediate values of redshift, and
456: finally, the {\it Type-2 AGN} are the low-z objects in this subsample. 
457: 
458: 
459: \citet{Alonso04} found that $\sim$25\% of their X-ray and 24 \micron~selected sources in both the EGS and the
460: Lockman Hole (45 in total) show pure type-1 AGN SEDs, while more than half of the sample have stellar 
461: emission-dominated 
462: or obscured SEDs. \citet{Franceschini05} detected 99 AGN in the X-rays and mid-infrared with Spitzer
463: in the SWIRE survey \citep{Lonsdale03}, sorting them in three main groups:  Type-1 AGN (39\%), Type-2 AGN (23\%),
464: and normal and starburst galaxies (38\%).
465: By adding \citet{Piccinotti82} and \citet{Kuraszkiewicz03} samples, there are 32 AGN 
466: with z$\le$0.12, also selected both in the
467: hard X-rays and mid-infrared, with more than half of these sources being type-1 AGN according to their SEDs. 
468: Ours and other works (e.g., \citet{Alonso04,Franceschini05}) performed with SED classification of X-ray and 
469: mid-infrared selected AGN in a wide range of redshift (up to z $\sim$ 2-3),
470: when compared with the results obtained for local samples of AGN selected in the same bands, 
471: seem to indicate that the percentage of type-1 objects decreases with redshift, 
472: while the number of obscured AGN at high redshift increases. 
473: 
474: 
475: 
476: Although the data used in this paper do not allow a deep study of the AGN feedback phenomenon, 
477: it is worth noticing that a redshift sequence can be readily seen in Figure \ref{hyperz}. 
478: Indeed, the {\it Starburst-dominated AGN} would have the highest redshifts in a decreasing 
479: sequence that goes through the {\it Type-1 AGN}, {\it Starburst-contaminated AGN}, and {\it Type-2 AGN},
480: ending with the {\it Normal galaxy hosting AGN} group, that shows the lowest redshifts. This evolutionary
481: sequence has been noticed for early-type galaxies by \citet{Schawinski07}. According to this recent work, 
482: the starbursts would start and be the dominant player after its onset. Subsequently, as the BH accretes 
483: enough mass, the AGN feedback reveals itself as the BH competes for the gas reservoir with the starbursts 
484: eventually quenching the star formation. The starburst phase thus declines, the AGN becoming dominant.
485: The {\it Starburst-contaminated AGN} phase would be the transition phase mentioned by \citet{Schawinski07}.
486: This process continue through lower ionization phases and it will end with the more quiescent 
487: {\it Normal galaxy hosting AGN} phases at lower redshifts. 
488: 
489: 
490: 
491: %\begin{figure}[!h]
492: %\includegraphics[width=14cm]{f3.eps}
493: %\caption{\footnotesize{Distribution of photometric redshift for the 96 objects with unique detection in our sample.
494: %Around 58\% of the sample has z$<$1, with the rest of the sources distributed in a decreasing tail 
495: %reaching up to z=3. Mean photometric redshift and corresponding standard deviation, together with the 
496: %median redshift and the number of objects are labelled at the top right.}
497: %\label{hyperz_all}}
498: %\end{figure}
499: 
500: 
501: 
502: \subsection{Correlations}
503: 
504: \subsubsection{Correlations for the whole sample}
505: 
506: One of the main advantages of the sample we are discussing is the multiwavelength coverage of the data, 
507: which allows us to study for the first time various correlations between ultraviolet/optical/infrared 
508: luminosities and X-ray luminosities for such a big AGN sample and within this range of redshift. 
509: The aim is to understand the behaviour of these sources in the different wavelength ranges. 
510: 
511: Absolute magnitudes (M$_{ABS}$) computed by HyperZ in each filter using the photometric redshifts 
512: and the chosen cosmological parameters, 
513: are used here to derive luminosities for the 96 objects with unique detection.
514: %These M$_{ABS}$ include the K correction, 
515: %which is the difference between the observed magnitude of a galaxy of a
516: %given age measured at $\lambda_{1}$=$\lambda_{0}$/(1+$z$) and the magnitude of the same galaxy at the same time 
517: %but at $\lambda_{0}$. Thus, the K correction corresponds to the difference in magnitude, of two galaxies with
518: %the same spectrum, due to the redshift, excluding the intrinsic evolution of the spectrum due to the evolution of the
519: %stellar populations \citep{Poggianti97}. 
520: The HyperZ code provides the M$_{ABS}$ (including the K correction) in the
521: ultraviolet, optical, near- and mid-infrared filters considered. 
522: Regarding the X-ray data, the observed rest-frame hard and soft
523: X-ray luminosities are obtained from the equation $L_X = \frac{4 \pi d_{L}^{2} f_X}{(z+1)^{2-\Gamma}}$, where 
524: $d_{L}$ is the luminosity distance (cm), f$_{X}$ is the X-ray flux (ergs cm$^{-2}$ s$^{-1}$), and $\Gamma$ is the
525: photon index. In this
526: case, the K correction vanishes since we assume a photon index $\Gamma$ = 2 \citep{Krumpe07,Alexander03,Mainieri02},
527: which is the canonical value for unobscured AGN \citep{George00}. Obscured active nuclei have
528: considerably flatter effective X-ray spectral slopes, due to the energy-dependent photoelectric absorption of the
529: X-ray emission \citep{Risaliti99}. However, \citet{Mainieri02} find the same intrinsic 
530: slope of the X-ray spectrum for both type-1 and type-2 AGN whatever their absorption levels, with $\Gamma \sim$~2  
531: for an X-ray selected sample in the Lockman Hole. We therefore assume a photon index $\Gamma$ = 2 for either obscured
532: and unobscured AGN, and consequently no K correction is needed for the X-ray luminosities.
533: 
534: The first row of Table \ref{types} shows the fitting slopes and correlation coefficients ($r$) 
535: of each scatter diagram between the far-UV/near-UV/ugriz/JK/IRAC/MIPS luminosities and the hard/soft
536: X-ray luminosities for the fits including all the objects with unique detection. 
537: In all cases the Spearman's rank correlation test has been performed, 
538: confirming that all the correlations are significant (p$<$0.01). 
539: Examples of these correlations for the far-UV/near-UV/r/K/IRAC 4.5 \micron\ /MIPS 24 \micron~luminosities 
540: and the hard/soft X-ray luminosities are shown in Figure \ref{correlations}. 
541: 
542: The expected slopes for AGN-dominated 
543: objects should be close to unity, since if the active nucleus is the dominant emitting source 
544: at all wavelengths, tight linear correlations should be drawn. Reality is different, and AGN 
545: are actually hosted by different types of galaxies. As it has been seen in previous sections, 
546: these host galaxies contaminate or even mask the AGN emission, thus deviating correlations from linear and
547: worsening them. Both the X-ray and mid-infrared emissions are mostly dominated by the active nuclei, 
548: whereas the optical and, to a lesser extent, the near-infrared bands are more affected by extinction, 
549: by stellar emission from the host galaxy, or by both.
550: This is clearly reflected in the slopes and correlation coefficients (hereafter $\alpha$ and $r$) 
551: of the global fits (see first row of Table \ref{types}).
552: Although correlations are all significant, with both the slopes and correlation coefficients close 
553: to unity, they begin getting slightly blurred as wavelength increases from the bluest optical bands 
554: up to the K band, improving again in the mid-infrared. 
555: The blurring is more noticeable when soft instead of hard X-rays are considered, 
556: due to the higher obscuration that affects the lower energies. 
557: 
558: Correlations between ultraviolet and X-ray luminosities are also good. 
559: The slopes are $\alpha \sim$~1.2 and 1.1 for the far-UV versus both the hard and soft X-rays 
560: luminosities, respectively, in good agreement with early X-ray studies of AGN that find
561: correlations between X-ray and ultraviolet monocromatic luminosities with similar slopes: L$_{X}$ $\propto$
562: L$_{UV}^{\beta}$, with $\beta \sim 0.7-0.8$, thus $\alpha \sim$ 1.4-1.2 \citep{Wilkes94,Vignali03,Strateva05,Steffen06}.
563: Nevertheless, this range of $\alpha$ was determined by using 2 keV and 2500 \AA~luminosities, which correspond to soft
564: X-rays and Near-UV, respectively. The slopes measured by us for the near-UV versus both the hard and soft X-rays 
565: luminosities are $\alpha \sim$~0.8 and 0.7, that are lower than expected. Nevertheless, \citet{LaFranca95}
566: found a correlation consistent with $\alpha$ = 1 using a generalized orthogonal regression that is in better agreement
567: with our values. 
568: 
569: 
570: \begin{figure}[!h]
571: \includegraphics[width=12cm,angle=90]{f4.eps}
572: \caption{\footnotesize{Examples of luminosity-luminosity scatter diagrams for all the objects in our sample with unique
573: detections and published data in the considered bands. Logarithms of $\nu$L$_{\nu}$ in the Far-UV, near-UV, r, K, 
574: IRAC 4.5 \micron, and MIPS 24 \micron~bands (erg s$^{-1}$) are represented versus their hard and soft X-ray
575: counterparts. Symbols are the same as in Figure \ref{z}, indicating the template fitting classification.}
576: \label{correlations}}
577: \end{figure}
578: 
579: 
580: \subsubsection{Correlations for the main AGN groups}
581: 
582: 
583: We also report slopes and correlation coefficients for the five main groups considered 
584: in the last sections separately in Table \ref{types}. 
585: Spearman's rank correlation tests have been performed for all scatter diagrams. Thus,
586: the values reported in Table \ref{types} correspond only to objects showing significant correlations (p$<$0.01).
587: 
588: Looking at the {\it Type-1 AGN} in Table \ref{types} we conclude that they are the
589: less contaminated active nuclei of the sample. Since we are seeing a direct view of the central engine, 
590: the emission is dominated by the AGN at all wavelengths. Therefore, these objects draw
591: the most tight correlations between each photometric band and either the hard 
592: or the soft X-ray luminosities\footnote{Note that in order to check that the good correlation displayed by {\it Type-1 AGN} luminosities is not due to a
593: distance effect (this group of galaxies presents the largest spread in redshift, as shown in 
594: Figure \ref{hyperz}), we have also analysed the corresponding fluxes, instead of luminosities, 
595: for all the groups considered in
596: this section. This way, the distance effect is eliminated from the fits. 
597: We find the same linear and significant correlations
598: for {\it Type-1 AGN} fluxes, confirming that the correlations displayed for this group of AGN are only due to their
599: intrinsic properties. }. 
600: 
601: If we look at the {\it Starburst-contaminated AGN}, we find significant correlations between 
602: the hard X-rays and the ultraviolet and infrared bands, dissapearing  
603: for the soft X-rays, due to the higher obscuration in this wavelength range.
604: 
605: For the {\it Starburst-dominated AGN}, correlations including the hard X-ray data are better and 
606: more robust than those with the soft X-ray ones, for which both the slope and $r$ values are far from unity. 
607: This is certainly due to the higher obscuration affecting the soft X-ray emission in these objects.  
608: The host galaxy emission and the dust are indeed masking the AGN.  
609: The same, but more dramatically, happens with the {\it Type-2 AGN} group, for which all the correlations 
610: involving the soft X-ray emission are not significant.
611: 
612: It is worth to mention the worsening of the fits for the {\it Starburst-dominated AGN} when 
613: the hard X-ray and either the IRAC~8 \micron~or MIPS 24 \micron~emission are considered. The slopes and 
614: correlation coefficients of both fits move away from unity, something that is interpreted as due to 
615: the increasing importance of the starburst emission at these longer wavelengths.
616: If we look at the MIPS 24 \micron~luminosity-luminosity scatter diagrams (bottom of Figure \ref{correlations}), the
617: overall majority of the {\it Starburst-dominated AGN} are located above the fit line. This indicates that there 
618: is an excess of mid-infrared emission, as compared with the X-ray luminosity coming principally from 
619: the AGN. This mid-infrared excess comes from warm dust heated by the intense
620: star formation bursts taking place in the galaxy (in addition to the dust heated by the AGN), hence deviating the 
621: {\it Starburst-dominated AGN} group from the linear fit, and making the correlation non-significant when 
622: the MIPS 24 \micron~luminosity is considered.
623: 
624: The behaviour of {\it Normal galaxy hosting AGN} is completely different: correlations when either the soft or 
625: hard X-rays are considered are quite similar, improving towards longer wavelengths, 
626: where the AGN resurfaces. This group of galaxies include low-luminosity AGN hosted in normal 
627: galaxies that dominates 
628: the optical and near-infrared bands, but not the mid-infrared emission. This explain why in some fits 
629: performed with this subset of templates, the IRAC 8 and MIPS 24 \micron~are not completely well 
630: reproduced by the fit.
631: 
632: 
633: %Regarding the other groups of objects,  we find also significative
634: %correlations for the total fit (the group including all the 96 objects), except when MIPS 24 \micron~are 
635: %considered versus both the hard and soft
636: %X-rays. And for the rest of the groups, the overall majority of the correlations are not significant when fluxes instead
637: %luminosities are considered, except for the J, K, IRAC 3.6, and IRAC 4.5 \micron~versus hard X-ray fluxes for the 
638: %{\it Starburst-dominated AGN} and for K and all the IRAC bands versus hard X-ray fluxes for the 
639: %{\it Starburst-contaminated AGN} group.
640: 
641: \subsection{X-ray properties}
642: 
643: 
644: Looking at the hard and soft luminosity ranges (see Table \ref{luminosity}) for each of the five main 
645: groups described above, we find that 
646: {\it Type-1 AGN} present the largest spread in luminosity, together with the highest luminosity values 
647: in both bands (L$_{Hard}=10^{40-45}$ and L$_{Soft}=10^{39-45}~erg~s^{-1}$, not corrected for absorption).
648: \citet{Alonso06} found that the majority of galaxies in their sample of X-ray detected sources in the
649: CDF-S fitted with Broad-line AGN (BLAGN) QSO templates showed hard X-ray luminosities in the range 
650: 10$^{43-44}erg~s^{-1}$ (also not corrected for absorption).
651: The same has been found when a spectroscopic classification of the objects has been possible
652: \citep{Zheng04,Szokoly04,Barger05}. Our hard X-ray luminosity range for {\it Type-1 AGN} agrees with
653: the literature in the sense that the most luminous X-ray sources are enclosed in that range, whilst 
654: five sources show L$_{Hard} < 10^{43}~erg~s^{-1}$, and only three have 
655: L$_{Hard} < 10^{42}~erg~s^{-1}$, namely irac068644, irac027980, and irac018192, all of them with 
656: z$_{phot} < 0.2$. The most X-ray luminous AGN in our sample is irac040934, with a 
657: L$_{Hard} = 10^{45}~erg~s^{-1}$ and z$_{phot}$ = 2.42.
658: 
659: The behaviour of the {\it Starburst-contaminated AGN} is very similar to that of the majority
660: of the {\it Type-1 AGN} and exactly coincides with the hard X-ray luminosity range found by
661: \citet{Alonso06} for BLAGN-fitted objects. This indicate that, despite the starburst appearance of the
662: SEDs of these objects at longer wavelengths, in regard to their X-ray emission their AGN nature dominates.
663: 
664: The {\it Starburst-dominated AGN} are contained in a narrower interval of X-ray luminosities 
665: (L$_{Hard}=10^{42-44}~erg~s^{-1}$ and L$_{Soft}=10^{41-44}~erg~s^{-1}$), 
666: although reaching high values,  indicating that these 
667: galaxies are not only starbursts, but also masked-AGN that show strong in their X-ray emission. 
668: Indeed, very few {\it bona fide} starburst galaxies have L$_{X} > 10^{42}$ erg s$^{-1}$, even including
669: luminous sources at moderate redshifts \citep{Zezas01}. 
670: Only for warm ultraluminous infrared galaxies (ULIRGs)
671: luminosities of up to 10$^{42}$ erg s$^{-1}$ are expected \citep{Franceschini03}.
672: {\it Type-2 AGN} display hard X-ray luminosities ranging from 10$^{42}$ to 10$^{43}$ erg s$^{-1}$, 
673: staying in a much narrower range and with lower values than those of {\it Type-1 AGN}. 
674: The values of hard X-ray luminosities that we find for {\it Starburst-dominated AGN} and {\it Type-2 AGN}
675: coincide with those found by \citet{Alonso06} for their galaxies fitted with Narrow-line AGN 
676: (NLAGN)+ULIRG templates.
677: 
678: Finally, the {\it Normal galaxy hosting AGN} group shows the lowest luminosity range of any of the
679: groups (L$_{Hard} = L_{Soft} = 10^{40-43}~erg~s^{-1}$), which is consistent with 
680: the fact that they are hosting low-luminosity AGN \citep{Dunlop03,Kauffmann03,Grogin05,Pierce07}. 
681: The hard X-ray luminosity range 
682: of this group of objects coincides with typical luminosities (L$_{Hard} <$ 2 x 10$^{42}~erg~s^{-1}$) 
683: of the local cool ULIRGs population, except for four sources, namely irac045337, irac019616, irac016716, and
684: irac049420.
685: 
686: These results, together with the mean redshift of each group reported in Section 4.1.,   
687: point out that the evolution of AGN is luminosity-dependent, with low-luminosity AGN 
688: peaking at lower redshifts than luminous active nuclei 
689: \citep{Hasinger03,Hasinger05,Fiore03,Ueda03,LaFranca05,Brandt05,Bongiorno07}.
690: 
691: 
692: \subsection{Infrared and optical properties}
693: 
694: 
695: The IRAC mid-infrared colors have been used as a diagnostic tool to separate AGN from non-active 
696: galaxies and stars in
697: different samples \citep{Lacy04,Hatziminaoglou05,Stern05,Alonso06,Barmby06,Donley07}. Particularly, 
698: \citet{Stern05} show an IRAC color-color diagram for the AGES sample, with all their objects 
699: spectroscopically classified. They found that BLAGN are clearly separated from Galactic 
700: stars and ordinary galaxies in their diagram, with the NLAGN located both inside 
701: and outside of the active galaxies area. 
702: 
703: An IRAC colour-colour diagram for our sample is represented in Figure \ref{color}. The different symbols 
704: indicate the template fitting classification. The dashed line in Figure 
705: \ref{color} corresponds to the \citet{Stern05} empirical separation of AGN in their sample. 
706: In our case, this region includes all the {\it Type-1 AGN}, and all but one of the 
707: {\it Starburst-contaminated AGN}. This is expected, since five of the seven galaxies belonging to 
708: that group were fitted with the Sy1/SB/ULIRG template (SED type = 4, see Table \ref{photoz}),
709: while the one located outside the AGN region was fitted with the Sy2/SB template (SED type = 6). 
710: 
711: The only-galaxy classified as {\it Starburst-contaminated AGN} fitted with a Sy2/SB/ULIRG template 
712: (SED type = 13) that is contained in the \citet{Stern05} AGN region is irac046309, its
713: photometric redshift being $z$=2.45. 
714: The redshift of this source is mentioned here because, as 
715: \citet{Barmby06} discuss and ilustrate in their Figure 6, the AGN-dominated
716: templates have red mid-infrared colors and thus, lie inside the \citet{Stern05} region at all redshifts, whereas
717: the star-forming galaxy templates begin to move into this area as the redshift increases.
718: This explains why all {\it Type-1 AGN} are located inside the AGN region marked by the
719: dashed line, as well as the six {\it Starburst-contaminated AGN}: five are fitted with the 
720: Sy1/SB/ULIRG template, and the galaxy irac046309 is a high redshift Sy2/SB/ULIRG.
721: The {\it Normal galaxy hosting AGN}-fitted objects 
722: (that have the bluest IRAC colors of the sample) are excluded of this region (except for one of them). 
723: 
724: 
725: As shown in \citet{Stern05}, the active galaxy region is contaminated with Galactic stars and normal
726: galaxies, with the NLAGN located both inside and outside of this area. The same happens in our
727: Figure \ref{color}: 
728: {\it Starburst-dominated AGN} and {\it Type-2 AGN} are partly contained in this area, and partly not. We have estimated
729: the mean redshifts of both groups of galaxies for the in- and out-objects, finding that 
730: the {\it Starburst-dominated AGN} 
731: lying outside the pure-AGN region have a mean $z$ = 1.35$\pm$0.54, while those inside have a mean 
732: $z$ = 1.52$\pm$0.50. Following the same trend, the {Type-2 AGN} mean redshift is $z$ = 0.79$\pm$0.26 for the
733: outside objects, and $z$ = 0.97$\pm$0.32 for the galaxies included in the \citet{Stern05} region. This is again consistent
734: with the evolution of mid-infrared colors with redshift for star-forming galaxies \citep{Barmby06,Donley08}. However, 
735: these mean redshifts for {\it Starburst-dominated AGN} and {Type-2 AGN} lying inside and outside the \citet{Stern05} region 
736: are only orientative, since the differences between them are not statistically significant.
737:  
738: The reliability of these type of diagram (mid-infrared color selection) in selectioning AGN have been questioned in 
739: the literature \citep{Cardamone08,Donley08}. It seems that they fail to identify a large number of X-ray selected AGN, 
740: finding only the most luminous. In our work, the \citet{Stern05} region wraps all the Type-1 objects, all but one
741: of the {\it Starburst-contaminated AGN}, and half of the {\it Starburst-dominated AGN} and {Type-2 AGN}. 52\% of our 
742: sample is included in this area, but the low-luminosity AGN (most of them {\it Normal galaxy hosting AGN} and several 
743: {\it Starburst-dominated AGN} and {\it Type-2 AGN}) are excluded. \citet{Cardamone08} find that 76\% of their
744: spectroscopically-selected BLAGNs fall inside this region, but only 40\% of the X-ray selected objects are included. 
745: Summarizing, although the diagram in Figure \ref{color} only includes half of our sample in the \citet{Stern05} region, 
746: it seems very effective segregating the different AGN groups. 
747: 
748:  
749: %As can be seen in
750: %Figure \ref{fits},  {\it Starburst-dominated AGN} and {Type-2 AGN} SEDs are very similar, since most type-2 objects
751: %show strong stellar contamination.
752: 
753: 
754: 
755: \begin{figure}[!h]
756: \includegraphics[width=10cm,angle=90]{f5.eps}
757: \caption{\footnotesize{IRAC color-color plot for the 96 sources in our sample. Symbols are the same as in 
758: Figure \ref{z}, indicating the template fitting classification. The dashed line 
759: corresponds to the \citet{Stern05} empirical separation of AGN in their sample.}
760: \label{color}}
761: \end{figure}
762: 
763: 	The left pannel of Figure \ref{obscured1} shows the mid-infrared 24~\micron~to optical 
764: (r band) flux ratio versus the (r-z)$_{AB}$ color for galaxies with unique detection. 
765: The 24~\micron~to optical flux ratio is a rough estimator of obscured activity in galaxies, 
766: since the 24~\micron~sources with with faint optical counterparts should be luminous AGN obscured
767: by dust and/or gas in the optical range \citep{Fiore08}.   
768: The (r-z)$_{AB}$ color depends on the obscuration present in the galaxy.
769: 
770: As expected for pure AGN, we find a significant correlation between the 24~\micron~to $r$
771: flux ratio and (r-z)$_{AB}$ for the {\it Type-1 AGN} and {\it Type-2 AGN}, because the nuclear
772: emission dominates both in the optical and mid-infrared wavelengths \citep{Fiore08}. However, the 
773: correlation is not significant for the {\it Starburst-dominated AGN} group, since they have an excess
774: in their mid-infrared emission, coming from the dust heated by the starbursts, in addition to the 
775: dust heated by the AGN. {\it Normal galaxy hosting AGN} display also a correlation between the two quantities, 
776: but with a different slope and lower correlation coefficient than the pure AGN objects. 
777: The corresponding slopes and correlation coefficients are indicated in
778: Figure \ref{obscured1}, except for the  {\it Starburst-contaminated AGN} group, 
779: due to the low number of objects fitted with this set of templates. 
780: 
781: 
782: A segregation between the different groups is noticeable in the plot: the {\it Starburst-dominated AGN} and 
783: {\it Starburst-contaminated AGN} 
784: are shifted towards the highest values of the mid-infrared to optical ratio 
785: Log (24 \micron / r band flux) $>$ 1.6 , {\it Type-1 AGN} and 
786: {\it Type-2 AGN} are located at intermediate values, and the 
787: {\it Normal galaxy hosting AGN} have the lowest values of this ratio (Log [24 \micron / r band flux] $<$ 1.8).
788: Obscured AGN should be located towards the top right of Figure \ref{obscured1} (left pannel), 
789: since they have the reddest optical colors and the highest 24 \micron/r band flux ratios. 
790: {\it Starburst-dominated AGN}, 
791: {\it Starburst-contaminated AGN}, and {\it Type-2 AGN}-fitted objects 
792: are the most obscured galaxies in our sample, according to this diagram, although they are not as obscured as 
793: those in \citet{Fiore08}. We have chosen the (r-z)$_{AB}$ color instead of the most common 
794: (r-K)$_{AB}$ due to the lower number of objects that have available K magnitudes. 
795: 
796: 
797: In the right pannel of Figure \ref{obscured1} the same mid-infrared 24~\micron~/ r band flux ratio is shown
798: against the (r-IRAC 3.6 \micron)$_{AB}$ color for galaxies with unique detection.
799: As much as the (r-z)$_{AB}$ color is contaminated  by the host galaxy contribution, the 
800: (r-IRAC 3.6 \micron)$_{AB}$ color
801: is dominated by the hot dust emission heated by the AGN and/or intense star formation \citep{Brusa05}. 
802: In this case,  all the individual groups of objects as well as the whole sample show significant
803: and tight correlations. The segregation between
804: the different groups mentioned before is clear again in this graph. 
805: The {\it Starburst-dominated AGN}  and {\it Starburst-contaminated AGN} clearly show the reddest colors of the
806: sample (r-IRAC 3.6 \micron~$>$ 2.3), while the {\it Normal galaxy hosting AGN} display the bluest, concentrated around 
807: r-IRAC 3.6 \micron~$\sim$ 1.6. These objects occupy the left bottom corner of the right pannel of Figure 
808: \ref{obscured1} because the host galaxy outshines the AGN emission at all wavelengths (except in the X-rays). 
809: 
810: %In this case the enhanced emission in the K band with respect to the
811: %R band is probably due to the combination of strong dust ex-
812: %tinction in the optical and a contribution of the point-like emis-
813: %sion in the near-infrared.
814: 
815: \begin{figure}[!h]
816: \includegraphics[width=11cm,angle=90]{f6.eps}
817: \caption{\footnotesize{Log (MIPS 24 \micron~/ r band) flux ratio versus the r-z (AB) color for objects 
818: with unique detection in our
819: sample and available r and z magnitudes (left pannel) and the same ratio versus the r-IRAC 3.6 \micron~
820: color (right pannel). The slopes and correlation coefficients are labeled for the global fit and for each of the 
821: four groups.}
822: \label{obscured1}}
823: \end{figure}
824: 
825: \subsection{Objects with double detection in the optical bands.}
826: \label{double}
827: 
828: Twenty out of the 116 objects that comprise our full sample show double detection in the ground-based images, 
829: thanks to their better spatial resolution. Figs. \ref{cromo1} and \ref{cromo2} show
830: ACS V-band/HST images of both detections
831: (indicated with circles) for each pair of galaxies, except for objects irac053271 and irac038708, 
832: which do not have HST imaging, and for which optical CFHTLS r-band images are shown instead.
833: The CFHTLS magnitudes have been employed for the calculations with the HyperZ code, although we have 
834: chosen the
835: HST images for display purposes, because of their better resolution. These images help us classifying 
836: morphologically these 20 objects with double
837: detections as either interacting systems, different star forming regions of the same galaxy, 
838: or simple source confusion, as described in Table \ref{double}. 
839: 
840: In the case of these objects with double detection, for a single IRAC source, there are two 
841: counterparts in the ground-based images within the 2.5\arcsec~search radius. As described in Section \ref{sample},
842: the optical/near-infrared reference image is used to determine the positions of each source. 
843: The IRAC images are then deconvolved using the IRAC PSFs.
844: The sources can be resolved for separations $\sim$1\arcsec~from each other, and    
845: IRAC fluxes are then remeasured by fixing the positions of the objects in each pair, and by scaling the
846: flux of each object in an aperture of 0.9\arcsec~\citep{Perez05,Perez08}. 
847: The integrated magnitude is derived applying an aperture correction based on empirical IRAC PSFs (for the 
848: 0.9\arcsec~aperture the factors are 1.01$\pm$0.07, 1.02$\pm$0.08, 1.2$\pm$0.10, and 1.44$\pm$0.14 for the 
849: channels 3.6, 4.5, 5.8, and 8.0 \micron, respectively). See \citet{Perez08} appendix 
850: for more details. The flux contamination is found to be smaller than the 10\% 
851: in most cases, an even smaller for the non-infrared-bright sources.
852: 
853: 
854: Once we know the positions of each galaxy in a pair, we can check whether the
855: mid-infrared emission comes from both, or just from one of the objects in the IRAC and MIPS images. 
856: In the majority of the cases,
857: all the mid-infrared flux in a pair of galaxies comes from only one of the objects
858: (see Table \ref{double}), 
859: the other probably being a non-active object. Then, we assume that the X-ray emission comes 
860: from the same mid-infrared emitter, and we calculate photometric redshifts for the active objects only.
861: 
862: 
863: 
864: In those cases where the mid-infrared emission can not be allocated clearly to one of the objects 
865: (irac038708, irac056633, and irac046783), photometric
866: redshifts calculated by HyperZ for both sources in each pair have been obtained and they are reported in Table 
867: \ref{photoz2} together with their  $\chi_{\nu}^{2}$, probabilities, SED type and A$_{V}$. 
868: For the other 17 pairs of galaxies, for which the mid-infrared emission comes clearly from only one of the objects,
869: we calculate photometric redshifts only for the mid-infrared emitter. Spectroscopic redshifts 
870: from the DEEP database are also given, when available, together with their corresponding reliability flags. 
871: Unfortunately, this is the case for only four objects, and all of them with low reliability flags (1 or 2, see Table
872: \ref{photoz2}). Nevertheless, we can assume that the photometric redshifts, obtained as described in 
873: Section \ref{spectral}, are reasonably good, since we have followed the same methodology as for the 96 sources 
874: with single detections. 
875: 
876: As reported in Table \ref{double}, irac056633$_{-}$2, irac036704$_{-}$1, irac022060$_{-}$1, irac029343$_{-}$1,
877: and irac019604$_{-}$1 are interacting systems themselves, as can be seen in the HST images 
878: (Figs. \ref{cromo1} and 
879: \ref{cromo2}). These sources must be treated with caution, since their fluxes could be contaminated with extra-emission
880: coming from their companions. This fact explains the low probabilities of the HyperZ fits for objects 
881: irac056633$_{-}$2, irac029343$_{-}$1, and irac019604$_{-}$1, reported in Table \ref{photoz2}.
882: 
883: In the same way as we have done for the objects with unique detection in previous sections, 
884: we distribute here the 23 template fitted-objects with double detection in the same five main categories described before. 
885: The percentages for each group are: {\it Starburst-dominated AGN} (48 \% of the mid-infrared emitters), 
886: {\it Starburst-contaminated AGN} (0 \%),
887: {\it Type-1 AGN} (17 \%), {\it Type-2 AGN} (22 \%), and {\it Normal galaxy hosting AGN} (13 \%).
888: Note that for this subsample of objects with double detection, almost half of the objects are described by
889: starburst-type SEDs. If, as in Section 4.1., we split the objects into AGN-dominated and host-dominated 
890: galaxies, we find that 39\% show AGN-like SEDs while 61\% are host-dominated, a clear overrepresentation. 
891: This is expected since  
892: if the pairs of galaxies are interacting objects, the number of starbursts in this subsample of galaxies 
893: should consequently increase. 
894: 
895: 
896: 
897: \begin{figure}[!h]
898: \centering
899: {\par
900: \includegraphics[width=4cm,height=4cm,angle=90]{f7a.eps}
901: \includegraphics[width=4cm,height=4cm,angle=90]{f7b.eps}
902: \includegraphics[width=4cm,height=4cm,angle=90]{f7c.eps}
903: \includegraphics[width=4cm,height=4cm,angle=90]{f7d.eps}\par}
904: {\par
905: \includegraphics[width=4cm,height=4cm,angle=90]{f7e.eps}
906: \includegraphics[width=4cm,height=4cm,angle=90]{f7f.eps}
907: \includegraphics[width=4cm,height=4cm,angle=90]{f7g.eps}
908: \includegraphics[width=4cm,height=4cm,angle=90]{f7h.eps}\par}
909: {\par
910: \includegraphics[width=4cm,height=4cm,angle=90]{f7i.eps}
911: \includegraphics[width=4cm,height=4cm,angle=90]{f7j.eps}
912: \includegraphics[width=4cm,height=4cm,angle=90]{f7k.eps}
913: \includegraphics[width=4cm,height=4cm,angle=90]{f7l.eps}\par}
914: {\par
915: \includegraphics[width=4cm,height=4cm,angle=90]{f7m.eps}
916: \includegraphics[width=4cm,height=4cm,angle=90]{f7n.eps}
917: \includegraphics[width=4cm,height=4cm,angle=90]{f7o.eps}
918: \includegraphics[width=4cm,height=4cm,angle=90]{f7p.eps}\par}
919: {\par
920: \includegraphics[width=4cm,height=4cm,angle=90]{f7q.eps}
921: \includegraphics[width=4cm,height=4cm,angle=90]{f7r.eps}
922: \includegraphics[width=4cm,height=4cm,angle=90]{f7s.eps}
923: \includegraphics[width=4cm,height=4cm,angle=90]{f7t.eps}\par}
924: \figcaption{\footnotesize{Optical images (10\arcsec~x10\arcsec) of the sources with double detection
925: in our sample. 
926: For objects irac053271 and irac038708, R-band images from CFHTLS are shown. The rest are ACS V-band images:
927: irac056633, irac034779, irac031799, irac036704, irac028084, irac071060, irac046783, and irac042079.
928: Two stamps are shown for each pair of galaxies, the circle indicating the position of each source candidate.}
929: \label{cromo1}}
930: \end{figure}
931: 
932: 
933: 
934: \begin{figure}[!h]
935: \centering
936: \centering
937: {\par
938: \includegraphics[width=4cm,height=4cm,angle=90]{f8a.eps}
939: \includegraphics[width=4cm,height=4cm,angle=90]{f8b.eps}
940: \includegraphics[width=4cm,height=4cm,angle=90]{f8c.eps}
941: \includegraphics[width=4cm,height=4cm,angle=90]{f8d.eps}\par}
942: {\par
943: \includegraphics[width=4cm,height=4cm,angle=90]{f8e.eps}
944: \includegraphics[width=4cm,height=4cm,angle=90]{f8f.eps}
945: \includegraphics[width=4cm,height=4cm,angle=90]{f8g.eps}
946: \includegraphics[width=4cm,height=4cm,angle=90]{f8h.eps}\par}
947: {\par
948: \includegraphics[width=4cm,height=4cm,angle=90]{f8i.eps}
949: \includegraphics[width=4cm,height=4cm,angle=90]{f8j.eps}
950: \includegraphics[width=4cm,height=4cm,angle=90]{f8k.eps}
951: \includegraphics[width=4cm,height=4cm,angle=90]{f8l.eps}\par}
952: {\par
953: \includegraphics[width=4cm,height=4cm,angle=90]{f8m.eps}
954: \includegraphics[width=4cm,height=4cm,angle=90]{f8n.eps}
955: \includegraphics[width=4cm,height=4cm,angle=90]{f8o.eps}
956: \includegraphics[width=4cm,height=4cm,angle=90]{f8p.eps}\par}
957: {\par
958: \includegraphics[width=4cm,height=4cm,angle=90]{f8q.eps}
959: \includegraphics[width=4cm,height=4cm,angle=90]{f8r.eps}
960: \includegraphics[width=4cm,height=4cm,angle=90]{f8s.eps}
961: \includegraphics[width=4cm,height=4cm,angle=90]{f8t.eps}\par}
962: \figcaption{\footnotesize{Same as in Figure \ref{cromo1} (all the images from ACS V-band/HST), for objects 
963: irac061881, irac022060, irac052826, irac054493, irac029343, irac019604, irac27967, irac022761, irac017174, 
964: irac021943.}
965: \label{cromo2}}
966: \end{figure}
967: 
968: 
969: \section{Conclusions}
970: 
971: We present a reliable method of classification of hard X-ray and mid-infrared selected AGN, based on 
972: the fit of well-sampled multiwavelength spectral energy distributions with a complete set of AGN and
973: starburst galaxy templates. 
974: The sample studied in this paper consists of 96 AGN with unique detection, 
975: and 20 AGN with double detection in the EGS. 
976: The following results were found:
977: 
978: \begin{itemize}
979: 
980: 
981: \item{Photometric redshifts have been calculated by
982: using the HyperZ code. The measured mean discrepancy between our z$_{phot}$'s and a subsample of higly reliable
983: DEEP spectroscopic redshifts (flag = 3 or 4) is $\Delta$z = -0.03, with $\sigma_{z}$ = 0.11, 
984: and 3 outliers (8\%). 
985: We provide more accurate photometric redshifts than the spectroscopic ones for objects with DEEP flag = 1 or 2.}
986: 
987: \item{Five main population groups have been considered according to the set of templates employed. 
988: For the 96 objects in our sample with unique detection, the following percentages have been found: 
989: {\it Starburst-dominated AGN} (24 \% of the sample), {\it Starburst-contaminated AGN} (7 \%),
990: {\it Type-1 AGN} (21 \%), {\it Type-2 AGN} (24 \%), and {\it Normal galaxy hosting AGN} (24 \%). We find that
991: 52\% of the sample has AGN-dominated SEDs and the remaining 48\% host-dominated SEDs.}
992: 
993: \item{58\% of the 96 objects with unique detection in our sample have
994: z$_{phot}$$<$1, with the rest of the z$_{phot}$ of the sources distributed in a decreasing tail up 
995: to z$_{phot}$=3. The {\it Starburst-dominated AGN} constitute the high-redshift population of the host-dominated
996: AGN, whilst the {\it Normal galaxy hosting AGN} are concentrated at low redshifts. In the AGN-dominated group,
997: {\it Type-1 AGN} are randomly distributed in distance, the {\it Starburst-contaminated AGN} are located
998: at intermediate values of redshift, and the {\it Type-2 AGN} are the lowest-z objects.}
999: 
1000: \item{An evolutionary trend is noticed, in which the {\it Starburst-dominated AGN} would be the progenitors of the 
1001: {\it Type-1 AGN} and {\it Type-2 AGN}, via quenching of the starburst through the AGN feedback.}
1002: 
1003: \item{Correlations between hard/soft X-ray luminosities and ultraviolet/optical/infrared data are reported for 
1004: such a sample of AGN spanning a wide range of redshift, being in this way represented the behaviour of 
1005: the different AGN types in the various wavelengths considered.}
1006: 
1007: \item{{\it Type-1 AGN} show the highest values of hard and soft X-ray luminosities of the sample, together with
1008: the {\it Starburst-contaminated AGN}, whilst the {\it Normal galaxy hosting AGN} majority are concentrated
1009: at the lowest values, coinciding with the local cool ULIRGs typical hard X-ray luminosities. 
1010: {\it Starburst-dominated AGN} and {\it Type-2 AGN} present intermediate values of X-ray emission, very similar to
1011: those of warm ULIRGs. This is consistent with a luminosity-dependent evolution of AGN, with low-luminosity AGN
1012: peaking at lower redshifts than luminous active nuclei.}
1013: 
1014: \item{{\it Type-1 AGN} are all contained in the IRAC color-color diagram region empirically determined by
1015: \citet{Stern05} for spectroscopically selected AGN. There are many {\it Type-2 AGN} and 
1016: {\it Starburst-dominated AGN}
1017: inside this AGN region. These objects have higher mean redshifts than those in the same group but outside the 
1018: pure-AGN area, according with the evolution of the mid-infrared colors with redshift 
1019: for star-forming galaxies described in \citet{Barmby06}.}
1020: 
1021: \item{Mid-infrared 24 \micron~to optical r band flux ratio versus the
1022: (r-z)$_{AB}$ or the (r-IRAC 3.6 \micron)$_{AB}$ colors show a clear segregation of the different
1023: groups in both diagrams. {\it Starburst-dominated AGN} and {\it Starburst-contaminated AGN}
1024: are displaced towards the highest values of the mid-infrared to optical ratio and display the reddest colors.
1025: {\it Type-1 AGN} and {\it Type-2 AGN} are located at intermediate values, and the {\it Normal galaxy hosting AGN}
1026: have the lowest values of the 24 \micron/r flux ratio and the bluest colors. }
1027: 
1028: \item{A tentative classification of objects with double detection into the five main population 
1029: groups considered through this paper shows an increase of the {\it Starburst-dominated AGN} of up 
1030: to 48\%, while the others decrease. 
1031: 61\% of the fitted objects show AGN-like SEDs, while 39\% is host-dominated.}
1032: 
1033: 
1034: \end{itemize}
1035: 
1036: 
1037: \acknowledgments
1038: 
1039: NASA's Chandra X-Ray Observatory was launched in July 1999. The Chandra Data Archive (CDA) is part
1040: of the Chandra X-Ray Center (CXC) which is operated for NASA by the Smithsonian Astrophysical
1041: Observatory.
1042: 
1043: Based on observations obtained with XMM-Newton, an ESA science mission with instruments and 
1044: contributions directly funded by ESA Member States and NASA.
1045: 
1046: GALEX (Galaxy Evolution Explorer) is a NASA Small Explorer, launched in April 2003. We gratefully
1047: acknowledge NASA's support for construction, operation, and science analysis of the GALEX mission,
1048: developed in cooperation with the Centre National d'Etudes Spatiales of France and the Korean Ministry
1049: of Science and Technology.)
1050: 
1051: This work is based in part on observations made with the Spitzer Space Telescope, which is
1052: operated by the Jet Propulsion Laboratory, California, Institute of Technology under a contract with
1053: NASA.
1054: 
1055: Based on observations obtained with MegaPrime/MegaCam, a joint project of CFHT and CEA/DAPNIA, at
1056: the Canada-France-Hawaii Telescope (CFHT) which is operated by the National Research Council (NRC) of
1057: Canada, the Institut National des Science de l'Univers of the Centre National de la Reserche
1058: Scientifique (CNRS) of France, and the University of Hawaii. This work is based in part on data
1059: products produced at TERAPIX and the Canadian Astronomy data Centre as part of the CFHT Legacy Survey,
1060: a collaborative project of NRC and CNRS.
1061: 
1062: Based on observations obtained at the Hale Telescope, Palomar Observatory, as part 
1063: of a collaborative agreement between the California Institute
1064: of Technology, its divisions Caltech Optical Observatories and the Jet Propulsion Laboratory 
1065: (operated for NASA), and Cornell University.
1066: 
1067: Many images of this article are based on observations made with the NASA/ESA Hubble Space Telescope, 
1068: obtained from the data archive at the Space Telescope Science Institute (STScI). STScI is operated by the 
1069: Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-26555.
1070: 
1071: This work uses data obtained with support of the National Science Foundation grants 
1072: AST 95-29028 and AST 00-71198.
1073: 
1074: This work is partially funded by PN AYA2007-67965-C03-01, PN AYA2006-02358 and by the Spanish MEC 
1075: under the Consolider-Ingenio 2010 Program grant CSD2006-00070: First Science with the GTC 
1076: (http://www.iac.es/consolider-ingenio-gtc/). P.G.P.-G. acknowledges support from the Ram\'{o}n y Cajal Program
1077: financed by the Spanish Government and the European Union.
1078: C.R.A., J.R.E., G.B., J.G., and P.G.P.-G. acknowledge Roser Pell\'{o}, Antonio Cabrera Lavers, and Casiana Mu\~{n}oz Tu\~{n}\'{o}n 
1079: for their valuable help. We finally appreciate the very useful report of the anonymous referee.
1080: 
1081: 
1082: \clearpage
1083: 
1084: \begin{thebibliography}{}
1085: 
1086: \bibitem[Alexander et al.(2003)]{Alexander03} Alexander, D. M., et al. 2003, \aj, 125, 383
1087: 
1088: \bibitem[Alonso-Herrero et al.(2004)]{Alonso04} Alonso-Herrero, A., et al. 2004, \apjs, 154, 155
1089: 
1090: \bibitem[Alonso-Herrero et al.(2006)]{Alonso06} Alonso-Herrero, A., et al. 2006, \apj, 640, 167 
1091: 
1092: \bibitem[Babbedge et al.(2004)]{Babbedge04} Babbedge, T. S. R., et al. 2004, \mnras, 353, 654
1093: 
1094: \bibitem[Barger et al.(2001)]{Barger01} Barger, A. J., Cowie, L. L., Mushotzky, R. F., \& Richards, E. A. 2001,
1095: \aj, 121, 662
1096: 
1097: \bibitem[Barger et al.(2005)]{Barger05} Barger, A. J., Cowie, L. L., Mushotzky, R. F., Yang, Y., Wang, W.-H., 
1098: Steffen, A. T., \& Capak, P. 2005, \aj, 129, 578
1099: 
1100: \bibitem[Barmby et al.(2006)]{Barmby06} Barmby, P., et al. 2006, \apj, 642, 126
1101: 
1102: \bibitem[Bolzonella et al.(2000)]{Bolzonella00} Bolzonella, M., Miralles, J.-M., \& Pell\'{o}, R. 2000, 
1103: \aap, 363, 476
1104: 
1105: \bibitem[Bongiorno et al.(2007)]{Bongiorno07} Bongiorno, A., et al. 2007, \aap, 472, 443
1106: 
1107: \bibitem[Boulade et al.(2003)]{Boulade03} Boulade, O., et al. 2003, \procspie, 4841, 72
1108: 
1109: \bibitem[Brusa et al.(2005)]{Brusa05} Brusa, M., et al. 2005, \aap, 432, 69
1110: 
1111: \bibitem[Brandt \& Hasinger (2005)]{Brandt05} Brandt, W. N. \& Hasinger, G. 2005, \araa, 43, 827
1112: 
1113: \bibitem[Bundy et al.(2006)]{Bundy06} Bundy, K., et al. 2006, \apj, 651, 120
1114: 
1115: \bibitem[Bundy et al.(2008)]{Bundy08} Bundy, K., et al. 2008, \apj, 681, 931
1116: 
1117: \bibitem[Calzetti et al.(2000)]{Calzetti00} Calzetti, D. et al. 2000, \apj, 533, 682
1118: 
1119: \bibitem[Cardamone et al.(2008)]{Cardamone08} Cardamone, C. N., et al. 2008, \apj, 680, 130
1120: 
1121: \bibitem[Cocchia et al.(2007)]{Cocchia07} Cocchia, F., et al. 2007, \aap, 466, 31
1122: 
1123: %\bibitem[Comastri et al.(1995)]{Comastri95} Comastri, A., Setti, G., Zamorani, G., \& Hasinger, G. 1995, 
1124: %\aap, 296, 1
1125: 
1126: \bibitem[Davis et al.(2007)]{Davis07} Davis, M., et al. 2007, \apj, 660, 1 
1127: 
1128: %\bibitem[Dawson et al.(2003)]{Dawson03} Dawson, S., McCrady, N., Stern, D., Eckart, M. E., Spinrad, H.,
1129: %Liu, M. C., \& Graham, J. R. 2003, \aj, 125, 1236
1130: 
1131: \bibitem[Dickinson et al.(2001)]{Dickinson01} Dickinson, M., et al. 2001, \baas, 33, 820
1132: 
1133: \bibitem[Donley et al.(2005)]{Donley05} Donley, J. L., Rieke, G. H., Rigby, J. R., \& 
1134: P\'{e}rez-Gonz\'{a}lez, P. G. 2005, \apj, 634, 169
1135: 
1136: \bibitem[Donley et al.(2007)]{Donley07} Donley, J. L.,  Rieke, G. H., P\'{e}rez-Gonz\'{a}lez, P. G., 
1137: Rigby, J. R., \& Alonso-Herrero, A. 2007, \apj, 660, 167
1138: 
1139: \bibitem[Donley et al.(2008)]{Donley08} Donley, J. L., Rieke, G. H., P\'{e}rez-Gonz\'{a}lez, P. G., \& Barro, G. 2008,
1140: \apj in press, arXiv:0806.4610
1141: 
1142: \bibitem[Dunlop et al.(2003)]{Dunlop03} Dunlop, J. S., McLure, R. J., Kukula, M. J., Baum, S. A., O'Dea, C. P., 
1143: Hughes, D. H. 2003, \mnras, 340, 1095
1144: 
1145: \bibitem[Dye (2008)]{Dye08} Dye, S. 2008, \mnras in press, arXiv:0806.1222
1146: 
1147: \bibitem[Eckart et al.(2006)]{Eckart06} Eckart, M. E., et al. 2006, \apjs, 156, 35
1148: 
1149: \bibitem[Eisenhardt et al.(2004)]{Eisenhardt04} Eisenhardt, P., et al. 2004, \apjs, 154, 48
1150: 
1151: \bibitem[Fiore et al.(2003)]{Fiore03} Fiore, F., et al. 2003, \aap, 409, 79
1152: 
1153: \bibitem[Fiore et al.(2008)]{Fiore08} Fiore, F., et al. 2008, \apj, 672, 94 
1154: 
1155: \bibitem[Franceschini et al.(2003)]{Franceschini03} Franceschini, A. et al. 2003, \mnras, 343, 1181
1156: 
1157: \bibitem[Franceschini et al.(2005)]{Franceschini05} Franceschini, A. et al. 2005, \aj, 129, 2074
1158: 
1159: \bibitem[George et al.(2000)]{George00} George, I. M., Turner, T. J., Yaqoob, T., Netzer, H., Laor, A., 
1160: Mushotzky, R. F., Nandra, K., \& Takahashi, T. 2000, \apj, 531, 52
1161: 
1162: %\bibitem[Gilli et al.(1999)]{Gilli99} Gilli, R., Risaliti, G., \& Salvati, M. 1999, \aap, 347, 424
1163: 
1164: %\bibitem[Gilli et al.(2001)]{Gilli01} Gilli, R., Salvati, M., \& Hasinger, G. 2001, \aap, 366, 407
1165: 
1166: \bibitem[Granato et al.(2004)]{Granato04} Granato, G. L., De Zotti, G., Silva, L., Bressan, A. 
1167: \& Danese, L. 2004, \apj, 600, 580
1168: 
1169: \bibitem[Grogin et al.(2005)]{Grogin05} Grogin, N. A., et al. 2005, \apj, 627, 97
1170: 
1171: \bibitem[Groth et al.(1994)]{Groth94}Groth, E. J., et al. 1994, \baas, 26, 1403
1172: 
1173: \bibitem[Hasinger (2003a)]{Hasinger03} Hasinger, G. 2003a, in "The Emergence of Cosmic Structure", ed. S. S.Holt, \&
1174: C. S. Reynolds, AIP Conf. Proc., 666, 227
1175: 
1176: \bibitem[Hasinger (2003b)]{Hasinger03b} Hasinger, G. 2003b, in "The restless high energy universe", ed. E.
1177: P. J. van den Heuvel, J. J. M. in 't Zand, and R. A. M. J. Wijers, Nucl. Physics B. Suppl. Ser.
1178: (astro-ph/0310804)
1179: 
1180: \bibitem[Hasinger et al.(2005)]{Hasinger05}, Hasinger, G., Miyaji, T., \& Schmidt, M. 2005, \aap, 441, 417
1181: 
1182: \bibitem[Hatziminaoglou et al.(2005)]{Hatziminaoglou05} Hatziminaoglou, E., et al. 2005, \aj, 129, 1198
1183: 
1184: %\bibitem[Heavens et al.(2004)]{Heavens04} Heavens, A., Panter, B., Jim\'{e}nez, R., \& Dunlop, J. 2004, 
1185: %\nat, 428, 625
1186: 
1187: \bibitem[Hornschemeier et al.(2001)]{Hornschemeier01} Hornschemeier, A. E., et al. 2001, \apj, 554, 742
1188: 
1189: %\bibitem[Houck et al.(2005)]{Houck05} Houck, J. R., et al. 2005, \apj, 622, 105
1190: 
1191: \bibitem[Jannuzi \& Dey (1999)]{Jannuzi99} Jannuzi, B. T., \& Dey, A. 1999, in ASP Conf. Ser. 191, 
1192: Photometric Redshifts and the Detection of High Redshift Galaxies, ed. R. Weymann et al. (San Francisco: ASP),
1193: 11 
1194: 
1195: \bibitem[Kauffmann et al.(2003)]{Kauffmann03} Kauffmann, G., et al. 2003, \mnras, 346, 1055
1196: 
1197: \bibitem[Kitsionas et al.(2005)]{Kitsionas05} Kitsionas, S., Hatziminaoglou, E., Georgakakis, A., \& 
1198: Georgantopoulos, I. 2005, \aap, 434, 475
1199: 
1200: \bibitem[Kron (1980)]{Kron80} Kron, R. G. 1980, \apjs, 43, 305
1201: 
1202: \bibitem[Krumpe et al.(2007)]{Krumpe07} Krumpe, M., et al. 2007, \aap, 466, 41
1203: 
1204: \bibitem[Kuraszkiewicz et al.(2003)]{Kuraszkiewicz03} Kuraszkiewicz, J. K., et al. 2003, \apj, 590, 128
1205: 
1206: \bibitem[Lacy et al.(2004)]{Lacy04} Lacy, M., et al. 2004, \apjs, 154, 166
1207: 
1208: \bibitem[Lacy et al.(2007)]{Lacy07} Lacy, M., Petric, A. O., Sajina, A., Canalizo, G., Storrie-Lombardi, L. J.,
1209: Armus, L., Fadda, D., \& Marleau, F. R. 2007, \aj, 133, 186
1210: 
1211: \bibitem[LaFranca et al.(1995)]{LaFranca95} LaFranca, F., Franceschini, A., Cristiani, S., \& Vio, R. 1995, 
1212: \aap, 299, 19
1213: 
1214: \bibitem[LaFranca et al.(2005)]{LaFranca05} LaFranca, F., et al. 2005, \apj, 635, 864
1215: 
1216: \bibitem[Lonsdale et al.(2003)]{Lonsdale03} Lonsdale, C. J., et al. 2003, \pasp, 115, 897
1217: 
1218: %\bibitem[Madau et al.(1994)]{Madau94} Madau, P., Ghisellini, G., \& Fabian, A. C. 1994, \mnras, 270, 17
1219: 
1220: \bibitem[Mainieri et al.(2002)]{Mainieri02} Mainieri, V., et al. 2002, \aap, 393, 425
1221: 
1222: \bibitem[Mart\'\i nez-Sansigre et al.(2005)]{Martinez-Sansigre05} Mart\'\i nez-Sansigre, A., Rawlings, S., 
1223: Lacy, M., Fadda, D., Marleau, F. R., Simpson, C., Willott, C. J., \& Jarvis, M. J. 2005, \nat, 436, 666
1224: 
1225: \bibitem[Mart\'\i nez-Sansigre et al.(2007)]{Martinez-Sansigre07} Mart\'\i nez-Sansigre, A., et al. 2007, \mnras,
1226: 379, L6
1227: 
1228: \bibitem[Mushotzky (2004)]{Mushotzky04} Mushotzky, R. 2004, in Suppermassive Black Holes in the Distant
1229: Universe, ed. A. J. Barger (Dordrecht: Kluwer), 53
1230: 
1231: \bibitem[Nandra et al.(2005)]{Nandra05} Nandra, K., et al. 2005, \mnras, 356, 568
1232: 
1233: %\bibitem[Norman et al.(2002)]{Norman02} Norman, C., et al. 2002, \apj, 571, 218
1234: 
1235: \bibitem[Park et al.(2008)]{Park08} Park, S. Q., et al. 2008, \apj, 678, 744
1236: 
1237: \bibitem[P\'{e}rez-Gonz\'{a}lez et al.(2008a)]{Perez08} P\'{e}rez-Gonz\'{a}lez, P. G., et al. 2008, \apj, 675, 234
1238: 
1239: \bibitem[P\'{e}rez-Gonz\'{a}lez et al.(2008b)]{Perez08b} P\'{e}rez-Gonz\'{a}lez, P. G.,  Trujillo, I., 
1240: Barro, G., Gallego, J., Zamorano, J., \& Conselice, C. J. 2008, \apj, in press, arXiv:0807.1069
1241: 
1242: \bibitem[P\'{e}rez-Gonz\'{a}lez et al.(2005)]{Perez05} P\'{e}rez-Gonz\'{a}lez, P. G., et al. 2005, \apj, 630, 82
1243: 
1244: \bibitem[Peterson et al.(2006)]{Peterson06} Peterson, K. C., Gallagher, S. C., Hornschemeier, A. E., 
1245: Muno, M. P., \& Bullard, E. C. 2006, \aj, 131, 133
1246: 
1247: \bibitem[Piccinotti et al.(1982)]{Piccinotti82} Piccinotti, G., Mushotzky, R. F., Boldt, E. A., Holt, S. S., 
1248: Marshall, F. E., Serlemitsos, P. J., \& Shafer, R. A. 1982, \apj, 253, 485
1249: 
1250: \bibitem[Pierce et al.(2007)]{Pierce07} Pierce, C. M., et al. 2007, \apj, 660, 19
1251: 
1252: %\bibitem[Poggianti (1997)]{Poggianti97} Poggianti, B. M. 1997, \aaps, 122, 399
1253: 
1254: \bibitem[Polletta et al.(2007)]{Polletta07} Polletta, M. et al. 2007, \apj, 663, 81
1255: 
1256: \bibitem[Risaliti, Maiolino \& Salvati (1999)]{Risaliti99} Risaliti, G., Maiolino, R., \& Salvati, M. 1999, \apj, 522, 157
1257: 
1258: \bibitem[Schawinski et al.(2007)]{Schawinski07} Schawinski, K., et al. 2007, \mnras, 382, 1415
1259: 
1260: \bibitem[Springel et al.(2005)]{Springel05} Springel, V., Di Matteo, T., \& Hernquist, L. 2005, \mnras, 361,776
1261: 
1262: \bibitem[Steffen et al.(2006)]{Steffen06} Steffen, A. T., et al. 2006, \aj, 131, 2826
1263: 
1264: \bibitem[Steidel et al.(1999)]{Steidel99} Steidel, C. C., Adelberger, K. L., Giavalisco, M., Dickinson, M., 
1265: \& Pettini, M. 1999, \apj, 519, 1
1266: 
1267: \bibitem[Stern et al.(2005)]{Stern05} Stern, D., et al. 2005, \apj, 631, 163 
1268: 
1269: \bibitem[Strateva et al.(2005)]{Strateva05} Strateva, I. V., Brandt, W. N., Schneider, D. P., 
1270: Vanden Berk, D. G., \& Vignali, C. 2005, \aj, 130, 387
1271: 
1272: \bibitem[Szokoly et al.(2004)]{Szokoly04} Szokoly, G. P., et al. 2004, \apjs, 155, 271
1273: 
1274: \bibitem[Ueda et al.(2003)]{Ueda03} Ueda, Y., Akiyama, M., Ohta, K., \& Miyaji, T. 2003, \apj, 598, 886
1275: 
1276: \bibitem[Vignali et al.(2003)]{Vignali03} Vignali, C., Brandt, W. N., \& Schneider, D. P. 2003, \aj, 125, 433
1277: 
1278: \bibitem[Villar et al.(2008)]{Villar08} Villar, V., et al. 2008, \apj, 677, 169
1279: 
1280: \bibitem[Waddington et al.(2001)]{Waddington01} Waddington, I., Dunlop, J. S., Peacock, J. A., Windhorst,
1281: R. A. 2001, \mnras, 328, 882
1282: 
1283: \bibitem[Waskett et al.(2003)]{Waskett03} Waskett, T. J., et al. 2003, \mnras, 341, 1217
1284: 
1285: \bibitem[Weiner et al.(2005)]{Weiner05} Weiner, B. J., et al. 2005, \apj, 620, 595
1286: 
1287: \bibitem[Wilkes et al.(1994)]{Wilkes94} Wilkes, B. J., Tananbaum, H., Worrall, D. M., Avni, Y., 
1288: Oey, M. S., \& Flanagan, J. 1994, \apjs, 92, 53
1289: 
1290: %\bibitem[Worsley et al.(2005)]{Worsley05} Worsley, M. A., et al. 2005, \mnras, 357, 1281
1291: 
1292: \bibitem[Zezas, Alonso-Herrero \& Ward (2001)]{Zezas01} Zezas, A., Alonso-Herrero, A., \& Ward, M. J. 2001,
1293: \apss, 276, 601
1294: 
1295: \bibitem[Zheng et al.(2004)]{Zheng04} Zheng, W., et al. 2004, \apjs, 155, 73
1296: 
1297: \end{thebibliography}
1298: 
1299: \clearpage
1300: 
1301: \input{tab1.tex}
1302: 
1303: \clearpage
1304: 
1305: \input{tab2.tex}
1306: 
1307: \clearpage
1308: 
1309: \input{tab3.tex}
1310: 
1311: \clearpage
1312: 
1313: \input{tab4.tex}
1314: 
1315: \clearpage
1316: 
1317: \input{tab5.tex}
1318: 
1319: \end{document}
1320: 
1321: