1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass{emulateapj}
3:
4: \received{}
5: \accepted{}
6:
7: \slugcomment{Accepted by ApJ}
8: \shortauthors{Kuraszkiewicz et al.}
9: \shorttitle{PCA of the SED and Emission Line Properties of Red 2MASS AGN}
10:
11: \def\etal {{\it et~al.}}
12: \def\alpe {{$\alpha_E$}}
13: \def\nh {${\rm N_H}$}
14: \def\lax {${_<\atop^{\sim}}$}
15: \def\gax {${_>\atop^{\sim}}$}
16: \def\chandra {{\it Chandra}}
17: \def\jmk {{J$-$K$_S$}}
18: \def\O3 {{[O\,{\small III}]}}
19: \def\O2 {{[O\,{\small II}]}}
20: \newcommand{\kms}{\ifmmode~{\rm km~s}^{-1}\else ~km~s$^{-1}~$\fi}
21:
22: \def\plotfiddle#1#2#3#4#5#6#7{\centering \leavevmode
23: \vbox to#2{\rule{0pt}{#2}}
24: \special{psfile=#1 voffset=#7 hoffset=#6 vscale=#5 hscale=#4 angle=#3}}
25:
26: \begin{document}
27:
28:
29: \title{PCA of the Spectral Energy Distribution and Emission Line
30: Properties of Red 2MASS AGN}
31: \author{Joanna Kuraszkiewicz\altaffilmark{1}, Belinda
32: J. Wilkes\altaffilmark{1}, Gary Schmidt\altaffilmark{2},
33: Paul S. Smith\altaffilmark{2}, Roc Cutri\altaffilmark{3}, Bo\.zena
34: Czerny\altaffilmark{4}}
35: \altaffiltext{1}{Harvard-Smithsonian Center for Astrophysics, Cambridge, MA 02138}
36: \altaffiltext{2}{Steward Observatory, University of Arizona, Tucson, AZ 85721}
37: \altaffiltext{3}{IPAC, Caltech, MS 100-22, Pasadena, CA 91125}
38: \altaffiltext{4}{Nicolaus Copernicus Astronomical Center, Warsaw, Poland}
39:
40: \begin{abstract}
41:
42:
43: We analyze the spectral energy distributions (SEDs) and emission line
44: properties of the red (\jmk\ $> 2$) 2MASS AGN using principle
45: component analysis (PCA). The sample includes 44 low redshift AGN with
46: low or moderate obscuration (\nh\ $< 10^{23}$~cm$^{-2}$) as indicated
47: by X-rays and SED modeling. The obscuration of the AGN allows us to
48: see weaker emission components (host galaxy emission, AGN scattered
49: light) which are usually outshone by the AGN. The first four
50: eigenvectors explain 70\% of the variance in the sample. Eigenvector~1
51: (33\% variance in the sample) correlates with the ratios of the
52: intrinsic X-ray flux to the observed optical/IR fluxes and the
53: F(2$-$10keV)/F([O\,{\small III}]) ratio. We suggest that it is
54: primarily driven by the $L/L_{Edd}$ ratio and strengthened by
55: intrinsic absorption (both circumnuclear and galactic). Eigenvector~2
56: (18\% of variance) correlates with optical/IR colors (B$-$K$_S$,
57: B$-$R, \jmk ) and optical spectral type and depends on the
58: contribution of the host galaxy relative to the observed AGN emission.
59: Eigenvector~3 (12\% of variance) correlates with reddening indicators
60: obtained from the X-rays (hardness ratio, spectral index, \nh ) and
61: the narrow H$\alpha$/H$\beta$ emission line ratio. Their relation
62: suggests a common absorber for the optical/X-rays lying outside the
63: narrow-line region (NLR), possibly in a moderately inclined host
64: galaxy. Eigenvector~4 (8\% of variance) correlates with the degree of
65: polarization and the broad H$\alpha$/H$\beta$ ratio, indicating that
66: dust which scatters the nuclear emission (continuum and the broad-line
67: region emission) also reddens the broad-lines. Our analysis shows
68: that, although as suggested by unification schemes, the inclination
69: dependent obscuration (circumnuclear and the host galaxy) is important
70: in determining the AGN SEDs, the $L/L_{Edd}$ ratio is the most
71: important factor, followed by host galaxy emission.
72:
73:
74:
75: \end{abstract}
76:
77:
78: \keywords{galaxies: active --- quasars: general}
79:
80:
81: \section{Introduction}
82:
83: Principal Component Analysis (PCA) is a mathematical tool used to
84: reduce the dimensionality of a dataset without losing information. It
85: transforms the old coordinate system where variables correlate among
86: each other into a new coordinate system defined by a smaller number of
87: uncorrelated variables called principal components. The first
88: principal component accounts for the most variance in the data, and
89: the successive principal components explain as much of the remaining
90: variance in the data as possible (see e.g. an introduction to PCA by
91: Francis \& Wills 1999 or Murtagh \& Heck 1987).
92:
93: Since QSOs show a plethora of emission line and continuum properties,
94: PCA is ideal for studying their properties and understanding which are
95: most important. PCA was first introduced to the AGN world by Boroson
96: \& Green (1992; hereafter BG92) who analyzed the optical emission line
97: and continuum properties of a complete, low redshift, Bright Quasar
98: Survey (BQS) sample. It was found that the primary eigenvector (EV1)
99: is anticorrelated with Fe\,{\small II}\,$\lambda 4570$ strength
100: (equivalent width and Fe\,{\small II}/H$\beta$ ratio), correlated with
101: [O\,{\small III}]\,$\lambda5007$ strength (luminosity and peak) and
102: H$\beta$ FWHM, and anticorrelated with the blue asymmetry of the
103: H$\beta$ line. Hence BG92 concluded that EV1 is driven by a physical
104: parameter that ``... balances Fe\,{\small II} excitation against
105: illumination of the narrow line region''. The BG92 EV1 was also found
106: to correlate with the soft X-ray properties: luminosity and spectral
107: index (Boroson \& Green 1992; Corbin 1993; Laor et al. 1994, 1997,
108: Wang \etal\ 1996, Brandt \& Boller 1998) and the correlations were
109: found to be stronger than with the individual emission line parameters
110: (Brandt \& Boller 1998). This implied that EV1 has a more fundamental
111: physical meaning. Physical parameters that have been suggested to
112: drive EV1 include: the $L/L_{Edd}$ ratio, orientation, and black hole
113: spin. Since EV1 is correlated with [O\,{\small III}] emission,
114: thought to be an isotropic property, BG92 concluded that EV1 is not
115: driven by orientation. Although the isotropy of the [O\,{\small III}]
116: emission has since been called into question (Jackson \& Browne 1990,
117: Di Serego Alighieri et al. 1997, Tadhunter et al. 1998) the
118: correlation of EV1 with the inclination independent [O\,{\small
119: II}]\,$\lambda3727$ emission, originating at larger distances from the
120: nucleus than [O\,{\small III}], strongly supports the EV1 orientation
121: independence (Kuraszkiewicz \etal\ 2000), at least for the radio-quiet
122: subsample of the BQS. BG92 suggest that the $L/L_{Edd}$ ratio is the
123: primary driver of EV1.
124:
125: BG92 EV1 was also found to correlate with the UV emission line
126: properties: CIII] width, SiIII]/CIII] ratio, CIV and NV strength, CIV
127: line shift (Wills \etal\ 1999, Wills, Shang, \& Yuan 2000, Sulentic
128: \etal\ 2000), suggesting that EV1 is also linked to the physics of
129: fueling and outflow and that higher accretion rates are linked to
130: denser gas and possibly higher nuclear starburst activity (Wang
131: \etal\ 2006).
132:
133:
134: By combining observational data together with numerical simulations,
135: Marziani \etal\ (2001) and Sulentic \etal\ (2000) were able to
136: reproduce the correlations between optical emission line properties
137: found in EV1 and confirm that EV1 is related to $L/L_{Edd}$, but
138: suggest that, in their sample, orientation is also important. Their
139: sample includes both the BQS quasars from BG92 (face-on, low
140: obscuration), and inclined/reddened sources of intermediate type and
141: Type~2 AGN that have {\it FOS/HST} spectra.
142:
143: A different way to approach the analysis of AGN properties is spectral
144: PCA first applied by Francis et al. (1992) to the analysis of more
145: than 200 spectra from the Large Bright QSO Survey (LBQS; redshift
146: range: $0.2\le z < 3.4$)
147: %Their first three spectral components (75\% of variance in sample)
148: %included: emission--line core strength, continuum slope, and broad
149: %absorption features.
150: or by Shang \etal\ (2003) to the spectra of PG QSOs. Their analysis is
151: superseded by Yip \etal\ (2004) who analyze spectra of more than
152: 16,000 QSOs from the Sloan Digital Sky Survey (SDSS) with a wider
153: range of redshifts ($0<z<5.4$), inclinations and/or reddening
154: values. Their spectral PCA found the following principal components:
155: mean spectrum, host galaxy component, UV--optical continuum slope, and
156: correlations of Balmer emission lines. Although the results obtained
157: from spectral PCA and emission line+SED PCA are difficult to compare,
158: it is interesting to invoke them here for comparison with our results.
159:
160: In this paper we analyze a sample of red (\jmk\ $>2$) low-redshift
161: 2MASS AGN with moderate absorption \nh \lax $10^{23}$cm$^{-2}$. Their
162: SEDs were analyzed in detail in Kuraszkiewicz \etal\ (2008; hereafter
163: Paper~I) and the emission line properties are presented in
164: Kuraszkiewicz \etal\ (2008 in preparation; hereafter Paper~II). In
165: this paper we run PCA on the SED and optical emission line correlation
166: matrix and present our interpretation of the eigenvectors in this
167: moderately absorbed 2MASS AGN sample. It is important to state here
168: that the results of the PCA analysis are dependent on the selection of
169: the sample and chosen properties. We do not expect the same
170: eigenvectors in the analysis of the 2MASS AGN sample as those found by
171: BG92 in the unobscured, face-on PG QSO sample. The PG QSOs' first two
172: eigenvectors were dominated by the parameters of the central engine
173: ($L/L_{Edd}$ and accretion rate -- Boroson 2002), while our 2MASS AGN
174: sample, due to AGN obscuration, is well suited for studies of weaker
175: components such as the host galaxy emission and intrinsic absorption.
176:
177:
178:
179: \section{The Sample and its Properties}
180:
181: The sample includes 44 red (\jmk\ $>2$) 2MASS AGN selected to have
182: B$-$K$_S>4.3$, and $K_S<13.8$, and observed by \chandra\ (see Paper~I
183: for more details). The sources have K band luminosities comparable to
184: those of quasars (Cutri \etal\ 2002) and lie at low redshift ($z <
185: 0.37$). The sample shows a wide range of observed
186: $K_S$-band-to-X-ray(1keV) slopes (1.1\lax $\alpha_{KX}$\lax 2; Wilkes
187: \etal\ 2002), and a broad range of observed polarization fraction at R
188: band ($0\% < P < 13\%$; Smith \etal\ 2002, 2003). As found in Paper~I
189: the sample's median SED is redder than that of the blue
190: optically/radio selected QSOs (Elvis \etal\ 1994) or the hard--X--ray
191: selected AGN (Kuraszkiewicz \etal\ 2003), showing little or no big
192: blue bump and implying that near-IR color selection isolates the
193: reddest subset of AGN that can be classified optically. Modeling of
194: the near-IR/optical colors and X-ray spectral analysis in Paper~I
195: implies moderate amounts of reddening (\nh\ $< few \times
196: 10^{22}$~cm$^{-2}$). The more highly obscured AGN in this sample are
197: most likely viewed at intermediate angles, where the central engine is
198: obscured from our view by either circumnuclear dust or dust lying
199: further out in the host galaxy inclined to our line of sight. The
200: obscuration allows us to see/study the contributions of weaker
201: emission components such as host galaxy emission or AGN scattered
202: light emission which are normally outshone by the bright AGN. The
203: sample includes: 7 Type~1 AGN with optical spectra resembling NLS1 or
204: BALQSOs, 11 Type~2s with low (for a Type~2) \nh\ $\sim
205: 10^{22}$cm$^{-2}$, and 26 intermediate type sources. Many of the 2MASS
206: AGN studied here also show high optical polarization ($P > 4$\%).
207:
208:
209:
210: \section{Principle Component Analysis Results}
211:
212: In order to investigate the relative importance of various parameters
213: in determining the SED shapes and emission--line properties in the red
214: 2MASS sources, we analyze the relations between the continuum
215: parameters and emission-line parameters by running principal component
216: analysis (PCA) on the correlation matrix. The first PCA run included
217: all continuum properties analyzed in Paper~I (Sections~4,5,6): optical
218: and near-IR colors (B$-$R, B$-$K$_S$, \jmk ), optical type, optical
219: slope, $\alpha_{opt}$, from optical spectra, degree of polarization,
220: redshift, X-ray slope, intrinsic absorption \nh , X-ray hardness
221: ratio, X-ray (1keV) luminosity (all from \chandra\ spectral fitting),
222: bolometric luminosity from SEDs, various IR/optical/X-ray flux or
223: luminosity ratios\footnote{Flux/luminosity ratios included:
224: F(1keV)/F$_B$, F(1keV)/F$_R$, F(1keV)/F$_I$, F(1keV)/F$_K$,
225: L(0.2-0.4$\mu$m)/L(0.4-0.8$\mu$m), L(0.2-0.4$\mu$m)/L(0.8-1.6$\mu$m),
226: L(0.4-0.8$\mu$m)/L(0.8-1.6$\mu$m), L(1-10$\mu$m)/L(10-100$\mu$m),
227: L(3-60)$\mu$m)/L(60-100$\mu$m), L(0.8-1.6$\mu$m)/L(1-100$\mu$m),
228: L(0.8-1.6$\mu$m)/L(60-100$\mu$m), L(0.8-1.6$\mu$m)/L(3-60$\mu$m)}, and
229: the emission--line parameters from Paper~II: $W_{\lambda}$, FWHM,
230: shift from systemic redshift of: H$\beta$ (broad and narrow
231: components), H$\alpha$ (broad and narrow components), [O\,{\small
232: III}], [O\,{\small II}], optical Fe\,{\small II}, and the following
233: line ratios: [O\,{\small II}]/[O\,{\small III}],
234: H$\beta^{broad}$/[O\,{\small III}] , H$\beta^{narrow}$/[O\,{\small
235: III}], H$\alpha^{narrow}$/H$\beta^{narrow}$,
236: H$\alpha^{broad}$/H$\beta^{broad}$, Fe\,{\small II}/H$\beta^{broad}$,
237: Fe\,{\small II}/H$\beta^{narrow}$, Fe\,{\small II}/[O\,{\small
238: III}]. A total of 56 parameters were analyzed. The EW, FWHM, line
239: shifts, most line ratios, and optical--to--IR and IR--to--IR
240: luminosity ratios did not contribute to the first few eigenvectors and
241: running PCA with or without them gave similar results (i.e. nearly
242: identical eigenvectors). To simplify the analysis these parameters
243: were omitted from the final PCA run. The final set of parameters is
244: shown in the first column of Table~1. Columns 2--5 show
245: their relation with the first four eigenvectors (bold font denoting
246: parameters that dominate the eigenvector) and quote in percentages how
247: much variance they explain. Together the first four eigenvectors
248: explain 70.3\% of the variance in the sample. Their interpretation is
249: presented below.
250:
251:
252:
253: \subsection {Eigenvector~1}
254:
255: Most of the variance (32.9\%) in the sample is due to the correlations
256: between the ratios of the intrinsic 1~keV X-ray flux to the observed
257: optical/IR flux: $F(1keV)/F_B$, $F(1keV)/F_R$, $F(1keV)/F_I$,
258: $F(1keV)/F_J$, $F(1keV)/F_K$ (see Fig.~\ref{fig:Fx/FBvsFx/Fnn}). We
259: choose to use the observed optical and IR fluxes since, in this
260: moderately obscured/inclined sample, these fluxes depend on intrinsic
261: reddening and host galaxy emission, which are a priori unknown and
262: hence difficult to subtract. The X-ray fluxes, on the other hand, are
263: corrected for intrinsic absorption, which was estimated from the
264: \chandra\ X-ray spectral fitting in the following way: the higher
265: (\gax 80) count sources were fit with a power-law and rest-frame
266: absorption with both parameters free (hereafter referred to as the
267: ``{\bf C}'' fits), the medium ($\sim 30-80$) count sources, were fit
268: with a fixed power-law with an X-ray photon index $\Gamma=2$ and \nh\
269: free (``{\bf B}'' fits). For the lowest count sources (\lax 30
270: counts; ``{\bf A}'' fits), fits were made using a power-law with fixed
271: $\Gamma = 2$ and \nh\ $= 7.6\times10^{21}$~cm$^{-2}$ (the median
272: absorption from the {\it C} fits). The importance of the optical/X-ray
273: flux ratio can be understood, for example, in the disk-wind model
274: paradigm (Murray \& Chiang 1990, Proga \etal\ 2005, and references
275: therein), which predict coupling between processes producing X-ray and
276: optical/UV photons. EV~1 is also strongly anticorrelated with the
277: $F([O\,{\small III}])/F(2-10keV)$ ratio (see Fig.~\ref{fig:EV1}).
278:
279: To understand the meaning of EV1 we need to take a closer look at the
280: objects with extreme values of EV1. Their SEDs and optical spectra are
281: presented in Figures~\ref{fig:extEV1low} and \ref{fig:extEV1high}, and
282: their properties are summarized in Table~2. We find that the most
283: negative EV1 sources (Fig.~\ref{fig:extEV1low}) with lowest
284: $F(1keV)/F_{B,R,I,J,K}$ and highest F([O\,{\small III}])/F(2$-$10keV),
285: are Type~1--1.5 with blue optical spectra dominated by strong
286: Fe\,{\small II} and weak (relatively to H$\beta$) [O\,{\small III}]
287: emission, that resemble the spectra of NLS1s and BALQSOs. Although
288: their low S/N X-ray spectra ({\it A} fits) do not yield an estimate
289: for \nh , modeling of B$-$R and \jmk\ colors in Paper~I (Section~5)
290: gave low \nh\ \lax $5\times10^{21}$~cm$^{-2}$ (hence their X-ray
291: fluxes are likely overestimated since \nh\ $=
292: 7.6\times10^{21}$~cm$^{-2}$ was assumed for these {\it A} fits during
293: spectral fitting) and shows that the SEDs at optical wavelengths are
294: dominated by AGN emission ($>$94\% of total observed flux at R band is
295: due to the AGN in these sources).
296:
297: The most positive EV1 sources (Fig.~\ref{fig:extEV1high}) with highest
298: $F(1keV)/F_{B,R,I,J,K}$ and lowest F([O\,{\small III}])/F(2$-$10keV),
299: are Type~1.9--2, with red spectra showing strong [O\,{\small III}] and
300: [O\,{\small II}] emission, and weak Fe\,{\small II} and H$\beta$
301: emission and strong galactic features. X-ray spectral fitting gives
302: \nh\ $\sim 10^{22}$~cm$^{-2}$ consistent with reddening obtained from
303: the modeling of B$-$R and \jmk\ colors in Paper~I (Section~5), that
304: also shows high host galaxy contribution at optical wavelengths
305: ($>$61\% of the total observed flux at R band is due to host galaxy).
306:
307: The luminosity of the [O\,{\small III}]\,$\lambda$5007 emission line,
308: originating from the narrow-line region (NLR), has been suggested as
309: an indicator of the intrinsic nuclear luminosity of the AGN due to the
310: similarity of the [O\,{\small III}]-to-hard-X-ray flux ratio between
311: Seyfert~1 and Seyfert~2 galaxies (Mulchaey \etal\ 1994, Alonso-Herrero
312: \etal\ 1997, Turner \etal\ 1997). As presented in Fig.~\ref{fig:EV1}
313: the most negative EV1 sources do not follow the [O\,{\small III}] to
314: hard-X-ray relation showing higher F([O\,{\small III}])/F(2$-$10keV)
315: ratios. This indicates that in these sources the intrinsic X-rays are
316: weak. The most positive EV1 sources, on the other hand, have lower
317: F([O\,{\small III}])/F(2$-$10keV) indicating strong intrinsic X-rays.
318:
319:
320: The differences in X-ray output relative to optical/UV may be
321: explained for example by an accretion disk and corona model of Witt,
322: Czerny, \& \.Zycki (1997). In this model both the accretion disk and
323: corona accrete and generate energy through viscosity, and division of
324: the flow into optically thin (corona) and optically thick (disk)
325: regions results from the cooling instability discussed by Krolik,
326: McKee and Tarter (1981).
327: %Such a model is able to predict the fraction of the energy generated
328: %in the corona instead of adopting this quantity as a free parameter.
329: The model is defined by 3 parameters: the mass of the central black
330: hole, the ratio of the luminosity to the Eddington luminosity
331: $L/L_{Edd}$ (or the accretion rate \.M), and the viscosity parameter
332: ($\alpha_{vis}$) assumed to be the same in both the disk and the
333: corona. The model predicts a systematic change in the opt/UV/X-ray
334: SED as a function of $L/L_{Edd}$. For high $L/L_{Edd}$, the big blue
335: bump is stronger both in luminosity and relative to the X-rays
336: %so the optical--X-ray slope becomes steeper ($\alpha_{ox}$=1.5-1.6 at
337: %few tenths of the $L/L_{Edd}$)
338: so the $F(1keV)/F_B$ ratio is low (as in low EV1 sources). For low
339: $L/L_{Edd}$ the big blue bump is weaker and the $F(1keV)/F_B$ ratio
340: becomes high (as in high EV1 sources; see Fig.~\ref{fig:models}).
341: %The dominant role of $L/L_{Edd}$ ratio in determining the strength of the
342: %big blue bump is confirmed by reverberation mapping of AGN that change
343: %their optical type (e.g. NGC~5548 a Sy~1 to 1.8). These show a change
344: %in the broad emission lines time delay relative to the continuum
345: %(Bentz \etal\ 2007, Peterson \etal\ 2002) implying that not only the
346: %extinction but also the intrinsic optical/UV luminosity change.
347: Since the accretion disk and corona model predictions match those with
348: EV1, they suggest that EV1 is fundamentally related
349: (i.e. anti-correlated) to the $L/L_{Edd}$ ratio. This is further
350: strengthened by the fact that the most negative EV1 sources (with
351: lowest $F(1keV)/F_B$) show optical spectra that characterize NLS1s and
352: BALQSOs (strong Fe\,{\small II} and weak [O\,{\small III}] emission)
353: that are thought to have high $L/L_{Edd}$ (e.g. Pounds, Done \&
354: Osborne 1995, Boller, Brandt, Fink 1996, Kuraszkiewicz \etal\ 2000).
355:
356: From Table~1 and Fig.~\ref{fig:Fx/FBvsFx/Fnn}c,d we see that the
357: correlation between EV1 and the F(1keV)/F$_{B}$ ratio is stronger than
358: the correlation between EV1 and F(1keV)/F$_{K}$. The optical/IR
359: fluxes used to calculate these flux ratios are observed, not
360: intrinsic, so they are affected by dust reddening (and to a lesser
361: extent also by host galaxy emission). Fluxes at optical wavelengths
362: will be more affected than those at near-IR wavelengths. Since EV1
363: shows stronger correlation with the $F(1keV)/F_B$ ratio, which depends
364: both on the $L/L_{Edd}$ and reddening, and shows more scatter with the
365: $F(1keV)/F_K$ ratio, which depends mostly on the $L/L_{Edd}$, and less
366: on reddening, we conclude that EV1 is dominated by
367: $L/L_{Edd}$. However, dust obscuration also contributes as it
368: stretches the underlying $L/L_{Edd}$ driven correlation, at optical
369: wavelengths, in these moderately obscured 2MASS sources. If dust
370: obscuration is inclination dependent then this result is consistent
371: with Marziani \etal\ (2001) who suggest that their EV1 found for a
372: heterogeneous sample of mixed type (1, 2 and intermediate) AGN, is
373: related to the $L/L_{Edd}$ ratio convolved with orientation.
374:
375: %Since the big blue bump in high $L/L_{Edd}$ sources is strong, the AGN
376: %will dominate the host galaxy at optical/near-IR wavelengths. In the
377: %low--$L/L_{Edd}$ sources, where the big blue bump is much weaker, the
378: %AGN's contribution will be relatively lower when compared to the host
379: %galaxy. This is indeed the case in the red 2MASS AGN sample, where the
380: %negative EV1 sources have $B-R$ and \jmk\ colors dominated by the AGN,
381: %while the colors of the positive EV1 sources are dominated by the host
382: %galaxy.
383:
384:
385: \subsection{Eigenvector~2}
386:
387: Eigenvector~2 (hereafter EV2), which explains 17.8\% of the variance
388: in the sample, is dominated by the observed optical/IR colors (in
389: order of importance): B$-$K$_S$, B$-$R, \jmk , and optical spectral
390: type (see Table~1). In Paper~I we found that the optical B$-$R color,
391: and to a lesser extent the near-IR \jmk\ color in the 2MASS AGN
392: sample, are affected by reddening and host galaxy emission (and in a
393: few, highly polarized objects, also by scattered AGN light). We recall
394: the results of the optical/IR color modeling from Paper~I in
395: Fig.~\ref{fig:colors}. The colors of a pure reddened AGN (median QSO
396: SED from Elvis \etal\ 1994 reddened by Milky Way dust;
397: long-dash--short-dash line) are modified by host galaxy emission
398: (Buzzoni 2005 templates), where
399: %We use three host galaxy an elliptical galaxy with 15--Gyr
400: %stellar population (E15), Sd and Sa galaxies with 5--Gyr stellar
401: %population (Sd05 and Sa05), i.e. templates with the reddest, bluest
402: %and intermediate B$-$R color respectively.
403: a range of host galaxy strengths relative to the intrinsic, {\it
404: unreddened} AGN at R band was assumed (intrinsic AGN/host galaxy ratio
405: $= 5-40$). For $A_V$\lax 3 the optical/near--IR colors are dominated by
406: the AGN. At higher $A_V$ the B$-$R colors become bluer and more
407: consistent with the host galaxy colors (represented by thick solid
408: lines). The \jmk\ colors are much less dependent on reddening and
409: host galaxy, only visible if their contribution is strong i.e. A$_V >
410: 10$ and AGN/host galaxy ratio low (compare for example (Sd05;5) and
411: (Sd05;40) curves that have a high and low host galaxy contribution
412: respectively).
413:
414: In Table~2 we summarize the properties of extreme EV~2 objects. We
415: find that the most positive EV2 objects are the bluest in B$-$R,
416: implying a strong big blue bump. They also have the bluest B$-$K$_S$
417: colors in sample. They are Type~1s, with optical spectra showing
418: strong Fe\,{\small II} and weak [O\,{\small III}] emission, resembling
419: spectra of NLS1 and BALQSO. Column densities derived from X-ray
420: spectral fitting are low (\lax 1.5$\times 10^{21}$~cm$^{-2}$), and the
421: optical/IR colors are modeled with a pure, unreddened AGN with no host
422: galaxy contribution.
423:
424: The most negative EV2 objects are red in B$-$K$_S$, B$-$R and \jmk\
425: color. The red optical/IR colors are modeled with an AGN reddened by
426: dust with $A_V$=10--11~mag. Such reddening is sufficient to fully
427: obscure the AGN at optical wavelengths, giving way to a large (98\% at
428: R band) contribution from the host galaxy. At near-IR wavelengths the
429: reddened AGN colors still dominate in these sources. The red \jmk\ $>
430: 3$ color can only be produced by a reddened AGN and not the host
431: galaxy which has bluer colors: \jmk\ = 0.8$-$1 (see
432: Fig.~\ref{fig:colors}).
433:
434: In Fig.~\ref{fig:EV2} we divide the sources according to the amount of
435: host galaxy contribution relative to the total emission at R band
436: obtained from the optical/IR color modeling (numbers are from Table~7
437: in Paper~I). We find that the amount of host galaxy contribution
438: relative to the reddened AGN decreases with increasing EV2
439: values. This implies that EV2 depends on the amount of host galaxy
440: contribution, relative to the observed/reddened AGN, where at one
441: (negative) end of EV2 reside pure, unreddened AGN dominating the
442: output at optical/near-IR wavelengths, and on the other (positive) end
443: reside reddened AGN whose output at optical wavelengths is dominated
444: by host galaxy emission. A stronger correlation of EV2 with the B$-$R
445: color than the \jmk\ color demonstrates that the B$-$R color is much
446: more sensitive to host galaxy contamination than the \jmk\ color, as
447: shown by the color-color modeling (see Fig.~\ref{fig:colors} or
448: detailed discussion in Paper~I). Yip \etal\ (2004), who analyze the
449: SDSS QSO spectra (sample with a wide range of redshifts, inclination
450: angles and reddening values) using {\it spectral} PCA, also find that
451: their second principle component is related to the host galaxy.
452:
453:
454:
455:
456: \subsection{Eigenvector~3}
457:
458: Eigenvector~3 (hereafter EV3; 11.6\% of the variance) is dominated (in
459: order of importance) by: X-ray hardness ratio, X-ray spectral index
460: $\Gamma_X$, column density \nh\ (measured only in the high and medium
461: S/N \chandra\ spectra i.e. in the {\it C} and {\it B} fits) and the
462: narrow H$\alpha$/H$\beta$ emission line ratio. In Fig.~\ref{fig:EV3}
463: we show two representative correlations between EV3 and the X-ray
464: hardness ratio and the narrow H$\alpha$/H$\beta$ emission line ratio.
465: All of these parameters are related to the amount of obscuration in
466: X-rays or optical. Their relation implies a common absorber for the
467: optical and X-rays, lying outside the NLR, possibly in a host galaxy
468: inclined to our line of sight (as in Polletta \etal\ 2008, Deo \etal\
469: 2007). This finding is supported by the HST/WFPC2 I-band images of the
470: 2MASS AGN, which show that these sources reside mostly in host
471: galaxies with inclination angles of $i=50-75^{o}$ (mean i=67$^{o}$;
472: Marble \etal\ 2003). We conclude that EV~3 is likely dominated by
473: host galaxy absorption.
474:
475: The presence of redshift as a component of EV3 (Table~1) is secondary
476: to the presence of \nh , for which measurements become less sensitive
477: as redshift increases and the absorption is shifted out of the soft
478: X-ray band. The correlation coefficient reduces significantly if \nh\
479: is corrected for redshift (assuming it is intrinsic).
480:
481:
482: %The 2MASS host galaxies were also found by Hutchings \etal\ (2003; R
483: %band images) to have a higher degree of tidal interaction when
484: %compared to low-z, blue QSOs, where 30\% of the 2MASS sources are
485: %highly interacting and 70\% show some sign of interaction, while only
486: %11\% of the low-z, blue QSOs are highly interacting, and 39\% show
487: %some sign of interaction. A higher degree of interaction in the 2MASS
488: %sources may imply higher amounts of obscuring dust/gas in these
489: %objects. Hutchings \etal\ finding was not confirmed by Marble \etal\
490: %(2003), but they had I band HST/WFPC2 images, with a smaller field of
491: %view, that could miss companion galaxies at high separation angles.
492:
493:
494: \subsection{Eigenvector~4}
495:
496: Eigenvector~4 (hereafter EV4; 8.0\% of the variance) is dominated by
497: the degree of optical polarization, broad H$\alpha$/H$\beta$ emission
498: line ratio and redshift. The relation between the polarization and the
499: broad H$\alpha$/H$\beta$ ratio implies that the same dust scatters the
500: intrinsic continuum+broad lines {\it and} reddens the broad emission
501: lines. Since there is no dependence on the narrow H$\alpha$/H$\beta$
502: ratio in this eigenvector, this dust must lie outside the BLR but not
503: outside the NLR, i.e., either on the BLR/NLR border, or within the
504: NLR. This is consistent with the conclusions reached by Smith \etal\
505: (2003) based on spectropolarimetry of the 2MASS AGN. The correlation
506: between EV4 and redshift is due to a correlation between redshift and
507: polarization (chance probability $\sim$ 2\%; there is no correlation
508: with the broad H$\alpha$/H$\beta$ ratio - chance probability
509: $>$60\%). This may be a selection effect as higher redshift sources
510: will have a higher measured degree of polarization due to the
511: wavelength dependence of dust scattering and lower host galaxy
512: dilution as one moves into restframe UV where the stellar contribution
513: decreases.
514:
515:
516:
517:
518: \section{Conclusions}
519:
520: PCA analysis of the SED and emission line parameters of a sample of 44
521: red \jmk\ $> 2$, moderately obscured (\nh\ $< 10^{23}$~cm$^{-2}$)
522: 2MASS AGN shows that 70\% of the variance in the sample is explained
523: by four eigenvectors, each having a physical explanation:
524:
525: \begin{itemize}
526: \item{Eigenvector~1 (33\% of variance in sample) correlates with the
527: intrinsic X-ray to observed optical, IR and [O\,{\small III}] flux
528: ratios, and is strongly related to the $L/L_{Edd}$ ratio, as shown by
529: disk-corona models, and strengthed by intrinsic reddening.}
530: \item{Eigenvector~2 (18\% of variance) correlates with optical/IR
531: colors and optical type, and, as shown by the optical/IR color
532: modeling, is related to the relative strength of the host galaxy to
533: the observed/reddened AGN.}
534: \item{Eigenvector~3 (12\% of variance) correlates with reddening
535: indicators obtained from X-rays and narrow emission lines, indicating
536: a common absorber that affects both the X-rays and narrow emission
537: lines and probably lies in a moderately inclined host galaxy.}
538: \item{Eigenvector~4 (8\% of variance) correlates with the degree of
539: polarization and broad H$\alpha$/H$\beta$ ratio implying the presence
540: of dust (different from the host galaxy absorber in EV3) that scatters
541: both the BLR and the continuum emission and reddens the broad lines.}
542:
543: \end{itemize}
544:
545:
546: PCA analysis of the SED and/or emission line properties of unobscured
547: (BG92) and moderately obscured AGN (Marziani \etal\ 2001, red 2MASS
548: sample presented here) points to the $L/L_{Edd}$ ratio as the most
549: important parameter in determining the SED and emission line
550: properties in AGN. Since the red 2MASS sample has moderate
551: obscuration, EV1 is also slightly effected by reddening, as is the EV1
552: found by Marziani \etal\ (2001) in their heterogeneous sample of
553: moderately obscured AGN. Successive eigenvectors pick up components
554: depending on the sample chosen: unobscured AGN samples (BG92) probe
555: the properties of the bright central engine (BG92 EV2 is related to
556: accretion rate - see Boroson 2002). Moderately obscured samples, such
557: as the red 2MASS AGN, pick up weaker components, like the host galaxy
558: emission, which become important as the reddening obscures the AGN
559: emission. Spectral PCA of the moderately obscured SDSS QSOs (Yip
560: \etal\ 2004) also finds its EV2 to be related to the host galaxy.
561: Successive eigenvectors in the red 2MASS sample are related to dust
562: obscuration, separated into circumnuclear and host galaxy
563: absorption. Our analysis is consistent with the orientation-dependent
564: unification scheme for AGN, where a range of inclination dependent
565: obscuration and host galaxy contribution (as suggested by Haas \etal\
566: 2008) and most importantly $L/L_{Edd}$ ratios (as suggested by our
567: PCA) can explain the properties of an AGN.
568:
569:
570:
571: \acknowledgements
572:
573:
574: We wish to thank the referee, Gordon Richards, for comments that helped
575: improve the paper and Martin Gaskell for useful discussions. BJW and
576: JK gratefully acknowledge the financial support of NASA Chandra
577: grants: GO1-2112A, GO3-4138A and NASA XMM-Newton grants: NNG04GD27G,
578: NNG05GM24G, which supported various aspects of this work. We also
579: gratefully acknowledge the financial support of grants: NAS8-39073,
580: GO-09161.05-A (HST). PSS acknowledges support from NASA/JPL contract
581: 1256424. This publication makes use of data products from the Two
582: Micron All Sky Survey, which is a joint project of the University of
583: Massachusetts and the Infrared Processing and Analysis
584: Center/California Institute of Technology, funded by the National
585: Aeronautics and Space Administration and the National Science
586: Foundation. We wish to thank F. Murtagh for making his PCA program
587: available on the web.
588:
589:
590:
591:
592:
593: \begin{references}
594:
595: \reference{} Alonso-Herrero, A., Ward, M. J., Kotilainen, J. K. 1997,
596: MNRAS, 288, 977
597:
598: %\reference{} Bentz, M. C., Denney, K. D., Cackett, E. M.,
599: %Dietrich, M., Fogel, J. K. J., Ghosh, H., Horne, K. D., Kuehn, C.,
600: %Minezaki, T., Onken, C. A., Peterson, B. M., Pogge, R. W.,
601: %Pronik, V. I., Richstone, D. O., Sergeev, S. G., Vestergaard, M.,
602: %Walker, M. G., Yoshii, Y. 2007, ApJ, 662, 205
603:
604:
605: \reference{} Boller, Th., Brandt, W. N., \& Fink, H. 1996, A\&A, 305, 53
606:
607: \reference{} Boroson, T. A., \& Green, R. F. 1992, ApJS, 80, 109 (BG92)
608:
609: \reference{} Boroson, T. A. 2002, ApJ, 565, 78
610:
611: \reference{} Brandt, N., \& Boller, Th. 1998, AN, 319, 163
612:
613: \reference{} Buzzoni, A. 2005, MNRAS, 361, 725
614:
615: \reference{} Corbin, M. R. 1993 ApJ, 403, L9
616:
617: \reference{} Cutri, R., Nelson, B. O., Francis, P. J. \& Smith, P. S.
618: 2002, ``AGN Surveys'', Edited by R. F. Green, E. Y. Khachikian, and
619: D. B. Sanders, Proceedings of IAU Colloquium 184, ASP Conference
620: Proceedings, Vol. 284, p. 127
621:
622: \reference{} Deo, R. P., Crenshaw, D. M., Kraemer, S. B., Dietrich,
623: M., Elitzur, M., Teplitz, H., Turner, T. J. 2007, ApJ, 671, 124
624:
625: \reference{} Elvis M., Wilkes, B. J., McDowell, J. C., Green, R. F.,
626: Bechtold, J., Willner, S. P., Oey, M. S., Polomski, E., \& Cutri, R. 1994,
627: ApJS, 95, 1
628:
629: \reference{} Francis, P.J. \& Wills, B. J., 1999, ``Quasars and
630: Cosmology'', eds. G. Ferland \& J. Baldwin, ASP Conference Series,
631: Vol. 162, 363
632:
633: \reference{} Francis, P. J., Hewett, P. C., Foltz, C. B., Chaffee,
634: F. H., 1992, ApJ, 398, 476
635:
636: \reference{} Haas, M., Willner, P. S., Heymann, F., Ashby M. L. N.,
637: Fazio, G. G., Wilkes B. J., Chini, R., Siebenmorgen, R., 2008,
638: astro--ph/0807.3966
639:
640: \reference{} Jackson, N., \& Browne, I. W. A. 1990, Nature, 343, 43
641:
642: \reference{} Krolik, J. H., McKee, C. F., Tarter, C. B. 1981, ApJ, 249,
643: 422
644:
645: \reference{} Kuraszkiewicz Wilkes, B. J., Brandt, W. N., \&
646: Vestergaard, M., 2000, ApJ, 542, 631
647:
648: \reference{} Kuraszkiewicz, J. K., Wilkes, B. J., Hooper, E. J.,
649: McLeod, K. K., Wood, K., Bjorkman, J., Delain, K. M., Hughes, D. H.,
650: Elvis, M. S., Impey, C. D., Lonsdale, C. J., Malkan, M. A.,
651: McDowell, J. C., \& Whitney, B. 2003, ApJ, 590, 128
652:
653: \reference{} Kuraszkiewicz, J. K., Wilkes, B., Schmidt, G., Ghosh, H.,
654: Smith, P. S., Cutri, R., Hines, D., Huff, E. M., McDowell, J. C.,
655: Nelson, B., 2008, ApJ submitted
656:
657: \reference{} Laor, A., Fiore, F., Elvis, M., Wilkes, B. J., \& McDowell,
658: J. C. 1994, ApJ, 435, 611
659:
660: \reference{} Laor, A., Fiore, F., Elvis, M., Wilkes, B. J., \& McDowell,
661: J. C. 1997, ApJ, 477, 93
662:
663: \reference{} Marble, A. R., Hines, D. C., Schmidt, G. D.,
664: Smith, P. S., Surace, J. A., Armus, L., Cutri, R. M., Nelson, B. 2003
665: ApJ, 590, 707
666:
667: \reference{} Marziani, P., Sulentic, J. W., Zwitter, T., Dultzin-Hacyan, D.,
668: Calvani, M., 2001, ApJ, 558, 553
669:
670: \reference{} Mulchaey, J. S., Koratkar, A., Ward, M. J.,
671: Wilson, A. S., Whittle, M., Antonucci, R. R. J., Kinney, A. L.,
672: Hurt, T. 1994, ApJ, 436, 586
673:
674: \reference{} Murray, N. \& Chiang, J. 1997, ApJ, 494, 125
675:
676: \reference{} Murtagh, F. \& Heck, A. 1987, ``Multivariate Data
677: Analysis'' Springer-Verlag Berlin Heidelberg 1987, ISBN 90-277-2425-3
678:
679: %\reference{} Peterson, B. M., Berlind, P., Bertram, R., \etal\ 2002,
680: %ApJ, 581, 197
681:
682: \reference{} Polletta, M., Weedman, D., Hönig, S., Lonsdale, C. J.,
683: Smith, H. E., Houck, J. 2008, ApJ, 675, 960
684:
685: \reference{} Pounds, K. A., Done, C., \& Osborne J. 1995, MNRAS, 277, L5
686:
687: \reference{} Proga, D. 2005, ApJ, 630, L9
688:
689: \reference{} di Serego Alighieri, S., Cimatti, A., Fosbury, R. A. E.,
690: \& Hes, R. 1997, A\&A, 328, 510
691:
692: \reference{} Shang, Z., Wills, B. J., Robinson, E. L., Wills, D.,
693: Laor, A., Xie, B.; Yuan, J, 2003, ApJ, 586, 52
694:
695: \reference{} Smith, P. S., Schmidt, G. D., Hines, D. C., Cutri, R. M.,
696: Nelson, B. O. 2002, ApJ, 569, 23
697:
698: \reference{} Smith, P. S., Schmidt, G. D., Hines, D. C.,
699: Foltz, C. B. 2003, ApJ, 593, 676
700:
701: \reference{} Sulentic J. W., Zwitter, T., Marziani, P.,
702: Dultzin-Hacyan, D. 2000, ApJ, 536, L5
703:
704: \reference{} Tadhunter, C. N., Morganti, R., Robinson, A., Dickson, R.,
705: Villar-Martin, M., \& Fosbury, R. A. 1998, MNRAS, 298, 1035
706:
707: \reference{} Turner, T.J., George, I.M., Nandra, K. \& Mushotzky,
708: R.F. 1997, ApJ, 113, 23
709:
710: \reference{} Wang, T.-G., Brinkmann, W., \& Bergeron, J. 1996, A\&A, 309, 81
711:
712: \reference{} Wang, J., Wei, J. Y., \& He, X. T. 2006, ApJ, 638, 106
713:
714: \reference{} Wilkes, B. J., Schmidt, G. D., Cutri, R. M., Ghosh, H.,
715: Hines, D. C., Nelson, B., Smith, P. S. 2002, ApJ, 564, L65
716:
717: \reference{} Wills, B.J., Shang, Z., \& Yuan, J. M., 2000, NewAR, 44, 511
718:
719: \reference{} Wills, B.J., Laor, A., Brotherton, M. S., Wills, D.,
720: Wilkes, B. J., Ferland, G. J., Shang, Z., 1999, ApJ, 512, L53
721:
722: \reference{} Witt, H. J. Czerny, B., \& \.Zycki, P. T. 1997, MNRAS, 286, 848
723:
724: \reference{} Yip, C. W., Connolly, A. J., Vanden Berk, D. E., Ma, Z.,
725: Frieman, J. A., SubbaRao, M., Szalay, A. S., Richards, G. T.,
726: Hall, P. B., Schneider, D. P., 2004, AJ, 128, 2603
727:
728:
729:
730: \end{references}
731:
732:
733:
734: \newpage
735:
736: %%%%%%%%%%%%%%%%%%%%%%%%%
737: %%%%%% FIGURES %%%%%%%
738: %%%%%%%%%%%%%%%%%%%%%%%%%
739:
740:
741: \clearpage
742: \begin{figure}
743: \epsscale{0.8}
744: \plotone{f1.eps}
745: \caption{{\it a, b} $-$ Correlations between the intrinsic 1~keV X-ray
746: flux to observed B flux ratio and the intrinsic 1~keV X-ray flux to
747: observed R and K flux ratios. These are examples of correlations
748: found to compose EV1. The X-ray flux is corrected for Galactic and
749: intrinsic extinction while the B,R and K fluxes are corrected only
750: for Galactic extinction. Crosses represent the blue optically/radio
751: selected quasars from Elvis \etal\ (1994). The 2MASS objects in all
752: figures are delineated by their Chandra S/N classifications, with
753: circles, squares and filled triangles representing sources with {\it
754: C} (highest S/N), {\it B} (medium S/N) and {\it A} (low S/N) spectral
755: types respectively. The dashed line in Fig.~1b shows the
756: B$-$K$_S$=4.3 color cut used to select the red 2MASS AGN. 1258+2329
757: lies below this cut since SuperCOSMOS B magnitude (which is more
758: consistent with the SED) is used here instead of the USNO--A2
759: magnitude originally used for color selection. {\it c, d} $-$
760: Correlations between Eigenvector~1 and the ratio of intrinsic 1keV
761: X-ray flux to the observed flux in the B and K bands.}
762: \label{fig:Fx/FBvsFx/Fnn}
763: \end{figure}
764:
765:
766: \clearpage
767: \begin{figure}
768: \plotone{f2.eps}
769: \caption{Correlation between the F([O\,{\small III}])/F(2$-$10keV)
770: ratio and Eigenvector~1. X-ray fluxes are corrected for Galactic
771: and intrinsic reddening. The [O\,{\small III}] emission--line flux
772: is corrected for NLR reddening estimated using the narrow
773: $H\alpha/H\beta$ ratio. Symbols indicate X-ray S/N as in
774: Fig.~\ref{fig:Fx/FBvsFx/Fnn}. Dashed line indicates the value of
775: $\log F([O\,III])/F(2-10keV) = -$1.89$\pm$0.75 (3$\sigma$
776: uncertainty - dotted line) found by Mulchaey \etal\ (1994) for
777: Seyfert~1s and Compton-thin Seyfert~2s.}
778: \label{fig:EV1}
779: \end{figure}
780:
781:
782: \clearpage
783: \begin{figure}
784: \plotone{f3.eps}
785: \caption{SED and emission line properties of AGN with the lowest EV1
786: values. {\it Left column} $-$ Restframe far--IR to X-ray SEDs. X-ray
787: fluxes are corrected for Galactic and intrinsic absorption. IR-to-UV
788: wavelengths have not been corrected for reddening or host galaxy
789: contamination. SEDs show low F(1keV)/F$_{B,R,J,K}$ ratios and blue
790: B$-$R colors. {\it Right column} $-$ Observed-frame optical spectra
791: on $F_{\lambda}$ (erg~s$^{-1}$~cm$^{-2}$~\AA$^{-1}$) versus
792: $\lambda$ (\AA ) scale. Optical spectral type, degree of optical
793: polarization and EV1 value have been denoted on each spectrum.
794: Spectra resemble those of NLS1s and BALQSOs characterized by strong
795: Fe\,{\small II} and weak [O\,{\small III}] emission.}
796: \label{fig:extEV1low}
797: \end{figure}
798:
799:
800: \begin{figure}
801: \plotone{f4.eps}
802: \caption{SED and emission line properties of AGN with the highest EV1
803: values. {\it Left column} $-$ Restframe far--IR to X-ray SEDs (for
804: details see Fig.~\ref{fig:extEV1low}). SEDs show high
805: F(1keV)/F$_{B,R,J,K}$ ratios. {\it Right column} $-$ Observed-frame
806: optical spectra on $F_{\lambda}$ (erg~s$^{-1}$~cm$^{-2}$~\AA$^{-1}$)
807: versus $\lambda$ (\AA ) scale. Optical spectral type, degree of
808: optical polarization and EV1 value have been denoted on each
809: spectrum. Spectra are characterized by strong narrow lines and weak
810: Balmer and Fe\,{\small II} lines.}
811: \label{fig:extEV1high}
812: \end{figure}
813:
814:
815: \clearpage
816: \begin{figure}
817: \plotone{f5.eps}
818: \caption{The AGN spectral energy distribution predicted by the
819: accretion disk and corona model of Witt, Czerny \& \.Zycki (1996)
820: for a black hole mass 10$^8$~M$_{\odot}$, viscosity parameter
821: $\alpha=0.33$ and from top to bottom $L/L_{Edd}$=0.25, 0.1, 0.01,
822: 0.003, 0.0017, 0.0013 respectively. Higher $L/L_{Edd}$ give a more
823: pronounced big blue bump relative to the X-rays and lower
824: F(1keV)/F$_{B,R}$ ratios. Lower $L/L_{Edd}$ produce a weaker big
825: blue bump relative to the X-rays and higher F(1keV)/F$_{B,R}$
826: ratios. The dashed line shows the $L/L_{Edd}$=0.0013 model reddened
827: at optical/UV wavelengths by A$_V$=3 (a reddening value obtained
828: from IR/optical color modeling for 1051+3539, shown in lower insert;
829: X-ray flux is left intrinsic since the \chandra\ X-ray flux was
830: corrected for both Galactic and intrinsic absorption). Inserts show
831: examples of red 2MASS AGN SEDs with high and low Eigenvector~1 values.}
832: \label{fig:models}
833: \end{figure}
834:
835:
836: \clearpage
837: \begin{figure}
838: %\plotone{f6_color.eps} %colored figure
839: \plotone{f6.eps} % B&W figure
840: \caption{Colors of a reddened AGN combined with contributions from a
841: host galaxy. Long-dash-short-dash curve shows colors of a pure
842: reddened AGN (Elvis \etal\ 1994 median QSO SED reddened by Milky Way
843: dust). Dotted (green) lines represent AGN colors modified by
844: emission from an elliptical host with a 15--Gyr stellar population
845: (E15; reddest in B$-$R). Solid (blue) lines represent an AGN~+~Sd
846: host and 5--Gyr stellar population (Sd05; bluest in
847: B$-$R). Short-dashed (red) line: AGN~+~Sa host galaxy (B$-$R colors
848: intermediate between E15 and Sd05). Each curve starts at the colors
849: of an unreddened (A$_V$=0) AGN~+~host galaxy (bluest optical/IR
850: colors) and extends to the reddened (A$_V$=10) AGN~+~host galaxy in
851: steps of 1~mag. denoted by small triangles on the curves (curve
852: (Sd05;5) has been extended to A$_V$=20). Numbers in parenthesis give
853: the host galaxy type and age of stars followed by the intrinsic,
854: unreddened AGN/host galaxy flux ratio at R band (e.g. (E15;10):
855: elliptical host, 15--Gyr stars, intrinsic AGN/host = 10). Thick
856: solid curves indicate pure host galaxy (E and Sd) colors changing
857: with star age (from 1 to 15--Gyr). Overplotted are the observed
858: colors of the red 2MASS AGN, divided according to the amount of host
859: galaxy contribution at R band obtained from color-color modeling:
860: where $\le 20$\% (filled circles), 20-60\% (squares), and $>60$\%
861: (stars) of the total observed flux at R band is due to host galaxy
862: emission.}
863: \label{fig:colors}
864: \end{figure}
865:
866:
867: \clearpage
868: \begin{figure}
869: \plotone{f7.eps}
870: \caption{Correlations between Eigenvector~2 and the B$-$K$_S$ and
871: \jmk\ colors. Symbols as in Fig.~\ref{fig:colors} denote host galaxy
872: contribution.}
873: \label{fig:EV2}
874: \end{figure}
875:
876:
877: \clearpage
878: \begin{figure}
879: \plotone{f8.eps}
880: \caption{Correlations between Eigenvector~3 and the X-ray hardness
881: ratio and between Eigenvector~3 and H$\alpha$/H$\beta$ measured from
882: narrow emission lines. Symbols indicate X-ray S/N, as in Fig.~\ref{fig:Fx/FBvsFx/Fnn}.}
883: \label{fig:EV3}
884: \end{figure}
885:
886:
887: \clearpage
888: \begin{figure}
889: \plotone{f9.eps}
890: \caption{Correlations between Eigenvector~4 and the degree of optical
891: polarization and broad H$\alpha$/H$\beta$. Symbols indicate X-ray
892: S/N, as in Fig.~\ref{fig:Fx/FBvsFx/Fnn}.}
893: \label{fig:EV4}
894: \end{figure}
895:
896:
897:
898:
899: %%%%%%%%%%%%%%%%%%%%
900: %%%% TABLES %%%%%%%%
901: %%%%%%%%%%%%%%%%%%%%
902:
903:
904: \clearpage
905: \input{tab1}
906: \begin{table}
907: %\dummytable
908: \label{tab:PCA} % table 1
909: \end{table}
910:
911:
912: %\clearpage
913: %\input{tab2}
914: %\begin{table}
915: %%\dummytable
916: %\label{tab:extremeEV} % table 2
917: %\end{table}
918:
919: \clearpage
920: \begin{figure}
921: % \plotfiddle{PSFILE}{VSIZE}{ROTANG}{HSCALE}{VSCALE}{HTRANS}{VTRANS}
922: \plotfiddle{tab2.ps}{6in}{0.}{90}{90}{-240}{-250}
923: \label{tab:extremeEV} % table 2
924: \end{figure}
925:
926:
927:
928:
929:
930: \end{document}
931: