1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass{emulateapj}
3:
4: \newcommand{\myemail}{riechers@mpia.de}
5: \usepackage{amssymb}
6:
7: %% -------Author macros----------------------
8: %% DO NOT USE IN MATH MODE!
9: \def\ts {\thinspace}
10: \def\kms {\ts km\ts s$^{-1}$}
11: \def\etal {{\rm et\ts al.}}
12: \def\msol {$M_{\odot}$}
13: \def\lsol {$L_{\odot}$}
14: \def\ci {C\ts {\scriptsize I}}
15: \def\hi {H\ts {\scriptsize I}}
16: \def\aco {{\rm CO}($J$=1$\to$0)}
17: \def\bco {{\rm CO}($J$=2$\to$1)}
18: \def\cco {{\rm CO}($J$=3$\to$2)}
19: \def\dco {{\rm CO}($J$=4$\to$3)}
20: \def\eco {{\rm CO}($J$=5$\to$4)}
21: \def\fco {{\rm CO}($J$=6$\to$5)}
22: \def\gco {{\rm CO}($J$=7$\to$6)}
23: \def\hco {{\rm CO}($J$=8$\to$7)}
24: \def\ico {{\rm CO}($J$=9$\to$8)}
25: \def\jco {{\rm CO}($J$=10$\to$9)}
26: \def\kco {{\rm CO}($J$=11$\to$10)}
27: \def\cho {{\rm H$_2$O}(3$_{13}$$\to$2$_{20}$)}
28: \def\ahcn {{\rm HCN}($J$=1$\to$0)}
29: \def\bhcn {{\rm HCN}($J$=2$\to$1)}
30: \def\chcn {{\rm HCN}($J$=3$\to$2)}
31: \def\ahco {{\rm HCO$^+$}($J$=1$\to$0)}
32: \def\bri {BRI\,1335--0417}
33: \def\pss {PSS\,J2322+1944}
34:
35: %%-------------------------------------------
36:
37: \slugcomment{draft version \today, Accepted for Publication in the Astrophysical Journal}
38:
39: \shorttitle{High--Resolution CO(2--1) Imaging of PSS\,J2322+1944}
40: \shortauthors{Riechers et al.}
41:
42: \begin{document}
43:
44: \title{A Molecular Einstein Ring at $z$=4.12:\\ Imaging the Dynamics
45: of a Quasar Host Galaxy Through a Cosmic Lens}
46:
47: \author{Dominik A. Riechers\altaffilmark{1,2,7}, %et al.}
48: Fabian Walter\altaffilmark{1}, Brendon J. Brewer\altaffilmark{3},
49: Christopher L. Carilli\altaffilmark{4}, \\
50: Geraint F. Lewis\altaffilmark{3}, Frank Bertoldi\altaffilmark{5},
51: and Pierre Cox\altaffilmark{6}}
52:
53:
54: \altaffiltext{1}{Max-Planck-Institut f\"ur Astronomie, K\"onigstuhl
55: 17, Heidelberg, D-69117, Germany}
56:
57: \altaffiltext{2}{Astronomy Department, California Institute of
58: Technology, MC 105-24, 1200 East California Boulevard, Pasadena, CA
59: 91125}
60:
61: \altaffiltext{3}{Institute of Astronomy, School of Physics, A28,
62: University of Sydney, NSW 2006, Australia}
63:
64: \altaffiltext{4}{National Radio Astronomy Observatory, PO Box O,
65: Socorro, NM 87801}
66:
67: \altaffiltext{5}{Argelander-Institut f\"ur Astronomie, Universit\"at
68: Bonn, Auf dem H\"ugel 71, Bonn, D-53121, Germany}
69:
70: \altaffiltext{6}{Institut de RadioAstronomie Millim\'etrique, 300 Rue
71: de la Piscine, Domaine Universitaire, F-38406 Saint Martin
72: d'H\`eres, France}
73:
74: \altaffiltext{7}{Hubble Fellow}
75:
76: \email{dr@caltech.edu}
77:
78:
79: \begin{abstract}
80: We present high-resolution (0.3$''$) Very Large Array (VLA) imaging
81: of the molecular gas in the host galaxy of the high redshift quasar
82: PSS\,J2322+1944 ($z=4.12$). These observations confirm that the
83: molecular gas (CO) in the host galaxy of this quasar is lensed into
84: a full Einstein ring, and reveal the internal dynamics of the
85: molecular gas in this system. The ring has a diameter of
86: $\sim$1.5$''$, and thus is sampled over $\sim$20 resolution elements
87: by our observations. Through a model-based lens inversion, we
88: recover the velocity gradient of the molecular reservoir in the
89: quasar host galaxy of PSS\,J2322+1944. The Einstein ring lens
90: configuration enables us to zoom in on the emission and to resolve
91: scales down to $\lesssim$1\,kpc. From the model-reconstructed
92: source, we find that the molecular gas is distributed on a scale of
93: 5\,kpc, and has a total mass of $M({\rm H_2})=1.7 \times
94: 10^{10}\,$\,\msol. A basic estimate of the dynamical mass gives
95: $M_{\rm dyn} = 4.4 \times 10^{10}\,\sin^{-2}\,i$\,\msol, that is,
96: only $\sim$2.5 times the molecular gas mass, and $\sim$30 times the
97: black hole mass (assuming that the dynamical structure is highly
98: inclined). The lens configuration also allows us to tie the optical
99: emission to the molecular gas emission, which suggests that the
100: active galactic nucleus (AGN) does reside within, but not close to
101: the center of the molecular reservoir. Together with the (at least
102: partially) disturbed structure of the CO, this suggests that the
103: system is interacting. Such an interaction, possibly caused by a
104: major `wet' merger, may be responsible for both feeding the quasar
105: and fueling the massive starburst of 680\,\msol\,yr$^{-1}$ in this
106: system, in agreement with recently suggested scenarios of quasar
107: activity and galaxy assembly in the early universe.
108: \end{abstract}
109:
110: \keywords{galaxies: active, starburst, formation, high redshift --- cosmology: observations
111: --- radio lines: galaxies}
112:
113: \section{Introduction}
114:
115: A fundamental aspect in studies of galaxy formation and evolution is
116: to understand the connection between AGN and starburst activity. The
117: existence of a physical connection between both processes is suggested
118: by the finding that present day galaxies show a strong relationship
119: between the mass of their central supermassive black holes (SMBHs) and
120: the mass and concentration of their stellar spheroids (Magorrian
121: \etal\ \citeyear{mag98}; Ferrarese \& Merritt \citeyear{fer00};
122: Gebhardt \etal\ \citeyear{geb00}; Graham \etal\ \citeyear{gra01}). If
123: these relations were due to a coevolution of both components during
124: the early assembly of a galaxy, high-redshift quasars and their
125: associated host galaxies would be ideal objects to study the active
126: formation of both SMBHs and bulge stars.
127:
128: Studies of molecular gas (most commonly rotational transitions of CO),
129: the prerequisite material that fuels star formation, have become an
130: important tool to probe distant quasar host galaxies, and revealed
131: large molecular gas reservoirs of $>$10$^{10}$\,\msol\ in a number of
132: these sources (see Solomon \& Vanden Bout \citeyear{sv05} for a
133: general review). These galaxies typically show huge far-infrared
134: (FIR) luminosities in excess of 10$^{13}$\,\lsol, which are thought to
135: be powered by starbursts (and possibly a central AGN; e.g., Omont et
136: al.\ \citeyear{omo01}; Wang et al.\ \citeyear{wan07}). Observations
137: of molecular gas trace the regions that can host massive starbursts.
138: In addition, the velocity structure of molecular line emission has the
139: potential to constrain the dynamical state of galaxies out to the
140: earliest epochs.
141:
142: Rotational molecular line emission typically emerges at FIR to radio
143: wavelengths, i.e., in the limited wavelength regime where the AGN in
144: distant quasars does not outshine all other emission. However, the
145: cosmological distances of high redshift quasars make it difficult to
146: resolve the faint emission from their host galaxies a such long
147: wavelengths. The physical resolution of such observations is in some
148: cases boosted by gravitational lenses acting as natural
149: telescopes. The gravitational lensing effect also magnifies the
150: observed flux of the background galaxy, in particular for systems in
151: Einstein ring configurations. Due to the compactness of the AGN,
152: optical quasars in Einstein ring lens configurations are rare. Due to
153: their greater extent, the host galaxies of quasars are much more
154: likely to cross the inner Einstein ring caustic of a gravitational
155: lens.
156:
157: In this paper, we report on high (0.3$''$) angular resolution Very
158: Large Array (VLA)\footnote{The Very Large Array is a facility of the
159: National Radio Astronomy Observatory, operated by Associated
160: Universities, Inc., under a cooperative agreement with the National
161: Science Foundation.} observations of CO in the host galaxy of the
162: $z$=4.12 quasar PSS\,J2322+1944, one of only two known $z$$>$4
163: galaxies that are both gravitationally lensed and detected in
164: molecular gas emission (the other being BRI\,0952--0115 at $z$=4.43;
165: Guilloteau \etal\ \citeyear{gui99}). This galaxy was identified in a
166: spectroscopic follow-up study of the Palomar Sky Survey (DPOSS;
167: Djorgovski et al.\ \citeyear{djo00}), and found to be a strongly
168: lensed optical quasar (S.~G.\ Djorgovski, private communication). It
169: was subsequently detected in hard X-ray (Vignali et al.\
170: \citeyear{vig05}), FIR dust (Omont et al.\ \citeyear{omo01}; Isaak et
171: al.\ \citeyear{isa02}) and radio continuum emission (Carilli et al.\
172: \citeyear{car01}), as well as molecular line emission (Cox et al.\
173: \citeyear{cox02}; Carilli et al.\ \citeyear{car02a}). It follows the
174: radio-FIR correlation of star-forming galaxies (Carilli et al.\
175: \citeyear{car01}; Beelen et al.\ \citeyear{bee06}), indicating that its
176: FIR continuum emission is dominated by intense star formation. In
177: spite of the fact that this source shows only two unresolved quasar
178: images in the optical, previous CO observations have shown that the
179: molecular gas reservoir in its host galaxy is lensed into an Einstein
180: ring (Carilli \etal\ \citeyear{car03}; hereafter:\ C03). These
181: observations were also used to derive a first lensing model for this
182: source. Based on the dynamical structure revealed by our new, higher
183: resolution observations of PSS\,J2322+1944, we have developed a new
184: lensing model, which enables us to reconstruct the velocity gradient
185: in the spatially resolved gas reservoir. We use a concordance, flat
186: $\Lambda$CDM cosmology throughout, with $H_0$=71\,\kms\,Mpc$^{-1}$,
187: $\Omega_{\rm M}$=0.27, and $\Omega_{\Lambda}$=0.73 (Spergel \etal\
188: \citeyear{spe03}, \citeyear{spe07}).
189:
190:
191: \section{Observations}
192:
193: We observed the \bco\ transition ($\nu_{\rm rest} = 230.53799\,$GHz)
194: towards \pss\ using the VLA in B configuration between 2006 June 19
195: and July 3, and in C configuration between 2002 October 21 and
196: November 15 (these short spacings data were published in the original
197: study by C03). The total on-sky integration time in the 11 observing
198: runs amounts to 70.5\,hr. At $z=4.119$, the line is redshifted to
199: 45.0351\,GHz (6.66\,mm). Observations were performed in
200: fast-switching mode (see, e.g., Carilli \& Holdaway \citeyear{ch99})
201: using the nearby source J23307+11003 for secondary amplitude and phase
202: calibration. Observations were carried out under very good weather
203: conditions with 26 or 27 antennas. The phase stability in all runs was
204: excellent (typically $<$20$^\circ$ phase rms for the longest
205: baselines). The phase coherence was checked by imaging the a
206: calibrator (J23207+05138) with the same calibration cycle as that used
207: for the target source. For primary flux calibration, 3C48 was
208: observed during each run. Due to the restrictions of the VLA
209: correlator, one 50\,MHz intermediate frequency (IF) with seven
210: 6.25\,MHz channels was centered at the \bco\ line frequency, leading
211: to an effective bandwidth of 43.75\,MHz (corresponding to 291\kms\ at
212: 45.0\,GHz). This encompasses almost the full CO line width as measured
213: in the \eco\ transition ($\Delta v_{\rm FWHM} = 273 \pm 50$\,\kms, Cox
214: \etal\ \citeyear{cox02}). Earlier observations set a 2\,$\sigma$ limit
215: of 150$\mu$Jy on the continuum emission at the line frequency (Carilli
216: \etal\ \citeyear{car02a}).
217:
218: For data reduction and analysis, the AIPS package was used. All data
219: were mapped using the CLEAN algorithm and natural weighting. The
220: synthesized clean beam using all data has a size of
221: 0.33$''$$\times$0.30$''$.\footnote{Imaging B array data only gives a
222: resolution of 0.17$''$$\times$0.15$''$ (natural weighting), or
223: 0.12$''$$\times$0.11$''$ (uniform weighting); however, most of the
224: emission is outresolved in such maps.} The final rms over a bandwidth
225: of 37.5\,MHz (250\,\kms, excluding the noisy edge channel) is
226: 48\,$\mu$Jy beam$^{-1}$. In addition, seven velocity channel maps
227: (6.25\,MHz, or 42\,\kms\ each) of the \bco\ emission were created. The
228: rms after Hanning smoothing is 77\,$\mu$Jy\,beam$^{-1}$. Convolving
229: the data to a linear spatial resolution of 0.5$''$ leads to slightly
230: higher rms values of 49 and 80\,$\mu$Jy beam$^{-1}$ for maps at 250
231: and 42\,\kms\ velocity resolution.
232:
233:
234:
235: \section{Results}
236:
237: \begin{figure}
238: \epsscale{1.2}
239: \vspace*{-7mm}
240:
241: \plotone{f1.ps}
242: \vspace*{-12mm}
243:
244: \caption{VLA map of the \bco\ emission toward \pss\ (integrated over
245: the central 37.5\,MHz, or 250\,km\,s$^{-1}$). Contours are shown at
246: (-3, -2, 1, 2, 3, 4, 5)$\times\sigma$ (1$\sigma = 48\,\mu$Jy
247: beam$^{-1}$). The beam size (0.33$''$$\times$0.30$''$) is shown in
248: the bottom left corner. The large crosses show the positions of the
249: quasar images at $\lambda_{\rm obs} = 1.6\,\mu$m ($\lambda_{\rm r} =
250: 314$\,nm), and the small cross shows the position of the lensing
251: galaxy at the same $\lambda$.
252: \label{f1}}
253: \end{figure}
254:
255: In Figure \ref{f1}, the velocity-integrated \bco\ emission over the
256: central 250\,\kms\ is shown. The emission is clearly resolved over
257: multiple beams, and extended on a scale of $\sim$1.5$''$. The
258: distribution of the gas is reminiscent of a full, almost circular
259: molecular Einstein ring, consistent with previous indications of such
260: a structure in lower resolution observations (C03). The emission
261: varies in intensity along the ring, showing clear substructure.
262:
263: The apparent surface brightness variations along the ring set strong
264: constraints on the geometry of the lens configuration. The brightest
265: CO peak on the ring can be used to set a lower limit on the intrinsic
266: brightness temperature of the molecular gas. The peak strength of
267: 280$\pm$48\,$\mu$Jy\,beam$^{-1}$ corresponds to a beam-averaged,
268: rest-frame brightness temperature of $T_{\rm b}$=8.7$\pm$1.5\,K. This
269: is by a factor of a few lower than the kinetic gas temperature $T_{\rm
270: kin}$ as predicted from CO line excitation models of this source
271: (Riechers et al.\ \citeyear{rie06}; A.~Wei\ss\ et al., in prep.), and
272: indicates that the substructure of the molecular reservoir is not
273: fully resolved by the observations.
274:
275: We derive a spatially integrated \bco\ line peak flux density of 2.50
276: $\pm$ 0.32\,mJy, which is consistent with that found by Carilli \etal\
277: (\citeyear{car02a}). Assuming constant $T_{\rm b}$ between \bco\ and
278: \aco, and a CO luminosity to H$_2$ mass conversion factor for
279: ultraluminous infrared galaxies
280: [$\alpha$=0.8\,\msol\,(K\,\kms\,pc$^2$)$^{-1}$; Downes \& Solomon
281: \citeyear{ds98}], this corresponds to a total molecular gas mass of
282: $M({\rm H_2})=9.0 \times 10^{10}\,\mu_L^{-1}$\,\msol\ (where $\mu_L$
283: is the lensing magnification factor, see below). Within the
284: uncertainties, this is in agreement with the value found by Riechers
285: \etal\ (\citeyear{rie06}) based on the \aco\ luminosity.
286:
287: To increase the peak signal--to--noise ratio, Figure \ref{f2} shows
288: the \bco\ emission convolved to 0.5$''$ linear resolution. This
289: contour map is shown overlaid on a greyscale image of the (rest-frame)
290: 314\,nm continuum emission of the source, as observed by the Hubble
291: Space Telescope (HST/NICMOS-2 F160W; image adopted from Peng et al.\
292: \citeyear{pen06}; C.~Y.\ Peng 2008, private communication). The two
293: brightest spots in the optical image are unresolved lensed images of
294: the AGN (henceforth `A' and `B',\footnote{By convention, A is the
295: brightest image.} also indicated as large crosses in Figure
296: \ref{f1}). The optical emission is likely dominated by the broad-line
297: region of the AGN, and thus expected to emerge from a compact
298: circumnuclear pc-scale region. The quasar images have a brightness
299: ratio of $b$(B,A)=0.181. They coincide within 0.10$''$ with the
300: positions measured by Keck at 430\,nm (2.2\,$\mu$m observed frame;
301: C03), i.e., within the relative astrometric errors. These positions
302: however are clearly offset from the brightest peaks of the molecular
303: line emission. The third spot in the optical image is the lensing
304: galaxy (`G', also indicated as a small cross in Figure \ref{f1}). The
305: lensing galaxy lies on the axis connecting images A and B, but by more
306: than a factor of 2 closer to image B. It however lies in the very
307: center of the molecular Einstein ring, as expected. At the position of
308: the lensing galaxy, we measure a 45.0\,GHz radio continuum flux of 79
309: $\pm$ 47\,$\mu$Jy. This corresponds to only 1.7$\sigma$, and thus has
310: to be considered tentative at best.
311:
312: In the adopted cosmology, the optical brightness of PSS\,J2322+1944
313: corresponds to an apparent bolometric luminosity of $L_{\rm bol} = 2.1
314: \times 10^{14}\,\mu_{L,{\rm opt}}^{-1}$\,\lsol\ (Isaak et al.\
315: \citeyear{isa02}). Assuming Eddington accretion, this corresponds to
316: an apparent black hole mass of $M_{\rm BH} = 7.0 \times
317: 10^{9}\,\mu_{L,{\rm opt}}^{-1}$\,\msol. Note that this may be
318: considered a lower limit, as black holes in $z \sim 4$ quasars are
319: found to have typical accretion rates of $\dot{M}={L_{\rm bol}/L_{\rm
320: Edd}} = 0.3-0.4$ (e.g., Shen et al.\ \citeyear{she07}).
321:
322: \begin{figure}
323: \vspace{-10mm}
324:
325: \epsscale{1.2}
326: \plotone{f2.ps}
327: \vspace*{-17.5mm}
328:
329: \caption{Contours of the \bco\ emission as shown in Figure \ref{f1},
330: but convolved to a linear resolution of 0.5$''$ and overlaid on a HST
331: NICMOS-2 image at $\lambda_{\rm obs} = 1.6\,\mu$m (position crosses in
332: Figure \ref{f1}). Contours are shown at (-3, -2, 2, 3, 4, 5, 6, 7,
333: 8)$\times\sigma$ (1$\sigma = 49\,\mu$Jy beam$^{-1}$). Note that the
334: HST image was not cleaned, and thus shows Airy rings around the
335: unresolved quasar images (Peng et al.\ \citeyear{pen06}; C.~Y.\ Peng
336: 2008, private communication).
337: \label{f2}}
338: \end{figure}
339:
340: \begin{figure*}
341: \epsscale{1.2}
342: \vspace{-10mm}
343:
344: \plotone{f3.eps}
345: \vspace{-22mm}
346:
347: \caption{Channel maps of the \bco\ emission at
348: 0.5$''$ resolution (beam size is shown in the bottom left corner of
349: the first panel). The same region is shown as in Fig.~\ref{f2}. One
350: channel width is 6.25\,MHz, or 42\,\kms\ [frequencies increase with
351: channel number and are shown at 45016.35, 45022.60, 45028.85,
352: 45035.10, 45041.35, 45047.60, and 45053.85\,MHz (red to blue)].
353: Contours are shown at (--3, --2, 2, 3, 4, 5, 6)$\times \sigma$
354: (1$\sigma = 80\,\mu$Jy\,beam$^{-1}$). Note that the noise in channel 7
355: is intrinsically higher by a factor of $\sqrt{2}$ relative to the
356: other channels.
357: \label{f3}}
358: \end{figure*}
359:
360: In Figure \ref{f3}, seven 42\,\kms\ wide velocity channels of the
361: \bco\ emission line are shown (channel 7 was omitted from the data
362: shown in Figures \ref{f1} and \ref{f2} due to higher noise). Emission
363: along the Einstein ring is detected in all channels. Clearly, the
364: emission is moving systematically along the ring from the red part of
365: the \bco\ line to the blue. Note that there are many peaks of similar
366: surface brightness (peak fluxes of 450--550\,$\mu$Jy\,beam$^{-1}$, or
367: $T_{\rm b}$=6--7\,K) that are found at different positions at
368: different velocities. This indicates, to first order, a dynamical
369: structure of uniform surface brightness, where the components at
370: different velocities get projected to different positions in the lens
371: plane. This puts us in the unique situation to model the
372: gravitational lens configuration in this system in detail, and to
373: recover the intrinsic dynamical structure of the quasar host galaxy in
374: the background.
375:
376:
377: \section{Gravitational Lens Inversion}
378:
379: In a previous attempt to model the gravitational lensing effect toward
380: the molecular gas reservoir of \pss, C03 adopted a generic model based
381: on the strongly lensed $z$=0.84 radio galaxy MG\,1131+0456. By
382: comparing various source configurations in the assumed lens potential
383: and based on plausibility arguments, a model was found that described
384: the overall properties of the observed ring structure. To enable us to
385: describe the intrinsic properties of \pss\ in more detail, we here
386: present a direct model reconstruction and inversion of the lensing
387: effect in this system. This modeling is based on the new CO
388: observations presented in the previous section.
389:
390:
391: \subsection{Method: Bayesian Inference}
392:
393: Due to the remaining observational uncertainties, it is not possible
394: to derive a {\em unique} solution for the inversion of the
395: gravitational lens. We thus explored the parameter space permitted by
396: the data (and our modeling assumptions) to find the best possible
397: solution; if some property of the models is consistent throughout this
398: permitted volume, it can be considered well-constrained at high
399: confidence. This parameter study thus follows a Bayesian approach
400: \citep{gre05}, using
401: the Markov Chain Monte Carlo (MCMC) code by \cite{bl06}. This
402: algorithm does not simply aim at minimizing $\chi^2$ to find the best
403: model (which may ``overfit'' the data by fitting part of the noise),
404: but explores the whole range of plausible fits. By doing this, we can
405: use a large number of parameters (e.g. source pixel values) in our
406: model without overfitting the noise (since a broad region in parameter
407: space with higher $\chi^2$ can outweigh a particular solution that has
408: very low $\chi^2$ if the volume of the parameter space near the very
409: low $\chi^2$ solution is very small).
410:
411: The Bayesian analysis thus encodes the phase space of possible source
412: distributions allowed by the data into a probability distribution
413: (e.g., Gregory \citeyear{gre05}). The lensed image of the source was
414: reconstructed based on the integrated CO emission line map and the
415: structure detected in the velocity channels, and then used to derive a
416: common model of the lensing galaxy's projected density profile that
417: reproduces the emission in all channels simultaneously. Due to the
418: differential structure among the velocity channels, this implies that
419: the reconstructed source components after lens inversion will be
420: different for each velocity channel, and will reproduce the velocity
421: gradient across the source.
422:
423: In this model description, the unknowns to be inferred from the data
424: are the seven source profiles $\{s_i\}_{i=1}^7$, one for each velocity
425: channel, where $s_i$ is shorthand for a large number of (unknown)
426: pixel values. The unknown lens model parameters are denoted
427: collectively by $L$. Given observed data $D$, the posterior
428: distribution for the unknown parameters is proportional to the product
429: of the prior distribution and the likelihood function:
430: \begin{equation}
431: p(\{s_i\}, L | D) \propto p(\{s_i\}, L)p(D | \{s_i\}, L ) .
432: \end{equation}
433: Here, $D$ consists of seven extended images (one for each velocity
434: channel). We make the following standard assumption for $p(D |
435: \{s_i\}, L )$: Assuming that we know the source and lens properties, we
436: would predict the observed image to be the lensed, blurred image of
437: that source, plus additive Gaussian noise:
438: \begin{equation}\label{likelihood}
439: p(D | \{s_i\}, L ) \propto \exp\left[-\frac{1}{2}\sum_{i=1}^7\chi_i^2\right],
440: \end{equation}
441: where
442: \begin{equation}\label{chisq}
443: \chi_i^2 = \sum_{j=1}^{N_{\rm pixels}}\left(\frac{D_{i,j} - M_j(s_i, L)}{\sigma_i}\right)^2.
444: \end{equation}
445: Here, $D_{i,j}$ is the $j$'th pixel of the $i$'th image, and $M(s_i,
446: L)$ is the model image calculated by lensing and blurring the $i$'th
447: source with the proposed lens model $L$, and $\sigma_i$ is the noise
448: standard deviation in the $i$'th image estimated from the outer `blank
449: sky' regions of each image, i.e., distant from any detected structure.
450:
451: In this study, the lens was parameterized as a singular isothermal
452: elliptical potential with five free parameters, which are the strength
453: $b$ of the lens, the ellipticity $q$ of the potential, the central
454: position $(x_c,y_c)$ of the lensing source, and the angle of
455: orientation $\theta$ of the projected density profile. For this model,
456: the lensing potential is
457: \begin{equation}
458: \phi(x,y) = b\sqrt{qx'^2 + y'^2/q} \, ,
459: \end{equation}
460: where
461: \begin{equation}
462: \left(\begin{array}{c}
463: x' \\
464: y'
465: \end{array}\right) =
466: \left(\begin{array}{cc}
467: \cos\theta & \sin\theta \\
468: -\sin\theta & \cos\theta
469: \end{array}\right)\left(\begin{array}{c}
470: x-x_c \\
471: y-y_c
472: \end{array}\right) .
473: \end{equation}
474: The source plane position $(x_s,y_s)$ corresponding to any lens plane position $(x,y)$
475: is then given by the lens equation:
476: \begin{eqnarray}
477: x_s = x - \frac{\partial\phi}{\partial x} \, , \nonumber \\
478: y_s = y - \frac{\partial\phi}{\partial y} \, .
479: \end{eqnarray}
480:
481: The optical position of the lensing galaxy was {\em not} used as an
482: initial model constraint to allow for a conservative treatment of the
483: errors. However, the optical data is used in a second step to better
484: constrain some of the source's intrinsic properties, as described in
485: more detail below.
486:
487: \subsection{Application to Interferometric Data}
488:
489: Interferometer maps are reconstructed from visibility data using a
490: scale that samples one synthesized interferometer beam (i.e.,
491: resolution element) with multiple pixels. This means that the noise is
492: not independent for all pixels, as assumed by
493: Equation~\ref{chisq}. Correct modelling of correlated noise is
494: computationally very expensive, and makes the evaluation of the
495: likelihood (Equation~\ref{likelihood}) very slow. In interferometric
496: images, the scale of the synthesized beam, or point-spread function
497: (PSF), usually is the same as the noise correlation scale (depending
498: on the interferometer baseline weighting function used in the imaging
499: Fourier transform). A good PSF model thus will allow for a proper, but
500: faster treatment of the noise properties.
501:
502: The correlation length scale of the noise was measured at a clear
503: distance from the detected molecular structure and used as the length
504: scale for constructing an optimized Gaussian PSF. This length scale
505: predicts that 1/62 of the pixels are effectively independent at the
506: scale of the images (0.03$''$\,pixel$^{-1}$). This corresponds to 106
507: pixels on the scale of the Einstein ring. To not ``overfit'' the image
508: due to noise, only this fraction of information can be used to
509: computationally determine the lens properties. This means that MCMC
510: calculations have to be run at an ``annealing temperature'' of 62, or
511: alternatively, the $\sigma$'s can be artificially increased by a
512: factor $\sqrt{62}$ (e.g., Gregory \citeyear{gre05}). The validity of
513: this short-cut procedure was confirmed by ensuring that the residuals
514: of the model images have the same statistical properties as the
515: background in the observed maps.
516:
517:
518: \subsection{Priors}
519:
520: The prior distribution for the lens parameters is chosen to be
521: diffuse, but the final result is independent from this selection. For
522: the seven unknown sources representing the source distribution in the
523: velocity channels, we use independent ``massive inference'' priors
524: \citep{0047}. To generate a random source,
525: a moderate number of ``atoms'' of a certain brightness are added to a
526: blank source plane. These atoms are given a uniform probability
527: distribution in position and an exponential probability distribution
528: in flux. The width of each atom in pixels is chosen at random from
529: one of three values to allow for the expectation that pixel values
530: should be correlated with their neighbours. This procedure generates a
531: random source in which most pixels are dim, and is a more appropriate
532: prior for astronomical sources than most conventional regularizers
533: \citep{bl06}.
534:
535: \subsection{Modeling Results}
536:
537: \begin{figure*}
538: \epsscale{1.2}
539: \plotone{f4.eps}
540: \caption{Reconstructed model images (posterior mean) of the 42\,\kms\ \bco\ velocity
541: channels, and source morphologies after lens inversion. The {\em
542: bottom} row shows the observed \bco\ emission in the velocity
543: channels as shown in Figure \ref{f3}, but without applying a beam
544: convolution (i.e., same resolution as Figure \ref{f1}, thus
545: 1$\sigma$=77\,$\mu$Jy\,beam$^{-1}$). The {\em middle} row shows the
546: model-reconstructed images of the channel maps in the lens plane
547: (same scale as bottom). The {\em top} row shows the reconstructed
548: images in the source plane after lens inversion. To account for the
549: magnification of physical scales (and thus in effective spatial
550: resolution), the model reconstruction in the source plane (same
551: region as shown in Figure \ref{f4}; scale in arcsec given on top) is
552: shown zoomed-in by a factor of 2 relative to the lens plane.
553: \label{f5}}
554: \end{figure*}
555:
556:
557: \subsubsection{Lens Parameters}
558:
559: From the MCMC run, we find that the lens has a strength $b$ =
560: 0.745$''$ $\pm$ 0.014$''$, an ellipticity $q$ = 0.969 $\pm$ 0.014, and
561: a position ($x_c$ = 0.074$''$ $\pm$ 0.024$''$; $y_c$ = --0.109$''$
562: $\pm$ 0.029$''$) [coordinates are relative to the center of the model
563: images at $\alpha$=$23^{\rm h}22^{\rm m}07^{\rm s}.176$,
564: $\delta$=$+19^\circ44'22''.16$]. Due to the fact that $q \simeq 1$,
565: the lens potential is close to circular, as is the projected lens
566: density profile. Even though the observed position of the lens
567: ($x_c^{\rm obs}$ = 0.105$''$; $y_c^{\rm obs}$ = --0.080$''$) was not
568: taken as an input parameter, the model naturally reproduces its
569: position within the errors. The posterior probability distribution for
570: $\theta$ is bimodal; with a probability of 62\% (38\%), the angle of
571: orientation of the potential is $\theta$ = 109.2$^\circ$ $\pm$
572: 7.1$^\circ$ (62.7$^\circ$ $\pm$ 6.5$^\circ$). The lens parameters are
573: constrained well by the molecular data alone, so the marginal
574: posterior distributions are close to Gaussian (except for the bimodal
575: $\theta$ distribution, which is well approximated by a mixture of two
576: Gaussians). Thus, all of the estimates and uncertainties quoted above
577: are of the form (mean $\pm$ standard deviation).
578:
579: The lensing model described above was derived based on the
580: distribution of the lensed CO emission only to allow for a
581: conservative treatment of the errors, and to avoid a main systematic
582: source of error:\ the remaining astrometric uncertainties between the
583: optical and radio reference frames. The results of our study indicate
584: that the optical position of the lens is reproduced well, and thus,
585: that the astrometric offset appears to be small. Due to the fact that
586: gravitational lensing is achromatic, we thus can use the model derived
587: based on the distribution of the lensed CO emission only to also
588: constrain the intrinsic optical properties of the source. We also can
589: use the positions of the lensed images of the quasar in the optical to
590: further constrain the allowed parameter space, and to estimate the
591: position of the AGN within the deprojected molecular gas reservoir.
592:
593: Based on the sample of lenses that fit the molecular line data, the
594: optical positions of the quasar were ray-traced back into the source
595: plane. As the optical emission of the source is compact, only those
596: models that map both quasar images onto the same position in the
597: source plane within the errors can be considered valid. We thus
598: discarded all models that did not fulfill this extra criterion.
599:
600: We find that this extra constraint only marginally changes previous
601: results for the strength, ellipticity, and position of the
602: lens. However, it does impact the solution for the angle of
603: orientation of the potential, and supports $\theta \sim 109^\circ$.
604: As the source reconstructions presented in this section are not
605: strongly sensitive to the extra constraint, only results (and the more
606: conservative errors implied) from simulations produced without using
607: the optical data are shown unless stated otherwise.
608:
609:
610: \subsubsection{Source Profiles}
611:
612: For the seven source profiles in the different velocity channels, a
613: ``best'' estimate was obtained by taking the average of all sources
614: encountered by the MCMC run. This gives the posterior mean for each
615: source, which is an optimal estimate, as it minimizes the expected
616: squared error. As no unique solution exists for the lens inversion,
617: the full source sample has a certain diversity. However, by taking
618: the mean, only reproducible features are retained from the full
619: sample. Due to the fact that both the resolution and sensitivity of
620: the observations are finite, the resulting surface brightness profiles
621: are expected to be smooth on a certain critical scale. Substructure
622: may appear on smaller scales, but will be smoothed out in the final
623: model due to the larger uncertainites involved in reproducing such
624: small structures across the permitted volume in parameter space.
625:
626: In Figure \ref{f5}, the model-reconstructed (posterior mean) molecular
627: gas distribution in the seven CO velocity channels ({\em columns}) are
628: shown in the lens plane ({\em middle row}), and after lens inversion
629: ({\em top row}), together with the observations (not convolved, i.e.,
630: at full spatial resolution; {\em bottom row}). Since the lens model
631: is common to all velocity channels, the changing distribution of
632: molecular gas between the velocity channels is due to intrinsic
633: velocity structure in the background source (note the difference in
634: scale between the lensed and the unlensed source).
635:
636: In Figure \ref{f7}, a color composite map of the modeled distribution
637: of the CO in the velocity channels is shown. Orange corresponds to
638: the redshifted part of the emission line, green corresponds to the
639: central part, and blue corresponds to the blueshifted part. In the
640: {\em right} panel, the image in the lens plane is displayed, and in
641: the {\em left} panel, the image in the source plane is shown after
642: lens inversion. A clear velocity gradient is seen in both images. The
643: brightest part of the emission shows a systemic velocity structure
644: that would be in agreement with a highly inclined, almost edge-on
645: rotating disk of $\sim$2.5\,kpc radius within the uncertainties of the
646: data and modeling. This structure encloses a dynamical mass of
647: $M_{\rm dyn} = 4.4 \times 10^{10}\,\sin^{-2}\,i$\,\msol\ (assuming a
648: linewidth of 273\,\kms\ as given above), corresponding to only 2.5
649: times the lensing-corrected molecular gas mass (for highly inclined
650: dynamical structures). However, there is also indication for a
651: second, possibly tidal component (blue/green). This structure is
652: slightly bent, and also has an extent of almost 5\,kpc, similar to the
653: brighter component. Each component extends over at least 4-5
654: resolution elements in the source plane, and thus is clearly detected
655: and resolved.
656:
657: \begin{figure*}
658: \epsscale{1.2}
659: \vspace{-3mm}
660:
661: \plotone{f5.eps}
662: \vspace{-5.5mm}
663:
664: \caption{RGB composite color map of the model-reconstructed \bco\
665: velocity channels shown in Figure \ref{f5}, with three colors
666: encoding the velocity range of the emission [orange: redshifted
667: (channels 1-2), green: central (channels 3-5), blue: blueshifted
668: (channels 6-7)]. The grey lines indicate the caustics. The bar at
669: the bottom indicates the scale in the source plane. {\em Right}:
670: Lensed, blurred image of the Einstein ring in the lens plane. {\em
671: Left}: Image of the model-reconstructed quasar host galaxy in the
672: source plane. \label{f7}}
673:
674: \end{figure*}
675:
676: The remaining uncertainties in the source reconstruction can be
677: quantified by examining individual MCMC samples to calculate a
678: probabilistic description of different properties of the source. A
679: manual comparison of models considered to be ``good'' by our
680: statistical analysis however shows that all share the same principal
681: features, and only show significant differences in substructure on
682: scales that are close to the resolution and detection limit, as
683: expected. It thus is plausible that the overall structure and velocity
684: gradient along the reconstructed source are real.
685:
686: To quantify the uncertainties in the structure of the molecular gas
687: reservoir of \pss\ in the source plane, the posterior probability $P$
688: that the flux in each pixel is nonzero is shown in
689: Figure~\ref{f4}. The detection probability is shown over the full
690: velocity range (i.e., summed up over all velocity channels). Due to
691: the fact that observational data contains noise (i.e., finite positive
692: or negative flux in every pixel of any velocity channel), the
693: detection probability is nonzero in every pixel. Thus, a detection
694: probability below a certain threshold has to be considered
695: ``background''. The detection probability is consistently
696: significantly higher than the background along the structures
697: predicted by the ``best model'' (see Figures \ref{f5} and
698: \ref{f7}).
699:
700: The overall lensing magnification of the CO emission as predicted by
701: the posterior distribution for the lens model is $\mu_L = 5.34 \pm
702: 0.34$. Figure \ref{f6} shows the differential flux magnification
703: between the velocity channels, estimated by calculating flux(lensed
704: image)/flux(source) for the MCMC samples of each velocity
705: channel. Although consistent with unity within the modeling
706: uncertainties, there is tentative evidence that the flux is more
707: gravitationally enhanced in the line center than in the outer
708: channels. If real, this would suggest that the profile of the \bco\
709: emission line is distorted by the lensing effect.
710:
711: The model-predicted lensing magnification of the 314\,nm continuum
712: emission from the AGN as derived from models based on both optical and
713: CO data is $\mu_L^{\rm opt} = 4.7 \pm 0.4$. Excluding the optical
714: constraints would result in slightly larger errors: $\mu_L^{\rm opt} =
715: 4.8 \pm 0.6$. The overlay in Figure \ref{f2} suggests that the
716: optical AGN is offset from the peak position of the molecular line
717: emission. In the red CO line wing, the northern peak of the Einstein
718: ring is brighter (as is the case for the two optical images), while in
719: the blue linewing, the southern peak is brighter. This, together with
720: the optical quasar image positions relative to the lens position,
721: suggests that the AGN is located close to the orange peak of the
722: reconstructed molecular gas distribution shown in Figure
723: \ref{f7}. This finding is supported by the model-based inversion of
724: the optical and CO data, although the determination of the exact
725: location within that part of the molecular structure is limited by the
726: remaining model uncertainties and the limited accuracy of the relative
727: astrometry of the optical and radio data. If the AGN is indeed
728: located in the upper part of the disklike structure rather than in the
729: center, one may speculate that the spatial and velocity structure of
730: the reconstructed source is more likely to be due to interaction than
731: due to a rotating disk.
732:
733:
734: \section{Discussion}
735:
736: We have imaged and modeled a molecular Einstein ring of a galaxy at
737: $z$=4.12. Our high resolution \bco\ maps of the lensed quasar host
738: galaxy of PSS\,J2322+1944 (a double image optical quasar) reveal
739: spatially resolved structure that shows a clear velocity gradient in
740: the CO emission line. By performing a model-based lens inversion of
741: the Einstein ring that is consistent with the data, we are able to
742: reconstruct the velocity structure of this distant quasar host galaxy.
743: The gravitational lensing effect acts as a natural telescope, and
744: allows us to zoom in on the molecular gas reservoir down to linear
745: scales of only $\lesssim$1\,kpc, sufficient to reveal velocity
746: structure over almost 10 resolution elements in the source plane. Our
747: novel modeling of this system reveals how the molecular gas crosses
748: the central caustic (causing the appearance of the Einstein ring)
749: moving from the redshifted to the blueshifted molecular emission. We
750: also find evidence that the optical quasar may be associated with the
751: redshifted part of the molecular reservoir. The full reservoir has a
752: mass of $M({\rm H_2})=1.7 \times 10^{10}\,$\,\msol\ (corrected for
753: lensing magnification). The molecular gas mass alone could account
754: for almost half of the dynamical mass in this system if the galaxy
755: were to be seen close to edge-on ($M_{\rm dyn}$\,sin$^2\,i$/$M({\rm
756: H_2})$ $\simeq$ 2.5). Due to the large spatial extent of the CO
757: emission, and due to the fact that the AGN is probably largely offset
758: from the center of the reservoir, we conclude that the molecular gas
759: and dust are likely dominantly heated by star formation.
760:
761: From the FIR luminosity of the source in the adopted cosmology
762: ($L_{\rm FIR} = 2.4 \times 10^{13}\,{\mu_L}^{-1}$\,\lsol; Cox \etal\
763: \citeyear{cox02}), we derive a star formation rate
764: \footnote{Assuming
765: SFR=1.5$\times$10$^{-10}\,L_{\rm FIR}$(\msol\,yr$^{-1}$/\lsol)
766: (Kennicutt \citeyear{ken98a}). The dust temperature and spectral index
767: of \pss\ indicate that the FIR continuum emission is dominated by star
768: formation, in agreement with the finding that the source follows the
769: radio-FIR correlation for star-forming galaxies (Beelen et al.\
770: \citeyear{bee06}).} (SFR) of 680\,\msol\,yr$^{-1}$.
771: At least part of the CO emission of the reconstructed source does not
772: appear to follow a systemic trend in velocity. In this picture, this
773: structure may be due to interaction, possibly caused by a major
774: merger. Such an event could both feed the AGN and fuel the starburst,
775: and thus be responsible for the coeval assembly of a supermassive
776: black hole and the stellar bulge in this system. Future observations
777: of the FIR continuum at comparable spatial resolution may shed more
778: light on this situation.
779:
780: Motivated by these results, the dynamical mass derived from the
781: molecular line observations in the host galaxy of the $z$=4.12 quasar
782: \pss\ can be used in an attempt to constrain the relationship between
783: the central SMBH mass and the stellar bulge mass ($M_{\rm
784: BH}$--$M_{\rm bulge}$) in high-$z$ AGN galaxies. Such a relation has
785: been proposed for different types of galaxies in the local universe,
786: and appears to hold over more than three orders of magnitude in SMBH
787: mass, essentially independent of galaxy type (predicting $M_{\rm
788: bulge} \simeq 700\,M_{\rm BH}$, e.g., Kormendy \& Gebhardt
789: \citeyear{kg01}).
790: Assuming the optical lensing factor of $\mu_L^{\rm opt} = 4.7$ derived
791: from our model of \pss\ and the Eddington limit derived in Section 3
792: gives $M_{\rm BH} = 1.5 \times 10^9\,$\msol. With the further
793: assumption that the dynamical molecular structure is seen close to
794: edge-on, and that it traces a major fraction of the gravitational
795: potential that hosts the stellar bulge, we thus find that $M_{\rm
796: bulge} \simeq 30\,M_{\rm BH}$ (subtracting out the black hole and gas
797: masses would even give $M_{\rm bulge} < 20\,M_{\rm BH}$). This value
798: is by more than an order of magnitude offset from the local $M_{\rm
799: BH}$--$M_{\rm bulge}$ relation, but in good agreement with results
800: obtained for other high-$z$ quasars, which show similar or larger
801: offsets (e.g., Walter et al.\ \citeyear{wal04}; Wei\ss\ et al.\
802: \citeyear{wei07}; Riechers et al.\ \citeyear{rie07}; see also Shields
803: et al.\ \citeyear{shi06}).
804:
805: \begin{figure}
806: \epsscale{1.4}
807: \plotone{f6.eps}
808: \caption{Map of the detection probability $P$(pixel $>$ 0)
809: in the source plane after lens inversion, as derived from the full
810: MCMC parameter study. A region of 1.21$''$$\times$1.21$''$ size is
811: shown.
812: \label{f4}}
813: \end{figure}
814:
815: \begin{figure}
816: \epsscale{1.2}
817: \plotone{f7.eps}
818: \caption{Model-predicted differential gravitational magnification
819: between the CO velocity channels shown in Figures \ref{f3} and
820: \ref{f5}. The magnification is shown in magnitudes (i.e., no
821: magnification corresponds to $m_L=0$). The error bars include the
822: modeling uncertainties. The numbers on top indicate the magnification
823: in each velocity channel. The dashed line indicates the total
824: magnification of $\mu_L$=5.34 in the integrated emission line map.
825: \label{f6}}
826: \end{figure}
827:
828:
829: In the case of \pss, one may attempt to account for this offset by
830: assuming that the dynamical structure is seen close to face-on (i.e.,
831: $i < 12^\circ$). This would however predict a large intrinsic CO
832: linewidth ($\Delta v_{\rm FWHM} > 1300$\,\kms ). Such a linewidth
833: would significantly exceed the velocity dispersions observed in the
834: spheroids of massive present day elliptical galaxies (which \pss\ will
835: likely eveolve into), but is of the same order of magnitude as those
836: observed toward some high-$z$ submillimeter galaxies (e.g., Carilli \&
837: Wang \citeyear{cw06}). Such large molecular linewidths are consistent
838: with those predicted by simulations of the hierarchical buildup of
839: massive quasar host galaxies at high redshift (e.g., Narayanan et al.\
840: \citeyear{nar07}), and thus compatible with the possible merger nature
841: of \pss. However, in interacting or merging systems, molecular lines
842: are likely more broad due to the fact that the dynamical molecular
843: structure is not fully virialized yet. The molecular line widths in
844: these galaxies thus may be by a factor of a few higher than the actual
845: virial velocity of the host halo, which would lead to an
846: overprediction of the bulge mass. It thus appears difficult to
847: explain the offset from the local $M_{\rm BH}$--$M_{\rm bulge}$
848: relation by simply assuming a small inclination angle toward the line
849: of sight. Also, assuming that the AGN accretes at sub-Eddington
850: rates, and/or taking into account that more than a third of the
851: dynamical mass derived above is likely not stellar, but accounted for
852: by gas, dust ($<$1\% of the gas mass), and the black hole further
853: increases this offset. Together with previous such examples, our
854: results for \pss\ thus appear to indicate that the black holes in
855: massive galaxies at high redshift assemble earlier than a large
856: fraction of their stellar bulges.
857:
858: The observations and modeling presented herein demonstrate the power
859: of spatially and dynamically resolved molecular gas studies in
860: strongly lensed, distant AGN-starburst systems to provide direct
861: evidence for the scenarios of quasar activity and galaxy assembly in
862: the early universe as suggested by recent cosmological simulations
863: (e.g., Springel \etal\ \citeyear{spr05}). The boost in line intensity
864: and spatial resolution provided by Einstein ring lens configurations
865: are currently the only means by which to probe the dynamical structure
866: of the most distant star-forming galaxies at (sub-)kiloparsec
867: resolution. Such observations provide an important foundation for
868: future observations of molecular gas and dust in the early universe
869: with the Atacama Large Millimeter/submillimeter Array (ALMA), which
870: will be able to probe more typical galaxy populations at high redshift
871: to comparable and higher physical resolution, even without the aid of
872: gravitational lensing.
873:
874: \acknowledgments
875: The authors would like to thank Dennis Downes for helpful discussions,
876: and Chien Y.~Peng for providing an HST image of PSS\,J2322+1944. DR
877: acknowledges support from the Deutsche Forschungsgemeinschaft (DFG)
878: through Priority Program 1177, and from NASA through Hubble Fellowship
879: grant HST-HF-01212.01-A awarded by the Space Telescope Science
880: Institute, which is operated by the Association of Universities for
881: Research in Astronomy, Inc., for NASA, under contract NAS 5-26555. CC
882: acknowledges support from the Max-Planck-Gesellschaft and the
883: Alexander von Humboldt-Stiftung through the Max-Planck-Forschungspreis
884: 2005. We thank the anonymous referee for a thorough reading of the
885: manuscript.
886:
887: \begin{thebibliography}{}
888:
889: \bibitem[Beelen et al.(2006)]{bee06} Beelen, A., Cox, P., Benford, D.~J.,
890: et al.\ 2006, ApJ, 642, 694
891: \bibitem[Brewer \& Lewis(2006a)]{bl06} Brewer B.~J., \& Lewis G.~F.,
892: 2006a, ApJ, 637, 608
893: \bibitem[Brewer \& Lewis(2006b)]{0047} Brewer, B.~J., \& Lewis, G.~F.\ 2006b, \apj, 651, 8
894: \bibitem[Carilli \& Holdaway(1999)]{ch99} Carilli, C.~L., \& Holdaway,
895: M.~A.\ 1999, Radio Science, 34, 817
896: \bibitem[Carilli \etal\ (2001)]{car01} Carilli, C.~L., Bertoldi, F.,
897: Omont, A., Cox, P., McMahon, R.~G., \& Isaak, K.~G.\ 2001, AJ, 122,
898: 1679
899: \bibitem[Carilli \etal\ (2002)]{car02a} Carilli, C.~L., Cox, P.,
900: Bertoldi, F., et al.\ 2002, ApJ, 575, 145
901: \bibitem[Carilli et al.(2003)]{car03} Carilli, C.~L., Lewis, G.~F.,
902: Djorgovski, S.~G., et al.\ 2003, Science, 300, 773 [C03]
903: \bibitem[Carilli \& Wang(2006)]{cw06} Carilli, C.~L., \& Wang, R.\
904: 2006, AJ, 131, 2763
905: \bibitem[Cox et al.(2002)]{cox02} Cox, P., Omont, A., Djorgovski, S.
906: G., et al.\ 2002, A\&A, 387, 406
907: \bibitem[Djorgovski et al.(2000)]{djo00} Djorgovski, S.~G., Gal, R.~R.,
908: Mahabal, A., et al.\ 2000, AAS, 197, 116.07
909: \bibitem[Downes \& Solomon(1998)]{ds98} Downes, D., \& Solomon, P.~M.\
910: 1998, ApJ, 507, 615
911: \bibitem[Ferrarese \& Merritt(2000)]{fer00} Ferrarese, L., \&
912: Merritt, D.\ 2000, ApJ, 539, L9
913: \bibitem[Gebhardt et al.(2000)]{geb00} Gebhardt, K., Bender, R.,
914: Bower, G., \etal\ 2000, ApJ, 539, L13
915: \bibitem[Graham et al.(2001)]{gra01} Graham, A.~W., Erwin, P., Caon,
916: N., \& Trujillo, I.\ 2001, ApJ, 563, L11
917: \bibitem[Gregory(2005)]{gre05} Gregory, P.~C.\ 2005, ``Bayesian
918: Logical Data Analysis for the Physical Sciences: A Comparative
919: Approach with `Mathematica' Support'', Cambridge University Press
920: \bibitem[Guilloteau et al.(1999)]{gui99} Guilloteau, S., Omont, A.,
921: Cox, P., McMahon, R.~G., \& Petitjean, P.\ 1999, A\&A, 349, 363
922: \bibitem[Isaak et al.(2002)]{isa02} Isaak, K.~G., Priddey, R.~S.,
923: McMahon, R.~G., \etal\ 2002, MNRAS, 329, 149
924: \bibitem[Kennicutt(1998)]{ken98a} Kennicutt, R.~C.\ 1998, ApJ, 498,
925: 541
926: \bibitem[Kormendy \& Gebhardt(2001)]{kg01} Kormendy, J., \& Gebhardt,
927: K.\ 2001, AIP conf. proc., 586, 363
928: \bibitem[Magorrian et al.(1998)]{mag98} Magorrian, J., \etal\ 1998,
929: AJ, 115, 2285
930: \bibitem[Narayanan et al.(2008)]{nar07} Narayanan, D., \etal\ 2008,
931: ApJS, 174, 13
932: \bibitem[Omont et al.(2001)]{omo01} Omont, A., Cox, P., Bertoldi, F.,
933: \etal\ 2001, A\&A, 374, 371
934: \bibitem[Peng et al.(2006)]{pen06} Peng, C.~Y., Impey, C.~D., Rix,
935: H.-W., \etal\ 2006, ApJ, 649, 616
936: \bibitem[Riechers et al.(2006)]{rie06} Riechers, D.~A., Walter, F.,
937: Carilli, C.~L., \etal\ 2006, ApJ, 650, 604
938: \bibitem[Riechers et al.(2007)]{rie07} Riechers, D.~A., Walter, F.,
939: Carilli, C.~L., \& Lewis, G.~F.\ 2007, ApJ, submitted
940: \bibitem[Shen et al.(2007)]{she07} Shen, Y., Greene, J.~E., Strauss,
941: M.~A., Richards, G.~T., \& Schneider, D.~P.\ 2007, ApJ, submitted
942: (arXiv:0709.3098)
943: \bibitem[Shields et al.(2006)]{shi06} Shields, G.~A., Menezes, K.~L.,
944: Massart, C.~A., \& Vanden Bout, P.\ 2006, ApJ, 641, 683
945: \bibitem[Solomon \& Vanden Bout(2005)]{sv05} Solomon, P.~M., \& Vanden
946: Bout, P.~A.\ 2005, ARA\&A, 43, 677
947: \bibitem[Spergel et al.(2003)]{spe03} Spergel, D.~N., Verde, L.,
948: Peiris, H.~V., \etal\ 2003, ApJS, 148, 175
949: \bibitem[Spergel et al.(2007)]{spe07} Spergel, D.~N., Bean, R.,
950: Dor\'e, O., \etal\ 2007, ApJS, 170, 377
951: \bibitem[Springel et al.(2005)]{spr05} Springel, V., Di Matteo, T.,
952: Hernquist, L.\ 2005, MNRAS, 361, 776
953: \bibitem[Vignali et al.(2005)]{vig05} Vignali, C., Brandt, W.~N.,
954: Schneider, D.~P.; \& Kaspi, S.\ 2005, AJ, 129, 2519
955: \bibitem[Walter et al.(2004)]{wal04} Walter, F., Carilli, C.,
956: Bertoldi, F., \etal\ 2004, ApJ, 615, L17
957: \bibitem[Wang et al.(2008)]{wan07} Wang, R., Wagg, J., Carilli, C.~L.,
958: \etal\ 2008, AJ, 135, 1201
959: \bibitem[Wei\ss\ et al.(2007)]{wei07} Wei\ss, A., Downes, D., Neri,
960: R., \etal\ 2007, A\&A, 467, 955
961: \end{thebibliography}
962:
963: \end{document}
964: