1: \documentclass[usenatbib]{mn2e}
2: %
3: % input macros for figures
4: \usepackage{psfig}
5: %
6:
7: \def\gs{\mathrel{\raise0.35ex\hbox{$\scriptstyle >$}\kern-0.6em
8: \lower0.40ex\hbox{{$\scriptstyle \sim$}}}}
9: \def\ls{\mathrel{\raise0.35ex\hbox{$\scriptstyle <$}\kern-0.6em
10: \lower0.40ex\hbox{{$\scriptstyle \sim$}}}}
11:
12: \date{Accepted 2008 June 16. Received 2008 June 13; in original form 2008 March 9}
13:
14: \begin{document}
15:
16: \title[KX-selected QSOs in the UKIDSS/UDS field]{A Pilot Survey for
17: KX QSOs in the UKIDSS Ultra Deep Survey Field}
18:
19: \author[Smail et al.]
20: {Ian Smail,$^{\!1\,*}$
21: Rob Sharp,$^{\! 2}$
22: A.\,M.\ Swinbank,$^{\! 1}$
23: M. Akiyama,$^{\! 3}$
24: Y.\ Ueda$^{4}$
25: S.\ Foucaud,$^{\! 5}$
26: \and
27: O.\ Almaini,$^{\! 5}$
28: S.\ Croom$^{6}$
29: \vspace*{6pt} \\
30: $^1$Institute for Computational Cosmology, Department of
31: Physics, Durham University, South Road, Durham, DH1 3LE, UK \\
32: $^2$Anglo-Australian Observatory, Epping, NSW 1710, Australia\\
33: $^3$Subaru Telescope, National Astronomical Observatory of Japan, 650 North A'ohoku Place, Hilo, HI 96720, USA\\
34: $^4$Department of Astronomy, Kyoto University, Kyoto 606-8502, Japan\\
35: $^5$School of Physics and Astronomy, University of Nottingham,
36: University Park, Nottingham NG7 2RD\\
37: $^6$School of Physics, University of Sydney, NSW 2006, Australia\\
38: $^*$Email: ian.smail@durham.ac.uk \\
39: }
40:
41: \maketitle
42:
43: \begin{abstract}
44: We have undertaken a pilot survey for faint QSOs in the UKIDSS Ultra
45: Deep Survey Field using the KX selection technique. These observations
46: exploit the very deep near-infrared and optical imaging of this field
47: from UKIRT and Subaru to select candidate QSOs based on their $VJK$
48: colours and morphologies. We determined redshifts for 426 candidates
49: using the AAOmega spectrograph on the AAT in service time. We identify
50: 17 QSOs ($M_B\ls -23$) in this pilot survey at $z=1.57$--3.29. We
51: combine our sample with an X-ray selected sample of QSOs in the same
52: field (a large fraction of which also comply with our
53: KX selection) to constrain the surface density of QSOs with $K\leq 20$,
54: deriving limits on the likely surface density of 85--150\,deg$^{-2}$.
55: We use the good image quality available from our near-infrared imaging
56: to detect a spatially extended component of the QSO light which
57: probably represents the host galaxies. We also use our sample to
58: investigate routes to improve the selection of KX QSOs at faint limits
59: in the face of the significant contamination
60: by compact, foreground galaxies. The brightest examples from our
61: combined QSO sample will be used in conjunction with a large VLT VIMOS
62: spectroscopic survey of high redshift galaxies in this region to study
63: the structures inhabited by gas, galaxies and growing super-massive
64: black holes at high redshifts in the UKIDSS UDS.
65: \end{abstract}
66:
67: \begin{keywords}
68: galaxies: active -- quasars: general.
69: \end{keywords}
70:
71: %
72: %
73: %
74: \section{Introduction}
75:
76: Simulations of the distribution of baryons in the Universe predict that
77: much material is spread out in a ``cosmic web'' between the galaxies,
78: groups and larger structures visible in conventional surveys. The main
79: components of this material are the Ly$\alpha$ forest, which traces
80: highly ionized hydrogen of low neutral column density and low chemical
81: enrichment distributed through the lower density regions, and metal
82: absorption lines which trace massive galaxy halos (and hence higher
83: density regions) via their metal-enriched gas. To probe this material
84: and investigate its 3-dimensional distribution we must exploit QSOs as
85: bright background sources to trace the web through its absorption.
86:
87: An optimal survey of the cosmic web would use a grid of distant bright
88: QSOs as probes of the intervening matter, with the redshift and
89: equivalent width of these absorbers providing information on the
90: relative distribution and clustering of gas around galaxies and larger
91: structures, as well as providing insights into the chemical enrichment
92: and heating of this material (e.g.\ Morris \& Januzzi 2006). Studies
93: comparing the results from high-redshift galaxy surveys around
94: individual bright QSOs have provided unique insights into the effect of
95: star formation on the gas surrounding young galaxies (e.g.\ Adelberger
96: et al.\ 2005). However, these studies are observationally expensive --
97: it would be much more efficient if a large number of galaxies and QSOs
98: could be compared within a single field as the number of
99: absorber--galaxy pairs scales as N$^2$. The difficulty is that the
100: surface density of bright QSOs is low and hence faint QSOs are required
101: to set up a dense grid of probes needed to sample the full range of
102: cosmic structures in a single field (Prescott et al.\ 2006). Nevertheless,
103: with enough QSOs in a single field the long integration time required
104: to undertake absorption-line analysis on relatively faint background
105: sources (AGN or galaxies) become worthwhile.
106:
107: We are undertaking a highly-sampled survey of the cosmic web in a
108: uniquely well-studied field: the UKIDSS (Lawrence et al.\ 2007)
109: Ultra Deep Survey (UDS) field. The UDS is the deepest, panoramic
110: near-infrared observations so far undertaken with the goal of reaching
111: $K\sim 23$ and $J\sim 25$ over $\sim 0.8$ sq.\ degrees (Foucaud et al.\
112: 2007). Equally deep $BV\! RIz$ imaging is available across the whole
113: field from Subaru/XMM Deep Survey (SXDS, Furusawa et al.\ 2008) and this
114: region has also been extensively studied at X-ray wavelengths as
115: part of the same project (Ueda et al.\ 2008).
116: These data are being analysed to study the build-up of massive galaxies
117: out to $z\sim3$ and its variation with environment (Foucaud et al.\
118: 2007; Lane et al.\ 2007; Cirasulo et al.\ 2007). This field is also
119: the target of a VLT Large Programme (PI: O.\ Almaini) to use the VIMOS
120: spectrograph to obtain redshifts for $\sim 5$,000 high-redshift
121: $K$-selected galaxies. In this paper we present a parallel pilot
122: survey to expand the size of the QSO sample in this field to allow us
123: to relate these luminous galaxies and structures to their surrounding
124: enriched and unenriched gas from a highly-sampled grid of sightlines.
125: Combining these two unique datasets will yield powerful insights into
126: the influence of galaxies and AGN (at the peak era in their activity)
127: on low-density gas in their environments, material which is required
128: for the long-term fueling of the star formation in galaxies.
129:
130: To construct a highly-sampled grid of QSOs in this field we have used
131: $VJK$-band imaging from UKIDSS/UDS to apply the KX method (Warren et al.\
132: 2000; Croom et al.\ 2001; Sharp et al.\ 2002) to identify a sample of
133: $\sim$\,800 $V<23$ potential AGN candidates in this region. The good seeing
134: available in the UDS imaging, 0.8$''$ FWHM across the whole field,
135: yields relatively clean catalogs of point sources -- which are a mix of
136: AGN, stars and compact galaxies. The KX method relies on the power-law
137: shape of AGN continuua across the $VJK$ bands, compared to the $H$-band
138: ``bump'' arising from the opacity minimum in stellar atmospheres, to
139: distinguish between those point sources which are AGN and those which
140: are stars. This color selection has the additional advantage that it
141: provides an unbiased selection of QSOs independent of their dust
142: reddening out to high redshifts (Warren et al.\ 2000), compared to the
143: classical ``UVX'' selected samples.
144:
145: We use a cosmology with $H_{0}=70$\,km\,s$^{-1}$, $\Omega_{M}=0.3$ and
146: $\Omega_{\Lambda}=0.7$ in which 1$''$ corresponds to 8.2\,kpc at
147: $z=2.5$ and 8.5\,kpc at $z=1.5$. All quoted magnitudes are Vega
148: and as the galactic reddening in our field is low, $E(B-V)=0.02$,
149: we have not applied any extinction corrections to the colours or magnitudes
150: in this paper.
151:
152:
153: %
154: %
155: %
156: \section{Observations and Reduction}
157:
158: %
159: % Figure 1
160: %
161: \begin{figure}
162: \centerline{\psfig{file=f1.ps,angle=90,width=4.0in}}
163: \caption{A plot of the FWHM in the $R$-band versus $R$-band magnitude
164: for a random 1-in-3 subset of sources in the UKIDSS UDS. The stellar
165: sequence is visible at $\sim 0.8''$ (image saturation causes the turn
166: up in the sequence at bright magnitudes). To ensure that we are not
167: biased against QSOs with bright host galaxies or very close companions
168: -- we broaden our selection to FWHM\,$\leq 1.25''$. We also plot the
169: new QSOs identified from our survey, which show a modest range in $R$-band
170: FWHM, as well as the existing X-ray selected AGN which comply with our
171: selection and which were used to tune the criteria.
172: }
173: \end{figure}
174:
175:
176: \subsection{Photometric Selection}
177:
178: Our sample selection starts from the matched near-infrared/optical
179: catalogue of sources in the UKIDSS UDS field. The near-infrared data
180: come from the UKIDSS Data Release 1 (Warren et al.\ 2007) and the
181: optical photometry are from an early version of the Subaru catalogues
182: (O.\ Almaini, priv.\ comm.) published by Furusawa et al.\ (2008).
183: The combined catalogue comprises a seeing-matched
184: $K$-band selected sample which is then matched to the seeing-matched
185: optical Subaru catalogues (see Furusawa et al.\ 2008) through the $i$-band. The
186: methods used to create this combined
187: optical--near-infrared photometric catalogue are described more fully
188: in Foucaud et al.\ (2007) and the DR1 catalogue itself is
189: described Almaini et al.\ (in prep). For our analysis we
190: select only those objects lying in areas with good photometry, without
191: bright neighbours and which are unsaturated/etc, leaving us with over
192: 36,000 objects with $BV\! RIzJK$ photometry across a 0.572 degree$^2$
193: field. The 3-$\sigma$ photometric limit of the catalogue in our
194: key bands are $V=27.8$, $R=27.0$, $J=23.4$ and $K=22.2$.
195:
196: In the following we also employ the existing spectroscopic survey of
197: X-ray sources in the SXDF/UDS from Akiyama et al.\ (2008). This provides
198: a useful sample of known AGN to assess our selection
199: criteria. The Akiyama et al.\ survey targets the 952 X-ray sources detected either in the 0.5--2 or
200: 2--10\,keV bands from the area of the {\it XMM-Newton} survey of the SXDF (Ueda et al.\ 2008)
201: which is covered by the deep optical imaging
202: in Furusawa et al.\ (2008). Of these 952 X-ray sources, 648 have optical
203: counterparts brighter than $R_{AB}=24$ in Furusawa et al.\ (2008). Intensive optical spectroscopic observations of these optical counterparts have been performed
204: with Subaru/FOCAS and VLT/VIMOS and to date $\sim 60$\% of the 648 X-ray sources
205: have been spectroscopically identified. The majority of these are moderate redshift
206: AGN, with a tail of QSOs extending out to beyond $z\sim 5$.
207: As we discuss below, in addition to using this catalogue as a training set for our
208: sample selection, we have culled from our catalogue any source with spectroscopy
209: in Akiyama et al., although we have include all the QSOs with
210: $M_B\leq -23.0$ and brighter than $K_{tot}=20.0$ from their survey
211: in our final estimate of the total surface density of QSOs in the $K$-band.
212:
213: We first identify the locus of stars in the FWHM--magnitude plane. We
214: use the Subaru $R$-band FWHM for this purpose for two reasons.
215: Firstly, we do not want to bias our selection against QSOs with
216: detectable host galaxies, as these hosts are likely to be redder than
217: the QSOs this argues for using the bluer optical bands in the
218: morphological selection, rather than our near-infrared data. Secondly,
219: the image quality in the $R$-band provides the best combination of
220: image quality and depth from our available data. We plot $R$-band FWHM
221: versus $R$-band magnitude in Fig.~1 and show our FWHM\,$\leq1.25''$
222: selection limit. This limit was chosen by comparison to the existing
223: spectroscopic sample in this field, with the aim of yielding a sample
224: well-matched to the grasp of the AAOmega spectrograph with modest
225: contamination. Setting the FWHM limit to $\leq 1.0''$ yielded a sample
226: of 218 sources complying with our KX-selection (see below). Of these,
227: 43 matched sources in the existing spectroscopically identified
228: catalogue (Akiyama et al.\ 2008) of which 41
229: were classified as broad emission line sources (although not all of
230: these have $M_B\leq -23$, our definition of a QSO), with just 2 were narrow
231: emission line sources. This selection of AGN thus appeared very pure,
232: but is not necessarily complete and moreover would not fully populate
233: the AAOmega fibres. Extending the FWHM cut to $\leq 1.25''$ increased
234: the sample of sources with the appropriate colours to 758 (well matched
235: to AAOmega), yielding 89 matches in the existing spectroscopic catalog,
236: from which 47 were broad emission line sources, 21 are narrow emission line sources and the
237: remainder were classified as absorption line spectra. Thus a cut on
238: $R$-band FWHM at $\leq 1.25''$ yielded a sufficiently large sample of
239: KX-selected targets to fill all the AAOmega fibres with a moderate
240: level of contamination (estimated from the existing spectroscopic
241: sample). This relatively relaxed constraint on image size also ensures
242: that we do not reject QSOs in which the host galaxy is detectable.
243:
244: We then use the 2$''$-diameter aperture photometry in the $V\! JK$
245: bands to isolate the stellar sequence in colour space (Fig.~2). We
246: limit our sample to $K_{tot}\leq 22.0$ to ensure completeness in the
247: $K$-band catalogue and then required targets complied with: $V_{ap}\leq
248: 23.0$; $(J-K)_{ap}\geq 1.10$; $(J-K)_{ap}\geq 0.25*(V-J)_{ap}+0.40$.
249: This selection is a variation on that used by Croom et al.\ (2001),
250: where the changes reflect small differences in the magnitude systems.
251: We note that as our photometric catalogue is seeing-matched, we do not
252: apply any aperture corrections to any of the colours in this paper. We
253: have also ignored the effect of variability on our measured colours and
254: magnitudes, and caution that this may be responsible for throwing QSOs
255: out of our selection. Finally, we also stress that given the typical
256: colours of QSOs, $(V-K)\sim 2.5$--3.5, the relatively bright $V$-band
257: magnitude limit required by our spectroscopic follow-up means that the
258: QSO sample is effectively limited at $K\sim 20$ so all have very
259: good $K$-band detections, $>20$-$\sigma$.
260:
261: As a final step before our spectroscopic observations, we remove from
262: our sample the 89 objects which are in the existing redshift catalog of
263: Akiyama et al.\ (2008), including their X-ray QSOs which comply with our
264: KX-selection and have $M_B\leq -23.0$, which we discuss further in
265: \S3.2. After removing these sources we have a final input catalogue
266: comprising 670 targets. In our subsequent spectroscopy we gave
267: priority to sources with $V<22.0$ above those with $V<22.5$ and
268: $V<23.0$. We model the effects of this variable completeness on our
269: results.
270:
271: %
272: % Figure 2
273: %
274: \begin{figure}
275: \centerline{\psfig{file=f2.ps,angle=90,width=4.0in}}
276: \caption{$(V-J)$ versus $(J-K)$ for objects from the UKIDSS UDS. We
277: plot as the background the distribution of colours for
278: the $K_{tot}\leq 20.0$ sources in our field. The stellar
279: sequence is clearly visible and can be fairly well separated from the AGN
280: (and compact galaxies) using the colour selections marked. We identify
281: the subset of sources selected on the basis of their morphologies and
282: colours to be targeted for AAOmega spectroscopy and the 17 new QSOs which
283: are identified by this spectroscopy. As can be seen these QSOs are well
284: distributed across the colour plane. We also indicate the potential
285: influence of dust reddening (either intrinsic or from foreground
286: absorption systems) on the colours of QSOs at $z\sim 1$--3. We also
287: mark on the selection boundaries (dashed lines) for sources with $V<23$ and either $K<20$
288: or $K<19$, and the expected variation in apparent colour of
289: a galaxy with a non-evolving E/S0 SED from $z=0$ (top) to $z=0.7$ (dot-dashed line).
290: }
291: \end{figure}
292:
293:
294:
295: %
296: % Figure 3
297: %
298: \begin{figure*}
299: \centerline{\psfig{file=f3.ps,angle=90,width=7.0in}}
300: \bigskip
301: \caption{Restframe spectra of the 17 new KX-selected QSOs from the AAOmega
302: observations of the UKIDSS/UDS. These are order in increasing redshift
303: (from the upper-left) and have been smoothed with a 7\AA-FWHM Gaussian
304: for display purposes.
305: In the final panel we show two combined spectra
306: for the redder and bluer QSOs in the sample, divided at $(B-K)=3.5$.
307: We mark on each spectrum the expected wavelengths of common emission
308: features: Ly$\beta$\,$\lambda$1026, Ly$\alpha$\,$\lambda$1216, N{\sc
309: v}\,$\lambda$1243, C{\sc iv}\,$\lambda$1549, C{\sc iii}]\,$\lambda$1908
310: and Mg{\sc ii}\,$\lambda$2800. The red end of the spectrum of UDS17477
311: shows interference fringing at the fiber/prism interface on the 2dF
312: fibre positioner. We stress that the spectra have not been accurately
313: flux calibrated (\S2.2) and so the continuua shape should be treated
314: with caution.
315: }
316: \end{figure*}
317:
318: \subsection{AAOmega Observations}
319:
320: The spectroscopic observations were undertaken as part of the service
321: programme on the Anglo-Australian Telescope (AAT). The total
322: integration time was 6\,hrs spread equally over three nights (2006
323: Aug.\ 31, Sept.\ 28 and 2007 July 15). The conditions for the first
324: two runs were good, with seeing $<1.5''$ FWHM and good transparency,
325: although the third run suffered from poor seeing, $3.0''$ FWHM, and
326: yielded few additional redshifts.
327:
328: Our observations used the AAOmega spectrograph and the multi-object
329: fibre feed from the 2dF fibre positioner system (Saunders et al.\ 2004;
330: Smith et al.\ 2004; Sharp et al.\ 2006). The dual-beam AAOmega
331: spectrograph was used in it's default low resolution configuration, the
332: 580V and 385R VPH gratings delivering a 3.4 pixel resolution element
333: and R$\sim$1300 over the wavelength range $\lambda= 3700$--8600\AA\ (the
334: lower limit essentially set by the transmission limit of the 38 meter
335: fibre optic feed, the upper limit constrained by the available CCD
336: coverage). The 5700\AA\ dichroic mirror was used to separate the twin
337: beams, the change over occurring in the region $5700\pm200$\AA.
338:
339: The observations where broken up into 2 hour blocks to minimize losses
340: from atmospheric effects\footnote{For details on these considerations
341: see http://www.aao.gov.au/AAO/2df/aaomega/aaomega\_CVD.html}. Each
342: observation block consists of a quartz-halogen flat-field exposure
343: (used both to trace fibre footprint on the CCD and also for the
344: relative response of the fibres), a composite arc lamp frame (utilizing
345: CuAr, FeAr hollow cathode arc-lamps and neutral density filtered Helium
346: and Neon lamps) used for primary wavelength calibration, and four
347: science exposures (each of 1800\,s). Twenty five of the $\sim$370
348: fibres available on the 2dF positioner at the time of observation were
349: allocated to sky positions for sky subtraction. These positions were
350: hand-picked from the Subaru imaging. The contemporaneous sky
351: observations are reduced as part of the science fibre reduction, and
352: then combined to make a high signal-to-noise sky spectrum free from
353: cosmic rays and CCD defects for each science exposure.
354:
355: As is usual for AAOmega observations, the data where processed using
356: the {\sc 2dfdr} data reduction package\footnote{Available from the AAO
357: web site at
358: http://www.aao.gov.au/AAO/2df/aaomega/aaomega\_software.html}. Data
359: from the blue and red arms of the spectrograph are processed
360: independently. In addition to the processing steps common to all fibre
361: spectrograph systems (overscan correction, fibre \emph{tramline map}
362: generation, spectral extraction, fibre relative response and wavelength
363: calibration and sky subtraction) {\sc 2dfdr} implements a Laplace
364: filtering cosmic ray rejection to identify cosmic rays before spectral
365: extraction (van Dokkum 2001) and an iterative sky subtraction involving
366: minimizing the sky subtraction residual by iteratively scaling for:
367: small relative wavelength shifts between the composite sky spectrum and
368: each science spectrum; slight degradation of the science spectrum
369: resolution to match the composite sky; and the intensity of the sky
370: spectrum.
371:
372: The multiple science spectra for each object are then combined using a
373: weighting derived from the average flux recorded in a subset of the
374: brighter objects in each field (to account for seeing and transparency
375: variations during observation). The blue and red arm data were then
376: spliced together to give a continuous spectrum using an archival
377: transfer function to correct for broad band sensitivity. No detailed
378: flux calibration is attempted. Data from all previous nights were then
379: stacked to build up signal on the fainter targets for which redshift
380: had not been determined. Between each night, all low-redshift ($z\ll
381: 1$) objects for which redshifts were reliably measured were replaced in
382: subsequent observations with alternate targets. This maximized the
383: yield for the final science exposure. In total 626 sources from our
384: candidate catalogue of 670 were observed.
385:
386: %
387: %
388: %
389: \begin{table*}
390: \begin{center}
391: {\footnotesize
392: {\centerline{\sc Table 1.}}
393: {\centerline{\sc Properties of new KX-selected QSOs}}
394: % \smallskip
395: \begin{tabular}{lcccccccccc}
396: \hline
397: \noalign{\smallskip}
398: ID & R.A.\ & Dec.\ & $K_{tot}$ & $(B-V)_{ap}$ & $(V-J)_{ap}$ & $(J-K)_{ap}$ & $z$ &
399: M$_B$ & $f_X^a$ & $L_X^b$ \\
400: & \multispan2{ ~(J2000) } &&&&&&& (0.5--4.5\,keV) & (2--10\,keV)\\
401: \hline
402: UDS2074 & 02\,17\,03.86 & $-05$\,31\,40.6 & 18.64& $-0.00\pm 0.01$~ & $1.47\pm 0.01$ & $1.40\pm 0.14$ & 2.286& $-24.2$ & $5.9\pm 0.9$ &1.9\\
403: UDS5596 & 02\,19\,10.00 & $-05$\,29\,30.8 & 18.62& $0.43\pm 0.01$ & $1.22\pm 0.01$ & $1.13\pm 0.16$ & 2.665& $-24.2$ & $2.6\pm 0.8$&1.1\\
404: UDS7821 & 02\,18\,45.94 & $-05$\,28\,06.5 & 18.47& $0.01\pm 0.01$ & $1.39\pm 0.01$ & $1.53\pm 0.14$ & 1.570& $-23.2$ & $8.4\pm 0.9$&1.0 \\
405: UDS10725$^*$ & 02\,18\,50.17 & $-05$\,26\,22.6 & 18.63& $0.60\pm 0.02$ & $2.41\pm 0.02$ & $1.61\pm 0.15$ & 1.979& $-23.3$ & $<1.5$&$<0.8$ \\
406: UDS15871 & 02\,18\,47.62 & $-05$\,23\,19.9 & 19.95& $0.75\pm 0.01$ & $1.08\pm 0.03$ & $1.11\pm 0.26$ & 3.020& $-23.1$ & $2.6\pm 0.5$&1.4\\
407: UDS17477 & 02\,17\,30.58 & $-05$\,22\,22.9 & 19.08& $0.41\pm 0.01$ & $0.90\pm 0.02$ & $1.70\pm 0.18$ & 2.498& $-23.3$ & $5.7\pm 0.6$&2.0\\
408: UDS23719$^*$ & 02\,18\,27.47 & $-05$\,18\,37.2 & 18.19& $0.83\pm 0.02$ & $2.52\pm 0.02$ & $1.88\pm 0.13$ & 2.988& $-24.4$ & $<1.3$&$<1.1$\\
409: UDS25968 & 02\,16\,20.71 & $-05$\,17\,18.7 & 18.85& $1.24\pm 0.01$ & $1.36\pm 0.01$ & $1.19\pm 0.14$ & 3.190& $-23.7$ & $7.7\pm 1.2$&7.6\\
410: UDS34882 & 02\,19\,24.29 & $-05$\,11\,49.6 & 18.66& $0.42\pm 0.01$ & $1.19\pm 0.02$ & $1.83\pm 0.10$ & 2.522& $-23.4$ & $6.3\pm 1.3$&6.4\\
411: UDS34994 & 02\,16\,55.23 & $-05$\,11\,46.7 & 19.50& $0.29\pm 0.01$ & $1.04\pm 0.02$ & $1.21\pm 0.19$ & 2.793& $-23.4$ & $3.4\pm 0.7$&1.5\\
412: UDS66644 & 02\,18\,08.62 & $-04$\,53\,54.6 & 18.35& $0.55\pm 0.01$ & $1.48\pm 0.01$ & $1.68\pm 0.12$ & 2.296& $-23.8$ & $<0.9$&$<0.6$\\
413: UDS70628 & 02\,18\,40.26 & $-04$\,51\,46.6 & 19.39& $0.34\pm 0.01$ & $0.78\pm 0.03$ & $1.74\pm 0.18$ & 2.550& $-23.2$ & $6.2\pm 1.0$&2.3\\
414: UDS71490 & 02\,17\,13.13 & $-04$\,51\,15.9 & 18.43& $0.40\pm 0.01$ & $1.24\pm 0.02$ & $1.41\pm 0.15$ & 2.504& $-24.2$ & $<2.1$&$<1.5$\\
415: UDS80102 & 02\,17\,14.04 & $-04$\,46\,12.6 & 18.70& $0.66\pm 0.01$ & $1.69\pm 0.01$ & $1.22\pm 0.15$ & 2.690& $-24.0$ & $4.2\pm 0.6$&1.7\\
416: UDS86884$^*$ & 02\,17\,35.95 & $-04$\,42\,33.6 & 18.77& $1.17\pm 0.02$ & $2.75\pm 0.02$ & $1.32\pm 0.15$ & 3.292& $-24.1$ & $1.7\pm 0.5$&1.4\\
417: UDS89180 & 02\,17\,32.87 & $-04$\,41\,17.6 & 19.15& $0.34\pm 0.01$ & $1.40\pm 0.02$ & $1.10\pm 0.18$ & 1.907& $-23.3$ & $3.8\pm 0.5$&0.7\\
418: UDS90590 & 02\,16\,47.28 & $-04$\,40\,30.0 & 18.76& $0.28\pm 0.01$ & $0.95\pm 0.02$ & $1.46\pm 0.10$ & 1.728& $-23.1$ & $7.7\pm 0.8$&1.5\\
419: \hline
420: \hline
421: %%%%%30499 z=1.9362 -22.67
422: \end{tabular}
423: \caption{Typical errors are $\pm 0.5''$ on positions, $\pm 0.05$ on
424: total $K$-band magnitudes, $\pm 0.1$--0.3 on $M_B$ and
425: $\pm 0.001$ on $z$. All redshifts have Q=1 except for those marked
426: $^*$ which have Q=2. a) units are 10$^{-15}$
427: erg\,s$^{-1}$\,cm$^{-2}$. b) units are 10$^{44}$\,erg\,s$^{-1}$.
428: }
429:
430: }
431: \end{center}
432: \medskip
433: \end{table*}
434:
435: %
436: %
437: %
438: \section{Analysis, Results and Discussion}
439:
440: \subsection{Spectroscopic Measurements}
441:
442: Redshift determination was performed using the {\sc autoz} package,
443: which is optimized for measuring QSO redshifts (Croom et al.\ 2001)
444: and the {\sc runz} package developed for the 2dFGRS (Colless et al.\
445: 2001) and updated for the 2SLAQ project (Cannon et al.\ 2006). All
446: redshifts were then visually checked. From the 626 objects observed,
447: our spectra yield a total of 426 sources with redshifts of quality (Q)
448: Q=1, unambiguous, or Q=2, probable. Of these 426, 11 are galactic
449: stars: 4 DA white dwarfs, 2 DB white dwarfs and five main sequence stars
450: (one each of A, B, F, G and M). Thus the colour selection from Fig.~2 has
451: significantly reduced, but not totally eliminated, the stellar contamination.
452:
453: We derive the absolute restframe $B$-band magnitudes for these extragalactic sources
454: using {\sc kcorrect} (Blanton \& Roweis 2007) by interpolating based on
455: a fit to the optical and near-infrared photometry spanning the
456: restframe $B$-band. These absolute magnitudes are based on the
457: aperture
458: photometry and so we correct them to total magnitudes using the mean
459: aperture correction for the sample, $\delta = -0.4\pm 0.1$, derived
460: in the $J$-band photometry.
461: We then apply a $M_B\leq -23.0$ cut to identify the
462: QSOs from our spectroscopic sample. This results in 17 QSOs
463: with $M_B\sim -23.3$ from our AAOmega observations in the UDS which
464: comply with our KX and various image selection criteria. Their
465: redshifts span $z=1.57$--3.29 and we show the spectra for these in
466: Fig.~3 and list their positions and photometric properties in Table~1.
467: The X-ray fluxes in the table come from Ueda et al.\ (2008) and are in
468: the observed 0.5--4.5\,keV band (assuming a power-law photon index of
469: 1.8 with no absorption).
470:
471: As expected, the spectra of the QSOs in our sample were all best
472: matched with the QSO (broad emission line) template spectrum employed by {\sc runz}
473: and they typically display broad emission lines, including
474: Ly$\alpha$\,$\lambda$1216, N{\sc v}\,$\lambda$1243, C{\sc
475: iv}\,$\lambda$1549, C{\sc iii}]\,$\lambda$1908 and Mg{\sc
476: ii}\,$\lambda$2800. In the lower-right panel of Fig.~3 we show the
477: composite spectra formed by combining the individual spectra for the
478: redder and bluer QSOs (divided at $(B-K)=3.5$) to demonstrate their
479: common features. We note that the redder QSOs show proportionally
480: higher equivalent width emission lines than the blue QSOs. Given the
481: small sample available, this could result from: 1) a selection effect
482: as the bluer QSOs tend to be lower redshift; 2) a selection effect
483: arising from the need for stronger features to be present in the redder
484: QSOs, which are typically fainter in the wavelength range of our
485: AAOmega observations; or 3) absorption and reddening of the underlying
486: continuum in the redder QSOs.
487:
488: The mean redshift for the new KX QSOs in our survey is $z=2.50\pm 0.50$.
489: This compares to a redshift range of $z=1.08$--4.55 and a mean of
490: $z=2.05\pm 0.71$ for the 23 X-ray selected sources from Akiyama et al.\
491: (2008) which comply with our KX selection and $M_B$ absolute magnitude
492: limit, but which were removed from our initial sample selection. As expected the
493: redshift distributions for these two samples are statistically
494: indistinguishable and the combined sample has a mean redshift of
495: $z=2.30\pm 0.81$.
496:
497:
498:
499: %
500: % Figure 4
501: %
502: \begin{figure}
503: \centerline{\psfig{file=f4.ps,angle=90,width=3.5in}}
504: \caption{The variation in $(B-K)$ colour as a function of $z$ for the
505: QSOs in the UDS, as well as the galactic contaminants. We plot the
506: trend in the median colour for the AAOmega KX-sample (solid) and for the
507: X-ray selected QSOs in the field (dashed). For comparison to the
508: galaxy colours we plot the track of a non-evolving E/S0 galaxy
509: (dot-dashed), which defines the red-envelope of the low-redshift
510: contamination. The horizontal dotted line shows the dividing line for
511: ``red'' QSOs, with $(B-K)\geq 3.5$: $\sim 35$\% of the new KX-selected
512: sample are redder than this limit. The histograms at the bottom of the figure show
513: the redshift distribution for the KX-selected QSOs from our AAOmega
514: observations and the existing X-ray selected QSO sample in the UKIDSS UDS field.
515: }
516: \end{figure}
517:
518: \subsection{QSO Surface Density}
519:
520: We can use our survey to place a lower limit on the number density of
521: QSOs brighter than our effective magnitude limit of $K_{tot}\leq 20$.
522: Correcting for the weighting in our spectroscopic sample selection
523: suggests our KX selection would have yielded more than 17 QSOs if we
524: had observed all of the candidates (this does not include QSOs where
525: the spectra failed to yield an ID). This correction implies 18.2 QSOs in our
526: 0.572 degree$^{2}$ field.
527:
528: To determine the total number density of QSOs in the field, we have to
529: add in the 30 $M_B\leq
530: -23.0$ QSOs brighter than $K_{tot}=20.0$ which were removed from our target list as
531: they had existing spectroscopy from Akiyama et al.\ (2008).
532: These include the 23 QSOs which comply with
533: our KX and morphological selection, and a further seven which
534: are not identified by the KX
535: colour and morphology selections. Four of these seven are more extended
536: (in $R$, and $K$) than our FWHM cut, one falls outside the KX color
537: region and two fall in regions with poor photometric coverage.
538: Combining these two samples yields a surface density of $\geq 85$ QSOs per degree$^{-2}$
539: brighter than $K_{tot}=20.0$.
540:
541: The main uncertainty in the estimate above is the incompleteness in our
542: spectroscopic identifications, which is significant for $V>22$ sources.
543: Correcting for this incompleteness is highly uncertain as it requires
544: an assumption about the spectral properties of any undetected QSOs.
545: Assuming that all of these systems exhibit strong emission lines would
546: suggest little incompleteness in our identification of QSOs, leading to
547: an actual surface density close to the limit quoted above. However,
548: such an assumption is at odds with the spectral properties of even the
549: detected QSOs (Fig.~3). An alternative approach is to assume that the
550: incompleteness of the QSO sample is similar to the overall
551: incompleteness of our identifications for the full spectroscopic sample
552: at the same $V$-band magnitudes (i.e.\ that the incompleteness simply reflects
553: continuum signal-to-noise in the spectra). Adopting this assumption we
554: estimate a surface density of QSOs with $K_{tot}\leq 20.0$ in our
555: survey of 56 degree$^{-2}$ and correcting for our sampling and adding
556: in the known X-ray selected QSOs, raises this to $\sim 150$
557: degree$^{-2}$.
558:
559: Our limit on the QSO surface density are broadly consistent with the
560: published limit of 325$^{+316}_{-177}$\, deg$^{-2}$ at $K\ls 19.5$ by
561: Croom et al.\ (2001), based on the detection of 3 QSOs in a small
562: 48\,arcmin$^2$ field. As expected, our observed QSO surface density is
563: significantly higher than that estimated at $K\sim 15$ by Glikman et
564: al.\ (2004, 2007), of just $\sim 0.1$ deg$^{-2}$. A more useful
565: comparison is to the recently completed $K$-band QSO survey by Maddox
566: et al.\ (2008). They surveyed 12.8 degree$^2$ to $K=17$ and determine
567: a surface density of $15.3\pm 1.1$ deg$^{-2}$, indicating that QSO
568: counts increase roughly linearly with $K$-band flux fainter than
569: $K=17$.
570:
571: In terms of theoretical predictions, Maddox \& Hewett (2006) provide
572: estimates for the number counts of QSOs in the $K$-band based on a
573: standard model of the evolution of the QSO Luminosity Function, but
574: with the addition of a model for the influence of host galaxy light on
575: the apparent magnitudes of the QSOs. Their pure-QSO model (no
576: contribution from the host galaxy) predicts $\sim 100$ degree$^{-2}$
577: QSOs with $M_B\leq -23.0$ QSOs and $K\leq 20.0$, while the inclusion of
578: light from the host galaxy in their models increases this to $\sim 110$
579: degree$^{-2}$. We therefore conclude that our observational limits are
580: consistent with the predicted number counts of QSOs in the
581: $K$-band from Maddox \& Hewett (2006) and depending upon the degree of
582: incompleteness in our survey may support a modest contribution from the
583: host galaxies in the $K$-band light from our QSO sample.
584:
585: \subsection{QSO Properties}
586:
587: %
588: % Figure 5
589: %
590: \begin{figure}
591: \centerline{\psfig{file=f5.ps,angle=90,width=3.5in}}
592: \caption{The radial $K$-band light profiles for the average QSO in our
593: KX-selected sample and a spatially- and magnitude-matched sample of
594: colour-selected stars. The measured FWHM from Moffat fits to the
595: average QSO and star are $0.80\pm 0.05''$ and $0.67\pm 0.05''$
596: respectively, showing that the QSOs are on average more extended than
597: nearby stars. Subtracting a scaled star from the composite QSO suggest
598: the residual light comprises 25--40\% of the light of the QSO,
599: equivalent to $K_{tot}\sim 20.3$, and has an extent of several kpc.
600: The profiles of the average QSOs which are redder and bluer than
601: $(B-K)=3.5$ indicates that the bluer QSOs are more compact, $0.75\pm
602: 0.05''$, although this is not statistically significant.
603: }
604: \end{figure}
605:
606:
607: We find a range of $(B-K)= 2.5$--5.5 for our QSO sample (Fig.~4), less
608: than found for lower-$z$ QSOs (Glikman et al.\ 2004). This moderate
609: range in colour likely reflects the limit imposed by our spectroscopic
610: incompleteness which becomes substantial beyond $V=22$ (equivalent to
611: $K\sim 18$), with only the bluest QSOs having adequate $V$-band signal-to-noise for
612: us to measure a redshift at $K\sim 20$. Nevertheless, we have 6/17
613: QSOs ($\sim35$\%) in our sample redder than the $(B-K)=3.5$, frequently
614: used to identify red QSOs (e.g.\ Jurek et al.\ 2008). Although we note
615: that we have no examples of very red QSOs with $(V-J)\geq 3$ in our
616: AAOmega sample (Croom et al.\ 2001), again likely due to the
617: incompleteness for fainter $V$-band sources in our spectroscopy,
618: but such red QSOs do exist in the X-ray sample in this field.
619:
620: The new KX-selected sample of QSOs in Fig.~4 displays a trend to redder $(B-K)$
621: colours for higher redshift QSOs. This most likely reflects increasing
622: absorption by the Ly$\alpha$ forest effecting the $B$-band emission of
623: more distant QSOs (the X-ray sample exhibits a similar trend, Fig.~4).
624:
625: We can exploit the high-quality multiwavelength data for the UDS region
626: to investigate the properties of the QSOs we have discovered. Of our
627: 17 new QSOs, 13 fall within 5$''$ of X-ray sources in the {\it
628: XMM-Newton} catalogue of this field (Ueda et al.\ 2008) and we list
629: their X-ray fluxes in Table~1. Indeed, we note that 31 sources from
630: our spectroscopic sample of 426 objects are matched to the 952 sources
631: in the X-ray catalog within 5$''$, and that of the 13 of these X-ray
632: sources which lie at $z>1.5$, 12 appear in our KX QSO catalogue. The
633: thirteenth was removed as the spectra were of too low quality. For the
634: 13 X-ray detected QSOs we derive colours in the observed 0.5--4.5\,keV
635: band for these sources. For 12/13 their colours are consistent with no
636: or little absorption, indicating that these are all type-1 AGNs.
637: However, the X-ray color of UDS34882 appears hard in the 0.5--4.5\,keV
638: band, and suggests an absorbing column of N$_H$ of $2 \times 10^{23}$\,cm$^{-2}$
639: (assuming a photon index of 1.8). This source therefore appears to be a
640: candidate of an ``X-ray absorbed'' QSO.
641:
642: Comparing the redshift distributions of the X-ray undetected and
643: detected subsamples in Table~1, we see that the X-ray undetected and detected QSOs
644: have similar redshifts distributions ($z=2.44\pm 0.18$ versus
645: $z=2.52\pm 0.15$, where the errors are bootstrap estimates). Looking
646: at the X-ray luminosities and limits of the detected and undetected
647: QSOs, and comparing these to their optical absolute $B$-band
648: magnitudes, it is clear that many of the X-ray detected QSOs lie close
649: to the limits of the X-ray data and hence the X-ray undetected QSOs may
650: represent the X-ray-faint wing of the distribution, rather than an
651: intrinsically different population. This underlines the effectiveness
652: of deep X-ray surveys for identifying QSOs.
653:
654: We find that none of our 17 new QSOs has detectable radio
655: emission brighter than 100$\mu$Jy at 1.4\,GHz (Simpson et al.\ 2006),
656: so they are all radio-quiet.
657:
658: Finally, we have also compared the image profiles of the $K$-band light in the
659: QSOs to nearby stars. We identify stars based on their $V\! JK$ colours in
660: Fig.~2 and select a subsample with the same range in $K$-band
661: magnitudes and lying within 120$''$ of the QSOs. We stacked the images
662: of 17 QSOs and comparison stars to produce composite profiles and we plot
663: these in Fig.~5. This shows that the QSOs are more spatially extended in
664: the $K$-band than nearby stars of similar magnitudes, with FWHM
665: measured from Moffat fits of $0.80\pm 0.03''$ for the QSOs and $0.67\pm
666: 0.03''$ for the stars. Subtracting a scaled star from the composite
667: QSO image suggest the extended component of the $K$-band light comprises
668: 25--40\% of the light of the QSO, equivalent to $K_{tot}\sim 20.3$, and
669: appears to have a spatial extent of $\sim 2$--5\,kpc. The crude sizes
670: and brightnesses of this component are similar to that measured from AO
671: imaging of high-redshift QSO hosts (e.g.\ Falomo et al.\ 2008) and the
672: fraction of the total light emitted in the restframe $R$-band is also
673: similar to that seen in local QSOs (Maddox \& Hewett 2006). Thus this
674: extended component thus has many of the features expected for the QSO
675: host galaxy (Maddox \& Hewett 2006).
676:
677: \subsection{Foreground Galaxies}
678:
679: We show in Fig.~4 the colour distribution for the large sample of
680: foreground galaxies identified in our survey. There are 302 galaxies
681: in our AAOmega sample with Q=1 spectra, their redshifts span
682: $z=0.03$--1.62 and a mean of $z=0.43\pm 0.22$. The spectral mix of
683: this population includes sources whose spectra are best-fit by broad-line AGN
684: (53), narrow-line AGN (13), narrow emission-line galaxy (23),
685: absorption-line galaxy (75) and spiral galaxy (138) templates. The
686: $(B-K)$ colour distribution for these galaxies extends to very red
687: colours, $(B-K)\gs 7$, and show an upper envelope consistent with the
688: expected colours of a non-evolving $L^*$ E/S0 galaxy (Fig.~4).
689:
690: Eleven of these foreground galaxies, at redshifts of $z=0.20$--1.38,
691: are matched to the {\it XMM-Newton} X-ray source catalogue (Ueda et
692: al.\ 2008) within 5$''$. Ten of these 11 have spectra which are best
693: fit by broad-line (QSO) spectral templates, although their absolute
694: restframe $B$-band luminosities are fainter than $M_B\sim -23$.
695:
696:
697: \subsection{Improving the sample selection}
698:
699: Our identification of 17 new QSOs from spectroscopy of over 400
700: sources indicates a modest rate of return. This is mitigated
701: somewhat by the fact that we removed from our input catalogue those
702: QSOs already known from the existing spectroscopy of X-ray sources in
703: this field. At best these would have doubled our success rate.
704: Nevertheless, it is clear that even with the grasp of the AAOmega
705: spectrograph a substantially larger future KX QSO survey would benefit
706: from reducing the contamination from foreground compact galaxies.
707:
708: We have therefore investigated two routes we could have potentially
709: improved the yield of QSOs in our spectroscopic survey: 1) reducing the
710: limit on object compactness and 2) including a cut based on $(B-V)$
711: colour. To illustrate this we show the distribution of $(B-V)$ colours
712: and $R$-band FWHM for the new QSOs from our AAOmega sample, the X-ray
713: selected QSOs in this field and the contaminating population of
714: foreground galaxies in Fig.~6.
715:
716: A reduction in contamination could be achieved by adopting a $(B-V)\leq
717: 1.2$ cut, which reduces contamination by $\sim 30$\% but at the cost of
718: potentially removing red QSOs and those at higher-redshifts. Similarly
719: using $B$-band FWHM (cut at $\leq 1.1''$) can reduce contamination by
720: $\sim 40$\%. Combining these two selections would reduce contamination
721: by $\sim 60$\% overall and result in a final QSO yield of 10\%.
722:
723: Adopting a more rigorous limit on FWHM, say $\leq 1.0''$, would
724: significantly reduce the contamination, by $\sim 80$\%, at the cost of losing a
725: small number of QSOs with resolved hosts ($\sim 10$\%). For certain experiments,
726: especially those targeting high-redshift QSOs, such an approach might be
727: appropriate as the host galaxies of such systems are unlikely to be
728: individually detected and in this mode KX-selection would yield very
729: high completeness.
730:
731: We caution that our sample is far from ideal for identifying
732: an optimal selection as the selection criteria will be tuned for
733: selecting QSOs similar to those that we have found: at $z\ls 3.5$ and
734: relatively blue. A better training set would be provided by a more
735: complete spectroscopic sample, including near-infrared spectroscopy
736: with FMOS, to overcome the bias towards optically bright QSOs in the
737: current sample.
738: We also note that further detailed studies of the contaminating
739: population may be a productive route to identify criteria to remove
740: them from the KX QSO sample.
741:
742:
743: %
744: % Figure 6
745: %
746: \begin{figure}
747: \centerline{\psfig{file=f6.ps,angle=90,width=3.5in}}
748: \caption{$(B-V)$ versus FWHM in the $R$-band for spectroscopically
749: identified QSOs in the UKIDSS UDS (both the X-ray and KX-selected) and
750: the contaminating population of foreground galaxies. We plot
751: histograms showing the distribution in $(B-V)$ colour and FWHM for each
752: of the three samples: new KX QSOs (solid), existing AGN (dotted) and galaxies
753: (dashed). These distributions illustrate the challenge of isolating
754: the QSO populations from the compact, faint galaxies. We conclude that
755: complete samples of QSOs cannot be efficiently
756: constructed from simple selection based on image compactness or colour,
757: although more stringent limits on image compactness would yield
758: significant reductions in contamination, at relatively modest cost in
759: terms of incompleteness.
760: }
761: \end{figure}
762:
763:
764:
765:
766: %
767: %
768: %
769: \section{Conclusions}
770:
771: We present the results from a pilot survey for QSOs in the UKIDSS UDS
772: field. We employ the KX selection method of Warren et al.\ (2000) to
773: select a sample of faint candidate QSOs from the very deep optical and
774: near-infrared imaging available for this field. We have then followed
775: up these sources with the 400-fibre AAOmega spectrograph in service
776: time on the AAT to identify the high-redshift QSOs in the $\sim
777: 700$ KX-selected compact sources in our catalogue.
778: We identify 17 QSOs with M$_B\leq -23$ in this pilot survey.
779: Supplemented by existing spectroscopy of X-ray selected AGN in this
780: field (Akiyama et al.\ 2008) this yields a combined QSO sample with a
781: surface density of $\sim 85$--150\,degree$^{-2}$ at
782: $K_{tot}\leq 20$ (where the range
783: reflects the uncertainty in the spectroscopic incompleteness in our
784: sample). This sample is roughly $4\times$ deeper than previous
785: KX-selected surveys over comparable areas (Jurek et al.\ 2008) and
786: $40\times$ larger than the previous surveys at similar $K$-band
787: depth (Croom et al.\ 2001).
788:
789: We have used the good image quality of the UKIDSS UDS data to
790: create composite $K$-band images of our QSO sample and a matched
791: sample of stars. The light profiles of the two composites show
792: that the QSOs are more spatially extended. The extended component
793: comprises some 25--40\% of the total $K$-band light of the QSO and has
794: a spatial extent of $\sim 2$--5\,kpc. This most likely represents
795: the host galaxy of the QSO.
796:
797: We analyse our sample to attempt to refine the KX selection technique
798: to reduce the considerable contamination by low redshift, $z\ls 0.6$,
799: compact galaxies. We find that colour or morphological selections
800: focusing on bluer passbands (e.g.\ an additional $(B-V)$ cut or limits
801: on image FWHM in the $B$-band) would reduce the contamination in our
802: parent sample. However, these approaches would bias the resulting
803: sample against redder or higher-redshift QSOs and so may not be
804: desirable for certain applications where completeness is
805: important. Alternatively, if the goal is to maximise
806: the number of high-redshift
807: QSOs detected, then the
808: KX selection using a more conservative FWHM limit, $\leq 1.0''$
809: in the case of the $0.8''$-FWHM UDS imaging, can significantly
810: reduce the foreground contamination while at the same time yielding
811: high completeness ($\sim 90$\%).
812:
813: From our sample we cull a subsample of $\sim 21$ QSOs at $z>2.5$ and $V<23$
814: which are suitable for an absorption line
815: study (see Adelberger et al.\ 2005) employing deep blue spectroscopy
816: with VIMOS on VLT. In addition, the full sample of QSOs (including
817: those at lower redshifts) provide a map of the distribution of rapidly
818: growing black holes within our survey volume, enabling us to relate
819: these to the growth of mass-selected samples of galaxies and the
820: surrounding absorption-line systems. These luminous AGN are expected
821: to have measurable feedback effects on the gas -- an issue of great
822: interest to current models of galaxy formation (in which such forms of
823: feedback are an essential part, Bower et al.\ 2006).
824:
825: %
826: %
827: %
828: \section*{acknowledgments}
829:
830: We thank the members of the UDS Working Group, in particular Chris
831: Simpson, and Natasha Maddox, Paul Hewett and Dave Alexander for help
832: and useful discussions. We thank the Referee for their clear
833: and insightful report, which helped improve the accuracy and
834: presentation of this paper.
835: IRS acknowledges support from the Royal
836: Society. AMS acknowledges support from STFC. This work is based in
837: part on data obtained as part of the UKIRT Infrared Deep Sky Survey.
838: We acknowledge the contributions of the staff of UKIRT to the
839: implementation UKIDSS survey and the Cambridge Astronomical Survey Unit
840: and the Wide Field Astronomy Unit in Edinburgh for processing the
841: UKIDSS data. The United Kingdom Infrared Telescope is operated by the
842: Joint Astronomy Centre on behalf of the Science and Technology
843: Facilities Council of the U.K. This paper uses data obtained through
844: the Service Programme at the Anglo-Australian Telescope. This work was
845: based in part on observations made with the Anglo-Australian
846: Telescope. We warmly thank all the present and former staff of the
847: Anglo-Australian Observatory for their work in building and operating
848: the 2dF and AAOmega facilities.
849:
850:
851: \begin{thebibliography}{}
852: \bibitem{} Adelberger, K., Shapley, A.E., Steidel, C.C., Pettini, M., Erb, D.K., Reddy, N.A., 2005, ApJ, 629, 636
853: \bibitem{} Akiyama, M., et al., 2008, in prep.
854: \bibitem{} Almaini, O., et al., 2008, in prep.
855: \bibitem{} Blanton, M.R., Roweis, S., 2007, AJ, 133, 734
856: \bibitem{} Bower, R.G., Benson, A.J., Malbon, R., Helly, J.C., Frenk, C.S., Baugh, C.M., Cole, S., Lacey, C.G., 2006, MNRAS, 370, 645
857: \bibitem{} Cannon, R., Drinkwater, M., Edge, A., Eisenstein, D., Nichol, R., Outram, P., Pimbblet, K., de Propris, R., Roseboom, I., 2006, MNRAS, 372, 425
858: \bibitem{} Cirasuolo, M., McLure, R.J., Dunlop, J.S., Almaini, O.,
859: Foucaud, S., Smail, I., Sekiguchi, K., et al., 2007, MNRAS, 380, 585
860: \bibitem{} Colless, M.M., Dalton, G., Maddox, S., Sutherland, W.,
861: Norberg, P., Cole, S., Bland-Hawthorn, J., Bridges, T.,
862: et al., 2001, MNRAS, 328, 1039
863: \bibitem{} Croom, S.M., Warren, S.J., Glazebrook, K., 2001, MNRAS, 328,
864: 150
865: \bibitem{} Croom, S.M., et al., 2001, MNRAS, 322, L29
866: \bibitem{} van Dokkum, P., 2001, PASP, 113, 1420
867: \bibitem{} Falomo, R., Treves, A., Kotilainen, J., Scarpa, R., Uslenghi,
868: M., 2008, ApJ, 673, 694
869: \bibitem{} Foucaud, S., Almaini, O., Smail, I., Conselice, C.J., Lane,
870: K. P.; Edge, A.C., Simpson, C., Dunlop, J.S., et al., 2007, MNRAS, 376, L20
871: \bibitem{} Furusawa, H., Kosugi, G., Akiyama, M., Takata, T., Sekiguchi, K., Tanaka, I., Iwata,
872: I., Kajisawa, M., et al., 2008, ApJS, 176, 1
873: \bibitem{} Glikman, E., Gregg, M.D., Lacy, M., Helfand, D.J., Becker, R.H., White, R.L., 2004, ApJ, 607, 60
874: \bibitem{} Glikman, E., Helfand, D.J., White, R.L., Becker, R.H., Gregg, M.D., Lacy, M., 2007, ApJ, 667, 673
875: \bibitem{} Jurek, R.J., Drinkwater, M.J., Francis, P.J., Pimbblet, K.A., 2008, MNRAS, 383, 673
876: \bibitem{} Lane, K., Almaini, O., Foucaud, S., Simpson, C., Smail, I.,
877: McLure, R .J., Conselice, C.J., Cirasuolo, M., et al., 2007, MNRAS, 379, L25
878: \bibitem{} Lawrence, A., Warren, S.J., Almaini, O., Edge, A.C.,
879: Hambly, N.C., Jameson, R.F., Lucas, P., et al., 2007, MNRAS, 379, 1599
880: \bibitem{} Maddox, N., Hewett, P.C., 2006, MNRAS, 367, 717
881: \bibitem{} Maddox, N., Hewett, P.C., Warren, S.J., Croom, S.M., 2008,
882: MNRAS, 386, 1605
883: \bibitem{} Morris, S., Januzzi, B., 2006, 367, 1261
884: \bibitem{} Prescott, M.K.M., Impey, C.D., Cool, R.D., Scoville, N.Z.,
885: 2006, ApJ, 644, 100
886: \bibitem{}Saunders, W., Bridges, T.,
887: Gillingham, P., Haynes, R., Smith, G.A., Whittard, J.D.,
888: Churilov, V., Lankshear, A., et al., 2004, SPIE, 5492, 389
889: \bibitem{} Sharp, R., et al., 2002, 337, 1153
890: \bibitem{} Sharp, R., Saunders, W., Smith, G., Churilov, V., Correll,
891: D., Dawson, J., Farrel, T., Frost, G., et al., 2006, SPIE, 6269, 14
892:
893: \bibitem{} Simpson, C., Martinez-Sansigre, A., Rawlings, S., Ivison, R.,
894: Akiyama, M., Sekiguchi, K., Takata, T., Ueda, Y., Watson, M., 2006,
895: MNRAS, 372, 741
896: \bibitem{} Smith, G.A., Saunders, W., Bridges, T., Churilov, V.,
897: Lankshear, A.,; Dawson, J., Correll, D., et al., 2004, SPIE, 5492, 410
898: \bibitem{} Ueda, Y., et al., 2008, in prep
899: \bibitem{} Warren, S., Hewett, P.C., Foltz, C.B., 2000, MNRAS, 312, 827
900: \bibitem{} Warren, S., Hambly, N.C., Dye, S., Almaini, O., Cross,
901: N.J.G., Edge, A.C., Foucaud, S., Hewett, P.C., et al., 2007, MNRAS, 375, 213
902:
903: \end{thebibliography}
904:
905: \end{document}
906: