0801.4827/ms.tex
1: %---------------------------------------------------------------------
2: %
3: % Title
4: % =====
5: %   Clustering Segregation with Ultraviolet and Optical Luminosities
6: %   of Lyman-Break Galaxies at z~3
7: %
8: % LaTeX
9: % =====
10: %   AASTeX v5.x LaTeX 2e macros
11: %
12: %---------------------------------------------------------------------
13: \documentclass[preprint2]{aastex}
14: %\doublespace
15: %\usepackage{mathrsfs}
16: %\usepackage{color}
17: 
18: %--- definitions of new commands -------------------------------------
19: \newcommand{\uv}{\mbox{$U\!-\!V$}}     % U-V
20: \newcommand{\vz}{\mbox{$V\!-\!z'$}}    % V-z'
21: \newcommand{\rz}{\mbox{$R\!-\!z'$}}    % R-z'
22: \newcommand{\zk}{\mbox{$z'\!-\!K$}}    % z'-K
23: %\newcommand{\gr}{\mbox{$G\!-\!\mathscr{R}$}}   % G-R
24: %\newcommand{\rrii}{\mbox{$\mathscr{R}\!-\!I$}} % R-I
25: \newcommand{\gr}{\mbox{$G\!-\!\mathfrak{R}$}}   % G-R
26: \newcommand{\rrii}{\mbox{$\mathfrak{R}\!-\!I$}} % R-I
27: 
28: \newcommand{\reffig}[1]{Figure~\ref{fig:#1}}
29: \newcommand{\reftbl}[1]{Table~\ref{tbl:#1}}
30: 
31: %---------------------------------------------------------------------
32: %---------------------------------------------------------------------
33: \begin{document}
34: 
35: \title{%
36: The Subaru/\textit{XMM-Newton} Deep Survey (SXDS) -- VII.\\
37: Clustering Segregation with Ultraviolet and Optical Luminosities\\
38: of Lyman-Break Galaxies at $z\sim 3$
39: \altaffilmark{1}
40: }
41: 
42: \author{%
43: Makiko Yoshida~\altaffilmark{2},
44: Kazuhiro Shimasaku~\altaffilmark{2,3},
45: Masami Ouchi~\altaffilmark{4,5},
46: Kazuhiro Sekiguchi~\altaffilmark{6},
47: Hisanori Furusawa~\altaffilmark{7},
48: and
49: Sadanori Okamura~\altaffilmark{2,3}
50: }
51: 
52: \email{myoshida@astron.s.u-tokyo.ac.jp}
53: 
54: \altaffiltext{1}{%
55: Based on data collected at the Subaru Telescope,
56: which is operated by the National Astronomical Observatory of Japan.}
57: \altaffiltext{2}{%
58: Department of Astronomy, Graduate School of Science,
59: The University of Tokyo, Tokyo 113-0033, Japan}
60: \altaffiltext{3}{%
61: Research Center for the Early Universe, Graduate School of Science,
62: The University of Tokyo, Tokyo 113-0033, Japan}
63: \altaffiltext{4}{%
64: Observatories of the Carnegie Institution of Washington,
65: 813 Santa Barbara Street, Pasadena, CA 91101, USA}
66: \altaffiltext{5}{%
67: Carnegie Fellow}
68: \altaffiltext{6}{%
69: Optical and Infrared Astronomy Division,
70: National Astronomical Observatory of Japan,
71: 2-21-1, Osawa, Mitaka, Tokyo 181-8588, Japan}
72: \altaffiltext{7}{%
73: Subaru Telescope, National Astronomical Observatory of Japan, 
74: 650 N. A'ohoku Place, Hilo, HI 96720, USA}
75: 
76: 
77: %---------------------------------------------------------------------
78: \begin{abstract}
79: 
80: We investigate clustering properties of Lyman-break galaxies (LBGs)
81: at $z\sim 3$ based on deep multi-waveband imaging data
82: from optical to near-infrared wavelengths
83: in the Subaru/\textit{XMM-Newton} Deep Field.
84: The LBGs are selected by \uv and \vz\ colors in one contiguous area
85: of 561 arcmin$^2$ down to $z'=25.5$.
86: We study the dependence of the clustering strength on rest-frame
87: UV and optical magnitudes, which can be indicators of star formation
88: rate and stellar mass, respectively.
89: The correlation length is found to be a strong function of
90: both UV and optical magnitudes with brighter galaxies being
91: more clustered than faint ones in both cases.
92: Furthermore, the correlation length is dependent on
93: a combination of UV and optical magnitudes in the sense that
94: galaxies bright in optical magnitude have large correlation lengths
95: irrespective of UV magnitude,
96: while galaxies faint in optical magnitude have correlation lengths
97: decreasing with decreasing UV brightness.
98: These results suggest that galaxies with large stellar masses
99: always belong to massive halos in which they can have various
100: star formation
101: rates, while galaxies with small stellar masses reside in less massive
102: halos only if they have low star formation rates.
103: There appears to be an upper limit to the stellar mass
104: and the star formation
105: rate which is determined by the mass of hosting dark halos.
106: 
107: \end{abstract}
108: 
109: %---------------------------------------------------------------------
110: \keywords{%
111: cosmology: observations ---
112: cosmology: large-scale structure of universe ---
113: galaxies: evolution ---
114: galaxies: high-redshift}
115: 
116: %---------------------------------------------------------------------
117: %---------------------------------------------------------------------
118: \section{INTRODUCTION} \label{sec:intro}
119: 
120: Galaxy evolution in dark halos is a central issue of cosmology
121: in the Cold Dark Matter (CDM) universe.
122: A critical key to the problem is to reveal
123: what kind of dark halos host what kind of galaxies,
124: specifically, the relationship between the mass of dark halos
125: and fundamental quantities of galaxies
126: such as star formation rate and stellar mass.
127: The clustering strength of galaxies can be used to infer
128: the mass of hosting dark halos,
129: since in the standard theoretical framework of galaxy evolution,
130: the large-scale distribution of galaxies
131: is determined by the distribution of underlying dark halos,
132: whose spatial clustering depends on their mass,
133: with more massive halos having stronger spatial clustering
134: \citep{mo1996}.
135: 
136: A number of studies have examined clustering properties of
137: various high-redshift galaxies
138: \citep{giavalisco1998,giavalisco2001,foucaud2003,daddi2003,
139: ouchi2004a,ouchi2005,adelberger2005a,lee2006,kashikawa2006,
140: hildebrandt2007,quadri2007,ichikawa2007}.
141: Measuring clustering strength requires a large galaxy sample
142: from a wide sky coverage.
143: One considerable case in which the relationship
144: between dark halos and galaxies has been successfully explored
145: by means of clustering analysis
146: is that of Lyman-break galaxies (LBGs).
147: LBGs, which are selected by their continuum features in rest-frame UV
148: spectra (the Lyman-break, the Ly$\alpha$ forest, and otherwise nearly
149: flat continuum longward of the Ly$\alpha$) redshifted into optical
150: bandpasses \citep{guhathakurta1990, steidel1996},
151: are normal, young star-forming galaxies with modest dust extinction.
152: Requiring only optical imaging in a few bands,
153: the LBG selection method
154: has yielded the largest and well-controlled samples
155: of galaxies in the young universe at $2\lesssim z\lesssim 6$
156: \citep[e.g.,][]{steidel1999,steidel2003,ouchi2004b,giavalisco2004,
157: dickinson2004,sawicki2006,yoshida2006,iwata2007,bouwens2007}.
158: LBGs are as numerous as the present-day galaxies,
159: suggesting that they play a significant role
160: in the early stage of galaxy evolution.
161: For a review of LBGs including the history of their discovery,
162: see \citet{giavalisco2002}.
163: 
164: An important piece of evidence which links
165: the mass of hosting dark halos to physical properties of LBGs
166: has been first discovered for the UV luminosity;
167: the clustering strength of LBGs increases strongly with UV luminosity
168: \citep[e.g.,][]{giavalisco2001,ouchi2004a,adelberger2005a,
169: lee2006,hildebrandt2007}.
170: Since UV luminosity is sensitive to star formation rate,
171: this implies that the star formation activity of LBGs
172: is somehow controlled by the mass of dark halos which host them.
173: To make a next step forward in our understanding of
174: how the evolution of galaxies is related to the mass of dark halos,
175: it would be necessary to examine the clustering strength
176: as a function of other properties of galaxies
177: like stellar mass, age, and dust extinction.
178: In particular, stellar mass is a robust quantity
179: in terms of the star formation history of galaxies
180: compared to star formation rate, which can vary with time.
181: This kind of analysis is, however, not easy because
182: it requires deep and wide-field observational data
183: covering wavelengths simultaneously from optical to near-infrared.
184: Only the dependence of clustering strength on rest-frame
185: optical luminosity has been examined by \citet{adelberger2005b}
186: for 300 galaxies at $z\sim 2$ (BX objects)
187: selected by UV wavelength in a similar manner to LBGs.
188: 
189: Recently, clustering properties of near-infrared selected galaxies
190: at high redshift
191: have been examined against various quantities of galaxies
192: \citep{quadri2007,ichikawa2007}.
193: The selection based on a near-infrared magnitude allows us to
194: compile a nearly stellar mass-selected sample.
195: However, the studies at high redshift on the basis of near-infrared
196: selected galaxies have so far been inevitably limited to shallow
197: \citep[$K\lesssim 23.0$;][]{quadri2007}
198: or small-field \citep[$\lesssim 25$ arcmin$^2$;][]{ichikawa2007}
199: samples.
200: This is due to the lower sensitivity and smaller area coverage of
201: near-infrared cameras compared to optical ones.
202: 
203: This paper studies the dependence of
204: clustering strength on both UV (apparent $z'$ band) and
205: optical (apparent $K$ band) luminosities
206: for LBGs at $z\sim 3$
207: to investigate how the stellar mass and the star formation rate
208: of galaxies depend on the mass of dark halos which host them
209: at high redshift.
210: The redshift $z\sim 3$ is the best target for such studies,
211: since it is the highest redshift at which ground-based
212: near-infrared imaging can access rest-frame optical wavelengths.
213: To construct a large sample of LBGs at $z\sim 3$,
214: we performed $U$-band imaging observation with Subaru/Suprime-Cam
215: in the Subaru/\textit{XMM-Newton} Deep Field (SXDF),
216: which has a set of very deep optical multi-waveband imaging data
217: from the Subaru/\textit{XMM-Newton} Deep Survey
218: \citep[SXDS;][]{sekiguchi},
219: and deep $J$ and $K$ data
220: taken with the wide-field near-infrared camera
221: WFCAM on UKIRT by the UKIDSS Ultra Deep Survey
222: \citep[UDS;][]{lawrence2007,warren2007}.
223: 
224: The outline of this paper is as follows:
225: In \S\ref{sec:data}, a brief account of the observations
226: and the data is presented.
227: A large sample of LBGs at $z\sim 3$ is constructed in \S\ref{sec:lbg}.
228: Simulations to assess the redshift distribution function
229: and the contamination by interlopers of the sample
230: are also described.
231: Clustering analysis is made in \S\ref{sec:acf}.
232: A summary and conclusions are given in \S\ref{sec:sum}.
233: 
234: Throughout this paper, the photometric system is based on AB magnitude
235: \citep{oke1983}.
236: The cosmology adopted is a flat universe with
237: $\mathrm{\Omega}_m = 0.3$, $\mathrm{\Omega}_\mathrm{\Lambda} = 0.7$,
238: %{\bfseries
239: $\sigma_8 = 0.9$, baryonic density $\mathrm{\Omega}_b = 0.04$,
240: and a Hubble constant of $H_0 = 100\ h$ km s$^{-1}$ Mpc$^{-1}$
241: with $h=0.7$.
242: The correlation length is expressed in units of $h^{-1}$ Mpc
243: to facilitate comparison with previous results.
244: 
245: %---------------------------------------------------------------------
246: %---------------------------------------------------------------------
247: \section{DATA} \label{sec:data}
248: 
249: We carried out $U$-band imaging observations
250: in the southern part of the Subaru/\textit{XMM-Newton} Deep Field
251: (SXDF) [$02^h18^m00^s$, $-05^\circ00'00''$ (J2000)]
252: with Subaru/Suprime-Cam in 2002.
253: The survey field was covered with one single pointing of the
254: Suprime-Cam,
255: i.e., $34' \times 27'$, with a pixel scale of $0.''202$ pixel$^{-1}$.
256: The stacked image has a seeing size, PSF FWHM, of $1.35''$.
257: The total exposure time was $3.7$ hours
258: and the $1\sigma$ surface brightness fluctuation within a $1''$
259: diameter aperture is $29.07$ mag.
260: The $5\sigma$ limiting magnitude within a $2''$ diameter aperture
261: is $26.41$ mag.
262: The SXDF has a set of very wide and deep
263: multi-waveband optical imaging data
264: from the Subaru/\textit{XMM-Newton} deep survey project
265: \citep[SXDS;][]{furusawa2007}
266: taken with Subaru/Suprime-Cam in five standard broad-band filters,
267: $B$, $V$, $R$, $i'$, and $z'$.
268: We combine these data with the $U$-band data
269: to select LBGs at $z\sim 3$.
270: All of the images were aligned and smoothed with Gaussian kernels
271: so that all have the same PSF size as that of the $U$-band image.
272: The surface brightness limits ($1\sigma$ fluctuation
273: within a $1''$ diameter
274: aperture) are 30.94, 30.38, 30.17, 29.90, and 28.87 mag
275: in $B$, $V$, $R$, $i'$, and $z'$, respectively.
276: The $5\sigma$ limiting magnitudes within a $2''$ diameter aperture
277: are 27.88, 27.30, 27.09, 26.77, and 25.76 mag
278: in $B$, $V$, $R$, $i'$, and $z'$, respectively.
279: 
280: Part of the SXDF was imaged in the $J$ and $K$ bands
281: with UKIRT/WFCAM by the UKIDSS Ultra Deep Survey
282: \citep[UDS;][]{lawrence2007,warren2007}.
283: In this study we make use of these data (DR1)
284: to examine near-infrared properties of LBGs at $z\sim 3$.
285: The $J$-band and $K$-band images were also aligned
286: and smoothed to be matched with the optical images.
287: The surface brightness limits ($1\sigma$ fluctuation
288: within a $1''$ diameter
289: aperture) are 26.78 and 26.56 mag in $J$ and $K$, respectively.
290: The $5\sigma$ limiting magnitudes within a $2''$ diameter aperture
291: are 23.67 and 23.47 mag in $J$ and $K$, respectively.
292: 
293: Object detection and photometry were performed using SExtractor
294: version 2.3 \citep{bertin1996}.
295: We detected objects in the $z'$-band image,
296: and for each detected object
297: photometry was made in all the images at exactly the same position
298: by running SExtractor in ``double-image mode''.
299: We adopt MAG\_AUTO in SExtractor for total magnitudes,
300: and use magnitudes within a $2''$ diameter aperture to derive colors
301: \footnote{We also measure magnitudes within a larger aperture of
302: $3''$ diameter,
303: as the PSF FWHM of the final images is somewhat large.
304: However, since over 90 \% of the LBG candidates selected with
305: $2''$-aperture magnitudes overlap with those
306: selected with $3''$-aperture magnitudes,
307: we adopt $2''$-aperture magnitudes
308: in order to obtain colors of faint objects with better S/N.
309: In addition, we compare aperture magnitudes with isophotal magnitudes
310: and find that the isophotal areas of objects satisfying the magnitude
311: threshold for LBG selection ($23.0<z'\le 25.5$)
312: are mostly larger than the $2''$-diameter aperture.
313: This means that $2''$-aperture magnitudes are less
314: noisy than isophotal magnitudes for most objects.}
315: with an aperture correction of $-0.2$ mag for $U$ magnitudes
316: since the PSF shape in the $U$ band image was different from
317: those in the others.
318: The value $-0.2$ was determined so that the difference between
319: the aperture magnitudes and the total magnitudes of the $U$ band
320: became equal to those of the other bands for LBG candidates.
321: The magnitudes of objects were corrected for a small amount of
322: foreground Galactic extinction using the dust map of
323: \citet{schlegel1998}.
324: The reddening is $E(B-V)= 0.023$, corresponding to extinctions of
325: $A_U = 0.10$, $A_B = 0.09$, $A_V = 0.07$, $A_R = 0.06$, $A_{i'} = 0.05$,
326: $A_{z'} = 0.03$, $A_J = 0.02$, and $A_K = 0.01$.
327: 
328: Clustering analysis requires high uniformity in sensitivity
329: over the whole area, since fluctuations of sensitivity can produce
330: spurious clustering signals
331: and bias measurements of clustering strength.
332: We examined the sensitivity variation over the area of the images
333: of respective bands by dividing them into small meshes
334: and estimating the sky noise
335: in each of the meshes.
336: Based on these sky-noise maps,
337: we carefully defined a high-quality region
338: in which the sensitivity is good and uniform,
339: trimming the edges of the images where sky noise was systematically
340: larger due to dithering observation.
341: The effective area with a complete coverage in all of the six
342: optical bands amounts to 740 arcmin$^2$,
343: and the overlapping area of the optical images and
344: the near-infrared images
345: is 561 arcmin$^2$ after the low-quality regions are discarded.
346: 
347: Spectroscopic follow-up observations have been carried out
348: for many objects in the SXDF with Subaru/FOCAS and VLT/VIMOS
349: \citep{akiyama,simpson,saito2007}.
350: The number of spectroscopically observed objects
351: which are located in the region used in this study
352: and have magnitudes in the range of $23.0<z'\le 25.5$,
353: within which we select LBGs,
354: is 63.
355: 
356: 
357: %---------------------------------------------------------------------
358: %---------------------------------------------------------------------
359: \section{LYMAN-BREAK GALAXY SAMPLE AT $z\sim 3$}
360: \label{sec:lbg}
361: 
362: \subsection{Selection of Lyman-break Galaxies}
363: \label{subsec:lbg-select}
364: 
365: \begin{figure}[t]
366:   \begin{center}
367:     \includegraphics[scale=0.4]{f1.eps}
368:   \end{center}
369:   \caption{%
370: The $U$, $V$, and $z'$ bandpasses
371: overplotted on the spectrum of a generic $z=3$ galaxy
372: (thick line), illustrating the utility of color
373: selection technique using these three bandpasses for
374: locating $z\sim 3$ galaxies.
375: \label{fig:sed}}
376: \end{figure}
377: 
378: \begin{figure}[p]
379:   \begin{center}
380:     %\includegraphics[scale=0.4]{f2.eps}
381:     \includegraphics[height=7cm]{f2_small.eps}
382:   \end{center}
383:   \caption{%
384: \uv\ vs \vz\ diagram for the detected objects with $23.0 < z' < 25.5$.
385: When the $U$ and/or $V$ magnitude of an object is fainter
386: than the $1\sigma$ magnitude of the band, the $1\sigma$
387: magnitude is assigned to the object.
388: The predicted colors of model galaxies and stars are overplotted.
389: The blue and red solid lines indicate the tracks for model spectra of
390: young star-forming galaxies of age 0.1 Gyr and 1.5 Gyr, respectively,
391: with reddening of $E(B-V)=0$, $0.15$,
392: and $0.3$ (from left to right).
393: The redshift range is from $z=2$ to higher redshifts,
394: and the circles on the track mark the redshift interval of 0.1
395: with the enlarged circles corresponding to $z=2.5$, $3.0$, and $3.5$.
396: The green dotted, dashed, and dot-dashed lines delineate the tracks for
397: model spectra of local elliptical, spiral, and irregular galaxies,
398: respectively, redshifted from $z=0$ to $2$ without evolution.
399: The circles on each track mark $z=0$, $1$, and $2$.
400: The yellow asterisks represent the colors of 175 Galactic stars
401: given by \citet{gunn1983}.
402: The thick orange line indicates the boundary which we adopt for
403: the selection of $z\sim 3$ LBGs.
404: \label{fig:color}}
405: \end{figure}
406: 
407: \begin{figure}[t]
408:   \begin{center}
409:     \includegraphics[scale=0.4]{f3.eps}
410:   \end{center}
411:   \caption{%
412: UV luminosity function for LBGs at $z\sim 3$.
413: Our data are shown by the filled circles.
414: The open squares and crosses are from \citet{steidel1999}
415: and \citet{sawicki2006}, respectively.
416: \label{fig:lf}}
417: \end{figure}
418:   
419: \begin{figure}[t]
420:   \begin{center}
421:     \includegraphics[scale=0.4]{f4.eps}
422:   \end{center}
423:   \caption{%
424: Sky distribution of the 962 LBGs in our sample.
425: The region outlined by the thick line indicates the area observed
426: in the $J$ and $K$ bands,
427: while the rest is the area observed only in optical bands,
428: after removal of low-quality regions.
429: The large and small circles represent LBGs with $23.0<z'\le 24.5$
430: and $24.5<z'\le 25.5$, respectively.
431: LBGs having $K$ magnitudes are further distinguished
432: by $z'$ and $K$ magnitudes with different colors,
433: where red, green, cyan, and blue represent LBGs with
434: $23.0<z'\le 24.5$ and $K\le 23.46$, $24.5<z'\le 25.5$ and $K\le 23.46$,
435: $23.0<z'\le 24.5$ and $K>23.46$, $24.5<z'\le 25.5$ and $K>23.46$,
436: respectively.
437: The projected comoving scale of 10 Mpc at $z=3.3$
438: is shown at the top left.
439: North is up and east is to the left in this image.
440: \label{fig:space}}
441: \end{figure}
442:   
443: We find that a combination of $U$, $V$, and $z'$ bands works best
444: to select LBGs at $z\sim 3$ among our bandpasses set (\reffig{sed}).
445: Note that the $U$-band filter of the Suprime-Cam is significantly
446: redder than standard $U$-band filters.
447: Consequently, the mean redshift of LBGs selected is
448: higher than those of traditional $U$-drop LBG samples
449: (see \S\ref{subsec:lbg-comp}).
450: In \reffig{color}, we show the distribution of detected objects
451: with $23.0 < z' \le 25.5$ in the \uv vs \vz\ diagram,
452: as well as predicted positions of high-redshift galaxies
453: and foreground objects (lower-redshift galaxies and Galactic stars).
454: When the magnitude of an object in $U$ and/or $V$ is fainter
455: than the $1\sigma$ magnitude,
456: the $1\sigma$ magnitude is assigned.
457: 
458: We set the selection criteria for LBGs at $z\sim 3$ as:
459: \begin{mathletters}
460:   \begin{eqnarray}
461:       && 23.0 < z' \le 25.5, \\
462: %      && U-V \ge 0.8,\ \ \ V-z' \le 2.7,\ \ \ U-V \ge 1.8(V-z') + 1.6.
463:       && U-V \ge 0.8,\ \ \ V-z' \le 2.7, \nonumber \\
464:       && \hspace{0.5cm} U-V \ge 1.8(V-z') + 1.6.
465:   \end{eqnarray}
466: \end{mathletters}
467: The boundaries on the \uv\ vs \vz\ diagram
468: defined by these color criteria
469: are outlined with the thick orange line in \reffig{color}.
470: The number of LBG candidates selected is 962 in total, among which
471: $708$ are found in the region observed in the $J$ and $K$ bands.
472: 
473: Among the 63 spectroscopic objects, 4 objects
474: satisfy the selection criteria and all of them are identified to be
475: at $z>2.9$.
476: On the other hand,
477: there are two additional objects which are found
478: in the redshift range of $2.9<z<3.7$ but do not pass the
479: criteria.
480: These two are
481: fainter in $U$ than the $1\sigma$ magnitude and are not red enough
482: in $U_{1\sigma}-V$ color.
483: The possibility of missing targeted galaxies for this reason
484: is taken into account in a simulation for estimating the completeness
485: of our sample described in \S\ref{subsec:lbg-comp}.
486: The missed objects have $z'$ magnitudes of $z'=24.93$ and $25.13$.
487: In fact, the simulation shows that
488: a small fraction of $z\sim$ 40\% of galaxies
489: at $2.9<z<3.7$ with these magnitudes is selected by the criteria.
490: 
491: The luminosity function for the LBG sample at $z\sim 3$ is
492: derived in the same manner as in \citet{yoshida2006}
493: using the completeness and the contamination fraction
494: obtained in the following subsections
495: (\S\ref{subsec:lbg-comp} and \S\ref{subsec:lbg-contam}).
496: \reffig{lf} shows the luminosity function
497: in comparison with those in the literature
498: \citep{steidel1999,sawicki2006}.
499: Note that the absolute UV luminosities of our galaxies
500: are based on apparent
501: $z'$ magnitudes, while others are based on apparent
502: $R$ magnitudes.
503: The agreement of our luminosity function with
504: those in previous studies is very good,
505: particularly when one considers different sample selections.
506: \reffig{space} shows the sky distribution of the LBG sample.
507: 
508: \subsection{Redshift Distribution Function} \label{subsec:lbg-comp}
509: 
510: \begin{figure}[t]
511:   \begin{center}
512:     \includegraphics[scale=0.4]{f5.eps}
513:   \end{center}
514:   \caption{%
515: Redshift distribution functions, $N(z)$, of the LBGs
516: with different apparent magnitudes
517: estimated from Monte Carlo simulations.
518: The red, magenta, green, cyan, and blue lines denote the $N(z)$
519: for $z'=23.25$, 23.75, 24.25, 24.75, 25.25, respectively.
520: The thick black line indicates the magnitude-weighted
521: redshift distribution function.
522: \label{fig:complete}}
523: \end{figure}
524:   
525: The redshift distribution function of the LBG sample is estimated
526: as a function of magnitude through a Monte Carlo simulation.
527: In the simulation, we generate artificial LBGs over an apparent
528: magnitude range of $23.0\le m_{z'}\le 25.5$
529: with an interval of $\Delta m=0.5$, and over a redshift range of
530: $2.0\le z\le 4.5$ with an interval of $\Delta z = 0.1$.
531: Model spectra of the artificial LBGs are constructed
532: using the stellar population synthesis code
533: developed by \citet{kodama1997}.
534: As model parameters, an age of 0.1 Gyr, a Salpeter initial mass
535: function, and a star-formation timescale of 5 Gyr are adopted,
536: and five values of reddening, $E(B-V) = 0.0$, $0.1$, $0.2$, $0.3$,
537: and $0.4$,
538: are applied using the
539: dust extinction formula for starburst galaxies by
540: \citet{calzetti2000}.
541: These values reproduce the average rest-frame ultraviolet-optical
542: spectral energy distribution
543: of LBGs observed at $z\sim 3$ \citep{papovich2001}.
544: The absorption due to the intergalactic medium is applied
545: following the prescription by \citet{madau1995}.
546: The colors are calculated by convolving thus
547: constructed model spectra with the response functions of the
548: Suprime-Cam filters.
549: We assume that the surface brightness distribution of LBGs is Gaussian
550: and assign apparent sizes to the artificial LBGs so that their size
551: distribution measured by SExtractor
552: matches that of the observed LBG candidates.
553: The artificial LBGs are then distributed randomly on the original
554: images after adding Poisson noise according to their magnitudes, and
555: object detection and photometry are performed in the same manner
556: as done for real objects.
557: A sequence of these processes is repeated 100 times to obtain
558: statistically accurate values of completeness.
559: In the simulation, the completeness for a given apparent magnitude,
560: redshift, and $E(B-V)$ value
561: can be defined as the ratio in number of the simulated LBGs
562: which are detected and also satisfy the selection criteria, to all
563: the simulated objects with the given magnitude, redshift,
564: and $E(B-V)$ value.
565: We calculate the completeness of the LBG sample
566: by taking a weighted average of the completeness
567: for each of the five $E(B-V)$ values.
568: The weight is taken using the $E(B-V)$ distribution function of
569: $z\sim 4$ LBGs derived by
570: \citet{ouchi2004b} (open histogram in the bottom panel of their Fig.20),
571: which has been corrected for
572: incompleteness due to selection biases.
573: The resulting completeness, $p(m,z)$, is shown in \reffig{complete}.
574: 
575: The magnitude-weighted redshift distribution function
576: of our LBG sample is derived from $p(m,z)$
577: by averaging the magnitude-dependent completeness
578: weighted by the number of LBGs in each magnitude bin
579: (a thick black line in \reffig{complete}).
580: The average redshift, $\bar{z}$, and its standard deviation, $s_z$,
581: are calculated to be $\bar{z}=3.3$ and $s_z=0.3$.
582: As we mentioned in \S\ref{subsec:lbg-select},
583: due to the redder $U$-band filter of the Suprime-Cam,
584: the redshift distribution function is biased toward
585: higher redshifts in comparison with traditional $U$-drop LBGs,
586: whose redshifts are distributed around $z\sim 3.0$
587: \citep[e.g.,][]{steidel2003}.
588: 
589: The measurement of the correlation length from the angular correlation
590: function relies on the estimation of the redshift distribution function
591: (\S\ref{sec:acf}).
592: In order to explore to what extent the redshift distribution function
593: and the resultant correlation length are affected
594: by the model assumption,
595: we recalculate the redshift distribution function
596: using a different model spectrum with an age of 1.5 Gyr.
597: In fact, some of the LBGs in our sample have red \zk colors explained
598: by such spectra.
599: The derived correlation lengths based on the two different model
600: spectra are found to be consistent within the errors.
601: 
602: \subsection{Contamination by Interlopers} \label{subsec:lbg-contam}
603: 
604: \begin{figure}[t]
605:   \begin{center}
606:     \includegraphics[scale=0.4]{f6.eps}
607:   \end{center}
608:   \caption{%
609: Fraction of interlopers as a function of magnitude for the LBG sample.
610: \label{fig:contam}}
611: \end{figure}
612: 
613: We estimate the fraction of low-redshift interlopers in the LBG
614: sample also by a Monte Carlo simulation as follows.
615: The adopted boundary redshift between interlopers and LBGs is
616: $z_0=2.9$.
617: We use objects in the Hubble Deep Field North (HDFN), for which
618: best-fit spectra and photometric redshifts are given by
619: \citet{furusawa2000}, as a template of the color, magnitude, and
620: redshift distribution of foreground galaxies, and generate 929
621: artificial objects which mimic the HDFN objects.
622: The apparent sizes of the artificial objects are adjusted so that
623: the size distribution recovered from the simulation is similar to
624: that of the real objects in our catalogs.
625: We distribute the artificial objects randomly on the original images
626: after adding Poisson noise according to their magnitudes, and
627: perform object detection and photometry in the same manner as
628: employed for real objects.
629: A sequence of these processes is repeated 100 times.
630: In the simulation, the number of interlopers can be defined as the
631: number of the simulated objects with low redshifts ($z<z_0$) which are
632: detected and also satisfy the selection criteria for LBGs.
633: The number of interlopers expected in the LBG sample can then be
634: calculated by multiplying the raw number by a scaling factor
635: which corresponds to the
636: ratio of the area of our field (740 arcmin$^2$) to
637: the area of the HDFN
638: multiplied by the repeated times (100 $\times $ 3.92 arcmin$^2$).
639: \reffig{contam} shows the fraction
640: of interlopers for our LBG sample as a function of magnitude.
641: The fraction is found to be at most 6\% at any magnitude.
642: 
643: %---------------------------------------------------------------------
644: %---------------------------------------------------------------------
645: \section{CLUSTERING PROPERTIES} \label{sec:acf}
646: 
647: \subsection{Method} \label{subsec:acf-method}
648: 
649: We measure the angular correlation function (ACF), $\omega(\theta)$,
650: using the estimator proposed by \citet{landy1993}:
651: \begin{eqnarray}
652:   \omega_\mathrm{obs}(\theta) &=&
653:     \frac{DD(\theta)-2DR(\theta)+RR(\theta)}{RR(\theta)},
654: \end{eqnarray}
655: where $DD(\theta)$, $DR(\theta)$, and $RR(\theta)$
656: denote the numbers of galaxy-galaxy, galaxy-random,
657: and random-random pairs, respectively,
658: with angular separations between $\theta-\delta\theta/2$
659: and $\theta+\delta\theta/2$.
660: The distribution of random points is subject to exactly
661: the same geometry
662: as the observed area, avoiding regions where galaxies
663: are not detected, e.g., in the vicinity of bright stars.
664: We generate 100 times as many random points as the number of galaxies
665: in order to reduce the uncertainties.
666: The errors are estimated by bootstrap resampling method
667: \citep{ling1986}.
668: As the contamination is very small for our LBG sample,
669: we do not apply any correction for it.
670: 
671: The ACF can be approximated as a power law given by
672: \begin{eqnarray}
673: \omega(\theta) &=& A_\omega\theta^{-\beta}.
674: \end{eqnarray}
675: However, since the average number of galaxies in a given field
676: is estimated
677: from a sample itself and
678: fluctuations on the scale of the field size are not accounted for,
679: the measured ACF is underestimated by a constant
680: known as the integral constraint, IC \citep{groth1977}:
681: \begin{eqnarray}
682: \omega_\mathrm{obs}(\theta) &=& \omega_\mathrm{true}(\theta)
683:   - \mathrm{IC}
684:   \ =\ 
685:   A_\omega\theta^{-\beta} - \mathrm{IC}.
686: \label{eq:omega}
687: \end{eqnarray}
688: The value of the integral constraint is equal to the
689: variance of the number of galaxies in the field:
690: \begin{eqnarray}
691: \mathrm{IC} &=& \frac{1}{N_\mathrm{gal}} + \sigma_\omega^2,
692: \label{eq:ic}
693: \end{eqnarray}
694: where the first term is the Poisson variance and the second term
695: accounts for an additional variance caused by clustering.
696: The variance caused by clustering can be estimated by
697: integrating $\omega_\mathrm{true}(\theta)$
698: over the survey area \citep{roche1999}:
699: \begin{eqnarray}
700: \sigma_\omega^2 &=&
701:   \frac{1}{\Omega^2}\int\int\omega_\mathrm{true}(\theta)\,
702:   d\Omega_1d\Omega_2 \nonumber \\
703:             &=&
704:   \frac{\sum_iRR(\theta_i)\omega_\mathrm{true}(\theta_i)}
705:        {\sum_iRR(\theta_i)} \nonumber \\
706:             &=&
707:   \frac{\sum_iRR(\theta_i)A_\omega\theta^{-\beta}}
708:        {\sum_iRR(\theta_i)}.
709: \end{eqnarray}
710: The quantity $\sigma_\omega^2/A_\omega$ is estimated directly
711: from the random point catalogs for a given value $\beta$.
712: Then the amplitude of the ACF, $A_\omega$, can be estimated
713: through $\chi^2$ fitting of $\omega_\mathrm{obs}(\theta)$
714: using equations (\ref{eq:omega}) and (\ref{eq:ic}).
715: Following recent results in clustering studies of LBGs
716: \citep{adelberger2005a,lee2006},
717: we assume $\beta = 0.6$ in what follows.
718: The fitting is made only at $\theta>10''$ since at $\theta<10''$
719: there may be a significant contribution from nonlinear small scale
720: clustering \citep{ouchi2005,lee2006}.
721: The error in $A_\omega$ is estimated from the range in which
722: an increase in $\chi^2$ from the best-fit value is less than unity.
723: 
724: We derive the spatial correlation function, $\xi(r)$,
725: by inverting $\omega(\theta)$ using the Limber transform
726: \citep{peebles1980}.
727: If the ACF is a power law, the spatial correlation function
728: also has to have a power law form:
729: \begin{eqnarray}
730:   \xi(r) &=& \left(\frac{r}{r_0}\right)^{-\gamma},
731: \end{eqnarray}
732: where $r_0$ is the spatial correlation length and $\gamma = \beta+1$.
733: The $A_\omega$ is related to $r_0$ as
734: \begin{eqnarray}
735:   A_\omega &=& Cr_0^\gamma\int F(z)D_\theta^{1-\gamma}(z)N(z)^2g(z)
736:                                            \,dz \nonumber \\
737:            & & \hspace{1.8cm}{}\times\left[\int N(z)\,dz\right]^{-2},
738: \label{eq:r0}
739: \end{eqnarray}
740: where $D_\theta$ is the angular diameter distance,
741: $N(z)$ is the redshift distribution function,
742: \begin{eqnarray}
743:   g(z) &=& \frac{H_0}{c}\left\{(1+z)^2(1+\Omega_0z+\Omega_\Lambda
744:               \left[(1+z)^{-2}-1)\right]^{1/2}\right\}, \nonumber
745: \end{eqnarray}
746: and
747: \begin{eqnarray}
748:   C &=& \sqrt{\pi}\frac{\Gamma[(\gamma-1)/2]}{\Gamma(\gamma/2)}.
749:   \nonumber
750: \end{eqnarray}
751: The function $F(z)$ describes the evolution of $\xi(r)$ with redshift.
752: The evolution is often modeled as
753: $F(z) = \left[(1+z)/(1+\bar{z})\right]^{-(3+\epsilon)}$.
754: We assume constant clustering in comoving units
755: in the redshift range of our LBG sample;
756: in this case the parameter $\epsilon$ is specified by
757: $\epsilon = \gamma-3$.
758: For the redshift distribution function, we use the one obtained by
759: the simulation in \S\ref{subsec:lbg-comp}.
760: 
761: The standard CDM model predicts that the clustering of dark halos
762: is correlated with halo mass \citep{mo1996}.
763: We use the observed spatial correlation function of LBGs to infer
764: the mass of dark halos hosting them
765: on the basis of the analytic model given by \citet{sheth2001},
766: which is derived from a fit to large N-body simulations.
767: According to the model, the bias of dark halos, $b_\mathrm{DH}$,
768: which relates the clustering of dark halos
769: to that of the overall dark matter, is calculated by
770: \begin{eqnarray}
771: b_\mathrm{DH} &=& 1 + \frac{1}{\delta_c}
772: \left[\nu'^2+b\nu'^{2(1-c)}\right. \nonumber \\
773: & & \hspace{0.5cm}\left.
774:     {}-\frac{\nu'^{2c}/\sqrt{a}}{\nu'^{2c}+b(1-c)(1-c/2)}\right],
775: \label{eq:bias_dh}
776: \end{eqnarray}
777: where $\nu' = \sqrt{a}\nu$, and the constants
778: $a = 0.707$, $b = 0.5$, $c = 0.6$.
779: Here, $\nu$ is defined by
780: \begin{eqnarray}
781: \nu &\equiv& \frac{\delta_c}{\sigma(M, z)}
782:     \ =\ \frac{\delta_c}{D(z)\sigma(M, 0)},
783: \label{eq:nu}
784: \end{eqnarray}
785: where $D(z)$ is the growth factor, $\sigma(M, z)$
786: is the relative mass fluctuation in spheres that contain
787: an average mass $M$,
788: and $\delta_c \approx 1.69$ represents the critical amplitude
789: of the perturbation for collapse.
790: We calculate $D(z)$ following \citet{carroll1992}
791: and $\sigma(M,0)$ from the initial power spectrum
792: with a power law index
793: of $n=1$ using the transfer function of \citet{bardeen1986}.
794: Since $\nu'$ is a function of redshift and mass,
795: the mass of dark halos is estimated from equation (\ref{eq:bias_dh}),
796: once the bias of the dark halos and the redshift are given.
797: We assume that the observed bias of galaxies at a large scale
798: reflects the bias of dark halos hosting them
799: (i.e., $b_\mathrm{gal}\simeq b_\mathrm{DH}$),
800: thereby obtain an estimate of the hosting halo mass.
801: The bias of galaxies against dark matter at a large scale
802: ($=8\ h^{-1}$Mpc) is measured by
803: \begin{eqnarray}
804: b_\mathrm{gal} &=& \sqrt{\frac{\xi(8\ h^{-1}\mathrm{Mpc})}
805:   {\xi_\mathrm{DM}(8\ h^{-1}\mathrm{Mpc})}} \nonumber \\
806:   &=& \sqrt{\frac{\left[(8\ h^{-1}\mathrm{Mpc})/r_0\right]^{-\gamma}}
807:        {\xi_\mathrm{DM}(8\ h^{-1}\mathrm{Mpc})}},
808: \label{eq:bias_gal}
809: \end{eqnarray}
810: where $\xi_\mathrm{DM}$ is the predicted spatial correlation function
811: of the overall dark matter and is computed by the nonlinear model of
812: \citet{peacock1996}.
813: 
814: \subsection{Results} \label{subsec:acf-result}
815: 
816: \begin{figure}[t]
817:   \begin{center}
818:     \includegraphics[scale=0.4]{f7.eps}
819:   \end{center}
820:   \caption{%
821: Angular correlation function, $\omega(\theta)$,
822: for the $z\sim 3$ LBG sample.
823: \label{fig:acf}}
824: \end{figure}
825: 
826: \reffig{acf} shows the ACF for the full LBG sample
827: containing 962 galaxies
828: with $23.0 < z'\le 25.5$.
829: We calculate the spatial correlation length from the amplitude of the
830: ACF using equation (\ref{eq:r0}) to find
831: $r_0 = 5.5^{+0.8}_{-0.9}\ h^{-1}$Mpc (comoving; \reftbl{clustering}).
832: The average mass of dark halos hosting these LBGs is estimated to be
833: $M_\mathrm{DM}\approx 2\times 10^{12}M_\odot$
834: from equations (\ref{eq:bias_dh}), (\ref{eq:nu}), (\ref{eq:bias_gal}).
835: We compare the measurement of the spatial correlation length
836: with those from other authors for samples
837: with similar redshift and luminosity ranges.
838: \citet{hildebrandt2007} found $r_0 = 5.0^{+0.2}_{-0.2}\ h^{-1}$Mpc
839: for LBGs with $23.5 < R \le 25.5$, which is consistent with ours.
840: On the other hand, \citet{adelberger2005a} and \citet{lee2006}
841: found slightly smaller values of
842: $r_0 = 4.0^{+0.6}_{-0.6}\ h^{-1}$Mpc for LBGs with $23.5 < R \le 25.5$
843: and $r_0 = 4.1^{+0.1}_{-0.2}\ h^{-1}$Mpc for LBGs with $R \le 25.5$,
844: respectively.
845: This may be due in part to the fact that the redshift distribution
846: function of our LBG sample is biased toward higher
847: redshifts
848: as discussed in \S\ref{subsec:lbg-comp}.
849: The average redshift of Adelberger et al.'s (2005a) sample
850: is $\bar{z}=2.9$,
851: while ours is $\bar{z}=3.3$.
852: At a given magnitude limit, this means that our sample includes
853: brighter LBGs, which are more strongly clustered.
854: In addition, at a fixed luminosity,
855: LBGs at higher redshifts are more strongly clustered.
856: We also note
857: that our sample is $z'$-magnitude limited,
858: while the others are $R$-magnitude limited.
859: The distribution in the $z'$ vs $R$ diagram of our sample implies that
860: there might be more galaxies with $z'>25.5$ than those with $R>25.5$,
861: suggesting that our sample may systematically miss more faint galaxies
862: than the other samples.
863: This could result in our larger $r_0$.
864: 
865: In the following,
866: we measure the spatial correlation length for various subsamples
867: to investigate the relationship between clustering strength
868: and galaxy properties.
869: 
870: \subsubsection{Dependence of Clustering Strength on UV Luminosity}
871: \label{subsubsec:acf_mag_z}
872: 
873: \begin{figure}[t]
874:   \begin{center}
875:     \includegraphics[scale=0.4]{f8.eps}
876:   \end{center}
877:   \caption{%
878: Angular correlation function, $\omega(\theta)$,
879: for three subsamples selected by $z'$ magnitude.
880: The red, green, and blue symbols indicate the $\omega(\theta)$ of
881: LBGs with $23.0 < z' \le 24.5$, $24.0 < z' \le 25.0$,
882: and $24.5 < z' \le 25.5$, respectively.
883: \label{fig:acf_mag_z}}
884: \end{figure}
885: 
886: \begin{figure}[h]
887:   \begin{center}
888:     \includegraphics[scale=0.4]{f9.eps}
889:   \end{center}
890:   \caption{%
891: Spatial correlation length, $r_0$, as a function of $z'$ magnitude.
892: \label{fig:r0_mag_z}}
893: \end{figure}
894: 
895: It is reported that the clustering strength of LBGs
896: depends on the rest-frame UV luminosity
897: in the sense that galaxies with higher UV luminosities
898: have larger spatial correlation lengths
899: \citep[e.g.,][]{giavalisco2001,ouchi2004a,adelberger2005a,
900: lee2006,hildebrandt2007}.
901: We examine the clustering segregation with respect to the
902: rest-frame UV luminosity
903: in our sample using subsamples selected by $z'$ magnitude
904: ($23.0<z'\le 24.5$, $24.0<z'\le 25.0$, and $24.5<z'\le 25.5$)
905: from the full sample (\reftbl{clustering}).
906: The ACF for each subsample is shown in \reffig{acf_mag_z}.
907: In \reffig{r0_mag_z}, the derived correlation length is plotted
908: as a function of $z'$ magnitude.
909: The magnitude of each point in \reffig{r0_mag_z} is the median
910: magnitude of the corresponding subsample.
911: In agreement with previous studies,
912: we find that the correlation length increases with rest-frame
913: UV luminosity,
914: suggesting that galaxies with higher star formation rates
915: are hosted by more massive dark halos.
916: 
917: \subsubsection{Dependence of Clustering Strength on Optical Luminosity}
918: \label{subsubsec:acf_mag_k}
919: 
920: \begin{figure}[t]
921:   \begin{center}
922:     \includegraphics[scale=0.4]{f10.eps}
923:   \end{center}
924:   \caption{%
925: Angular correlation function, $\omega(\theta)$,
926: for three subsamples selected by $K$ magnitude.
927: The red, green, and blue symbols indicate the $\omega(\theta)$ of
928: LBGs with $K \le 23.46$, $22.96 <K \le 23.96$  and $K > 23.46$,
929: respectively.
930: \label{fig:acf_mag_k}}
931: \end{figure}
932: 
933: \begin{figure}[h]
934:   \begin{center}
935:     \includegraphics[scale=0.4]{f11.eps}
936:   \end{center}
937:   \caption{%
938: Spatial correlation length, $r_0$, as a function of $K$ magnitude.
939: \label{fig:r0_mag_k}}
940: \end{figure}
941: 
942: \begin{figure}[h]
943:   \begin{center}
944:     \includegraphics[scale=0.4]{f12.eps}
945:   \end{center}
946:   \caption{%
947: \zk\ plotted against $K$ for the LBG sample.
948: When the $K$ magnitude of an LBG is fainter
949: than the $1\sigma$ magnitude,
950: the $1\sigma$ magnitude is assigned (open circles).
951: Just for a guide, the faint limit to the sample of
952: $z_\mathrm{total}\le 25.5$ is indicated with the dotted line.
953: Note that the \zk\ colors are measured with
954: 2"-aperture magnitudes and some objects lie beyound the line.
955: \label{fig:k_zk}}
956: \end{figure}
957: 
958: We divide the sample by $K$ magnitude ($K\le 23.46$,
959: $22.96<K\le 23.96$, and $23.46<K$)
960: to see clustering dependence on the rest-frame optical luminosity
961: ($\sim 5000$\AA),
962: which can be an indicator of stellar mass (\reftbl{clustering}).
963: The ACF for each subsample is shown in \reffig{acf_mag_k}.
964: \reffig{r0_mag_k} shows the relationship between
965: the derived correlation length and $K$ magnitude.
966: The magnitude of each point in \reffig{r0_mag_k} is the median
967: magnitude of the corresponding subsample.
968: It is found that there is a strong trend that LBGs
969: brighter in rest-frame optical are also more strongly clustered.
970: 
971: Our LBG sample shows a correlation between
972: $K$ magnitude and $z'-K$ color;
973: there is a significant deficit of bright LBGs with blue $z'-K$ colors
974: (\reffig{k_zk}).
975: Consequently, we find that correlation length increases
976: with $z'-K$ color from
977: $r_0 = 5.4^{+1.4}_{-1.7}\ h^{-1}$Mpc for $z'-K\le 0.8$ to
978: $r_0 = 8.8^{+1.6}_{-1.9}\ h^{-1}$Mpc for $z'-K>0.8$
979: (\reftbl{clustering}).
980: Both results on the dependence of the clustering strength imply that
981: galaxies with large stellar masses reside in massive dark halos.
982: \citet{shapley2005} suggested a relationship between stellar mass
983: and $R-K$ color as well as $K$ magnitude
984: for their sample of BX objects at $z\sim 2$,
985: and \citet{adelberger2005b} ascribed a similar segregation
986: of the clustering strength with $K$ magnitude and $R-K$ color
987: for the sample to an underlying correlation with stellar mass.
988: 
989: \subsubsection{Dependence of Clustering Strength \
990: on a Combination of UV and Optical Luminosity}
991: \label{subsubsec:acf_mag_zk}
992: 
993: \begin{figure}[t]
994:   \begin{center}
995:     \includegraphics[scale=0.4]{f13.eps}
996:   \end{center}
997:   \caption{%
998: Distribution of the LBG sample on the $z'$ vs $K$ plane.
999: When the $K$ magnitude of an LBG is fainter
1000: than the $1\sigma$ magnitude,
1001: the $1\sigma$ magnitude is assigned (open circles).
1002: The solid lines indicate the magnitude limit to the sample.
1003: The dashed lines indicate the boundary of four subsamples.
1004: \label{fig:zk}}
1005: \end{figure}
1006: 
1007: \begin{figure}[t]
1008:   \begin{center}
1009:     \includegraphics[scale=0.4]{f14.eps}
1010:   \end{center}
1011:   \caption{%
1012: Angular correlation function, $\omega(\theta)$,
1013: for four subsamples selected by $z'$ and $K$ magnitudes.
1014: Colors of the symboles for subsamples are designated as follows:
1015: (red) $23.0 < z' \le 24.5$ and $K \le 23.46$,
1016: (green) $24.5 < z' \le 25.5$ and $K \le 23.46$,
1017: (cyan) $23.0 < z' \le 24.5$ and $K > 23.46$,
1018: and (blue) $24.5 < z' \le 25.5$ and $K > 23.46$.
1019: \label{fig:acf_mag_zk}}
1020: \end{figure}
1021: 
1022: \begin{figure}[h]
1023:   \begin{center}
1024:     \includegraphics[scale=0.4]{f15.eps}
1025:   \end{center}
1026:   \caption{%
1027: Spatial correlation length, $r_0$, in dependence on $z'$ and $K$
1028: magnitudes.
1029: Colors of the symbols are the same as in \reffig{acf_mag_zk}.
1030: \label{fig:r0_mag_zk}}
1031: \end{figure}
1032: 
1033: To investigate in what manner the star formation activity of LBGs
1034: relates to the mass of dark halos in detail,
1035: we explore the dependence of the clustering strength
1036: on a combination of $z'$ and $K$ magnitudes,
1037: i.e., the behavior of the clustering strength on the $z'$ vs $K$  plane.
1038: The LBG sample is divided into four subsamples on that plane
1039: ($23.0 < z' \le 24.5$ and $K \le 23.46$,
1040: $24.5 < z' \le 24.5$ and $K \le 23.46$,
1041: $23.0 < z' \le 24.5$ and $K > 23.46$,
1042: and $24.5 < z' \le 24.5$ and $K > 23.46$;
1043: \reffig{zk}, \reftbl{clustering}).
1044: The magnitude of $K=23.46$ corresponds to $S/N=5$ in the $K$ band
1045: image.
1046: The ACFs for all the subsamples are presented in \reffig{acf_mag_zk}.
1047: In this figure, we see that only one subsample which is faint
1048: in both $z'$ and $K$ has a smaller correlation amplitude,
1049: whereas the other subsamples which are bright either in $z'$ or $K$
1050: have very similar large correlation amplitudes.
1051: The correlation lengths derived from the amplitudes are plotted
1052: as functions of $z'$ and $K$ magnitudes in \reffig{r0_mag_zk}.
1053: The magnitude of each point in \reffig{r0_mag_zk} is the median
1054: magnitude of the corresponding subsample.
1055: The correlation length of galaxies bright in either magnitude
1056: is large irrespective of the other magnitude.
1057: On the other hand, galaxies faint in one of the two magnitudes
1058: have correlation lengths decreasing with the other magnitude.
1059: The same trends are seen when we further divide the faintest subsample
1060: into two by each of the magnitudes,
1061: although the errors are large due to poor statistics.
1062: The mass of dark halos for these subsamples ranges from
1063: $2\times 10^{13} M_\odot$ for the subsamples with large
1064: correlation amplitudes
1065: to $1\times 10^{12} M_\odot$ for the subsample with small one.
1066: 
1067: The overall clustering behavior against magnitudes and a color
1068: on the $z'$ versus $K$ plane
1069: suggests that the clustering strength is primarily related to
1070: the combination of $K$ magnitude and $z'$ magnitude
1071: instead of the combination of $K$ magnitude and $z'-K$ color
1072: or $z'-K$ color alone.
1073: For reference to results obtained from
1074: near-infrared selected galaxies,
1075: \citet{quadri2007} examined clustering properties of $K$-selected
1076: galaxies ($K<22.8$)
1077: at $2<z<3.5$
1078: and claimed that the clustering strength
1079: seemed to be independent of $K$ magnitude for their sample.
1080: This does not, however, necessarily conflict with our result,
1081: since the limit of $K$ magnitude of their sample is brighter
1082: than the limit of our brightest subsample.
1083: This is true even for the absolute magnitudes
1084: where the slight difference of redshift range is taken into account.
1085: When we explore the dependence among our brightest subsample
1086: by splitting it into two,
1087: the difference in clustering strength between the two subsamples
1088: is found not to be significant
1089: though there are large errors due to poor statistics.
1090: \citet{ichikawa2007} measured the clustering strength of $K$-selected
1091: galaxies down to $K=25.0$ for a deeper but 28 arcmin$^2$ field
1092: and found that the clustering strength
1093: does increase with $K$-band luminosity at $K>23.0$.
1094: It is indicated that the stellar mass of galaxies may not be a strong
1095: function of the mass of dark halos in the most massive dark halos.
1096: As to the dependence on rest-frame UV luminosity, \citet{quadri2007}
1097: found that the optically brighter subsample clusters less
1098: strongly than the fainter subsample for the $K$-selected galaxies.
1099: However, the subsample that had a larger correlation length
1100: is fainter than $R=25.5$, which is close to the limit of our sample
1101: and we cannot confirm their results from our sample.
1102: They have not explored the dependence for $R<25.5$.
1103: We note here that the samples of \citet{quadri2007}
1104: and \citet{ichikawa2007} are rest-frame optical selected,
1105: while ours is rest-frame UV selected,
1106: and the differences in results may be partly caused by the
1107: difference in the way of selection.
1108: 
1109: One implication of the result in terms of the stellar mass assembly
1110: of galaxies is that massive halos can host any galaxies
1111: from ones with small stellar masses
1112: to ones which have accumulated large stellar masses,
1113: while in less massive halos,
1114: only galaxies with small stellar masses reside.
1115: There appears to be an upper limit to the stellar mass accumulated
1116: which is determined by the mass of hosting dark halos.
1117: Moreover, galaxies in massive halos can have various star formation
1118: rates whatever stellar masses they have at $z\sim 3$.
1119: On the other hand, galaxies in less massive halos accumulated only
1120: small stellar masses and have lower star formation rates at $z\sim 3$.
1121: \citet{ichikawa2007} suggest a similar tendency for galaxies
1122: with small stellar masses in their $K$-selected galaxy sample
1123: that galaxies red in rest-frame UV color
1124: (galaxies with passive star formation)
1125: in low-mass samples tend to belong to less massive halos,
1126: although their result was derived based on a small field
1127: and contains large errors.
1128: It is suggested that the mass of dark halos governs the
1129: current star formation activity as well as the past star formation
1130: history at $z\sim 3$.
1131: 
1132: %---------------------------------------------------------------------
1133: %---------------------------------------------------------------------
1134: \section{SUMMARY AND CONCLUSIONS} \label{sec:sum}
1135: 
1136: We have investigated clustering properties of Lyman-break galaxies
1137: (LBGs) at $z\sim 3$ based on deep multi-waveband imaging data
1138: from optical to near-infrared wavelengths
1139: in the Subaru/\textit{XMM-Newton} Deep Field.
1140: The LBGs are selected by \uv and \vz\ colors in one contiguous area
1141: of 561 arcmin$^2$ down to $z'=25.5$.
1142: The number of LBG candidates detected is 962 in total, among which
1143: $708$ are found in the region observed in the $J$ and $K$ bands.
1144: We use Monte Carlo simulations to estimate the redshift distribution
1145: function and the fraction of contamination by interlopers of the
1146: LBG samples.
1147: The fraction is found to be at most 6\% at any magnitude.
1148: 
1149: We explored the dependence of the clustering strength on rest-frame
1150: UV and optical magnitudes, which can be indicators of star formation
1151: rate and stellar mass, respectively.
1152: The correlation length is found to be a strong function of
1153: both UV and optical magnitudes with brighter galaxies being
1154: more clustered than faint ones in both cases.
1155: It is found that the correlation length also increases with
1156: $z'-K$ color, which is correlated with $K$ magnitude in our LBG sample.
1157: These results imply that galaxies with larger star formation rates
1158: and larger stellar masses are hosted by more massive dark halos.
1159: 
1160: Furthermore, the correlation length is interestingly dependent on
1161: a combination of UV and optical magnitudes in the sense that
1162: galaxies bright in optical magnitude have large correlation lengths
1163: irrespective of UV magnitude,
1164: while galaxies faint in optical magnitude have correlation lengths
1165: decreasing with UV magnitude.
1166: One implication of this result in terms of the stellar mass assembly
1167: of galaxies is that galaxies which have accumulated
1168: large stellar masses
1169: always belong to massive halos in which they can have
1170: various star formation
1171: rates, while galaxies with small stellar masses reside in less massive
1172: halos only if they have low star formation rates.
1173: To put it another way, massive halos can host any galaxies
1174: from ones with small stellar masses
1175: to ones with large stellar masses,
1176: while in less massive halos,
1177: only galaxies with small stellar masses reside.
1178: There appears to be an upper limit to the stellar mass accumulated
1179: which is determined by the mass of hosting dark halos.
1180: Moreover, galaxies in massive halos can have various star formation
1181: rates whatever stellar masses they have at $z\sim 3$.
1182: On the other hand, galaxies in less massive halos accumulated only
1183: small stellar masses and have lower star formation rates at $z\sim 3$.
1184: It is suggested that the mass of dark halos governs the
1185: current star formation acitivity as well as the past star formation
1186: history at $z\sim 3$.
1187: 
1188: \begin{deluxetable}{cccc}[p]
1189:   \tablecolumns{4}
1190:   \tablewidth{0pt}
1191:   \tablecaption{Clustering Measurements \label{tbl:clustering}}
1192:   \tablehead{
1193:     \colhead{\hspace{0.3cm} Sample\hspace{0.3cm} }
1194:     & \colhead{\hspace{0.5cm} $N$\tablenotemark{\ast}\hspace{0.5cm} }
1195:     & \colhead{\hspace{0.3cm} $A_\omega$\hspace{0.3cm} }
1196:     & \colhead{\hspace{0.3cm} $r_0$ ($h^{-1}$Mpc)\hspace{0.3cm} }
1197:   }
1198:   \startdata
1199: $23.0<z'\le 25.5$  & 962 & $1.17^{+0.29}_{-0.29}$ & \phn$5.5^{+0.8}_{-0.9}$ \\
1200: \\
1201: $23.0<z'\le 24.5$ & 288 & $3.44^{+1.00}_{-1.00}$ & $10.7^{+1.9}_{-2.1}$ \\
1202: $24.0<z'\le 25.0$ & 467 & $1.46^{+0.56}_{-0.56}$ & \phn$6.3^{+1.4}_{-1.6}$ \\
1203: $24.5<z'\le 25.5$ & 674 & $0.76^{+0.36}_{-0.35}$ & \phn$4.2^{+1.1}_{-1.3}$ \\
1204: \\
1205: \phantom{$22.96<$}$K\le 23.46$ & 171 & $3.85^{+1.28}_{-1.28}$ & $11.5^{+2.3}_{-2.6}$ \\
1206: $22.96<K\le 23.96$             & 225 & $1.80^{+0.92}_{-0.92}$ & \phn$7.1^{+2.1}_{-2.6}$ \\
1207: $23.46<K$\phantom{$\le 23.96$} & 537 & $0.96^{+0.39}_{-0.39}$ & \phn$4.8^{+1.1}_{-1.3}$ \\
1208: \\
1209: \zk$\le 0.8$ & 365 & $1.14^{+0.52}_{-0.52}$ & \phn$5.4^{+1.4}_{-1.7}$ \\
1210: \zk$>0.8$    & 343 & $2.50^{+0.78}_{-0.79}$ & \phn$8.8^{+1.6}_{-1.9}$ \\
1211: \\
1212: $23.0<z'\le 24.5$,\ \ \ $K\le 23.46$ & 105     & $4.63^{+2.14}_{-2.13}$ & $12.9^{+3.5}_{-4.1}$ \\
1213: $24.5<z'\le 25.5$,\ \ \ $K\le 23.46$ & \phn 66 & $3.98^{+3.29}_{-3.29}$ & $11.7^{+5.4}_{-7.8}$ \\
1214: $23.0<z'\le 24.5$,\ \ \ $23.46<K$    & \phn 96 & $4.41^{+2.49}_{-2.49}$ & $12.5^{+4.0}_{-5.1}$ \\
1215: $24.5<z'\le 25.5$,\ \ \ $23.46<K$    & 441     & $0.91^{+0.51}_{-0.52}$ & \phn$4.7^{+1.5}_{-1.9}$ \\
1216:   \enddata 
1217:   \tablenotetext{\ast\ }{Number of LBGs contained in the sample.}
1218: \end{deluxetable}
1219: 
1220: %---------------------------------------------------------------------
1221: %---------------------------------------------------------------------
1222: \acknowledgments
1223: We would like to thank the referee for valuable comments and
1224: suggestions.
1225: We are deeply grateful to the Subaru Telescope staff for their
1226: invaluable support in the observations.
1227: M.Y. acknowledges support from the Japan Society for the Promotion of
1228: Science (JSPS) through JSPS Research Fellowship for Young Scientists.
1229: 
1230: %---------------------------------------------------------------------
1231: %---------------------------------------------------------------------
1232: \clearpage
1233: \begin{thebibliography}{}
1234:   \bibitem[Adelberger et al.(2005a)]{adelberger2005a}
1235:     Adelberger, K. L., Steidel, C. C., Pettini, M., Shapley, A. E.,
1236:     Reddy, N. A., \& Erb, D. K.
1237:     2005a, \apj, 619, 697
1238: 
1239:   \bibitem[Adelberger et al.(2005b)]{adelberger2005b}
1240:     Adelberger, K. L., Erb, D. K., Steidel, C. C., Reddy, N. A.,
1241:     Pettini, M., \& Shapley, A. E.
1242:     2005b, \apj, 620, L75
1243: 
1244:   \bibitem[Akiyama et al., in preparation]{akiyama}
1245:     Akiyama, M., et al., in preparation
1246: 
1247:   \bibitem[Bardeen et al.(1986)]{bardeen1986}
1248:     Bardeen, J. M., Bond, J. R., Kiser, N., \& Szaley, A. S.
1249:     1986, \apj, 304, 15
1250: 
1251:   % SExtractor
1252:   \bibitem[Bertin \& Arnouts(1996)]{bertin1996}
1253:     Bertin, E. \& Arnouts, S.
1254:     1996, \aaps, 117, 393
1255: 
1256:   \bibitem[Bouwens et al.(2007)]{bouwens2007}
1257:     Bouwens, R. J., Illingworth, G. D., Franx, M., \& Ford, H.
1258:     2007, \apj, in press (arXiv:0707:2080)
1259: 
1260:   % dust extinction
1261:   \bibitem[Calzetti et al.(2000)]{calzetti2000}
1262:     Calzetti, D.,
1263:     Armus, L., Bohlin, R. C., Kinney, A. L.,
1264:     Koornneef, J., \& Storchi-Bergmann, T.
1265:     2000, \apj, 533, 682
1266:  
1267:   \bibitem[Carroll et al.(1992)]{carroll1992}
1268:     Carroll, S. M., Press, W. H., \& Turner, E. L.
1269:     1992, \araa, 30, 499
1270: 
1271:   \bibitem[Daddi et al.(2003)]{daddi2003}
1272:     Daddi, E., R\"ottgering, H. J. A., Labb\'e, I.,
1273:     Rudnick, G., Franx, M., Moorwood, A. F. M.,
1274:     Rix, H. W., van der Werf, P. P., \& van Dokkum, P. G.
1275:     2003, \apj, 588, 50
1276:  
1277:   \bibitem[Dickinson et al.(2004)]{dickinson2004}
1278:     Dickinson, M., et al.
1279:     2004, \apj, 600, L99
1280:  
1281:   \bibitem[Foucaud et al.(2003)]{foucaud2003}
1282:     Foucaud, S., McCracken, H. J., Le F\`evre, O.,
1283:     Arnouts, S., Brodwin, M., Lilly, S. J., Crampton, D.,
1284:     \& Mellier, Y.
1285:     2003, \aap, 409, 835
1286:    
1287:   \bibitem[Furusawa et al.(2000)]{furusawa2000}
1288:     Furusawa, H.,
1289:     Shimasaku, K., Doi, M., \& Okamura, S.
1290:     2000, \apj, 534, 624
1291: 
1292:   % SXDS
1293:   \bibitem[Furusawa et al.(2007)]{furusawa2007}
1294:     Furusawa, H., et al.
1295:     2007, \apjs, submitted
1296: 
1297:   \bibitem[Giavalisco et al.(1998)]{giavalisco1998}
1298:     Giavalisco, M., Steidel, C. C., Adelberger, K. L.,
1299:     Dickinson, M. E., Pettini, M., \& Kellogg, M.
1300:     1998, \apj, 503, 543
1301: 
1302:   \bibitem[Giavalisco \& Dickinson(2001)]{giavalisco2001}
1303:     Giavalisco, M. \& Dickinson, M.
1304:     2001, \apj, 550, 177
1305:   
1306:   \bibitem[Giavalisco (2002)]{giavalisco2002}
1307:     Giavalisco, M.
1308:     2002, \araa, 40, 579
1309:   
1310:   \bibitem[Giavalisco et al.(2004)]{giavalisco2004}
1311:     Giavalisco, M. et al.
1312:     2004, \apj, 600, L103
1313:   
1314:   \bibitem[Groth \& Peebles(1977)]{groth1977}
1315:     Groth, E. J., \& Peebles, P. J. E.
1316:     1977, \apj, 217, 385
1317:   
1318:   \bibitem[Guhathakurta et al.(1990)]{guhathakurta1990}
1319:     Guhathakurta, P., Tyson, J. A., \& Majewski, S. R.
1320:     1990, \apj, 357, L9
1321: 
1322:   % Gunn \& Stryker
1323:   \bibitem[Gunn \& Stryker(1983)]{gunn1983}
1324:     Gunn, J. E. \& Stryker, L. L.
1325:     1983, \apjs, 52, 121
1326: 
1327:   \bibitem[Hildebrandt et al.(2007)]{hildebrandt2007}
1328:     Hildebrandt, H., Pielorz, J., Erben, T., Schneider, P.,
1329:     Eifler, T., Simon, P., \& Dietrich, J. P.
1330:     2007, \aap, 462, 865
1331: 
1332:   \bibitem[Ichikawa et al.(2007)]{ichikawa2007}
1333:     Ichikawa, T., Suzuki, R., Tokoku, C., Katsuno-Uchimoto, Y.,
1334:     Konishi, M., Yoshikawa, T., Kajisawa, M.,
1335:     Ouchi, M., Hamana, T., Akiyama, M., Nishimura, T., Omata, K.,
1336:     Tanaka, I., \& Yamada, T.
1337:     2007, \pasj, submitted (astro-ph/0701820)
1338: 
1339:   \bibitem[Iwata et al.(2007)]{iwata2007}
1340:     Iwata, I., Ohta, K., Tamura, N., Akiyama, M., Aoki, K., Ando, M.,
1341:     Kiuchi, G., \& Sawicki, M.
1342:     2007, \mnras, 376, 155
1343: 
1344:   \bibitem[Kashikawa et al.(2006)]{kashikawa2006}
1345:     Kashikawa, N., et al.
1346:     2006, \apj, 637, 631
1347: 
1348:   % model
1349:   \bibitem[Kodama \& Arimoto(1997)]{kodama1997}
1350:     Kodama, T. \& Arimoto, N.
1351:     1997, \aap, 320, 41
1352: 
1353:   % ACF
1354:   \bibitem[Landy \& Szalay(1993)]{landy1993}
1355:     Landy, S. D. \& Szalay, A. S.
1356:     1993, \apj, 412, 64
1357: 
1358:   % UDS
1359:   \bibitem[Lawrence et al.(2007)]{lawrence2007}
1360:     Lawrence, A., et al.
1361:     2007, \mnras, 379, 1599
1362: 
1363:   \bibitem[Lee et al.(2006)]{lee2006}
1364:     Lee, K-S., Giavalisco, M., Gnedin, O. Y., Somerville, R. S.,
1365:     Ferguson, H. C., Dickinson, M., \& Ouchi, M.
1366:     2006, \apj, 642, 63
1367: 
1368:   % bootstrap error
1369:   \bibitem[Ling et al.(1986)]{ling1986}
1370:     Ling, E. N., Barrow, J. D., \& Frenk, C. S.
1371:     1986, \mnras, 223, P21
1372: 
1373:   % IGM apsorption
1374:   \bibitem[Madau(1995)]{madau1995}
1375:     Madau, P.
1376:     1995, \apj, 441, 18
1377: 
1378:   \bibitem[Mo \& White(1996)]{mo1996}
1379:     Mo, H. J. \& White, S. D. M.
1380:     1996, \mnras, 282, 347
1381: 
1382:   % magnitude
1383:   \bibitem[Oke \& Gunn(1983)]{oke1983}
1384:     Oke, J. B. \& Gunn, J. E.
1385:     1983, \apj, 266, 713
1386: 
1387:   \bibitem[Ouchi et al.(2004a)]{ouchi2004a}
1388:     Ouchi, M., et al.
1389:     2004, \apj, 611, 685
1390: 
1391:   \bibitem[Ouchi et al.(2004b)]{ouchi2004b}
1392:     Ouchi, M., et al.
1393:     2004, \apj, 611, 660
1394: 
1395:   \bibitem[Ouchi et al.(2005)]{ouchi2005}
1396:     Ouchi, M., et al.
1397:     2005, \apj, 635, L117
1398: 
1399:   \bibitem[Papovich et al.(2001)]{papovich2001}
1400:     Papovich, C., Dickinson, M., \& Ferguson, H. C.
1401:     2001, \apj, 559, 620
1402: 
1403:   \bibitem[Peebles(1980)]{peebles1980}
1404:     Peebles, P. J. E.
1405:     1980, The Large-Scale Structure of the Universe
1406:     (Princeton: Princeton Univ. Press)
1407: 
1408:   \bibitem[Peacock \& Dodds(1996)]{peacock1996}
1409:     Peacock, J. A., \& Dodds, S. J.
1410:     1996, \mnras, 280, L19
1411: 
1412:   \bibitem[Quadri et al.(2007)]{quadri2007}
1413:     Quadri, R., van Dokkum, P., Gawiser, E., Franx, M.,
1414:     Marchesini, D., Lira, P., Rudnick, G., Herrera, D., Maza, J.,
1415:     Kriek, M., Labb\'e, I., \& Francke, H.
1416:     2007, \apj, 654, 138
1417: 
1418:   \bibitem[Roche \& Earles(1999)]{roche1999}
1419:     Roche, N. \& Earles, S. A.
1420:     1999, \mnras, 307, 703
1421: 
1422:   \bibitem[Saito et al.(2007)]{saito2007}
1423:     Saito, T., Shimasaku, K., Okamura, S., Ouchi, M., Akiyama, M.,
1424:     Yoshida, M., \& Ueda, Y.
1425:     2007, \apj, submitted (astro-ph/0705.1494)
1426: 
1427:   \bibitem[Sawicki \& Thompson(2006)]{sawicki2006}
1428:     Sawicki, M. \& Thompson, D.
1429:     2006, \apj, 642, 653
1430: 
1431:   % Galactic extinction
1432:   \bibitem[Schlegel et al.(1998)]{schlegel1998}
1433:     Schlegel, D. J., Finkbeiner, D. P., \& Davis, M.
1434:     1998, \apj, 500, 525
1435: 
1436:   % SXDS
1437:   \bibitem[Sekiguchi et al., in preparation]{sekiguchi}
1438:     Sekiguchi, K., et al., in preparation
1439: 
1440:   \bibitem[Shapley et al.(2005)]{shapley2005}
1441:     Shapley, A. E., Steidel, C. C., Erb, D. K., Reddy, N. A.,
1442:     Adelberger, K. L., Pettini, M., Barmby, P., \& Huang, J.
1443:     2005, \apj, 626, 698
1444: 
1445:   \bibitem[Sheth et al.(2001)]{sheth2001}
1446:     Sheth, R. K., Mo, H. J., \& Tormen, G.
1447:     2001, \mnras, 323, 1
1448: 
1449:   \bibitem[Simpson et al., in preparation]{simpson}
1450:     Simpson et al., in preparation
1451: 
1452:   \bibitem[Steidel et al.(1996)]{steidel1996}
1453:     Steidel, C. C., Giavalisco, M., Pettini, M.
1454:     Dickinson, M., \& Adelberger, K. L.
1455:     1996, \apj, 462, L17
1456: 
1457:   \bibitem[Steidel et al.(1999)]{steidel1999}
1458:     Steidel, C. C., Adelberger, K. L., Giavalisco, M.,
1459:     Dickinson, M., \& Pettini, M.
1460:     1999, \apj, 519, 1
1461: 
1462:   \bibitem[Steidel et al.(2003)]{steidel2003}
1463:     Steidel, C. C., Adelberger, K. L., Shapley, A. E.,
1464:     Pettini, M., Dickinson, M., \& Giavalisco, M.
1465:     2003, 592, 728
1466: 
1467:   % UDS
1468:   \bibitem[Warren et al.(2007)]{warren2007}
1469:     Warren, S. J., et al.
1470:     2007, \mnras, 375, 213
1471: 
1472:   \bibitem[Yoshida et al.(2006)]{yoshida2006}
1473:      Yoshida, M., et al.
1474:      2006, \apj, 653, 988
1475: 
1476: \end{thebibliography}
1477: 
1478: \end{document}
1479: