1: %\documentclass[]{aastex}
2: %\documentclass[preprint]{aastex}
3: \documentclass[]{emulateapj}
4:
5: \shorttitle{HOD Modeling of LRG Clustering}
6: \shortauthors{Zheng, Zehavi, Eisenstein, Weinberg, \& Jing}
7:
8: \def\NavgM{\langle N(M)\rangle}
9: \def\Ncen{\langle N_{\rm cen}(M)\rangle}
10: \def\Nsat{\langle N_{\rm sat}(M)\rangle}
11: \def\Mmin{M_{\rm min}}
12: \def\sigM{\sigma_{\log M}}
13: \def\Msun{M_\odot}
14: \def\Mcut{M_{\rm cut}}
15: \def\hMsun{h^{-1}M_\odot}
16: \def\erf{{\rm erf}}
17: \def\hMpc{h^{-1}{\rm Mpc}}
18:
19: \begin{document}
20:
21: \title{Halo Occupation Distribution Modeling of Clustering of Luminous Red
22: Galaxies}
23: \author{
24: Zheng Zheng\altaffilmark{1,2,3},
25: Idit Zehavi\altaffilmark{4},
26: Daniel J. Eisenstein\altaffilmark{5},
27: David H. Weinberg\altaffilmark{6},
28: and Y.P. Jing\altaffilmark{7}
29: }
30: \altaffiltext{1}{
31: Institute for Advanced Study, Einstein Drive, Princeton, NJ 08540;
32: zhengz@ias.edu
33: }
34: \altaffiltext{2}{
35: Hubble Fellow
36: }
37: \altaffiltext{3}{
38: John Bahcall Fellow
39: }
40: \altaffiltext{4}{
41: Department of Astronomy, Case Western Reserve University,
42: 10900 Euclid Avenue, Cleveland, OH 44106; izehavi@astronomy.case.edu
43: }
44: \altaffiltext{5}{
45: Steward Observatory, University of Arizona, 933 N. Cherry Avenue,
46: Tucson, AZ 85121; deisenstein@as.arizona.edu
47: }
48: \altaffiltext{6}{
49: Department of Astronomy and Center for Cosmology and AstroParticle Physics,
50: Ohio State University, 140 W. 18th Avenue,
51: Columbus, OH 43210; dhw@astronomy.ohio-state.edu
52: }
53: \altaffiltext{7}{
54: Shanghai Astronomical Observatory, Joint Institute for Galaxy and Cosmology
55: (JOINGC) of SHAO and USTC, Nandan Road 80, Shanghai, 200030, China;
56: ypjing@shao.ac.cn
57: }
58:
59: \begin{abstract}
60: We perform Halo Occupation Distribution (HOD) modeling to interpret
61: small-scale and intermediate-scale clustering of 35,000 luminous
62: early-type galaxies and their cross-correlation with a reference imaging
63: sample of normal $L_*$ galaxies in the Sloan Digital Sky Survey.
64: The modeling results show that most of these luminous red galaxies (LRGs)
65: are central galaxies residing in massive halos of typical mass $M \sim$
66: a few times $10^{13}$ to $10^{14} \hMsun$, while a few percent of them
67: have to be satellites within halos in order to produce the strong
68: auto-correlations exhibited on smaller scales. The mean luminosity $L_c$
69: of central LRGs increases with the host halo mass, with a rough scaling
70: relation of $L_c \propto M^{0.5}$. The halo mass required to host on average
71: one satellite LRG above a luminosity threshold is found to be about 10 times
72: higher than that required to host a central LRG above the same threshold.
73: We find that in massive halos the distribution of $L_*$ galaxies roughly
74: follows that of the dark matter and their mean occupation number scales
75: with halo mass as $M^{1.5}$. The HOD modeling results also allows for
76: an intuitive understanding of the scale-dependent luminosity dependence
77: of the cross-correlation between LRGs and $L_*$ galaxies. Constraints
78: on the LRG HOD provide tests to models of formation and evolution
79: of massive galaxies, and they are also useful for cosmological parameter
80: investigations. In one of the appendices, we provide LRG HOD parameters with
81: dependence on cosmology inferred from modeling the two-point auto-correlation
82: functions of LRGs.
83: \end{abstract}
84: \keywords{cosmology: observations --- galaxies: halos ---
85: galaxies: statistics ---
86: galaxies: clusters: general --- galaxies: elliptical and
87: lenticular, cD --- galaxies: evolution}
88:
89: \section{Introduction}
90:
91: The clustering of galaxies depends on their properties, such as
92: morphology
93: (e.g.,
94: \citealt{hubble36,zwicky68,davis76,dressler80,postman84,guzzo97,willmer98,
95: zehavi02,goto03}),
96: luminosity
97: (e.g., \citealt{davis88,hamilton88,white88,park94,loveday95,guzzo97,
98: benoist96,norberg01,zehavi02,Zehavi05b,Coil06,Coil08}),
99: color
100: (e.g., \citealt{willmer98,brown00,zehavi02,Zehavi05b,Coil08}),
101: and spectral type
102: (e.g., \citealt{norberg02,budavari03,madgwick03}).
103: Galaxy clustering thus provides important clues to
104: the physics of galaxy formation. Often found to reside in galaxy groups and
105: clusters, luminous red galaxies (LRGs) constitute the bright end of the galaxy
106: luminosity function. Clustering of LRGs encodes information about their
107: environments, which are typically the central regions of groups and clusters.
108: The LRG redshift sample \citep{Eisenstein01} of the Sloan Digital Sky Survey
109: (SDSS; \citealt{York00}) provides an enormous data set with which to measure
110: the clustering of LRGs. In this paper, we perform theoretical modeling of
111: auto-clustering of LRGs and cross-clustering of LRGs with other types of
112: galaxies to understand the origin of their clustering properties, to learn
113: how they are distributed among massive dark matter halos, and to aid the
114: study of formation and evolution of massive galaxies.
115:
116: The theoretical understanding of galaxy clustering has been greatly enhanced
117: through the framework of the halo occupation distribution
118: (HOD, see, e.g., \citealt{Jing98a,Ma00,Peacock00,Seljak00,Scoccimarro01,
119: Berlind02}) and the closely related approach of the conditional luminosity
120: function (CLF, \citealt{Yang03}). The HOD formalism describes the bias
121: relation between galaxies and matter at the level of individual virialized
122: dark matter halos, whose distribution and properties can be readily predicted
123: by numerical simulations or analytic models given a cosmological model.
124: The key ingredients of this formalism are the probability distribution
125: $P(N|M)$ that a halo of mass $M$ contains $N$ galaxies of a given type
126: and spatial and velocity distributions of galaxies within halos. In the
127: CLF approach, the dependence of $P(N|M)$ on galaxy luminosity is implicitly
128: derived by inferring the conditional luminosity distribution of galaxies
129: as a function of halo mass $M$. Given the
130: HOD/CLF and the cosmology, any statistics of galaxy clustering on any
131: scales can be predicted, therefore the HOD/CLF method provides a complete
132: description of the bias relation between galaxies and dark matter.
133: HOD/CLF modeling has been applied to interpret galaxy clustering
134: measurements in several galaxy surveys (e.g., \citealt{Jing98b,Jing02,
135: Bullock02,Moustakas02,Bosch03,Magliocchetti03,Yan03,Zheng04,Yang05,Zehavi05b,
136: Lee06,Hamana06,Cooray06,Conroy06,White07,Zheng07,Blake08,Wake08}). HOD/CLF
137: analysis recasts galaxy clustering measurements into a form that is more
138: physically informative and conducive for testing galaxy formation theories
139: (see, e.g., \citealt{Berlind02,Berlind03,Bosch03,Zheng05}).
140: \citet{Zehavi05b} present clustering measurements and HOD modeling
141: for galaxies in the SDSS main galaxy
142: spectroscopic sample \citep{Strauss02}. The
143: HOD modeling results
144: of the luminosity and color dependence of galaxy clustering
145: are found to be in good qualitative
146: agreement with predictions from galaxy formation models (\citealt{Zheng05}).
147:
148: In this paper, we apply HOD modeling to interpret clustering in the SDSS
149: LRG spectroscopic sample, which uses color and magnitude cuts to effectively
150: select LRGs out to redshift $z\sim 0.45$ \citep{Eisenstein01}. The large
151: volume probed by the LRG sample has led to the detection of the baryon
152: acoustic peak in the two-point correlation function
153: (\citealt{Eisenstein05b}). \citet{Zehavi05a} report the measurements of
154: two-point correlation functions of 35,000 LRGs on scales of 0.3--40$\hMpc$.
155: They find that LRGs are highly clustered (correlation length $\sim$10$\hMpc$)
156: and that more luminous LRGs are more clustered. Clear deviations from a
157: power law are seen in the correlation functions, with a dip at $\sim$2$\hMpc$.
158: \citet{Eisenstein05a} measure the two-point cross-correlation between
159: 32,000 spectroscopic LRGs and 16 million galaxies in the SDSS imaging
160: sample. Since these reference galaxies have luminosities around
161: $L_*$, the characteristic luminosity of the \citet{Schechter76} luminosity
162: function, they are denoted as $L_*$ galaxies. \citet{Eisenstein05a} find a
163: strong luminosity dependence of the LRG--$L_*$ cross-clustering amplitude.
164: The form of the luminosity dependence
165: is itself dependent on scale,
166: with more variation in the
167: clustering amplitude on small scales. Understanding all these auto- and
168: cross-clustering features is one of the goals of our HOD modeling. Since
169: LRGs trace massive halos, the cross-correlation between LRGs and $L_*$
170: galaxies also allows us to study the HOD of $L_*$ galaxies in these massive
171: halos.
172:
173: The structure of this paper is as follows. In \S~2, we briefly describe the
174: SDSS LRG and $L_*$ galaxy samples we use and our modeling method.
175: In \S~3, we describe our HOD parameterization for LRG samples and $L_*$
176: samples. In \S~4, we perform HOD modeling for LRG two-point auto-correlation
177: functions and the two-point cross-correlation functions between LRGs and
178: $L_*$ galaxies. We show how LRGs occupy dark matter halos and how $L_*$
179: galaxies occupy massive halos. Based on the modeling results, we interpret
180: the luminosity dependence of the cross-clustering. Finally, we summarize
181: and discuss our results in \S~5. We also include three appendices in the
182: paper. In Appendix~\ref{sec:appendixA}, we investigate which HOD parameter
183: plays the major role in the departures of the galaxy two-point correlation
184: function from a pure power law. In Appendix~\ref{sec:appendixB}, we present
185: HOD modeling results for two luminosity threshold LRG samples for different
186: cosmological models. In Appendix~\ref{sec:appendixC}, we have a brief
187: discussion on the mass function of the most massive halos and the fluctuation
188: of the number of massive halos in the volume probed by the SDSS LRG samples.
189:
190: \section{Samples and Method}
191:
192: The LRG two-point auto-correlation functions in the SDSS have been measured
193: by \citet{Zehavi05a} with spectroscopic samples. The luminosity cuts of the
194: two LRG samples at $z\sim0.3$ we model in this paper are $-23.2<M_g<-21.2$
195: and $-23.2<M_g<-21.8$, where $M_g$ is the restframe $g$-band absolute
196: magnitude at $z=0.3$ computed from the observed $r$-band magnitude with
197: $k$ and passive evolution corrections \citep{Eisenstein05a}. Since galaxies
198: with $M_g<-23.2$ are extremely rare,
199: these can be regarded essentially as luminosity-threshold samples. For
200: brevity, we call them the $M_g<-21.2$ sample and the $M_g<-21.8$ sample, or
201: simply the faint sample and the bright sample. The comoving number densities
202: of the two samples are $9.73\times 10^{-5}h^3{\rm Mpc^{-3}}$ and
203: $2.40\times 10^{-5}h^3{\rm Mpc^{-3}}$, respectively. Because of the fiber
204: collision effect, the smallest scale can be probed by the spectroscopic LRG
205: samples is about 0.4$\hMpc$. \citet{Masjedi06} extend the measurements of
206: the two-point correlation function for the $M_g<-21.2$ sample down to a scale
207: of $\sim$0.015$\hMpc$ by cross-correlating the spectroscopic sample with the
208: imaging sample to avoid the fiber collision effect. They find that the
209: real-space two-point correlation function of the $M_g<-21.2$ sample roughly
210: follow an $r^{-2}$ profile from $\sim$100$\hMpc$ down to $\sim$0.01$\hMpc$.
211: In this paper, we focus on the measurements from the spectroscopic samples
212: in \citet{Zehavi05a} and limit ourselves to modeling the clustering on scales
213: above 0.3$\hMpc$.
214:
215: The two-point cross-correlation functions between spectroscopic LRG samples
216: and the imaging $L_*$ galaxy sample in the SDSS at $z\sim 0.3$ are measured by
217: \citet{Eisenstein05a}. The $L_*$ galaxy sample we model is the one defined
218: by the luminosity range $M^*-0.5$ to $M^*+1.0$, which is called the $M^*+1.0$
219: sample in \citet{Eisenstein05a}. For the LRG sample in the cross-clustering,
220: we adopt the luminosity-bin sample defined by $-21.7<M_g<-21.2$.
221:
222: We essentially follow the method presented in \citet{Zehavi05b} and adopt
223: the improvements of \citet{Tinker05} for theoretical modeling of the
224: two-point auto- and cross-correlation functions in the HOD framework. In
225: this paper, more general HOD parameterizations are used, as described
226: in the following sections. While the two-point auto-correlation functions
227: in \citet{Zehavi05a} are projected along the redshift direction, similar
228: to those in \citet{Zehavi05b}, the LRG--$L_*$ two-point cross-correlation
229: functions $\Delta$ measured in \citet{Eisenstein05a} are volume-averaged
230: real-space cross-correlation functions $\xi_\times$. The average is weighted
231: by a spherical window function,
232: \begin{equation}
233: \label{eqn:Delta}
234: \Delta(a)=\frac{1}{V} \int_0^{\infty} dr 4\pi r^2 \xi_\times(r)W(r;a),
235: \end{equation}
236: where the window function is of the form
237: \begin{equation}
238: \label{eqn:window}
239: W(r;a) = \frac{r^2}{a^2} \exp\left(-\frac{r^2}{2a^2}\right).
240: \end{equation}
241: The effective volume $V$ for the window function is $3(2\pi a^2)^{3/2}$.
242: In our modeling, we perform the same volume average to compute the predicted
243: cross-correlation function.
244:
245: When performing fits to the two-point auto- and cross-correlation functions,
246: we calculate values of $\chi^2$ using the full error covariance matrices,
247: inferred through the jackknife method (see \citealt{Eisenstein05a} and
248: \citealt{Zehavi05b} for details).
249:
250: Throughout the paper, we adopt the spatially flat ``concordance'' cosmological
251: model with the matter density parameter $\Omega_m=0.3$ and baryon density
252: parameter $\Omega_b=0.047$. We assume adiabatic Gaussian primordial
253: density fluctuations with a power-law index of the spectrum $n_s=1$.
254: The r.m.s. matter density fluctuation in spheres of radius $8\hMpc$ linearly
255: extrapolated to $z=0$ is assumed to be $\sigma_8=0.8$. The Hubble constant
256: we use is $h=0.7$ in unit of 100 km~s$^{-1}$~Mpc$^{-1}$. Appendix B presents
257: the dependence of the derived HOD parameters on $\sigma_8$ and $\Omega_m$ for
258: a five-parameter HOD model. The mean redshift of the above LRG and $L_*$
259: samples is around 0.3 and our model calculations take this into account,
260: i.e., all halo properties are calculated at $z=0.3$ for the adopted cosmology.
261: In general, the comoving unit is adopted for distance, but to be compatible
262: with \citet{Eisenstein05a}, we also use proper units with explicit mention
263: when discussing the cross-clustering.
264: Finally, we assume $h=1$ when quoting magnitudes throughout the paper.
265:
266: \section{HOD Parameterization}
267: \label{sec:hod}
268:
269: \subsection{The Mean Occupation Function of LRGs}
270: \label{sec:meanlrg}
271:
272: The two LRG samples for which the two-point auto-correlation functions are
273: modeled are nearly luminosity-threshold samples. The LRG samples defined
274: in the cross-correlation measurements in \citet{Eisenstein05a} are
275: luminosity-bin samples. We concentrate on the cross-correlation between
276: $L_*$ galaxies and $-21.7<M_g<-21.2$ LRGs, together with the two-point
277: auto-correlation functions of the two luminosity-threshold samples
278: with $M_g<-21.2$ and $M_g<-21.8$. The three LRG samples with
279: different luminosity cuts provide leverage to constrain the relation
280: between galaxy luminosity and halo mass. We therefore first parameterize
281: the LRG HOD in a way that includes the luminosity information.
282:
283: For HOD parameterization, it has been found to be useful to separate
284: contributions from central and satellite galaxies (\citealt{Kravtsov04,
285: Zheng05}). Central galaxy luminosity is correlated with the host halo mass.
286: Based on predictions of galaxy formation models (\citealt{Zheng05}),
287: we assume that at a fixed halo mass the central galaxy luminosity follows
288: a log-normal distribution and the mean luminosity of this distribution has
289: a power-law form $L_c=L_s(M/M_s)^p$, with $p$ independent of halo mass in
290: the range that LRGs probe. In terms of absolute magnitudes,
291: \begin{equation}
292: \label{eqn:cenmag}
293: M_{gc}=M_{gs}-2.5p\log(M/M_s),
294: \end{equation}
295: where $M_{gs}$ is the mean luminosity in halos with a pivot mass $M_s$ and
296: is simply set to be $-19.8$. We note that in this paper $M$ with a subscript
297: ``{\it g}'' stands for the $g$-band absolute magnitude, while that without the
298: subscript ``{\it g}'' is used for halo mass. With a constant standard deviation
299: $\sigma_{M_g}$ in magnitude, the luminosity distribution of central galaxies
300: in halos of mass $M$ is then
301: \begin{equation}
302: \label{eqn:cenclf}
303: \frac{d\Ncen}{dM_g} = \frac{1}{\sqrt{2\pi}\sigma_{M_g}}
304: \exp\left[-\frac{(M_g-M_{gc})^2}{2\sigma_{M_g}^2}\right].
305: \end{equation}
306: LRG samples are not defined purely through luminosity, since there is a color
307: cut in the LRG selection. Therefore, for an LRG sample with
308: $M_g<M_{g,{\rm thres}}$, the mean occupation function of central galaxies is
309: calculated as
310: \begin{equation}
311: \label{eqn:Ncen}
312: \Ncen = \int_{-\infty}^{M_{g,{\rm thres}}} dM_g \frac{d\Ncen}{dM_g} f(M_g),
313: \end{equation}
314: where $f(M_g)$ is the LRG selection function at a given luminosity. Based
315: on the LRG luminosity function and the total galaxy luminosity function,
316: we approximate the selection function as $f(M_g)=-2(M_g+20.45)/3$ in the
317: range $-21.95<M_g<-20.45$, 0 for $M_g>-20.45$, and 1 for $M_g<-21.95$.
318: The HOD of a luminosity-bin galaxy sample is just the difference of those of
319: two luminosity-threshold samples.
320:
321: The theoretically predicted mean occupation function of satellite galaxies
322: for a luminosity-threshold
323: sample is close to a power law (\citealt{Kravtsov04,Zheng05}). Here, we adopt
324: a more flexible parameterization, which can give us a better idea on
325: the constraining power of the two-point clustering on the HOD. For each
326: luminosity-threshold LRG sample,
327: we parameterize the HOD of satellite LRGs through the mean occupation numbers
328: at five mass scales, and the the mean occupation function is assumed to be a
329: cubic spline curve passing through the five points (see \citealt{Zheng07a}).
330: Linear extrapolations are used outside of the ranges of the five points. In
331: addition,
332: we impose the central galaxy cutoff profile $\Ncen$ to $\Nsat$. The five
333: mass scales are chosen to be $\log (M/\hMsun) = 14.10+0.35i$
334: ($i=0,1,2,3,4$).
335: The mean occupation function of satellites in
336: a luminosity-bin sample is obtained by interpolating the values of
337: $\log \Nsat$ for the two luminosity-threshold samples (linearly in $\log L$).
338: The satellite occupation number around the mean is assumed to follow
339: the Poisson distribution (\citealt{Kravtsov04,Zheng05}).
340:
341: The parameterization of the halo occupation of LRGs we present here is
342: a hybrid of the CLF framework (\citealt{Yang03}) and the HOD framework
343: (\citealt{Berlind02}) --- the distribution of central LRGs is put in a CLF
344: form, while that of the satellite LRGs is in the usual HOD form. This
345: combination allows us to infer information on how luminosity of central
346: galaxies changes with halo mass and at the same time to keep a reasonable
347: number of free parameters.
348:
349: With the above parameterized HOD, we can easily form the mean occupation
350: functions for the two luminosity-threshold LRG samples and the luminosity-bin
351: sample given their luminosity cuts. The thirteen free parameters are the pivot
352: mass scale $M_s$ and the slope $p$ in the luminosity-mass relation, the
353: scatter $\sigma_{M_g}$ of the luminosity distribution of central LRGs,
354: and the total of ten spline points of $\Nsat$ for the two luminosity-threshold
355: samples. In Appendix~\ref{sec:appendixB}, we provide HOD modeling results
356: for the two luminosity threshold samples and their cosmology dependence
357: based on a simpler five-parameter HOD model.
358:
359: \subsection{The Mean Occupation Function of $L_*$ Galaxies}
360:
361: The $L_*$ sample in \citet{Eisenstein05a} is at a mean redshift $z\sim0.3$.
362: To parameterize the HOD of this $L_*$ sample, we make use of the modeling
363: result in \citet{Zehavi05b} at $z\sim 0$. The $L_*$ galaxy sample is a
364: luminosity-bin sample $M^*-0.5<M_r<M^*+1.0$, which can be regarded as the
365: difference between two luminosity-threshold samples, $M_r<M^*+1.0$ and
366: $M_r<M^*-0.5$. As shown in \citet{Kravtsov04}, at low redshifts, for threshold
367: samples of a fixed (comoving) galaxy number density, the shape of the mean
368: occupation function as a function of redshift approximately remains unchanged.
369: We take the advantage of this presumed property and construct the mean occupation
370: function for $L_*$ galaxies at $z\sim0.3$ based on the HODs of $z\sim 0$
371: galaxies. We first find out the number densities of the two
372: luminosity-threshold samples with $M_r<M^*+1.0$ and $M_r<M^*-0.5$ at
373: $z\sim 0.3$. To do this, we adopt the evolution model of the
374: luminosity function in \citet{Blanton03} set to $z=0.3$ and obtain the two
375: number densities as $n_1=1.1\times 10^{-2}h^3{\rm Mpc^{-3}}$ and
376: $n_2=1.3\times 10^{-3} h^3{\rm Mpc^{-3}}$, respectively.
377: We find the luminosity thresholds $L_1$ and $L_2$ for $z\sim 0$ galaxies
378: that match the two number densities,
379: \begin{equation}
380: n(L>L_1; z\sim 0)=n_1
381: \end{equation}
382: and
383: \begin{equation}
384: n(L>L_2; z\sim 0)=n_2.
385: \end{equation}
386: We then use the results presented in
387: \citet{Zehavi05b} to infer the HODs at $z\sim 0$ for luminosity-threshold
388: samples with $L_1$ and $L_2$ as the thresholds. Finally, we take the
389: difference of these two HODs to infer the mean occupation function for
390: galaxies in the $L_1<L<L_2$ luminosity bin,
391: \begin{equation}
392: \langle N(L_1<L<L_2; M)\rangle =
393: \langle N(L>L_1; M)\rangle - \langle N(L>L_2; M)\rangle.
394: \end{equation}
395: We take the {\it shape} of this mean occupation function as that for
396: the $L_*$ galaxies at
397: $z\sim0.3$. The form of this mean occupation function is a sum of a square
398: window for central galaxies and a
399: power law with an index $1.10$ for satellites. With the shape fixed,
400: we allow the mean occupation function to shift in mass scales. Furthermore,
401: we add freedom to the high mass slope, as described below, because we
402: have better data to constrain it.
403:
404: As shown later, the small-scale two-point cross-correlation between $L_*$
405: galaxies and LRGs comes from satellite $L_*$ galaxies paired with LRGs.
406: As our goal here is to investigate the $L_*$ occupation in high mass halos,
407: it is necessary to introduce additional degrees of freedom in the mean
408: occupation function of the satellite $L_*$ galaxies in massive halos.
409: We only adopt the $z\sim0.3$ mean occupation function for $L_*$ satellites,
410: as constructed above, up to $10^{13}\hMsun$, and introduce another power
411: law to represent the mean occupation function of (satellite) $L_*$ galaxies
412: in halos more massive than $10^{13}\hMsun$. We assume this occupation
413: distribution follows the Poisson distribution with the mean in the form of
414: $N_t(M/M_t)^{\alpha_h}$. Here $M_t$ is a pivot mass scale fixed
415: to be $2.5\times10^{14}\hMsun$, $N_t$ is the mean occupation number of
416: $L_*$ galaxies in halos of this pivot mass, and $\alpha_h$ denotes the
417: slope at the high mass end. The scale of the pivot mass is chosen to
418: minimize the correlation between $N_t$ and $\alpha_h$. In order to have
419: more flexibility, we do not impose continuity at $10^{13}\hMsun$.
420:
421: At first glance, our construction of the $L_*$ HOD seems complex,
422: involving theoretical priors and introducing an artificial break in the
423: satellite occupation function. A more general HOD model for $L_*$ galaxies
424: could be obtained by parameterizing the HODs of two luminosity threshold
425: samples and taking their difference (e.g., \citealt{Tinker07}). However, we
426: only have a small number of data points from the LRG-$L_*$ cross-correlation
427: functions to constrain the $L_*$ HOD. The constraints on the low mass part of
428: the $L_*$ HOD (especially the mean occupation function for central galaxies)
429: come from the the number density of the $L_*$ sample and the large scale bias
430: of the cross-correlation. The shape of the low mass part of a general $L_*$
431: HOD can not be well constrained by either the number density or the large
432: scale bias, given that the halo bias factor in this mass range only increases
433: slowly. The constraints to the high mass part of the $L_*$ HOD come from the
434: small-scale cross-correlation with the LRGs.
435: Our procedure of the $L_*$ HOD parameterization reduces the degrees of freedom
436: in the low mass part of the HOD and allows some flexibility in the high
437: mass part of the HOD which is our focus. Although our parameterization
438: is a restricted one, it is suitable for our purpose of understanding
439: clustering properties in $L_*$-LRG cross-correlation.
440: To summarize, we allow the low mass end of the mean occupation function
441: (a square window plus
442: the low-mass power law) to have an overall horizontal shift with the shape
443: fixed. The main role of shifting the low mass part of the $L_*$ HOD is
444: to match the number density of the $L_*$ sample and the large scale bias
445: of the cross-correlation, while the high mass part of the occupation function
446: of $L_*$ galaxies that we are interested in determines the small-scale
447: cross-correlation with the LRGs.
448:
449:
450: The distribution profile of $L_*$ satellite galaxies inside halos is assumed
451: to follow the Navarro-Frenk-White (NFW) profile (\citealt{Navarro95,
452: Navarro96,Navarro97}) with a concentration parameter
453: $c=c_0(M/M_{\rm nl})^\beta$, where $M_{\rm nl}\sim 1.5\times 10^{12}\hMsun$
454: is the nonlinear mass at $z\sim 0.3$ and $\beta$ is fixed to -0.13
455: (\citealt{Bullock01}). As a whole, there are four free parameters in the
456: $L_*$ HOD, the overall horizontal shift of the low mass part, the high mass
457: amplitude $N_t$ and slope $\alpha_h$ of the mean occupation function, and
458: the halo concentration parameter $c_0$ extrapolated to the nonlinear mass
459: scale.
460: To compute the one-halo term of LRG--$L_*$ cross-correlation, we make the
461: simplifying assumption that the occupation number of LRGs and that of $L_*$
462: galaxies inside the same halo are uncorrelated. That is,
463: $\langle N({\rm LRG};M) N(L_*;M)\rangle =
464: \langle N({\rm LRG};M) \rangle
465: \langle N(L_*;M)\rangle$.
466: We discuss the effect of this assumption on our modeling results in \S~4.
467:
468: \begin{figure*}
469: \epsscale{1.0}
470: \plotone{f1.eps}
471: \epsscale{1}
472: \caption[]{
473: \label{fig:wp_LRG}
474: Projected two-point auto-correlation functions and best-fit HODs for the
475: two luminosity-threshold LRG samples.
476: Panel ({\it a}):
477: The measured two-point correlation functions (data points and error
478: bars) and the HOD model fits (solid curves). The two dashed curves for
479: each sample show the envelope of predictions from models with
480: $\Delta\chi^2<4$. The predicted one-halo and two-halo terms (dot-dashed
481: curves) are also shown for the sample with the lower luminosity
482: threshold.
483: Panel ({\it b}):
484: The mean occupation functions (solid curves) of LRGs from the best fits,
485: with contributions from central (dotted) and satellite (dashed) LRGs.
486: For each sample, the two sets of curves are the envelope from models
487: with $\Delta\chi^2<4$
488: and those for all galaxies are shaded.
489: Panel ({\it c}):
490: The marginalized distribution of the satellite fraction in each sample
491: with the central 68\% distribution marked by the two dashed lines.
492: Panel ({\it d}):
493: The probability distribution of halo masses for the LRGs in each sample
494: (solid lines), obtained from the occupation function shown in panel
495: ({\it b}) weighted by the differential halo mass function. The dotted
496: lines show the halo mass probability distribution from just the central
497: galaxies.
498: }
499: \end{figure*}
500:
501: \section{Modeling Results}
502:
503: We model simultaneously the two-point auto-correlation functions of the two
504: luminosity-threshold LRG samples ($M_g<-21.2$ and $M_g<-21.8$) and the
505: two-point cross-correlation function between the $L_*$ and the luminosity-bin
506: ($-21.7<M_g<-21.2$) LRG sample. Altogether, there are seventeen free
507: parameters, four for the $L_*$ HOD and thirteen for the LRG HOD.
508:
509: In addition
510: to the sum of $\chi^2$s from the two auto-correlation functions and the
511: cross-correlation function, we also add the number densities of the two
512: threshold LRG samples and the $L_*$ galaxies into the overall $\chi^2$.
513: That is,
514: \begin{eqnarray}
515: \chi^2 & = & \mathbf{(w_1-w_1^*)^T C_1^{-1} (w_1-w_1^*)} \nonumber \\
516: & + & \mathbf{(w_2-w_2^*)^T C_2^{-1} (w_2-w_2^*)} \nonumber \\
517: & + & \mathbf{(\Delta-\Delta^*)^T C_\times^{-1} (\Delta-\Delta^*)}
518: +\sum_{i=1}^3\frac{(n_i-n_i^*)^2}{\sigma_{n_i}^2},
519: \end{eqnarray}
520: where $\mathbf{w_1}$, $\mathbf{w_2}$, and $\mathbf{\Delta}$ are the vectors of
521: auto-correlation functions of the two LRG samples and the LRG--$L_*$
522: cross-correlation function, and $n_i$ ($i=1,2,3$) are the three number
523: densities. The observed values are denoted with a superscript $*$.
524: The full covariance matrix is used for each correlation function and
525: 10\% fractional errors are assumed for each of the galaxy number densities.
526:
527:
528: The three covariance matrices, $\mathbf{C_1}$, $\mathbf{C_2}$, and
529: $\mathbf{C_\times}$ are estimated using jackknife resampling. For each of
530: the two LRG samples, 104 jackknife subsamples are used \citep{Zehavi05a} for
531: estimating the covariance among 10 data points. For the LRG-$L_*$
532: cross-correlation, the covariance among 6 data points is estimated with
533: 50 jackknife subsamples \citep{Eisenstein05a}.
534: \citet{Zehavi05b} performed extensive tests with mock catalogs to access
535: the reliability of the jackknife error estimates in projected correlation
536: functions over a similar range of separations. They used 100 mock catalogs
537: with the same geometry and angular completeness as the SDSS sample and
538: similar clustering properties, created using the PTHalos method of
539: \citet{Scoccimarro02}. Their tests showed that
540: the jackknife method is a robust way to estimate the error covariance matrix,
541: especially for the relatively large volumes probed here. This holds
542: as long as the number of jackknife realizations, $n$, is significantly
543: larger than the dimension of the data vector, $p$.
544: \citet{Hartlap07} discusses related issues, pointing out a potential bias
545: in general model fitting which depends on the ratio of $p$ to $n$. As a
546: crude test of our error uncertainties, we incorporate their proposed method
547: to remove this bias by multiplying the inverse of the covariance matrix by
548: a factor of $(n-p-2)/(n-1)$. In our case, this factor is about 0.9. We
549: perform this for the $M_g<21.2$ LRG sample, resulting in a $\sim10\%$
550: increase in the uncertainties of the HOD parameters, but no noticeable
551: change in the best-fit values. We note
552: that the results presented later in this paper do not include such a
553: correction.
554:
555: We assume the different clustering measurements to be independent, ignoring
556: possible correlations between statistical errors across correlation
557: functions and number densities of different samples. Strictly speaking,
558: such statistical
559: correlations are not zero. For example, the $M_g<-21.2$ LRG sample includes
560: LRGs with $M_g<-21.8$, therefore the auto-correlation functions of the
561: $M_g<-21.2$ and $M_g<-21.8$ samples are partially correlated. These
562: correlations could be estimated with the jackknife technique, but they would
563: be noisy. Neglecting such correlations might make the constraints on the
564: HOD parameters somewhat tighter than they should be, which is a caveat for
565: interpreting our results.
566:
567: We adopt a Markov Chain Monte Carlo technique (MCMC;
568: e.g., \citealt{Gilks96}) to explore the HOD parameter space. At each point
569: of the chain, we take a random walk in the parameter space to generate a
570: new set of HOD parameters. The step-size of the random walk for each parameter
571: is drawn from a Gaussian distribution. The probability to accept the new set
572: of HOD parameters depends on the difference between $\chi^2_{\rm old}$ and
573: $\chi^2_{\rm new}$ (the values of $\chi^2$ for the new and old models): 1
574: for $\chi^2_{\rm new}\leq\chi^2_{\rm old}$ and $\exp[-(\chi^2_{\rm new}-
575: \chi^2_{\rm old})]$ for $\chi^2_{\rm new}>\chi^2_{\rm old}$.
576: Flat priors in logarithmic space are adopted for the LRG satellite occupation
577: numbers at the ten spline points and the pivot mass scale $M_s$ and for $N_t$
578: and mass scale shift of $L_*$ galaxies. Flat priors in linear space are used
579: for other HOD parameters.
580:
581: The total number of data points to model is 29 (10+10 auto-correlation
582: measurements for the two LRG samples, 6 LRG-$L_*$ cross-correlation
583: measurements, and 3 number densities). The number densities are well
584: reproduced by our model, with the median values 1.1-$\sigma$, 0.4-$\sigma$,
585: and 0.1-$\sigma$ away from the observed ones for the two LRG and the $L_*$
586: samples. Our model has 17 free parameters
587: and therefore the number of degree of freedom is 12. We find that the
588: best-fit model has $\chi^2=25$. The probability for a $\chi^2$ value higher
589: than our best-fit $\chi^2$ is 1.5\%. (Including the \citet{Hartlap07}
590: correction gives $\chi^2 \sim 22$ and the probability increases to $\sim$4\%).
591: The relatively-large $\chi^2$ values might be partly caused by our neglecting
592: the correlation between statistical errors across correlation functions and
593: number densities of different samples. It also indicates that the accuracy
594: in our analytical model of the two-point galaxy correlation function needs
595: to be improved and that our HOD parameterization is not perfect. With the
596: above caveats in mind, we present our modeling results below.
597:
598: \begin{figure*}[t]
599: \epsscale{1.0}
600: \plotone{f2.eps}
601: \epsscale{1.0}
602: \caption[]{
603: \label{fig:crosspar}
604: Constraints on HOD parameters from the LRG auto-correlation functions and
605: the LRG-$L_*$ cross-correlation function (see the text for details).
606: Panel ({\it a}):
607: The marginalized distribution of the parameter $\sigma_{M_g}$, which
608: is the width (in magnitude) of central galaxy luminosity distribution
609: at a fixed halo mass.
610: Panel ({\it b}):
611: The marginalized distribution of the parameter $p$, which characterizes
612: the relation between mean central galaxy luminosity and halo mass
613: through $L_c\propto M^p$.
614: In panels ({\it a}) and ({\it b}), the two dashed vertical lines
615: indicate the central 68\% of the distribution.
616: Panel ({\it c}):
617: Illustration of the effect of $\sigma_{M_g}$ and $p$ on the mean
618: occupation functions of LRGs. The thick dotted and dashed curves
619: are for the $M_g<-21.2$ LRG sample, with $\sigma_{M_g}$ varied from 0.37
620: to 0.43 (i.e., a $\pm 1\sigma$ change). The thin dashed and dotted curves
621: are for the $M_g<-21.8$ LRG sample, with $p$ varied from 0.46 to 0.51
622: (a $\pm 1\sigma$ change in $p$).
623: Panel ({\it d}):
624: The marginalized joint distribution of the concentration parameter
625: $c_0$ (normalized to that at the $z\sim 0.3$ nonlinear mass scale)
626: and the high mass slope $\alpha_h$ of the mean occupation function for
627: the $L_*$ galaxies. The contours show the 68\% and 95\% confidence
628: levels for two parameters.
629: }
630: \end{figure*}
631:
632: \subsection{Constraints on the HOD of LRGs}
633:
634: \subsubsection{HOD for the Luminosity-threshold LRG Samples}
635:
636: Figure~\ref{fig:wp_LRG} shows the fitting results for the two
637: luminosity-threshold LRG samples. Figure~\ref{fig:wp_LRG}{\it a} shows
638: the best-fit projected two-point correlation functions together with
639: the measurements. For the $M_g<-21.2$ sample, the one-halo and two-halo
640: components (dot-dashed curves) of the fit are also shown. In
641: Figure~\ref{fig:wp_LRG}{\it b}, the mean occupation function for each
642: sample is plotted, separating into contributions from central (dotted) and
643: satellite (dashed) galaxies.
644: The range of mean occupation functions with $\Delta\chi^2<4$ is denoted
645: by the shaded region ($\Delta\chi^2<4$ is chosen so that the envelopes are
646: sampled by a large number of MCMC points and the range delineated by the
647: envelopes can be clearly seen in the plot).
648: As in \citet{Zehavi05b}, the amplitude of the
649: high halo mass end of the mean occupation function is poorly constrained by
650: two-point correlation functions. Since the abundance of halos drops
651: exponentially at the high mass end, the two-halo pairs are mostly contributed
652: by lower mass halos. Although the number of one-halo pairs per halo rises
653: roughly as $\NavgM^2$, the exponential drop of the number of high mass halos
654: also makes most one-halo pairs come from lower mass halos. Overall, the
655: two-point correlation function is dominated by signals from halos of lower
656: mass where $\NavgM$$\sim$ a few, leading to the poor constraints on the HOD
657: at the very high mass end. The constraint on the high mass end
658: {\it slope} of $\Nsat$ for the faint LRG sample appears to be relatively
659: strong, and we discuss its possible implications in
660: \S~\ref{sec:discuss}.
661:
662: The results show that the host halos of LRGs are massive, above
663: $10^{13}\hMsun$, corresponding to large galaxy groups and clusters,
664: which is consistent with studies of the environment of luminous galaxies
665: (e.g., \citealt{Loh2003}). We find that as the threshold luminosity increases
666: by a factor of $\sim$1.7, the mass scale of host halos shifts by a factor of
667: $\sim 2.3$, which implies a correlation between the luminosity $L_c$ of the
668: central LRG and the mass $M$ of the host halo in the form of $L_c \propto
669: M^{0.66}$.
670:
671: LRGs are often thought to be the central elliptical galaxies of galaxy
672: groups of clusters. Our results clearly show that a fraction of the LRGs
673: have to be satellite galaxies. The main constraint for this comes from the
674: high amplitudes of the correlation function at small scales. Without
675: satellite LRGs, the projected correlation function would flatten out
676: toward small scales, similar in shape to the two-halo term shown in
677: Figure~\ref{fig:wp_LRG}{\it a}. To explain the rising small-scale
678: clustering, the one-halo term has to be introduced, hence we require the
679: existence of satellite LRGs. However, the fits imply that only a small
680: fraction of LRGs are satellites: 5.2--6.2\% and 2.3--3.2\%
681: in the $M_g<-21.2$ and $M_g<-21.8$ samples, respectively, as shown in
682: Figure~\ref{fig:wp_LRG}{\it c}. The steeper rise of the small scale
683: clustering makes the two-point correlation function of the
684: brighter LRG sample deviate from a power law more prominently. In
685: Appendix~\ref{sec:appendixA}, we investigate the key ingredients that
686: lead to the departure from a power law and provide insight as to
687: why the departure
688: becomes more clear for samples that are more luminous (as shown here) and
689: for samples at higher redshifts (e.g., \citealt{Ouchi05}).
690:
691: The mean occupation function tells us the mean number of LRGs as a function
692: of the halo mass. The result can be viewed differently by asking what the
693: probability distribution of masses of halos hosting such LRGs would be.
694: The probability is simply the product of the mean occupation function and
695: the differential halo mass function. We show such probability distributions
696: from the MCMC run in Figure~\ref{fig:wp_LRG}{\it d}.
697: The distribution of $M_g<-21.2$ LRGs peaks at $\sim 4.5\times 10^{13}\hMsun$,
698: and for $M_g<-21.8$ LRGs the peak shifts to higher mass, $\sim 10^{14}\hMsun$.
699: Both distributions span a large range in halo mass. The full width at half
700: maximum (FWHM) for either distribution is about $\Delta\log M=0.8$. Since
701: only a small fraction of LRGs are satellites, the probability distribution
702: is almost determined by the central LRGs. Only in halos more massive than
703: a few times $10^{14}\hMsun$ does the number of satellites in a halo become
704: significant (i.e., greater than one on average), and around this mass scale
705: a low amplitude shoulder in the distribution emerges. In FWHM sense, we
706: find that $M_g<-21.2$ and $M_g<-21.8$ LRGs reside in
707: halos of mass 2--13$\times 10^{13}\hMsun$ and 4--25$\times 10^{13}\hMsun$,
708: respectively.
709:
710: From the mean occupation functions in
711: Figure~\ref{fig:wp_LRG}{\it b},
712: one also notices that, compared with the samples of lower luminosities
713: in \citet{Zehavi05b}, the low mass cutoff profiles for the LRG samples
714: are better constrained. The relatively tight constraints come from
715: the steepening both in the halo mass function and in the mass dependence of
716: the halo bias factor
717: toward high halo mass. The former steepening makes the galaxy number
718: density more sensitive to the cutoff profile, while the latter one
719: increases the sensitivity of the large-scale galaxy bias factor to the
720: cutoff profile. With our HOD parameterization, the cutoff profile
721: of the LRG mean occupation function encodes information on the distribution
722: of central galaxy luminosity in halos of fixed mass and on how the mean
723: central galaxy luminosity scales with halo mass.
724: Figure~\ref{fig:crosspar}{\it a} shows the marginalized distribution of the
725: width $\sigma_{M_g}$ (in magnitude) of the (log-normal) central galaxy
726: luminosity distribution at a fixed halo mass. The clustering data require
727: a scatter of $\sim 0.16$ dex in the central galaxy luminosity in halos of
728: a given mass. The thick dotted and dashed curves in
729: Figure~\ref{fig:crosspar}{\it c} shows how $\sigma_{M_g}$ affects the
730: cutoff profile in $\Ncen$ of the $M_g<-21.2$ sample by varying
731: $\sigma_{M_g}$ by $\pm 1\sigma$. We note that the constraints on the
732: cutoff profile (and therefore $\sigma_{M_g}$) depend on cosmological
733: parameters, especially $\sigma_8$, in the sense of a larger $\sigma_{M_g}$
734: for a larger $\sigma_8$ (see Appendix~\ref{sec:appendixB} for more details).
735:
736: The mean central galaxy luminosity scales with halo mass, which is
737: characterized by the parameter $p$, $L_c\propto M^p$. LRG samples with
738: different luminosity cuts allow us to constrain the parameter $p$. The
739: marginalized distribution of $p$ is shown in Figure~\ref{fig:crosspar}{\it b}.
740: The thin dotted and dashed curves in Figure~\ref{fig:crosspar}{\it c} show
741: the effect $p$ on the mass scale shift of the $M_g<-21.8$ sample relative
742: to the $M_g<-21.2$ sample. The value of $p$ is around 0.48, which seems to
743: be inconsistent with the value 0.66 estimated from comparing the luminosity
744: and halo mass scale of the two threshold LRG samples. The reason is simple ---
745: in our parameterization, the parameters $p$ and $\sigma_{M_g}$ constrained
746: here correspond to all central galaxies, not only central LRG galaxies that
747: have a color selection criterion imposed.
748:
749: Our HOD parameterization here is rather flexible in the satellite HOD.
750: Again, in Appendix~\ref{sec:appendixB}, we present modeling results with
751: a five-parameter
752: HOD model and show their dependence on cosmology. In \S~\ref{sec:discuss},
753: we compare our HOD modeling results with those from other works and discuss
754: a few issues related to the modeling.
755:
756: We note that, on large scales, the bestfit $w_p$ curve in
757: Figure~\ref{fig:wp_LRG}{\it a} appears to be lower than that from the
758: five-parameter model (Fig.~\ref{fig:5par_sig8}{\it a}) for the $M_g<-21.2$
759: sample. The low mass cutoff profiles of $\Ncen$ for the bright and the faint
760: LRG samples are correlated in the parameterization adopted here, which means
761: that the HOD for central galaxies is more restrictive than that in the
762: five-parameter model. The LRG-$L_*$ cross-correlation also limits the range
763: of LRG HODs. The bestfit solution is a compromise in matching the large-scale
764: amplitudes of both the LRG auto-correlation functions and the LRG-$L_*$
765: cross-correlation function. The result implies that our HOD parameterization
766: is not perfect and that there is room to improve it (e.g., by allowing the
767: scatter in the central galaxy luminosity to vary with halo mass).
768:
769: \subsubsection{Mass Scales of Host Halos of Central and Satellite LRGs}
770:
771: Applying HOD modeling to galaxy samples with different luminosity thresholds,
772: \citet{Zehavi05b} find that there is a remarkable scaling relation between
773: the characteristic minimum mass $\Mmin$ of the host halos and the mass scale
774: $M_1$ of halos that on average host one satellite galaxy (above the luminosity
775: threshold) in addition to the central galaxy, $M_1\sim 23\Mmin$. With a
776: HOD parameterization close to what is used here, \citet{Zheng07} found the
777: relation is more like $M_1\sim 18\Mmin$.
778: Given the parameterization described in \S\ref{sec:meanlrg}, we
779: define $\Mmin$ to be the halo mass at which the expected number
780: of central galaxies above the luminosity threshold is 0.5, whether
781: or not the galaxy satisfies the LRG color criteria.
782: Theoretical studies of HODs of
783: subhalos in dissipationless dark matter simulations (\citealt{Kravtsov04})
784: and those predicted by SPH and semi-analytic galaxy formation
785: models (\citealt{Zheng05}) reveal a similar relation with the scaling factor
786: around 20. As shown by \cite{Berlind03}, the large gap between $M_1$ and
787: $\Mmin$ arises because in the low occupation regime, a more massive halo
788: tends to host a more massive central galaxy, rather than multiple smaller
789: galaxies. Does the $M_1$-$\Mmin$ scaling relation extend to massive halos
790: hosting LRGs?
791:
792: In Figure~\ref{fig:ML}, we plot $\Mmin$ and $M_1$ as a function of the
793: threshold luminosity from \citet{Zheng07}, corrected to be consistent
794: with $\sigma_8=0.8$ adopted in this paper. The luminosity in \citet{Zheng07}
795: is in $z=0.1$ $r$-band (see \citealt{Zehavi05b}). For comparison, we convert
796: the $K$-corrected and passively evolved $z=0.3$ $g$-band threshold
797: luminosities of the two LRG samples to the $z=0.1$ $r$-band ones by adopting
798: an apparent color $g-r=0.4$, and we obtain $M_r<-21.6$ and $M_r<-22.2$,
799: respectively.
800: We obtain the distribution of $\Mmin$ and $M_1$ for the two
801: luminosity-threshold LRG samples modeled in this paper by solving
802: $\Ncen=0.5$ and $\Nsat=1$ for each set of HOD parameters in the MCMC chain.
803: The results
804: are shown as the last two pairs of points. Roughly
805: speaking, they seem to follow the previous trend.
806: The mild discontinuity probably reflects an imperfect magnitude
807: conversion, which is not surprising as we are trying to account
808: for filter difference, $K$-correction, and stellar population evolution
809: from $z\sim 0.3$ to $z\sim 0.15$.
810: A more interesting difference is that the scaling
811: factors between $M_1$ and $\Mmin$ become 10.5$_{-1.5}^{+5.1}$ ($M_g<-21.2$)
812: and 7.2$_{-0.8}^{+0.9}$ ($M_g<-21.8$), respectively.
813: For comparison, the mean $M_1$-to-$\Mmin$ ratio for the MAIN galaxy samples
814: shown in Figure~\ref{fig:ML} is 18.3 with a mean 1$\sigma$ uncertainty of
815: $\sim$3.6.
816: This indicates that
817: the scaling relation may
818: break down in very massive halos.
819: In fact, the scaling relation from the
820: MAIN sample with the highest luminosity threshold ($M_r<-21.5$) in
821: Figure~\ref{fig:ML} has already shown a trend of such a decrease, with an
822: $M_1$-to-$\Mmin$ ratio of 13.2$_{-2.4}^{+2.7}$.
823: As a whole, these results show that the
824: $M_1$-$\Mmin$ scaling factor decreases for luminous galaxies that reside in
825: massive halos. We caution, however, that the definition of the $M_1/\Mmin$
826: ratio becomes more parameterization dependent at high galaxy luminosities
827: because the scatter between luminosity and halo mass is larger (i.e., the low
828: mass cutoff of the HOD is less sharp).
829:
830: The change of the $M_1$-$\Mmin$ scaling factor in massive halos can be
831: understood from the competition between accretion and destruction processes.
832: In general, massive halos accrete their satellites more recently than less
833: massive halos. While the rate of satellite accretion for low mass halos
834: peaks at $\sim 10$ Gyr ago, cluster-sized halos constantly accrete satellites
835: until recently (\citealt{Zentner05}). As a consequence, there is less time
836: for the orbit of satellites in a massive halo to decay through dynamical
837: friction and for them to merge with the central galaxy to form a larger
838: (brighter) central galaxy. In addition, the LRG samples we study are at
839: redshift $\sim0.3$, which makes the accretion even more dominant. Therefore,
840: the decrease in the ratio of $M_1$ and $\Mmin$ could be a manifestation of
841: the favor of accretion over destruction in massive halos.
842:
843: \begin{figure}
844: \epsscale{1.0}
845: \plotone{f3.eps}
846: \epsscale{1.0}
847: \caption[]{
848: \label{fig:ML}
849: Mass scales of the LRG HODs as a function of threshold luminosity. Shown
850: are the relation between the characteristic minimum mass $\Mmin$
851: at which 50\% of halos host central galaxies above the luminosity threshold
852: and the mass $M_1$ of halos that on average host one
853: satellite galaxy, as a function of the threshold absolute magnitude. The square
854: points are taken from \citet{Zheng07} for the SDSS main galaxy sample
855: (corrected to $\sigma_8=0.8$). Open and filled circles are the results
856: for the two LRG samples (note that the $z=0.3$ $g$-band luminosity
857: is converted to $z=0.1$ $r$-band by adopting an apparent color of $g-r=0.4$).
858: Dotted lines indicate different relations between luminosity of central galaxy
859: and mass of host halo.}
860: \end{figure}
861:
862: \begin{figure*}[t]
863: \epsscale{1.0}
864: \plotone{f4.eps}
865: \epsscale{1.0}
866: \caption[]{
867: \label{fig:crossNavg}
868: Mean occupation functions and cross-correlation functions of $L_*$ galaxies
869: and LRGs from the HOD modeling. {\it Left panels}: the predicted two-point
870: cross-correlation function between LRGs and $L_*$ galaxies separated into
871: contributions from central and satellite LRGs paired with $L_*$ galaxies within
872: same halos and LRGs paired with $L_*$ galaxies from different halos. The scale
873: $a$ is in units of proper $\hMpc$ to be consistent with that adopted
874: in \citet{Eisenstein05a}. The quantity $\phi_0=2.267\times 10^{-2}
875: h^3{\rm Mpc}^{-3}$ is the proper number density of $L_*$ galaxies at
876: the mean redshift $z\sim 0.3$. The lower panel shows the fractional difference
877: between model fits (thick for the flexible model and thin for the slope-fixed
878: model) and data (see the text).
879: {\it Right panel}: the mean occupation function for the $L_*$ galaxies in
880: massive halos (top solid curves) and that for the luminosity-bin LRGs.
881: (bottom curves). The square window at the low mass end represents the mean
882: occupation function of
883: central
884: $L_*$ galaxies. The envelopes of the
885: mean occupation functions are derived from models with $\Delta\chi^2<4$.
886: Dotted and dashed curves in the LRG mean occupation function are contributions
887: from central and satellite galaxies, respectively. The dot-dashed curve
888: shows the mean occupation function for the $L_*$ galaxies from a model with
889: the slope fixed to be 1.10 in the whole mass range.
890: }
891: \end{figure*}
892:
893: \subsubsection{HOD for the Luminosity-Bin LRG Sample}
894:
895: The cross-correlation between luminosity-bin LRG sample
896: ($-21.7<M_g<-21.2$) and the $L_*$ galaxies also leads to constraints on
897: the HOD of the luminosity-bin LRG sample. The lower (solid) curves in
898: the right panel of Figure~\ref{fig:crossNavg} are the $\Delta\chi^2<4$
899: envelope of the mean occupation function of the $-21.7<M_g<-21.2$ LRGs.
900: The bump (dotted) around $\sim 10^{14}\hMsun$ is the contribution from
901: central galaxies and the dashed curves represent contributions from
902: satellites. The result shows that most of the LRGs in this luminosity bin
903: are central galaxies in halos of mass $\sim$2--20$\times 10^{13}\hMsun$,
904: and a small fraction ($\sim$7\%) of them are satellites in more massive halos.
905:
906: \subsection{Constraints on the HOD of $L_*$ Galaxies}
907:
908: \begin{figure*}[t]
909: \epsscale{1.0}
910: \plotone{f5.eps}
911: \epsscale{1.0}
912: \caption[]{
913: \label{fig:crossL}
914: The luminosity dependence of the LRG--$L_*$ cross-correlation functions
915: at different scales. The six panels correspond to (proper) scales $a$=
916: 0.125 to 4$\hMpc$, respectively, as labeled in each panel. The quantity
917: $\phi_0=2.267\times 10^{-2} h^3{\rm Mpc}^{-3}$ is the proper number
918: density of $L_*$ galaxies at the mean redshift $z\sim 0.3$, and
919: $V=3(2\pi a^2)^{3/2}$ is the effective volume for the window function
920: [see eq.~(\ref{eqn:window})]. In each panel, the shaded regions are
921: predictions from the modeling results. Since the model predictions are based
922: on clustering information up to only $\sim 4L_*$, the shaded regions
923: below and above $4L_*$ are model interpolations and extrapolations,
924: respectively. The data points with error bars are the measurements
925: in \citet{Eisenstein05a}. Thick (thin) dotted curves are contributions
926: from central (satellite) LRGs paired with $L_*$ galaxies within common halos
927: (calculated from the best-fit HOD model), and dashed curves, which can be
928: clearly seen in panels ({\it d})--({\it f}),
929: represent the two-halo pair contribution.
930: }
931: \end{figure*}
932:
933: \citet{Zehavi05b} perform HOD modeling of the two-point auto-correlation
934: functions of the MAIN galaxy sample and show that the two-point function
935: can impose important constraints on the HOD of a sample of galaxies.
936: In general, the mean occupation function is tightly constrained around
937: $\NavgM\sim$ a few. It becomes loosely constrained towards higher halo
938: masses because the two-point correlation function is less sensitive to
939: the occupation distribution in these halos as a result of the steep drop
940: of the halo mass function. Therefore, the analyses of the MAIN
941: galaxy sample with low luminosity thresholds in \citet{Zehavi05b} cannot
942: quite reveal how these low luminosity galaxies reside in massive halos. Since
943: LRGs automatically pick out the massive halos, the cross-correlation between
944: MAIN sample galaxies and LRGs provides us a nice way to study the halo
945: occupation of MAIN sample galaxies in massive halos, enabling us to
946: better constrain the HOD of $L_*$ galaxies in massive halos.
947:
948: Figure~\ref{fig:crosspar}{\it d} shows constraints on the concentration
949: parameter $c_0$ and the high mass slope $\alpha_h$ of the mean occupation
950: function for $L_*$ galaxies, marginalized over the other parameters. These two
951: parameters are correlated in a sense that a higher $c_0$ corresponds to a
952: higher $\alpha_h$. Higher $\alpha_h$ means that more $L_*$ galaxies reside
953: in higher mass halos with lower concentrations and larger virial radius,
954: so to maintain the small-scale (cross-)clustering the distribution
955: of galaxies need to be more concentrated, i.e., a higher $c_0$. The high
956: mass slope of the mean occupation function of $L_*$ galaxies is
957: $1.49\pm 0.09$ (1-$\sigma$ range), steeper than the value 1.10 at lower
958: mass. The concentration parameter $c_0$ represents the value extrapolated to
959: $M_{\rm nl}\sim 1.5\times 10^{12}\hMsun$ according to
960: $c=c_0(M/M_{\rm nl})^{-0.13}$.
961: Its 1-$\sigma$ range is found to be $9.4\pm 1.7$, which translates to
962: $5.4\pm 1.0$ and $4.0\pm 0.7$ in halos of $10^{14}\hMsun$ and $10^{15}\hMsun$,
963: respectively, implying that the distribution of $L_*$ galaxies more or less
964: follows that of the dark matter, at radii the data can probe
965: ($\gtrsim 0.2\hMpc$ comoving). The mean occupation function of
966: $L_*$ galaxies in massive halos is shown as upper (solid) curves in the
967: right panel of Figure~\ref{fig:crossNavg}. These two curves are the envelope
968: determined by $\Delta\chi^2<4$. On average, about ten $L_*$ galaxies
969: are expected to reside in a halo of mass $\sim 10^{14}\hMsun$.
970:
971: Our fits show that the
972: mean occupation function of $L_*$ galaxies at high halo masses becomes
973: steeper than the slope 1.10 at intermediate masses (and lower redshift).
974: We have also run a model with fixed slope 1.10 for the satellite
975: mean occupation function in the whole mass range.
976: The fits to the
977: LRG auto-correlation function and the LRG-$L_*$ cross-correlation function
978: become much worse,
979: with the overall $\chi^2$ increasing by $\sim$50.
980: The resulting best fit to the cross-correlation function
981: is plotted in the left panels of Figure~\ref{fig:crossNavg}. The lower panel
982: shows the fractional difference between the fits (thick curve for the flexible
983: model and thin curve for the one with fixed slope) and the data.
984: The fit from the model with fixed slope is almost on top of the one from the
985: more flexible model, but the error bars in the measurement are
986: small and the change in $\chi^2$ is substantial (an increase of
987: $\sim10$ in the $\chi^2$ contributed by the cross-correlation function).
988: The corresponding high-mass
989: end of the
990: mean occupation function of $L_*$ galaxies from this more restricted model is
991: shown as the dot-dashed line in the right panel of Figure~\ref{fig:crossNavg}.
992: The single-slope model has
993: more $L_*$ satellites in halos less massive than
994: $10^{14}\hMsun$, and the best-fit concentration parameter for $L_*$
995: galaxies is about a factor of two smaller; these two effects compensate each
996: other to approximately maintain the amplitude of the small-scale
997: cross-correlation. Unless the error bars in the measurements were
998: underestimated (e.g., by a factor of two), the restricted model is highly
999: disfavored by its much worse fits to the data ($\Delta\chi^2\sim 50$).
1000:
1001: The cause and implication of
1002: the steep inferred high mass slope of the $L_*$ galaxy occupation function
1003: is not clear. It may be related to the selection of $L_*$ galaxies ---
1004: the sample used here is composed of galaxies in a bin of 1.5 magnitude around
1005: a redshift-dependent reference magnitude that is supposed to match
1006: $L_*$ at $z\sim 0.3$ \citep{Eisenstein05a}. Since the redshift range
1007: ($0.16<z<0.44$) is not narrow, the $L_*$ sample should be regarded as an
1008: effective sample, rather than a uniform sample. It may also be caused by
1009: the imperfection in the analytic model of the two-point cross-correlation
1010: function. The small measurement errors in the two-point cross-correlation
1011: function may require a more accurate model than the one used in this paper,
1012: and allowing for uncertainty in the model accuracy
1013: would increase the allowed range of the high mass slope.
1014: Finally, it may also be related to the assumption that the occupation numbers
1015: of LRGs and $L_*$ galaxies in the same halo are uncorrelated (\S~3.2). Since
1016: the effect of any correlation between the two occupation numbers becomes
1017: smaller as
1018: halo mass increases \citep{Simon09}, an anti-correlation between the LRG and
1019: $L_*$ occupation numbers would lead to a decrease in the high mass slope of
1020: the $L_*$ galaxy to conserve the number of LRG--$L_*$ pairs.
1021: As a whole, we therefore have higher confidence in the value of $\langle N\rangle \sim 10$
1022: at $M \sim 10^{14}\hMsun$, where the fits from the two models
1023: cross, than we have in the slope $\alpha_h$ itself.
1024:
1025: The modeling result also shows (see the left panel of
1026: Figure~\ref{fig:crossNavg}) that the LRG--$L_*$ cross-correlation function
1027: is dominated by central LRGs paired with satellite $L_*$ galaxies on scales
1028: less than $\sim 0.5 \hMpc$ (comoving), while above this scale the signal from
1029: satellite LRGs paired with satellite $L_*$ galaxies takes over until
1030: $\sim 1.5\hMpc$ (comoving), where the two-halo pairs start to dominate.
1031: We show below that the variation with scales in the contributing
1032: components is the key to understanding the scale dependence of the
1033: luminosity-dependent cross-clustering.
1034: In the one-halo term, the signal from central LRGs mainly comes from
1035: halos of mass a few times $10^{13}$ to $10^{14}\hMsun$, while the signal
1036: from satellite LRGs is from halos of $\sim 10^{14.5}\hMsun$, where the LRG
1037: occupation number is a few. Although the occupation numbers of
1038: LRGs and $L_*$ galaxies keep rising toward higher halo mass, more massive
1039: halos are too rare to make a significant contribution to the cross-correlation
1040: signal.
1041:
1042: \subsection{On the Scale Dependence of the Luminosity-Dependent
1043: Cross-Clustering}
1044:
1045: For the LRG--$L_*$ cross-correlation, \citet{Eisenstein05a} find a strong
1046: dependence on LRG luminosity. The clustering amplitude becomes higher for
1047: more luminous LRGs and varies by a factor of up to four over a factor of four
1048: in LRG luminosity (see their Fig.~2). Furthermore, the clustering amplitude
1049: increases more strongly with luminosity at smaller scales. We now show that
1050: these complex trends can be largely explained by the HOD results described
1051: above.
1052: We note that the HOD results are based on LRGs with luminosity $L<4L_*$,
1053: and while the LRG--$L_*$ cross-correlation in \citet{Eisenstein05a} is
1054: measured up to an LRG luminosity of $8L_*$, we focus on the $L<4L_*$ results.
1055: We emphasize that we do not intend to explain the data points for $L>4L_*$,
1056: for which we extrapolate our results.
1057: Our purpose here is to give a qualitative interpretation of the scale-dependent
1058: luminosity dependence of the cross-correlation between LRGs and $L_*$ galaxies.
1059:
1060: From our modeling results, the HOD for LRGs in narrow luminosity bins can be
1061: readily constructed, similar to what we do for the $-21.7<M_g<-21.2$ sample.
1062: We do not refit auto-correlation functions of different LRG samples; rather,
1063: we apply the value of $\sigma_{M_g}$ and the scaling $L_c\propto M^p$ derived
1064: from the faint and bright luminosity-threshold samples and interpolate (or
1065: extrapolate) the satellite occupation function from these samples.
1066: Figure~\ref{fig:crossL} compares the predicted dependence of the
1067: cross-correlation on LRG luminosity to the observed one. The plotted
1068: quantity $\phi_0 V\Delta$ is the average excess number of $L_*$ galaxies
1069: around an LRG in an effective spherical volume of $V$
1070: (see eq.~[\ref{eqn:Delta}]). We have already shown (Fig.~\ref{fig:crossNavg})
1071: that our HOD model reproduces the overall scale dependence of the
1072: cross-correlation accurately. Figure~\ref{fig:crossL} shows that the model
1073: also captures the trend of stronger luminosity dependence at smaller scales.
1074: Since the prediction is mostly based on modeling LRGs in the luminosity
1075: range of 2--4 $L_*$, it becomes less accurate at higher luminosities, where
1076: it is essentially an extrapolation.
1077:
1078: Close inspection of the observed points shows that, at the smallest scales,
1079: the clustering amplitude rises steeply and steadily from $\sim 2L_*$ to
1080: $\sim 8L_*$; while on larger scales, the luminosity dependence in the
1081: $2L_*$ -- $4L_*$ range is relatively flat, before steepening at higher
1082: luminosities. In addition, the overall luminosity dependence is weaker
1083: at larger scales. An intuitive understanding of these features can be built
1084: from our HOD modeling results.
1085:
1086: At very small scales, over all the LRG luminosity range, the cross-correlation
1087: signal is dominated by {\it central} LRG paired with satellite $L_*$ galaxies.
1088: The thick dotted curves in Figure~\ref{fig:crossL} show this component from
1089: the model. The slope of this component is estimated as follows.
1090: Approximating the mean occupation function of central LRGs in a narrow
1091: luminosity bin as a Dirac-$\delta$ function, the two-point cross-correlation
1092: function at a separation $r$ between $L_*$ galaxies and LRGs is simply
1093: proportional to the pair number $\langle N_*(M) \rangle f(r;M)$, where
1094: $\langle N_*(M) \rangle$ is the mean occupation function of $L_*$ galaxies
1095: in halos of mass $M$ and $f(r;M)$ is the fraction of $L_*$ galaxies located
1096: at a radius $r$ from the central LRG in halos of mass $M$. The function $f$
1097: is just the distribution profile of $L_*$ galaxies, $f(r;M)=\rho(r;M)r^2/\int
1098: \rho(r;M)r^2 dr$. Using a power law to approximate the inner profile,
1099: $\rho(r;M)\propto (r/R_{\rm vir})^\gamma $, we have $f(r;M)\propto
1100: R_{\rm vir}^{-(3+\gamma)} \propto M^{-(1+\gamma/3)}$, where $R_{\rm vir}$
1101: is the virial radius of the halo. Since $\langle N_*(M) \rangle \propto
1102: M^{\alpha_h}$, we see that the cross-correlation amplitude $\Delta \propto
1103: M^{\alpha_h-(1+\gamma/3)}$. Noting that the central luminosity $L \propto
1104: M^p$, the cross-correlation amplitude has the dependence on
1105: luminosity as $\Delta \propto L^{(\alpha_h-1-\gamma/3)/p}$. Inserting
1106: typical values of the model results, $\alpha_h=1.49$, $\gamma=-1$ (inner NFW
1107: profile), and $p=0.66$ to the expression, we obtain $\Delta \propto L^{1.2}$,
1108: close to the observational result, which is roughly $\Delta \propto L^{1.1}$.
1109:
1110: On larger scales (but still within the regime of one-halo pair domination),
1111: the contribution from {\it satellite} LRGs paired with satellite $L_*$
1112: galaxies starts to show up at the low luminosity end, as can be seen at
1113: scales 0.25 and 0.5 $\hMpc$ in Figure~\ref{fig:crossL}. At a {\it fixed} scale,
1114: most satellite-satellite pairs come from halos of a narrow mass range.
1115: As the luminosity in consideration increases, the occupation number of
1116: satellite LRGs at this halo mass decreases, leading to a decreasing
1117: contribution to the clustering amplitude from satellite-satellite pairs.
1118: The opposing dependences of clustering contributions from central LRGs and
1119: satellite LRGs on luminosity flattens the overall luminosity dependence at the
1120: low luminosity end, a feature seen in the observed clustering.
1121:
1122: On much larger scales (e.g., 4$\hMpc$), the two-halo pairs dominate the
1123: cross-correlation between LRGs and $L_*$ galaxies, and the signal is
1124: proportional to the large-scale bias factor of LRGs. If the LRG bias factor
1125: is approximated by the halo bias factor, the luminosity-dependent clustering
1126: simply reflects the dependence of the halo bias factor on the halo mass.
1127: Around $M\sim 10^{14}\hMsun$, the halo bias factor can be approximated by
1128: a power law with index $\alpha_b\sim$ 0.3--0.4. This, together with the
1129: relation between central luminosity and halo mass $L\propto M^p$
1130: ($p\sim 0.66$), gives a luminosity dependence of the cross-clustering
1131: following roughly as $L^{0.5}$, which agrees well with the observed trend.
1132:
1133: In modeling the LRG-$L_*$ cross-correlation, we make the assumption that
1134: there is no correlation between the occupation numbers of LRGs and $L_*$
1135: galaxies inside the same halo. Loosing this assumption would lead to
1136: changes in the contributions from one-halo central-satellite and
1137: satellite-satellite pairs. These changes would be at the level of fine
1138: details, and the above picture of the interplay among the three components
1139: for interpreting the scale-dependent luminosity dependence of the
1140: cross-correlation would remain valid.
1141:
1142: \section{Summary and Discussion}
1143: \label{sec:discuss}
1144:
1145: We have modeled the two-point auto-correlation functions
1146: of LRGs and the two-point cross-correlation functions between LRGs and $L_*$
1147: galaxies in the SDSS, within the HOD framework, obtaining results on the mean
1148: relation between central LRG luminosity and halo mass, the dispersion about
1149: this relation, the slope and amplitude of the satellite occupation function,
1150: and the abundance of $L_*$ galaxies in massive halos.
1151:
1152: The continuous rise toward small scales of two-point auto-correlation
1153: functions of LRGs implies that not all LRGs are the bright central galaxies
1154: in galaxy groups or clusters; a fraction of LRGs must be satellites
1155: to produce small scale, one-halo pairs. However, the satellite fraction is
1156: small and decreases with the LRG
1157: luminosity, e.g., $\sim$5--6\% for $M_g<-21.2$ and $\sim$2--3\% for
1158: $M_g<-21.8$ based on the HOD modeling.
1159: The characteristic minimum host halo mass of central LRGs
1160: ($\Mmin$, at which 50\% of halos host a galaxy above the luminosity
1161: threshold)
1162: is a few times $10^{13}\hMsun$ and increases with LRG luminosity.
1163:
1164: \citeauthor{Zehavi05b} (\citeyear{Zehavi05b};
1165: see also \citealt{Zheng07}) found a ratio
1166: $M_1/\Mmin\sim 20$ between the halo mass required to host a satellite above
1167: a luminosity threshold and the mass required to host a central galaxy above
1168: the same threshold. For these LRG samples, which populate higher mass halos
1169: and have a median redshift $z\sim 0.3$, we find a smaller ratio,
1170: $M_1/\Mmin\sim 10$. A similar drop is seen for the brightest sample
1171: ($M_r<-22$) in \citet{Zehavi05b}, which has a mean redshift $\sim0.16$.
1172: The decrease of the scaling factor reflects the balance between accretion
1173: and destruction
1174: of satellites (\citealt{Zentner05}) --- massive halos assemble more recently
1175: and their satellites have less time to merge with the central galaxy. The
1176: relatively higher redshift of the samples further strengthens this effect.
1177:
1178: The HOD of LRGs has been inferred using different methods in several recent
1179: investigations. To compare our results with others, one needs to pay
1180: attention to
1181: the differences in the sample definition, the underlying assumptions, and
1182: the modeling details. \citet{Mandelbaum06} present a mass determination of
1183: host halos for two SDSS LRG samples based on galaxy-galaxy lensing
1184: measurements. The construction of their LRG samples is not identical to
1185: ours, but the halo masses determined from their two samples appear to
1186: be consistent with those of the two luminosity-threshold LRG samples we
1187: model. The galaxy lensing directly measures the halo masses, while our
1188: results come from the mass distribution of halos and the galaxy assignment
1189: required to reproduce the observed clustering. The agreement between the
1190: two results is therefore encouraging.
1191: \citet{Wake08} model the projected
1192: two-point correlation functions for LRGs in the 2dF-SDSS LRG and QSO survey
1193: with a three-parameter model. For the SDSS $z=0.21$ LRG sample, their
1194: inferred mean occupation function is in general agreements with ours.
1195: \citet{Blake08} perform HOD modeling of the two-point angular
1196: correlation function of $0.4<z<0.7$ SDSS LRGs with photometric redshifts.
1197: Their parameterization is a slight variation of our five-parameter model
1198: presented in Appendix~\ref{sec:appendixB}. For samples with similar number
1199: densities, their inferred halo mass scales, cutoff widths of central
1200: galaxy occupation function, and high mass slopes of satellite occupation
1201: function are close to what we obtain.
1202: \citet{Kulkarni07} constrain the HOD
1203: of SDSS LRGs with redshift-space two-point and three-point correlation
1204: functions, by comparing the measurements to those from mock catalogs
1205: generated through populating halos identified in $N$-body simulations. They
1206: use a three-parameter HOD description and assume no velocity bias.
1207: They find that redshift-space three-point correlation functions
1208: favor a lower high mass slope ($\sim 1.4$) for the satellite occupation
1209: function. Since the $z=0$ outputs of $\sigma_8=0.9$ simulations are used
1210: in their modeling while the median redshift of LRGs is about 0.3, the
1211: effective $\sigma_8$ in their modeling is about 1.05. From
1212: Appendix~\ref{sec:appendixB} (eq.~[B3]), we see that the high mass slope
1213: from our modeling is 1.5 for such a high $\sigma_8$, which is close to the
1214: value favored by \citet{Kulkarni07}. However, we note that they use quite
1215: a different halo definition than ours (with a much lower overdensity
1216: threshold), which complicates the comparison. A more detailed comparison
1217: of the LRG modeling results with different methods and samples can be found
1218: in \citet{Brown08}.
1219:
1220: The inferred high mass slopes of the LRG occupation functions tend to
1221: be substantially larger than unity, either from our results or others (e.g.,
1222: \citealt{Blake08,Kulkarni07,Wake08}). This appears to differ from observational
1223: inferences and theoretical predictions for low luminosity
1224: samples (e.g., \citealt{Zehavi05b,Zheng07,Kravtsov04,Zheng05,Conroy06}).
1225: Is the steep slope of the inferred LRG occupation function a true feature
1226: or merely a result of modeling imperfections?
1227: First of all, the steep slope should not be a result of any restriction in
1228: our HOD parameterizations. Our five-parameter HOD model
1229: (Appendix~\ref{sec:appendixB}) introduces a cutoff in $\Nsat$ at the low
1230: mass end so that the connection between the high mass end slope and that at
1231: $\Nsat\sim$ a few is broken. For the fainter LRG sample, the constraint on
1232: the slope up to $M\sim 2\times 10^{15}\hMsun$ remains tight even with our
1233: more flexible
1234: parameterization [see Fig.~\ref{fig:wp_LRG}({\it b})]. In the LRG
1235: survey volume, there are not many halos that are more massive than
1236: $10^{15}\hMsun$. The true halo mass function in this volume can therefore
1237: deviate from the theoretical one used in the modeling. In
1238: Appendix~\ref{sec:appendixC}, we show that this fluctuation in the halo mass
1239: function does not seem to introduce any systematic biases in model
1240: fitting as it is already reflected in the covariance matrix of the data.
1241: To determine which features of the data drive the steep slope, we fit the
1242: faint LRG sample with the five-parameter model after excluding some data
1243: points and find that the two data points at $r_p\sim 1.7\hMpc$ and
1244: $2.7\hMpc$ play a large role
1245: --- if they are excluded, the slope drops from 1.8 to 1.6 (for $\sigma_8=0.8$).
1246: These scales are in the one-halo to two-halo transition region where the
1247: model is sensitive to the treatments of halo exclusion and scale-dependent
1248: halo bias. A more accurate scale-dependent halo bias with halos defined
1249: by spherical over-density (SO) could lead to a somewhat lower value of the high
1250: mass end slope (J.~L. Tinker, private communication). \citet{Reid09}
1251: constrain the LRG HOD with the counts-in-cylinders multiplicity function
1252: and the correlation function through populating halos in a
1253: simulation. They obtain a good fit by using SO halos and the high-mass end
1254: slope is found to be close to unity. Compared to friends-of-friends (FoF;
1255: \citealt{Davis85}) finder, the SO halo finder does not have the problem of
1256: linking two halos by a thin bridge of particles. We plan to pursue
1257: analytic models of galaxy clustering based on SO halo properties
1258: \citep{Tinker08} in future work.
1259:
1260: Because of the steep high mass slope, our model fits predict that massive
1261: clusters should host multiple LRGs. For example, the LRG occupation number
1262: of a $2\times 10^{15}\hMsun$ cluster would be about ten. Such a prediction
1263: can be tested with a cluster catalog if the mass can be determined. Using
1264: a sample of X-ray selected galaxy clusters at $0.2<z<0.6$, \cite{Ho07}
1265: assess cluster membership for LRGs based on their photometric redshifts
1266: and assign halo mass based on X-ray luminosity. They define halos as
1267: objects with mean density of 200 times
1268: the critical density rather than the mean
1269: background density as we do, and they
1270: assume $\Omega_m=0.238$. After correcting the
1271: differences in the halo definition and cosmology, our HOD result for the
1272: faint LRG sample matches theirs in the regime of $\NavgM \sim$ a few.
1273: (However,
1274: near the cutoff mass they find a much larger occupation number, while we have
1275: $\NavgM <1$.) In their catalog, there are only two very massive clusters with
1276: masses of $\sim 8\times10^{14}\hMsun$ and $\sim 1.1\times10^{15}\hMsun$
1277: (corrected to be consistent with our halo definition), and the (corrected)
1278: numbers of LRG members are about 8 and 14, respectively. We also performed
1279: a rough calculation to associate LRGs in the faint sample to MaxBCG
1280: clusters \citep{Koester07} in the overlapped sky region and redshift
1281: range. The mass of each cluster is estimated from the total number of
1282: MAIN galaxies inside the virial radius, with calibration by weak lensing
1283: \citep{Sheldon07}. For each cluster, we infer the halo radius and velocity
1284: dispersion from the halo mass corrected to match our halo definition.
1285: LRGs that fall in the projected halo radius and three times the velocity
1286: dispersion along the line-of-sight direction are assigned as cluster
1287: members (with no completeness and edge correction applied).
1288: The number of LRG members for clusters more massive than
1289: $6\times 10^{14}\hMsun$ is found to range from 0 to 7. The predicted number
1290: of LRGs in high mass halos (Fig.~\ref{fig:wp_LRG}) appears approximately
1291: consistent with the estimates from clusters. However, given the
1292: uncertainties in the mass estimator and the small number
1293: statistics, more work is needed to test our derived values of the high
1294: mass slope.
1295:
1296: \begin{figure*}
1297: \epsscale{1.0}
1298: \plotone{f6.eps}
1299: \epsscale{1.0}
1300: \caption[]{
1301: \label{fig:samcmp}
1302: Modified projected two-point correlation function and six-parameter HOD
1303: fit. Panel ($a$) shows the original $w_p(r_p)$ data points (filled circles)
1304: and the modified ones (open circles), which approximate those predicted by
1305: the semi-analytic model of \citet{Bower06} (see \citealt{Almeida08}).
1306: The thick curve is the best fit to the original data from the five-parameter
1307: HOD model and the dotted curve is that from the HOD model presented in \S~3.1.
1308: The thin curve is the best fit to the modified data from the six-parameter
1309: HOD model (see the text).
1310: The two curves are best HOD model fits. Panel ($b$) shows the best fit mean
1311: occupation functions. The thick curves are from fitting the original $w_p(r_p)$
1312: with the five-parameter model and the thin ones are the $\Delta\chi^2<1$
1313: envelopes of mean occupation functions from fitting the modified $w_p(r_p)$
1314: with a six-parameter model (see the text) and and those for all galaxies are
1315: shaded. Dashed, dotted, and solid curves
1316: are for central, satellite, and all galaxies, respectively.
1317: }
1318: \end{figure*}
1319:
1320: By modeling clustering of LRGs with different luminosities, we infer how
1321: the mean luminosity of central galaxies changes with the mass of their host
1322: halos, $L_c\propto M^p$ with $p\sim$ 0.46--0.51 for $M\sim 10^{14}\hMsun$.
1323: This is consistent with the relation $M\propto L^2$ found by
1324: \citet{Mandelbaum06} based on galaxy-galaxy lensing measurements.
1325: \citet{Padmanabhan09} model the clustering of a sample of photometrically
1326: selected SDSS LRGs that are fainter than those in the spectroscopic samples
1327: we use.
1328: Combining their HOD modeling results with ours for LRGs at $z\sim 0.3$,
1329: we find that the value of $p$ is in the range of 0.4--0.5 over a larger
1330: range of LRG luminosity.
1331: From a study of the halo occupation statistics of galaxies using galaxy
1332: groups identified in the Two Degree Field Galaxy Redshift Survey (2dFGRS;
1333: \citealt{Colless01}), \citet{Yang05} infer a $L_c$--$M$ relation that is
1334: well described by a broken power law with $p\sim$ 2/3 and 1/4 below and
1335: above $10^{13}\hMsun$, respectively. The inferred $L_c$--$M$ relation by
1336: \citet{Vale04} from matching the luminosity function from
1337: the 2dFGRS with the theoretical subhalo mass function has a similar
1338: high-mass slope $p\sim 0.28$. \citet{More09} constrain the high-mass slope
1339: to be $0.28_{-0.09}^{+0.07}$ for $z<0.072$ SDSS galaxies based on satellite
1340: kinematics. The high-mass slope extrapolated from the HOD modeling results
1341: of SDSS MAIN galaxies \citep{Zheng07} is also about 0.3.
1342: The value we obtain at the high-mass end from modeling the LRG clustering
1343: differs significantly from these results. It is interesting to see whether
1344: such a difference can be explained by the difference in the galaxy samples.
1345: The 2dFGRS luminosity is in the $b_J$ band, while we use the SDSS $g$-band
1346: luminosity. Since the wavelength coverages of these two bands are close,
1347: it seems unlikely that the band difference can cause the apparent
1348: discrepancy. The other
1349: thing to notice is that the 2dFGRS galaxies and the SDSS MAIN galaxies have
1350: a mean redshift
1351: $\sim 0.1$ and the LRG galaxies in our analysis are located around redshift
1352: 0.3. Could the discrepancy indicate an evolution effect over the intervening
1353: $\sim$2
1354: billion years? Through fitting restframe $B$-band galaxy luminosity functions
1355: at different redshifts using a CLF approach, \citet{Cooray05} finds that
1356: the data are compatible with a halo-mass-dependent central galaxy luminosity
1357: evolution, with the high mass slope $p$ increasing with redshift. The
1358: fitting results of \citet{Cooray05} imply that $p$ could be as high as 0.5
1359: at $z\sim 0.3$, close to our inferred value.
1360:
1361: If we take the inferred values of $p$ at $z\sim 0.3$ from our analysis and
1362: at $z\sim 0.1$ from the 2dFGRS studies at face value, they
1363: indicate that the luminosity evolution of central bright
1364: galaxies depends on halo mass in the sense that either galaxies in more
1365: massive halos fade more or the fraction of stars that were assembled into
1366: central galaxies from mergers between $z\sim 0.3$ and $z\sim0.1$ is smaller
1367: in more massive halos. \citet{Bernardi06}'s study of the properties of
1368: early-type galaxies in the SDSS as a function of local environment and
1369: redshift suggests that star formation in early-type galaxies happens
1370: earlier in dense regions and lasts over a shorter time-scale,
1371: which implies that central galaxies in more
1372: massive halos on average experience star formation at an earlier epoch.
1373: Since younger stellar populations fade faster than older ones, this seems
1374: to rule out the possibility that luminosity of central galaxies in more
1375: massive halos fades more between $z\sim 0.3$ and $z\sim 0.1$. We are
1376: left with the possibility that mergers in more massive halos are less
1377: efficient in adding stars to the central galaxies, which appears to be
1378: consistent with
1379: our finding of the drop of $M_1/\Mmin$ and its interpretation
1380: based on the competition between accretion and destruction as a function
1381: of halo mass. This is also in line with the results of LRG evolution
1382: in \citet{Brown08} from HOD modeling of their clustering from
1383: $z\sim 0.2$ to $z\sim 1.0$ in the NOAO Deep Wide-Field Survey.
1384:
1385: Since LRGs naturally separate out massive halos, our HOD modeling of the
1386: two-point cross-correlation functions between LRGs and $L_*$ galaxies
1387: circumvents the usual challenge in constraining the HOD of $L_*$ galaxies in
1388: massive halos from their two-point auto-correlation function. We show that
1389: the cross-correlation data are consistent with the case where the
1390: distribution of $L_*$ galaxies inside massive halos roughly follows that
1391: of the matter at distances greater than $\sim 0.2\hMpc$ and that the mean
1392: number of $L_*$ galaxies scales with halo mass as $M^{1.5}$. The slope of the
1393: mean occupation function is steeper than what is found for $z\sim 0$ galaxies,
1394: which is $\sim 1.10$. There may not be inconsistency between the results.
1395: The constraint for the slope for $z\sim 0$ galaxies from auto-correlation
1396: function is not sensitive to the occupation in very massive halos, while
1397: here from cross-correlation with LRGs, we have a better constraint on the
1398: $L_*$ occupation function in massive halos. On the other hand, the selection
1399: of $L_*$ galaxies and its redshift dependence make it hard to do a precise
1400: comparison between results at $z\sim 0$ and at $z\sim 0.3$.
1401:
1402: The luminosity
1403: dependence of the LRG--$L_*$ galaxy cross correlation depends on scale in
1404: a rather complex way \citep{Eisenstein05a}. By separating
1405: contributions from pairs of $L_*$ galaxies with central and satellite LRGs
1406: in common halos and in different halos, our
1407: HOD modeling results explain these trends, in a relatively transparent way.
1408: At a fixed scale, the luminosity dependence in the cross correlation reflects
1409: the luminosity-dependent HOD of LRGs. As the scale in consideration changes,
1410: the relative contributions of the one-halo central-satellite, the one-halo
1411: satellite-satellite, and the two-halo LRG-$L_*$ pairs to the cross-correlation
1412: function vary, which leads to the scale-dependent luminosity dependence of
1413: the cross correlation.
1414:
1415: Our LRG modeling results establish the relation between massive galaxies
1416: and dark matter halos at $z\sim 0.3$, which itself provides useful tests
1417: to models of formation of massive galaxies. \citet{Almeida08} present
1418: predictions for properties of LRGs in semi-analytic galaxy formation models.
1419: They show that the \citet{Bower06} model, which is based on the Millennium
1420: simulation \citep{Springel05}, predicts a $z=0.24$ LRG luminosity
1421: function that is in good agreement with the observation, although it fails at
1422: $z=0.5$. Compared to the measurement, the \citet{Bower06} model seems to
1423: predict the $z=0.24$ LRG two-point correlation function remarkably well on
1424: both small and large scales (see their Fig.13). However, the LRG HOD predicted
1425: in this model differs significantly from our modeling results presented in this
1426: paper: (a) in the \citet{Bower06} model, LRGs in the fainter sample can
1427: reside in halos of $10^{12}\hMsun$ (with mean occupation number of $\sim$0.01),
1428: much lower than the mass scale we infer; (b) the mass of halos that can on
1429: average host one LRG is about $3\times 10^{14}\hMsun$, higher than what we
1430: find (see Fig.11 in \citealt{Almeida08} and Fig.16 in \citealt{Wake08});
1431: (c) the probability distribution of LRG host halo masses is broad
1432: (see Fig.16 in \citealt{Wake08}), ranging from $10^{12}\hMsun$ to
1433: $10^{15}\hMsun$, rather than narrowly peaked around a few times
1434: $10^{13}\hMsun$ as we find (Fig.~\ref{fig:wp_LRG}$d$).
1435:
1436: Does the discrepancy between the theoretically predicted HOD and our
1437: observationally inferred HOD imply that our HOD parameterization is not
1438: generic enough to model LRG clustering? To investigate the problem,
1439: we modify the five-parameter HOD model (Appendix~\ref{sec:appendixB})
1440: to mimic the shape of the mean occupation function predicted in
1441: \citet{Bower06} model. The parameterization for the satellite mean occupation
1442: function remains unchanged (with three parameters). For the central galaxy
1443: occupation function, we model it as a linear (in logarithmic space) ramp
1444: going from $\langle N_{\rm min} \rangle$ at $\Mmin$ to unity at $M_u$ and
1445: staying at unity for $M>M_u$. In total, this parameterization has six free
1446: parameters. With the measured $w_p(r_p)$, we find that the best-fit HOD
1447: from this six-parameter model (not shown in Figure~\ref{fig:samcmp}$b$)
1448: closely follows the result from the five-parameter model (shown as thick curves in Figure~\ref{fig:samcmp}$b$). Therefore,
1449: change in the HOD parameterization does not solve the discrepancy and the
1450: five-parameter model is not inadequate in modeling LRG clustering. A close
1451: look at Figure~13 in \citet{Almeida08} shows that the predicted two-point
1452: correlation function does not match the data perfectly --- it is $\sim$15\%
1453: higher on scales of 0.1--2$\hMpc$ and $\sim 30\%$ lower on scales larger than
1454: 2$\hMpc$ (note that the vertical range of the plot is over eight orders of
1455: magnitude). We therefore modify the amplitude of the observed $w_p(r_p)$ data
1456: points to mimic the predicted correlation function and perform an HOD fit
1457: with the six-parameter model (by adopting the predicted number density,
1458: which is 10\% lower than the observed one). The results on the mean
1459: occupation functions are shown in panel ($b$) of Figure~\ref{fig:samcmp}
1460: as thin curves.
1461: Note that the shaded region represents the $\Delta\chi^2<1$ envelopes
1462: and the linear ramp for the central galaxy mean occupation number (thin dashed
1463: curves) has not reached unity at the highest mass in the plot.
1464: The mean occupation function can extend to halos of mass
1465: as low as $2\times 10^{12}\hMsun$ and reaches unity around
1466: $4\times 10^{14}\hMsun$, which appears to be approximately consistent with
1467: the prediction of the \citet{Bower06} model. From the above investigations,
1468: we conclude that the discrepancy
1469: between the theoretically predicted LRG HOD and the observationally inferred
1470: LRG HOD reflects the imperfection of the semi-analytic galaxy formation
1471: model --- the $15-30\%$ discrepancies with observed clustering are
1472: real and physically significant ---
1473: rather than limitations in our HOD parameterization.
1474: The results suggest that the mechanism of turning blue galaxies to red in
1475: the semi-analytic model is too efficient in halos of a few times
1476: $10^{12}\hMsun$ and not efficient enough in more massive halos.
1477: Our HOD modeling
1478: results thus provide important tests to galaxy formation theory.
1479:
1480: In combination with passive evolution
1481: of LRGs and a halo merger history (e.g., \citealt{White07,Seo08}), our
1482: modeling results can be used to predict the HOD of LRGs and the clustering
1483: of LRGs at lower or slightly higher redshifts. Supplemented with corresponding
1484: observations at these redshifts, we would be able to test our understanding
1485: of the formation and evolution of massive galaxies.
1486: Constraints on the HOD of LRGs are also useful for cosmological parameter
1487: investigations based on LRG clustering. For example, the most precise
1488: measurements of the large scale galaxy power spectrum have come from the
1489: SDSS LRG sample (\citealt{Tegmark06,Percival07}), and the principal
1490: limitation in interpreting these measurements is the uncertain level of
1491: scale-dependent bias between galaxy and matter power spectra in the mildly
1492: non-linear regime. With HOD constraints like those derived here, this
1493: scale-dependent bias can be calculated and corrected \citep{Yoo09}.
1494: As another example, HOD constraints could be combined with
1495: galaxy-galaxy lensing measurements of the LRG-mass cross-correlation
1496: (R. Mandelbaum et al., in preparation) to improve determinations of
1497: $\sigma_8$ and $\Omega_m$ \citep{Yoo06}. The role of LRGs in cosmological
1498: studies seems destined to grow with surveys that target large numbers of
1499: LRGs to measure baryon acoustic oscillations, including AAOmega LRG survey
1500: \citep{Ross08} and the Baryon Oscillation Spectroscopic Survey (BOSS), part of
1501: a proposed successor to SDSS-II. Understanding the evolving relation between
1502: LRGs and dark matter halos will be crucial to exploiting their power as
1503: cosmological probes and to revealing the physics that governs the formation
1504: of the most massive galaxies in the universe.
1505:
1506:
1507: \acknowledgments
1508: We are grateful to Jeremy Tinker and Jaiyul Yoo for valuable discussions.
1509: We thank Erin Sheldon for providing us with his cluster lensing results and
1510: acknowledge the Aspen Center for Physics, where discussions in a stimulating
1511: atmosphere led to a more complete presentation of this paper. We thank Tobias
1512: Baldauf for pointing out typos in an early draft. Finally, we
1513: thank the referee for a careful reading of the paper and for helpful comments
1514: that improved the paper.
1515:
1516: At an early stage of this work, ZZ was supported by NASA through Hubble
1517: Fellowship grants HF-01181.01-A, awarded by the Space Telescope Science
1518: Institute, which is operated by the Association of Universities for Research
1519: in Astronomy, Inc., for NASA, under contract NAS 5-26555. ZZ also
1520: acknowledges support from the Institute for Advanced Study through a John
1521: Bahcall Fellowship. IZ and ZZ acknowledge support from NSF grant
1522: AST-0907947 and DW acknowledges support from NSF grant AST-0707985.
1523: IZ was further supported by NASA through a contract issued by the Jet
1524: Propulsion Laboratory. YPJ is supported by NSFC (10533030), by the
1525: Knowledge Innovation Program of CAS (No. KJCX2-YW-T05), and by 973 Program
1526: (No.2007CB815402).
1527:
1528: Funding for the SDSS and SDSS-II has been provided by the Alfred P. Sloan
1529: Foundation, the Participating Institutions, the National Science Foundation,
1530: the U.S. Department of Energy, the National Aeronautics and Space
1531: Administration, the Japanese Monbukagakusho, the Max Planck Society, and
1532: the Higher Education Funding Council for England. The SDSS Web Site is
1533: http://www.sdss.org/.
1534:
1535: The SDSS is managed by the Astrophysical Research Consortium for the
1536: Participating Institutions. The Participating Institutions are the
1537: American Museum of Natural History, Astrophysical Institute Potsdam,
1538: University of Basel, Cambridge University, Case Western Reserve
1539: University, University of Chicago, Drexel University, Fermilab,
1540: the Institute for Advanced Study, the Japan Participation Group,
1541: Johns Hopkins University, the Joint Institute for Nuclear Astrophysics,
1542: the Kavli Institute for Particle Astrophysics and Cosmology,
1543: the Korean Scientist Group, the Chinese Academy of Sciences (LAMOST),
1544: Los Alamos National Laboratory, the Max-Planck-Institute for Astronomy (MPIA),
1545: the Max-Planck-Institute for Astrophysics (MPA), New Mexico State University,
1546: Ohio State University, University of Pittsburgh, University of Portsmouth,
1547: Princeton University, the United States Naval Observatory, and the
1548: University of Washington.
1549:
1550: %\clearpage
1551: \appendix
1552:
1553: %\renewcommand{\thesection}{\Alph{section}}
1554:
1555: \section{Role of Halo Mass Scales on the Departures from a Power
1556: Law in the Galaxy Two-Point Correlation Function}
1557: \label{sec:appendixA}
1558:
1559: \begin{figure}[h]
1560: \label{fig:appendixA}
1561: \plotone{fa1.eps}
1562: \caption[]{Impact of the $M_1/\Mmin$ and $\Mmin/M_{\rm nl}$ ratios on
1563: departures from a
1564: power law in the galaxy two-point correlation function. Each row (column)
1565: has the same $M_1/\Mmin$ ($\Mmin/M_{\rm nl}$) with the value marked on the
1566: right (top) of the plot. In each panel, the solid line shows the predicted
1567: projected correlation function, $w_p(r_p)$. The short and long dashed curves
1568: are the one-halo terms from central-satellite and satellite-satellite
1569: galaxy pairs, respectively, and the dot-dashed curve is the two-halo term.
1570: The dotted line is a power-law fit to $w_p(r_p)$ in the range of 1--10$\hMpc$
1571: to guide the eye.
1572: }
1573: \end{figure}
1574:
1575: Departures of the galaxy two-point correlation function from a pure
1576: power law have been observed for both low and high redshift galaxies
1577: (\citealt{Hawkins03,Zehavi04,Ouchi05,Coil06,Lee06}). In the two
1578: luminosity-threshold LRG samples studied in this paper, the departures are
1579: also clearly seen, shown as an upturn in the two-point correlation function
1580: at small scales (see also \citealt{Zehavi05a}). Such departures have been
1581: nicely explained by the HOD
1582: model as the transition from a regime dominated by one-halo pairs on small
1583: scales to that dominated by two-halo pairs on large scales \citep{Zehavi04}.
1584: The strength of the departures of the galaxy two-point correlation function
1585: from a pure power law is closely related to the amplitude of the one-halo
1586: term, which itself depends on the scatter in the occupation number and
1587: the halo mass function (\citealt{Benson00,Berlind02}). It would be helpful
1588: to
1589: gain a better understanding of the key factor that determines the strength
1590: of the departures.
1591:
1592: For luminosity-threshold samples, there are two mass scales, the
1593: characteristic minimum mass $\Mmin$ of halos that host central
1594: galaxies above the luminosity threshold and the mass scale $M_1$
1595: of halos that on average host one satellite galaxy above the
1596: luminosity threshold. An additional mass scale is the nonlinear mass
1597: $M_{\rm nl}$, marking a transition in the halo abundance from
1598: a power-law form to an exponential cutoff at high mass.
1599: We can thus identify two ratios that can shape the one-halo term: $M_1/\Mmin$
1600: and $\Mmin/M_{\rm nl}$. The $M_1/\Mmin$ ratio affects the shape of the galaxy
1601: occupation functions and tells us how quickly the transition
1602: from sub-Poisson to Poisson scatter occurs
1603: when going to higher halo masses. A smaller $M_1/\Mmin$ increases the
1604: importance of one-halo
1605: pairs by increasing the satellite fraction and thus results in a higher
1606: amplitude
1607: of the one-halo term. The $\Mmin/M_{\rm nl}$ ratio determines which part of the
1608: halo mass function the galaxy sample probes. The slope of the one-halo term
1609: reflects the drop of the halo mass function toward high masses. For a galaxy
1610: sample that probes the exponential tail of the halo mass function
1611: ($\Mmin/M_{\rm nl}\ga 1$),
1612: we expect a steep drop in the one-halo term and thus a sharp upturn around
1613: the one-halo to two-halo transition scale.
1614:
1615: To figure out the relative importance of $M_1/\Mmin$ and $\Mmin/M_{\rm nl}$
1616: on the
1617: departures from a power law in the galaxy two point correlation function,
1618: we calculate the predicted correlation functions on a $3\times 3$ grid
1619: of $M_1/\Mmin$ and $\Mmin/M_{\rm nl}$, as shown in
1620: Figure~\ref{fig:appendixA}.
1621: In each panel, the short dashed curve is the one-halo term from
1622: central-satellite pairs, the long dashed curve is the one-halo term from
1623: satellite-satellite pairs, and the dot-dashed curve is the two-halo term.
1624: The dotted line is a power-law fit to $w_p(r_p)$ in the range of 1--10$\hMpc$
1625: for comparison. The satellite-satellite pair contributions are similar at
1626: a fixed $M_1/\Mmin$, since $\Nsat$ is assumed to be proportional to halo mass.
1627: In a massive halo of fixed mass, however, the contribution of
1628: central-satellite pairs relative to satellite-satellite pairs is larger for
1629: a higher $\Mmin/M_{\rm nl}$ sample because of its low satellite occupation
1630: number. This leads to a larger slope across the central-satellite and
1631: satellite-satellite contributions for a higher $\Mmin/M_{\rm nl} $ sample, thus
1632: a steeper upturn in the correlation
1633: function\footnote{We thank Alison Coil,
1634: Jeremy Tinker, and Risa Wechsler for helpful discussions that led us
1635: to separately investigate the central-satellite and satellite-satellite pair
1636: contributions.}. In more detail, the slope
1637: across the central-satellite and satellite-satellite contributions should
1638: depend on the $M_1$ mass scale, but $M_1/\Mmin$ should not vary by an
1639: extremely large factor at least for a threshold galaxy sample. So a dependence
1640: on $M_1$ can be translated to that on $\Mmin/M_{\rm nl}$.
1641: When probing the
1642: exponential tail of the halo mass function for observed galaxy samples,
1643: we expect a steeper slope in both the central-satellite
1644: and satellite-satellite one-halo terms. This in turn leads to a more
1645: prominent rise at small scales.
1646:
1647: From the results, we conclude that overall the $\Mmin/M_{\rm nl}$ ratio plays
1648: a much more important role than $M_1/\Mmin$ in driving departures from a
1649: power law in galaxy
1650: two-point correlation functions. That is, a steeper drop in the halo mass
1651: function is translated to a steeper radial cutoff in the one-halo term,
1652: leading to a stronger inflection where the one-halo and two-halo term join.
1653: This explains why the departure is
1654: stronger for more luminous galaxy samples (e.g., \citealt{Zehavi05a,Zehavi05b}) that
1655: probe the high mass end of the halo mass function. This also explains why
1656: the departures are more prominent at higher redshifts (e.g.,
1657: \citealt{Ouchi05}) --- the fast drop of the nonlinear mass $M_{\rm nl}$
1658: toward high
1659: redshift makes the observed (bright) galaxies more likely to probe the
1660: exponential tail of the halo mass function.
1661: A more detailed theoretical investigation of the departure from a power
1662: law in the two-point correlation function of galaxies and its dependence
1663: on galaxy properties will be presented elsewhere.
1664:
1665: %\clearpage
1666: \section{LRG HODs with a Five-Parameter Model and Dependence on Cosmology}
1667: \label{sec:appendixB}
1668:
1669: \begin{figure}
1670: \plotone{fb1.eps}
1671: \caption[]{
1672: \label{fig:5par_sig8}
1673: Dependence of the LRG HOD on cosmology. Panel ({\it a}) shows the measured
1674: two-point correlation function (data points with error bars) of the
1675: $M_g<-21.2$ sample and the best HOD fits (solid curves) for cosmological
1676: models differing only in $\sigma_8$. Note that the best fits overlap with
1677: each other. Panel ({\it b}) shows
1678: the mean occupation functions (solid curves) corresponding to the best fits,
1679: separated into central (dotted) and satellite (dashed) contributions. From
1680: left to right, the HODs correspond to $\sigma_8$ values increasing from
1681: 0.65 to 1.00 with a step-size 0.05. The two bottom panels are similar, but
1682: for the $M_g<-21.8$ LRG sample.
1683: }
1684: \end{figure}
1685:
1686: The LRG clustering modeling in the main text assumes a flexible HOD
1687: parameterization, which reveals details of the constraining power of
1688: the two-point correlation functions on the HOD. In general, the HODs for
1689: luminosity-threshold samples predicted by galaxy formation models
1690: can be well described by a less flexible (i.e., more restricted) parametric
1691: form. The mean occupation
1692: function is well characterized by a step-like function for central
1693: galaxies and a power law-like function for satellite galaxies
1694: (\citealt{Kravtsov04,Zheng05}). We present here the modeling results for
1695: the two luminosity-threshold LRG samples with an alternative, five-parameter
1696: HOD model and give their dependence on cosmology. This set of results would
1697: be useful for comparisons with other work and for making mock catalogs
1698: in a wide range of cosmological models.
1699:
1700: The mean occupation function of a luminosity-threshold LRG sample,
1701: being the sum of central and satellite mean occupation functions, is
1702: parameterized as (see \citealt{Zheng05,Zheng07})
1703: \begin{equation}
1704: \langle N(M)\rangle=\frac{1}{2}
1705: \left[1+\erf\left(\frac{\log M-\log\Mmin}{\sigM}\right)\right]
1706: \left[1+\left(\frac{M-M_0}{M_1^\prime}\right)^\alpha\right],
1707: \end{equation}
1708: where $\erf$ is the error function
1709: \begin{equation}
1710: \label{eqn:erf}
1711: {\rm erf}(x)=\frac{2}{\sqrt{\pi}} \int_0^{\rm{x}} e^{-t^2} dt .
1712: \end{equation}
1713: The distribution of the occupation number of central galaxies and satellite
1714: galaxies are assumed to follow the nearest-integer and Poisson distributions,
1715: respectively, as usual. The five free parameters are the mass scale $\Mmin$
1716: and width $\sigM$ of the cutoff profile for the mean occupation function of
1717: central galaxies and the cutoff mass scale $M_0$, normalization $M_1^\prime$,
1718: and high mass slope $\alpha$ of the mean occupation function of
1719: satellite galaxies.
1720:
1721: We vary the normalization $\sigma_8$ (at $z=0$) of the matter fluctuation
1722: power spectrum from 0.65 to 1.00 with a step-size 0.05. Other
1723: cosmological parameters are assumed to be
1724: ($\Omega_m$, $\Omega_\Lambda$, $\Omega_b$, $n_s$, $h$)=(0.24, 0.76, 0.04, 0.95, 0.73).
1725: We perform an MCMC run
1726: for each LRG sample under each cosmological model and obtain the marginalized
1727: distribution for each of the five HOD parameters.
1728:
1729: In general, for a higher $\sigma_8$, the nonlinear mass increases and there
1730: are more high mass halos. To conserve the galaxy number density, the cutoff
1731: mass scale $\Mmin$ needs to increase. The increase in $\Mmin$ turns out to be
1732: slower than that in $M_{\rm nl}$, which leads to a lower halo bias that is
1733: necessary to maintain the large scale clustering of galaxies (i.e., the
1734: square of $b_g\sigma_8$). The width $\sigM$ of the cutoff increases so that
1735: some LRGs are populated to lower mass halos to further adjust the large scale
1736: bias factor and the galaxy number density. The mean occupation function of
1737: satellites also
1738: responds to the $\sigma_8$ change to match the small scale clustering
1739: amplitude: the mass scale $M_1^\prime$ increases and the slope $\alpha$
1740: decreases with increasing $\sigma_8$. For a large range of $\sigma_8$,
1741: our HOD modeling yields almost identical best fits to the data points,
1742: similar to what is found and discussed in \citet{Zheng07a}.
1743:
1744: While the dependence of
1745: $\log \Mmin$, $\sigM$, $\log M_0$, $\log M_1^\prime$ or $\alpha$ on
1746: $\sigma_8$ appears to be quite close to linear, we fit the results
1747: by adding a quadratic term for a better accuracy.
1748: The HOD parameters for the $M_g<-21.2$ LRG sample are
1749: \begin{eqnarray}
1750: \log \Mmin &=& 13.673 +1.419(\sigma_8-0.8)-1.706(\sigma_8-0.8)^2, \nonumber \\
1751: \sigma_{\log M} &=& \,\,\,0.621 +0.908(\sigma_8-0.8)-0.935(\sigma_8-0.8)^2, \nonumber \\
1752: \log M_0 &=& 12.339 +0.658(\sigma_8-0.8)+9.206(\sigma_8-0.8)^2, \nonumber \\
1753: \log M_1^\prime &=& 14.533 +1.248(\sigma_8-0.8)-1.394(\sigma_8-0.8)^2, \nonumber \\
1754: \alpha &=& \,\,\,1.832 -1.326(\sigma_8-0.8)+0.523(\sigma_8-0.8)^2.
1755: \end{eqnarray}
1756: The typical 1--$\sigma$ uncertainties for the five HOD parameters are
1757: 0.06, 0.07, 0.6, 0.025, and 0.08, respectively.
1758: The HOD parameters for the $M_g<-21.8$ LRG sample are
1759: \begin{eqnarray}
1760: \log \Mmin &=& 14.304 +1.694(\sigma_8-0.8)\,\,\,-1.810(\sigma_8-0.8)^2, \nonumber \\
1761: \sigma_{\log M} &=& \,\,\,0.797 +0.761(\sigma_8-0.8)\,\,\,-0.614(\sigma_8-0.8)^2, \nonumber \\
1762: \log M_0 &=& 12.491 -1.476(\sigma_8-0.8)+29.983(\sigma_8-0.8)^2, \nonumber \\
1763: \log M_1^\prime &=& 14.946 +1.616(\sigma_8-0.8)\,\,\,-0.712(\sigma_8-0.8)^2, \nonumber \\
1764: \alpha &=& \,\,\,1.717 -0.589(\sigma_8-0.8)\,\,\,-7.437(\sigma_8-0.8)^2.
1765: \end{eqnarray}
1766: The typical 1--$\sigma$ uncertainties for the these five HOD parameters are
1767: 0.06, 0.055, 0.7, 0.1, and 0.4, respectively.
1768: All the masses are in units of $\hMsun$.
1769:
1770: For results with cosmological models with $\Omega_m$ different than 0.24, one
1771: only needs to change the three mass scales by a factor of $\Omega_m/0.24$
1772: (see \citealt{Zheng02,Zheng07a}). This scaling assumes that the {\it shape}
1773: of the linear power spectrum stays fixed, with a change in $h$ or $n_s$
1774: compensating the impact of changing $\Omega_m$. Therefore, the solutions
1775: here cover a wide range of cosmological models in the ($\Omega_m$,$\sigma_8$)
1776: plane.
1777:
1778: \section{Halo Number Fluctuation in the LRG Survey Volume and the Covariance
1779: Matrix}
1780: \label{sec:appendixC}
1781:
1782: \begin{figure}[h]
1783: \label{fig:mf_fluctuation}
1784: \plottwo{fc1.eps}{fc2.eps}
1785: \caption[]{
1786: {\it (Left):} Halo mass function and its scatter from the
1787: \citet{Jing07} simulation, for a volume similar in size to the LRG survey
1788: volume. The green solid curve is the mean halo mass function in the
1789: simulation, the green dashed curves mark the scatter and the red dashed
1790: curves denote a Poisson scatter around the mean. The black solid, dotted,
1791: and dashed curves are the Jenkins, Sheth-Tormen, and Press-Schechter mass
1792: functions, respectively (see the text).
1793: {\it (Right):} Contribution of LRGs in massive halos to small-scale
1794: $w_p$ and the variation due to the fluctuation of halo mass function in a
1795: finite volume. The green solid curve is the best fit from a five-parameter
1796: HOD model to the measurements (green points). The green dotted and dashed
1797: curves are the one-halo and two-halo terms, respectively. The red solid
1798: curve is $w_p$ calculated by excluding halos more massive than $10^{15}\hMsun$.
1799: The dotted red curve shows the one-halo term from halos below
1800: $10^{15}\hMsun$, and the blue curve shows it for halos above this value.
1801: The black solid curves are the one-halo term
1802: contributions from halos more massive than $10^{15}\hMsun$ in each of the
1803: 32 simulation subvolumes that have the same size as the LRG survey volume.
1804: The bottom panel shows the comparison of the diagonal jackknife errors and
1805: uncertainties introduced by the fluctuation of the number of
1806: massive halos in the LRG survey volume. See text for more details.
1807: }
1808: \end{figure}
1809:
1810: In our HOD modeling of the LRG two-point correlation functions, we
1811: use a theoretical halo mass function \citep{Jenkins01}, which is a fitting
1812: formula based on $N$-body simulations. In the LRG survey volume,
1813: which is 0.72$h^3{\rm Gpc}^{-3}$ \citep{Eisenstein05b}, fluctuations in the
1814: number of massive halos (e.g., with mass above $10^{15}\hMsun$) are expected.
1815: Since LRGs reside in massive halos, the details of their small scale
1816: clustering may be sensitive to such fluctuations. In this Appendix, we
1817: investigate whether these fluctuations introduce any systematic effect on
1818: the HOD modeling.
1819:
1820: \citet{Jing07} performed $N$-body simulations with $1024^3$ particles in
1821: a box of 1800$\hMpc$ on a side. For our investigation, we make use of
1822: the catalog of massive halos ($> 10^{15}\hMsun$) identified in the
1823: $z=0.274$ outputs from four independent simulations. The volume of
1824: each realization is divided into eight octants. Each octant has a volume
1825: similar to the LRG survey volume, so altogether we have 32 sub-volumes to
1826: investigate the mass function variations.
1827:
1828: The left panel of Figure~\ref{fig:mf_fluctuation} shows the fluctuation of
1829: the number of massive halos among the 32 sub-volumes. The green solid curve
1830: is the mean halo mass function in a sub-volume and the green dashed curves
1831: mark the scatter around the mean. The scatter turns out to closely follow
1832: that of a Poisson distribution (the two dashed red curves).
1833: The fluctuation in the number of halos is about 15\% for $M>10^{15}\hMsun$
1834: and increases with halo mass. For comparison, we also plot three
1835: frequently used theoretical functions for the same cosmological model
1836: assumed in the simulation. The Jenkins mass function \citep{Jenkins01}, shown
1837: as the black solid curve, gives a good description of the mean halo mass
1838: function at $10^{15}\hMsun$ and becomes a factor of two higher at
1839: $2\times 10^{15}\hMsun$. The Sheth-Tormen mass function (\citealt{Sheth99};
1840: black dotted curve) is slightly higher than the Jenkins mass function, while
1841: the Press-Schechter mass function (\citealt{Press74}; black dashed curve)
1842: underestimates the mass function by a factor of five in the mass range
1843: considered here.
1844:
1845: The right panel of Figure~\ref{fig:mf_fluctuation} shows the effect of the
1846: fluctuation in the number of massive halos on the small scale clustering
1847: of the $M_g<-21.2$ LRGs. The green solid curve is the best fit to the
1848: measured $w_p$ (green points) from the five-parameter HOD model
1849: (Appendix~\ref{sec:appendixB}) with the same cosmological model as used
1850: in the simulation. The green dotted and dashed
1851: curves are the contributions from the one-halo and two-halo term,
1852: respectively. With the best-fit HOD model fixed, the red solid curve
1853: show the predicted $w_p$ when only keeping halos less massive than
1854: $10^{15}\hMsun$ in the calculation. The red dotted curve is the
1855: one-halo term from halos less massive than $10^{15}\hMsun$.
1856:
1857: The blue solid curve is the mean contribution to the one-halo term from
1858: halos of $M>10^{15}\hMsun$. The fluctuation in the number of massive
1859: halos would lead to a fluctuation around this mean contribution. To
1860: see this, we populate the massive halos in the 32 sub-volumes according
1861: to the best-fit HOD model and measure the one-halo $w_p$ in each sub-volume.
1862: Black solid curves show the measurements for individual sub-volumes.
1863: Such a fluctuation in the one-halo term introduces uncertainties in
1864: the small scale clustering. By adding the one-halo term from halos
1865: with $M<10^{15}\hMsun$ (red dotted), that from halos above $10^{15}\hMsun$
1866: in each sub-volume (black solid) and the two halo-term (green dashed),
1867: we obtain the black points with error bars reflecting the fluctuation in
1868: the number of massive halos. On small scales, these error bars appear to
1869: have similar (somewhat smaller) amplitude to those from the diagonal elements
1870: of the covariance matrix (see the comparison in the lower panel), which
1871: is computed through jackknife technique.
1872:
1873: We see that the fluctuation of the number of massive halos in the LRG survey
1874: volume can be large, e.g., $\sim$15\% for $M>10^{15}\hMsun$ halos and
1875: $\sim$50\% for $M>2\times10^{15}\hMsun$ halos. This introduces a
1876: fluctuation in the small scale clustering of LRGs. However, our
1877: investigation suggests that such a fluctuation in $w_p$ should be correctly
1878: reflected in the jackknife covariance matrix. Therefore, it is sufficient
1879: to use the mean mass function in modeling the two-point correlation functions
1880: of LRGs.
1881:
1882: \begin{thebibliography}{}
1883:
1884: \bibitem[Almeida et al.(2008)]{Almeida08}
1885: Almeida, C., Baugh, C.~M., Wake, D.~A., Lacey, C.~G., Benson, A.~J., Bower,
1886: R.~G., \& Pimbblet, K.\ 2008, \mnras, 386, 2145
1887: % Luminous Red Galaxies in hierarchical cosmologies
1888:
1889: \bibitem[Benoist et al.(1996)]{benoist96}
1890: Benoist, C., Maurogordato, S., da Costa, L. N., Cappi, A.,
1891: \& Schaeffer, R. 1996, \apj, 472, 452
1892: % luminosity dependence of bias in SSRS2, flat below L_* rising above
1893:
1894: \bibitem[Benson et al.(2000)]{Benson00}
1895: Benson, A. J., Cole, S., Frenk, C. S., Baugh, C. M., \& Lacey, C. G. 2000,
1896: \mnras, 311, 793
1897: % Nature of galaxy bias and clustering
1898:
1899: \bibitem[Berlind \& Weinberg(2002)]{Berlind02}
1900: Berlind, A.\ A.\ \& Weinberg, D.\ H.\ 2002, \apj, 575, 587
1901: % HOD
1902:
1903: \bibitem[Berlind et al.(2003)]{Berlind03}
1904: Berlind, A.~A., et al.\ 2003, \apj, 593, 1
1905: % The Halo Occupation Distribution and the Physics of Galaxy Formation
1906:
1907: \bibitem[Bernardi et al.(2006)]{Bernardi06}
1908: Bernardi, M., Nichol, R.~C., Sheth, R.~K., Miller, C.~J., \& Brinkmann, J.\
1909: 2006, \aj, 131, 1288
1910: % Evolution and enviroment of early-type galaxies
1911:
1912: \bibitem[Blake et al.(2008)]{Blake08}
1913: Blake, C., Collister, A., \& Lahav, O.\ 2008, \mnras, 385, 1257
1914: % Halo-model signatures from 380,000 SDSS Luminous Red Galaxies with
1915: % photometric redshifts
1916:
1917: \bibitem[Blanton et al.(2003)]{Blanton03}
1918: Blanton, M.~R., et al.\ 2003, \apj, 592, 819
1919: % The Galaxy Luminosity Function and Luminosity Density at Redshift z = 0.1
1920:
1921: \bibitem[Bower et al.(2006)]{Bower06}
1922: Bower, R.~G., Benson, A.~J., Malbon, R., Helly, J.~C., Frenk, C.~S., Baugh,
1923: C.~M., Cole, S., \& Lacey, C.~G.\ 2006, \mnras, 370, 645
1924: % Breaking the hierarchy of galaxy formation
1925:
1926: \bibitem[Brown, Webster \& Boyle(2000)]{brown00}
1927: Brown, M.\ J.\ I., Webster, R. L., \& Boyle, B.\ J.\ 2000, \mnras, 317, 782
1928: %% the clustering of color-selected galaxies
1929:
1930: \bibitem[Brown, Zheng, White, et al. (2008)]{Brown08}
1931: Brown, M.\ J.\ I., Zheng, Z., White, M., Dey, A., Jannuzi, B.~T., et al.\
1932: 2008, \apj, 682, 937
1933: %% red galaxy evolution and the HOD
1934:
1935: \bibitem[Budavari et al.(2003)]{budavari03}
1936: Budavari, T., et al., 2003, \apj, 595, 59
1937: %%angular clustering lum and color dependence
1938:
1939: \bibitem[Bullock et al.(2001)]{Bullock01}
1940: Bullock, J.\ S., Kolatt, T.\ S., Sigad, Y., Somerville, R.\ S., Klypin, A.\ A.,
1941: Primack, J.\ R., Dekel, A.\ 2001, \mnras, 321, 559
1942: % concentration c as a function of halo mass
1943:
1944: \bibitem[Bullock, Wechsler, \& Somerville(2002)]{Bullock02}
1945: Bullock, J.\ S., Wechsler, R.\ H., \& Somerville, Rachel, S.\ 2002, \mnras,
1946: 329, 246
1947: %% HOD @ high z
1948:
1949: \bibitem[Coil et al.(2006)]{Coil06}
1950: Coil, A.~L., Newman, J.~A., Cooper, M.~C., Davis, M., Faber, S.~M., Koo,
1951: D.~C., \& Willmer, C.~N.~A.\ 2006, \apj, 644, 671
1952: % The DEEP2 Galaxy Redshift Survey: Clustering of Galaxies as a Function
1953: % of Luminosity at z=1
1954:
1955: \bibitem[Coil et al.(2008)]{Coil08}
1956: Coil, A.~L., et al.\ 2008, \apj, 672, 153
1957: % The DEEP2 Galaxy Redshift Survey: Color and Luminosity Dependence of
1958: % Galaxy Clustering at z~1
1959:
1960: \bibitem[Colless et al.(2001)]{Colless01}
1961: Colless, M., et al.\ 2001, \mnras, 328, 1039
1962: % 2dFGRS
1963:
1964: \bibitem[Conroy et al.(2006)]{Conroy06}
1965: Conroy, C., Wechsler, R.~H., \& Kravtsov, A.~V.\ 2006, \apj, 647, 201
1966: % modeling lum-dependent galaxy clustering with sub-halo simulations
1967:
1968: \bibitem[Cooray(2005)]{Cooray05}
1969: Cooray, A.\ 2005a, \mnras, 364, 303
1970: % A brighter past: galaxy luminosity function at high redshifts
1971:
1972: \bibitem[Cooray(2006)]{Cooray06}
1973: Cooray, A.\ 2006, \mnras, 365, 842
1974: % Halo Model at Its Best: Constraints on Conditional Luminosity Functions
1975: % from Measured Galaxy Statistics
1976:
1977: \bibitem[Davis \& Geller(1976)]{davis76}
1978: Davis, M., \& Geller, M.\ J. 1976, \apj, 208, 13
1979: % Galaxy correlations as a function of morphological type
1980: %(with the Uppsala catalog)
1981:
1982: \bibitem[Davis et al.(1985)]{Davis85}
1983: Davis, M., Efstathiou, G., Frenk, C.\ S., \& White, S.\ D.\ M.\ 1985, \apj,
1984: 292, 371
1985: % FoF
1986:
1987: \bibitem[Davis et al.(1988)]{davis88}
1988: Davis, M., Meiksin, A., Strauss, M.\ A., da Costa, L.\ N., \& Yahil, A.
1989: 1988, \apj, 333, L9
1990: % CfA correlations for volume-limited catalogs
1991:
1992: \bibitem[Dressler(1980)]{dressler80}
1993: Dressler, A. 1980, \apj, 236, 351
1994: %% morphological segregation
1995:
1996: \bibitem[Eisenstein et al.(2001)]{Eisenstein01}
1997: Eisenstein, D.~J., et al.\ 2001, \aj, 122, 2267
1998: % Spectroscopic Target Selection for the Sloan Digital Sky Survey:
1999: % The Luminous Red Galaxy Sample
2000:
2001: \bibitem[Eisenstein et al.(2005a)]{Eisenstein05a}
2002: Eisenstein, D.~J., Blanton, M., Zehavi, I., Bahcall, N., Brinkmann, J.,
2003: Loveday, J., Meiksin, A., \& Schneider, D.\ 2005, \apj, 619, 178
2004: % The Small-scale Clustering of Luminous Red Galaxies via Cross-Correlation
2005: % Techniques
2006:
2007: \bibitem[Eisenstein et al.(2005b)]{Eisenstein05b}
2008: Eisenstein, D.~J., et al.\ 2005, \apj, 633, 560
2009: % Detection of the Baryon Acoustic Peak in the Large-Scale Correlation
2010: % Function of SDSS Luminous Red Galaxies
2011:
2012: \bibitem[Gilks, Richardson, \& Spiegelhalter(1996)]{Gilks96}
2013: Gilks, W.\ R., Richardson, S., \& Spiegelhalter, D.\ J.\ 1996, Markov
2014: Chain Monte Carlo in Practice (London: Chapman and Hall)
2015: % MCMC
2016:
2017: \bibitem[Goto et al.(2003)]{goto03}
2018: Goto, T., Yamauchi, C., Fujita, Y., Okamura, S., Sekiguchi, M., Smail, I.,
2019: Bernardi, M., \& Gomez, P.\ L. 2003, \mnras 346, 601
2020: % SDSS morphology-density relation
2021:
2022: \bibitem[Guzzo et al.(1997)]{guzzo97}
2023: Guzzo, L., Strauss, M.\
2024: A., Fisher, K.\ B., Giovanelli, R., \& Haynes, M.\ P.\ 1997, \apj, 489, 37
2025: % redshift space distortion & real space clustering vs. morphology
2026:
2027: \bibitem[Hamana et al.(2006)]{Hamana06}
2028: Hamana, T., Yamada, T., Ouchi, M., Iwata, I., \& Kodama, T.\ 2006, \mnras,
2029: 369, 1929
2030: % Masses of high-z galaxy hosting haloes from angular clustering and
2031: % their evolution in the CDM model
2032:
2033: \bibitem[Hamilton(1988)]{hamilton88}
2034: Hamilton, A. J. S. 1988, \apj, 331, L59
2035: % biasing of L>L_* galaxies in CfA survey
2036:
2037: \bibitem[Hartlap et al.(2007)]{Hartlap07}
2038: Hartlap, J., Simon, P., \& Schneider, P.\ 2007, \aap, 464, 399
2039: % Why your model parameter confidences might be too optimistic.
2040: % Unbiased estimation of the inverse covariance matrix
2041:
2042: \bibitem[Hawkins et al.(2003)]{Hawkins03}
2043: Hawkins, E., et al.\ 2003, \mnras, 346, 78
2044: % The 2dF Galaxy Redshift Survey: correlation functions, peculiar
2045: % velocities and the matter density of the Unive
2046:
2047: \bibitem[Ho et al.(2007)]{Ho07}
2048: Ho, S., Lin, Y.-T., Spergel, D., \& Hirata, C.~M.\ 2007, ArXiv Astrophysics
2049: e-prints, arXiv:0706.0727
2050: %Luminous Red Galaxy Population in Clusters at $0.2\le z\le 0.6$
2051:
2052: \bibitem[Hubble(1936)]{hubble36}
2053: Hubble, E.P. 1936, The Realm of the Nebulae (Oxford University Press: Oxford),
2054: 79
2055: % page ref is for ellipticals cluster more strongly than spirals
2056:
2057: \bibitem[Jenkins et al.(2001)]{Jenkins01}
2058: Jenkins, A., Frenk, C.\ S., White, S.\ D.\ M., Colberg, J.\ M., Cole, S.,
2059: Evrard, A.\ E., Couchman, H.\ M.\ P., \& Yoshida, N.\ 2001, \mnras, 321, 372
2060:
2061: \bibitem[Jing, Mo, \& B\"orner(1998)]{Jing98a}
2062: Jing, Y. P., Mo, H. J., \& B\"orner, G. 1998, \apj, 494, 1
2063: % sigma_v from Las Campagnas
2064: % cluster-weighted model, initial idea of HOD
2065:
2066: \bibitem[Jing \& B\"{o}rner(1998)]{Jing98b}
2067: Jing, Y.\ P. \& B\"{o}rner, G.\ 1998, \apj, 503, 37
2068: % application of cluster-weighted model, (HOD)
2069: % LCRS Three-Point Correlation
2070:
2071: \bibitem[Jing, B\"{o}rner, \& Suto(2002)]{Jing02}
2072: Jing, Y.\ P., B\"{o}rner, G., \& Suto, Y.\ 2002, \apj, 564, 15
2073: % application of cluster-weighted model, (HOD)
2074: % Spatial Correlation Functions and the Pairwise Peculiar Velocity
2075: % Dispersion of Galaxies in the Point Source Catalog Redshift Survey
2076:
2077: \bibitem[Jing et al.(2007)]{Jing07}
2078: Jing, Y.~P., Suto, Y., \& Mo, H.~J.\ 2007, \apj, 657, 664
2079: % The Dependence of Dark Halo Clustering on Formation Epoch and
2080: % Concentration Parameter
2081:
2082: \bibitem[Koester et al.(2007)]{Koester07}
2083: Koester, B.~P., et al.\ 2007, \apj, 660, 239
2084: % A MaxBCG Catalog of 13,823 Galaxy Clusters from the Sloan Digital Sky Survey
2085:
2086: \bibitem[Kravtsov et al.(2004)]{Kravtsov04}
2087: Kravtsov, A.~V., Berlind, A.~A., Wechsler, R.~H., Klypin, A.~A.,
2088: Gottl{\" o}ber, S., Allgood, B., \& Primack, J.~R.\ 2004, \apj, 609, 35
2089: % The Dark Side of the Halo Occupation Distribution
2090:
2091: \bibitem[Kulkarni et al.(2007)]{Kulkarni07}
2092: Kulkarni, G.~V., Nichol, R.~C., Sheth, R.~K., Seo, H.-J., Eisenstein,
2093: D.~J., \& Gray, A.\ 2007, \mnras, 378, 1196
2094: % The three-point correlation function of luminous red galaxies in the
2095: % Sloan Digital Sky Survey
2096:
2097: \bibitem[Lee et al.(2006)]{Lee06}
2098: Lee, K., Giavalisco, M., Gnedin, O.~Y., Somerville, R., Ferguson, H.,
2099: Dickinson, M., \& Ouchi, M.\ 2006, \apj, 642, 63
2100: % The Large-scale and Small-scale Clustering of Lyman-Break Galaxies at
2101: % 2.5 < z< 5.5 from the GOODS survey
2102:
2103: \bibitem[Loh(2003)]{Loh2003}
2104: Loh, Y.~S.\ 2003, Ph.D. thesis, Princeton Univ.
2105:
2106: \bibitem[Loveday et al.(1995)]{loveday95}
2107: Loveday, J., Maddox, S. J., Efstathiou, G., \& Peterson, B. A. 1995,
2108: \apj, 442, 457
2109: % luminosity and morphology dependence of Stromlo-APM clustering
2110:
2111: \bibitem[Ma \& Fry(2000)]{Ma00}
2112: Ma, C., \& Fry, J. N. 2000, \apj, 543, 503
2113: % Mass power spectrum from halo model
2114:
2115: \bibitem[Madgwick et al.(2003)]{madgwick03}
2116: Madgwick, D.~S. et al. 2003, \mnras, 344, 847
2117: % spectral type dependence of 2dF galaxy clustering
2118:
2119: \bibitem[Magliocchetti \& Porciani(2003)]{Magliocchetti03}
2120: Magliocchetti, M., \& Porciani, C.\ 2003, \mnras, 346, 186
2121: %% HOD for 2dF galaxies (astro-ph/0304003)
2122:
2123: \bibitem[Mandelbaum et al.(2006)]{Mandelbaum06}
2124: Mandelbaum, R., Seljak, U., Cool, R.~J., Blanton, M., Hirata, C.~M.,
2125: \& Brinkmann, J.\ 2006, \mnras, 372, 758
2126: % Density profiles of galaxy groups and clusters from SDSS galaxy-galaxy
2127: % weak lensing
2128:
2129: \bibitem[Masjedi et al.(2006)]{Masjedi06}
2130: Masjedi, M., et al.\ 2006, \apj, 644, 54
2131: % Very Small Scale Clustering and Merger Rate of Luminous Red Galaxies
2132:
2133: \bibitem[More et al.(2009)]{More09}
2134: More, S., van den Bosch, F.~C., Cacciato, M., Mo, H.~J., Yang, X., \& Li, R.\
2135: 2009, \mnras, 392, 801
2136: %% Satellite Kinematics II: The Halo Mass-Luminosity Relation of
2137: %% Central Galaxies in SDSS
2138:
2139: \bibitem[Moustakas \& Somerville(2002)]{Moustakas02}
2140: Moustakas, L.\ A., \& Somerville, R.\ S.\ 2002, \apj, 577, 1
2141: %% The Masses, Ancestors, and Descendants of Extremely Red Objects:
2142: %% Constraints from Spatial Clustering (HOD model of EROs)
2143:
2144: \bibitem[Navarro, Frenk, \& White(1995)]{Navarro95}
2145: Navarro, J.\ F., Frenk, C.\ S., \& White, S.\ D.\ M. 1995, \mnras, 275, 56
2146: %NFW profile 1
2147:
2148: \bibitem[Navarro, Frenk, \& White(1996)]{Navarro96}
2149: Navarro, J.\ F., Frenk, C.\ S., \& White, S.\ D.\ M. 1996, \apj, 462, 563
2150: %NFW profile 2
2151:
2152: \bibitem[Navarro, Frenk, \& White(1997)]{Navarro97}
2153: Navarro, J.\ F., Frenk, C.\ S., \& White, S.\ D.\ M. 1997, \apj, 490, 493
2154: %NFW profile 3
2155:
2156: \bibitem[Norberg et al.(2001)]{norberg01}
2157: Norberg, P., et al. 2001, \mnras, 328, 64
2158: % 2dF luminosity dependence of galaxy clustering
2159:
2160: \bibitem[Norberg et al.(2002)]{norberg02}
2161: Norberg, P., et al. 2002, \mnras, 332, 827
2162: % 2dF galaxy clustering dependence on lum and spectral type
2163:
2164: %\bibitem[Norberg et al.(2009)}{norberg09}
2165: %Norberg, P., Baugh, C.\ M., Gaztanage, E., \& Croton, D.\ J. 2009,
2166: %\mnras, 396, 19
2167: % jackknife errors issues
2168:
2169: \bibitem[Ouchi et al.(2005)]{Ouchi05}
2170: Ouchi, M., et al.\ 2005, \apjl, 635, L117
2171: % Definitive Identification of the Transition between Small- and
2172: % Large-Scale Clustering for Lyman Break Galaxies
2173:
2174: \bibitem[Padmanabhan et al.(2009)]{Padmanabhan09}
2175: Padmanabhan, N., White, M., Norberg, P., \& Porciani, C.\ 2009, \mnras, 959
2176: % The real-space clustering of luminous red galaxies around z < 0.6
2177: % quasars in the Sloan Digital Sky Survey
2178:
2179: \bibitem[Park et al.(1994)]{park94}
2180: Park, C., Vogeley, M. S., Geller, M. J., \& Huchra, J. P. 1994,
2181: \apj, 431, 569
2182: % P(k), xi(r), and tests for luminosity bias in CfA2
2183:
2184: \bibitem[Peacock \& Smith(2000)]{Peacock00}
2185: Peacock, J. A., \& Smith, R. E. 2000, \mnras, 318, 1144
2186: % Halo bias model
2187:
2188: \bibitem[Percival et al.(2007)]{Percival07}
2189: Percival, W.~J., et al.\ 2007, \apj, 657, 645
2190: % The Shape of the Sloan Digital Sky Survey Data Release 5 Galaxy Power Spectrum
2191:
2192: \bibitem[Postman \& Geller(1984)]{postman84}
2193: Postman, M., \& Geller, M. J. 1984, \apj, 281, 95
2194: % morphology-density relation
2195:
2196: \bibitem[Press \& Schechter(1974)]{Press74}
2197: Press, W.\ H.\ \& Schechter, P.\ 1974, \apj, 187, 425
2198:
2199: \bibitem[Reid \& Spergel(2009)]{Reid09}
2200: Reid, B.~A., \& Spergel, D.~N.\ 2009, \apj, 698, 143
2201: % Constraining the Luminous Red Galaxy Halo Occupation Distribution
2202: % Using Counts-In-Cylinders
2203:
2204: \bibitem[Ross et al.(2008)]{Ross08}
2205: Ross, N.~P., Shanks, T., Cannon, R.~D., Wake, D.~A., Sharp, R.~G.,
2206: Croom, S.~M., \& Peacock, J.~A.\ 2008, \mnras, 387, 1323
2207: % Luminous Red Galaxy Clustering at z ~ 0.7 - First Results using AAOmega
2208:
2209: \bibitem[Schechter(1976)]{Schechter76}
2210: Schechter, P.\ 1976, \apj, 203, 297
2211: %% Schechter LF
2212:
2213: \bibitem[Scoccimarro et al.(2001)]{Scoccimarro01}
2214: Scoccimarro, R., Sheth, R. K., Hui, L., \& Jain, B. 2001, \apj, 546, 20
2215: % Halo bias model
2216:
2217: \bibitem[Scoccimarro \& Sheth(2002)]{Scoccimarro02}
2218: Scoccimarro, R., \& Sheth, R.~K.\ 2002, \mnras, 329, 629
2219: %PTHALOS
2220:
2221: \bibitem[Seljak(2000)]{Seljak00}
2222: Seljak, U. 2000, \mnras, 318, 203
2223: % Mass power spectrum from halo model
2224:
2225: \bibitem[Seo et al.(2008)]{Seo08}
2226: Seo, H.-J., Eisenstein, D.~J., \& Zehavi, I.\ 2008, \apj, 681, 998
2227: % Passive Evolution of Galaxy Clustering
2228:
2229: \bibitem[Sheldon et al.(2007)]{Sheldon07}
2230: Sheldon, E.~S., et al.\ 2007, ArXiv e-prints, 709, arXiv:0709.1162
2231: % Cross-correlation Weak Lensing of SDSS Galaxy Clusters III:
2232: % Mass-to-light Ratios
2233:
2234: \bibitem[Sheth \& Tormen(1999)]{Sheth99}
2235: Sheth, R.\ K.\ \& Tormen, G.\ 1999, \mnras, 308, 119
2236: %% halo mass function
2237:
2238: \bibitem[Simon et al.(2009)]{Simon09}
2239: Simon, P., Hetterscheidt, M., Wolf, C., Meisenheimer, K., Hildebrandt, H.,
2240: Schneider, P., Schirmer, M., \& Erben, T.\ 2008, arXiv:0805.3459
2241: % Relative clustering and the joint halo occupation distribution of
2242: % red-sequence and blue-cloud galaxies in COMBO-17
2243:
2244: \bibitem[Springel et al.(2005)]{Springel05}
2245: Springel, V., et al.\ 2005, \nat, 435, 629
2246: % Simulations of the formation, evolution and clustering
2247: % of galaxies and quasars
2248:
2249: \bibitem[Strauss et al.(2002)]{Strauss02}
2250: Strauss, M.~A., et al.\ 2002, \aj, 124, 1810
2251: % Spectroscopic Target Selection in the Sloan Digital Sky Survey:
2252: % The Main Galaxy Sample
2253:
2254: \bibitem[Tegmark et al.(2006)]{Tegmark06}
2255: Tegmark, M., et al.\ 2006, \prd, 74, 123507
2256: % Cosmological constraints from the SDSS luminous red galaxies
2257:
2258: \bibitem[Tinker et al.(2005)]{Tinker05}
2259: Tinker, J.~L., Weinberg, D.~H., Zheng, Z., \& Zehavi, I.\ 2005, \apj, 631, 41
2260: % halo bias factor
2261:
2262: \bibitem[Tinker et al.(2007)]{Tinker07}
2263: Tinker, J.~L., Norberg, P., Weinberg, D.~H., \& Warren, M.~S.\ 2007, \apj, 659, 877
2264: % On the Luminosity Dependence of the Galaxy Pairwise Velocity Dispersion
2265:
2266: \bibitem[Tinker et al.(2008)]{Tinker08}
2267: Tinker, J., Kravtsov, A.~V., Klypin, A., Abazajian, K., Warren, M., Yepes, G.,
2268: Gottl{\"o}ber, S., \& Holz, D.~E.\ 2008, \apj, 688, 709
2269: % SO halos
2270:
2271: \bibitem[Vale \& Ostriker(2004)]{Vale04}
2272: Vale, A., \& Ostriker, J.~P.\ 2004, \mnras, 353, 189
2273: % Linking halo mass to galaxy luminosity
2274:
2275: \bibitem[van den Bosch et al.(2003)]{Bosch03}
2276: van den Bosch, Frank, C., Yang, X.\ H., \& Mo, H.\ J.\ 2003, \mnras, 340, 77
2277: %% apply HOD to 2dF (with comparison to galaxy formation model)
2278:
2279: \bibitem[Wake et al.(2008)]{Wake08}
2280: Wake, D.~A., et al.\ 2008, \mnras, 387, 1045
2281: % The 2dF-SDSS LRG and QSO Survey:
2282: % evolution of the clustering of luminous red galaxies since z = 0.6
2283:
2284: \bibitem[White, Tully, \& Davis(1988)]{white88}
2285: White, S.\ D.\ M., Tully, R.\ B., \& Davis, M.\ 1988, \apjl, 333, L45
2286: % dependence of circular velocity on local density
2287:
2288: \bibitem[White et al.(2007)]{White07}
2289: White, M., Zheng, Z., Brown, M.~J.~I., Dey, A., \& Jannuzi, B.~T.\ 2007,
2290: \apjl, 655, L69
2291: % Evidence for Merging or Disruption of Red Galaxies from the Evolution of
2292: % Their Clustering
2293:
2294: \bibitem[Willmer, da Costa \& Pellegrini(1998)]{willmer98}
2295: Willmer, C. N. A., da Costa, L. N., \& Pellegrini, P. S. 1998, \aj, 115, 869
2296: % biasing in SSRS
2297:
2298: \bibitem[Yan, Madgwick, \& White(2003)]{Yan03}
2299: Yan, R., Madgwick, D.\ S., \& White, M.\ 2003, \apj, 598, 848
2300: %% CLF application to DEEP2 results (astro-ph/0307248)
2301:
2302: \bibitem[Yang, Mo, \& van den Bosch(2003)]{Yang03}
2303: Yang, X.\ H., Mo, H.\ J., \& van den Bosch, F.\ C.\ 2003, \mnras, 339, 1057
2304: % CLF HOD
2305:
2306: \bibitem[Yang et al.(2005)]{Yang05}
2307: Yang, X., Mo, H.~J., Jing, Y.~P., \& van den Bosch, F.~C.\ 2005, \mnras,
2308: 358, 217
2309: % Galaxy occupation statistics of dark matter haloes: observational results
2310:
2311: \bibitem[Yoo et al.(2006)]{Yoo06}
2312: Yoo, J., Tinker, J.~L., Weinberg, D.~H., Zheng, Z., Katz, N., \& Dav{\'e},
2313: R.\ 2006, \apj, 652, 26
2314: % From Galaxy-Galaxy Lensing to Cosmological Parameters
2315:
2316: \bibitem[Yoo et al.(2009)]{Yoo09}
2317: Yoo, J., Weinberg, D.~H., Tinker, J.~L., Zheng, Z., \& Warren, M.~S.\ 2009,
2318: \apj, 698, 967
2319: %% P(k) recovery
2320:
2321: \bibitem[York et al.(2000)]{York00}
2322: York, D.\ G., et al.\ 2000, \apj, 120, 1579
2323: %% SDSS
2324:
2325: \bibitem[Zehavi et al.(2002)]{zehavi02}
2326: Zehavi, I., Blanton, M.\ R., Frieman, J. A., Weinberg, D.\ H., Mo, H.\ J.,
2327: et al.\ 2002, \apj, 571, 172
2328: % early galaxy clustering paper
2329:
2330: \bibitem[Zehavi et al.(2004)]{Zehavi04}
2331: Zehavi, I., et al.\ 2004, \apj, 608, 16
2332: % On Departures from a Power Law in the Galaxy Correlation Function
2333:
2334: \bibitem[Zehavi et al.(2005a)]{Zehavi05a}
2335: Zehavi, I., et al.\ 2005, \apj, 621, 22
2336: % The Intermediate-Scale Clustering of Luminous Red Galaxies
2337:
2338: \bibitem[Zehavi et al.(2005b)]{Zehavi05b}
2339: Zehavi, I., et al.\ 2005, \apj, 630, 1
2340: % The Luminosity and Color Dependence of the Galaxy Correlation Function
2341:
2342: \bibitem[Zentner et al.(2005)]{Zentner05}
2343: Zentner, A.~R., Berlind, A.~A., Bullock, J.~S., Kravtsov, A.~V., \& Wechsler,
2344: R.~H.\ 2005, \apj, 624, 505
2345: % The Physics of Galaxy Clustering. I. A Model for Subhalo Populations
2346:
2347: \bibitem[Zheng et al.(2002)]{Zheng02}
2348: Zheng, Z., Tinker, J.\ L., Weinberg, D.\ H., \& Berlind, A.\ A.\ 2002, \apj,
2349: 575, 000
2350: % halo population and cosmology
2351:
2352: \bibitem[Zheng(2004)]{Zheng04}
2353: Zheng, Z.\ 2004, \apj, 610, 61
2354: %% high-z powerlaw departure (astro-ph/0307030)
2355:
2356: \bibitem[Zheng et al.(2005)]{Zheng05}
2357: Zheng, Z., et al.\ 2005, \apj, 633, 791
2358: % Theoretical Models of the Halo Occupation Distribution:
2359: % Separating Central and Satellite Galaxies
2360:
2361: \bibitem[Zheng \& Weinberg(2007)]{Zheng07a}
2362: Zheng, Z., \& Weinberg, D.~H.\ 2007, \apj, 659, 1
2363: % Breaking the degeneracies between cosmology and galaxy bias
2364:
2365: \bibitem[Zheng, Coil, \& Zehavi(2007)]{Zheng07}
2366: Zheng, Z., Coil, A.~L., \& Zehavi, I.\ 2007, \apj, 667, 760
2367: % HOD modeling of DEEP2 and SDSS galaxy clustering
2368:
2369: \bibitem[Zwicky et al.(1968)]{zwicky68}
2370: Zwicky, F., Herzog, E., Wild, P., Karpowicz, M., \& Kowal, C.,
2371: 1961-1968, {\it Catalog of Galaxies and of Clusters of Galaxies},
2372: Vols. 1-6, (Pasadena: California Institute of Technology)
2373:
2374: \end{thebibliography}{}
2375:
2376: \end{document}
2377: