0807.0001/ms.tex
1: \documentclass[useAMS,usenatbib]{mn2e}
2: \usepackage{epsfig}
3: \usepackage{color}
4: 
5: \title[{The Colours of Satellite Galaxies}]{The Colours of Satellite Galaxies in Groups and Clusters}
6: 
7: \author[A.~S.~Font et~al.]{A. S. Font$^{1}$\thanks{E-mail:
8: andreea.font@durham.ac.uk},  R.~G.~Bower$^{1}$, I.~G.~McCarthy$^{1}$,
9:   A.~J.~Benson$^{2}$, C.~S.~Frenk$^{1}$, 
10: \newauthor J.~C.~Helly$^{1}$, C.~G.~Lacey$^{1}$, C.~M.~Baugh$^{1}$, S.~Cole$^{1}$\\
11: $^{1}$Department of Physics, University of Durham, South Road, Durham DH1 3LE\\
12: $^{2}$Theoretical Astrophysics, Caltech, MC130-33, 1200 E. California Blvd., Pasadena CA 91125, USA}
13: \begin{document}
14: 
15: \date{Accepted 2008, July 8. Received 2008, June 7; in original form 2008 March 19}
16: 
17: \pagerange{\pageref{firstpage}--\pageref{lastpage}} \pubyear{2008}
18: 
19: \maketitle
20: 
21: \label{firstpage}
22: 
23: \begin{abstract}
24: Current models of galaxy formation predict satellite galaxies in groups and
25: clusters that are redder than observed. We investigate the effect on the colours
26: of satellite galaxies produced by the ram pressure stripping of their hot
27: gaseous atmospheres as the satellites orbit within their parent halo. We
28: incorporate a model of the stripping process based on detailed hydrodynamic
29: simulations within the Durham semi-analytic model of galaxy formation. The
30: simulations show that the environment in groups and clusters is less
31: aggressive than previously assumed. The main uncertainty in the model is
32: the treatment of gas expelled by supernovae. With reasonable
33: assumptions for the stripping of this material, we find that satellite
34: galaxies are able to retain a significant fraction of their hot gas for several
35: Gigayears, thereby replenishing their reservoirs of cold, star forming gas and
36: remaining blue for a relatively long period of time. A bimodal distribution of
37: galaxy colours, similar to that observed in SDSS data, is established and the
38: colours of the satellite galaxies are in good agreement with the data. In
39: addition, our model naturally accounts for the observed dependence of 
40: satellite colours on environment, from small groups to high mass clusters.
41: \end{abstract}
42: 
43: \begin{keywords}
44: galaxies: clusters: general; galaxies: evolution -- galaxies: 
45: fundamental parameters: colours -- galaxies: luminosity function
46: \end{keywords}
47: 
48: \section{Introduction}
49: 
50: Recent multiwavelength imaging with the Sloan Digital Sky 
51: survey (SDSS; \citealt{york00}) has convincingly demonstrated 
52: that the colour-magnitude distribution of galaxies is bimodal  
53: \citep{strateva01,hogg02,blanton03,baldry06}. Other galaxy 
54: properties, such as the star formation rates, disk-to-bulge 
55: ratios, concentrations, stellar surface mass densities and gas 
56: mass fractions also show bimodal distributions 
57: \citep{kauffmann03,brinchmann04,balogh04,hogg04,kannappan04}. 
58: Elucidating the origin of these distributions requires 
59: understanding whether they arise in situ or whether they 
60: are environmentally driven (i.e., the ``nature'' or ``nurture'' 
61: dichotomy).
62:  
63: While they are not the only means by which theorists attempt to 
64: tackle the problem of galaxy formation, at present semi-analytic 
65: models offer the best hope for understanding the origin 
66: of the bimodality. The level of sophistication of these models 
67: has grown rapidly in recent years and so too has their ability 
68: to match a wide variety of observational data  
69: \citep{kauffmann93,lacey93,cole94,kauffmann99,somerville99,cole00,benson02,benson03,baugh05,croton06,bower06}.  The recent addition of feedback from active galactic 
70: nuclei (AGN) in the semi-analytic models (e.g., 
71: \citealt{granato04,bower06,cattaneo06,croton06,menci06,kang06}) 
72: has provided an explanation for three well-known, yet 
73: puzzling observational results: (1) the steep cut-off at 
74: the bright end of the  galaxy luminosity function; (2) the fact that 
75: most massive galaxies today tend to be dominated by old, red stellar 
76: populations; and (3) the absence of classical cooling flows in 
77: the centres of massive X-ray clusters.  In addition, some of 
78: these models (e.g., that of \citealt{bower06}, hereafter 
79: B06, which we adopt as the baseline model in this paper) are 
80: also able to reproduce the evolution of the K-band luminosity 
81: and the galaxy stellar mass functions out to high redshift (see 
82: \citealt{baugh06} and references therein for a more in-depth 
83: discussion of the successes and limitations of semi-analytic models).
84: 
85: In terms of galaxy colours, semi-analytic models with AGN 
86: feedback are able to produce a clear bimodal separation of 
87: colours at high luminosities, match the slope of the 
88: red and blue sequences and explain the absence of massive 
89: blue galaxies. However, it has recently become clear that these 
90: models are unable to match the relative numbers of 
91: faint satellite galaxies on the red and blue sequences. 
92: Specifically, \citet{weinmann06b} and \citet{baldry06} have 
93: shown that the faint satellite galaxies in groups and clusters 
94: modelled semi-analytically are on average too red in comparison to 
95: satellite galaxies of similar luminosity in the SDSS sample. 
96: Using DEEP2 data, \citet{coil08} have shown that a similar 
97: problem may exist at higher redshifts  ($z \sim 0.7-0.9$), where 
98: the semi-analytic models predict a stronger clustering of red 
99: galaxies than is observed. These results indicate that 
100: one or more physical processes (either internal or 
101: environmentally-driven) that affect the ability of the galaxies 
102: to form stars are not yet accurately accounted for in the 
103: models.  
104: 
105: In the present study, we concentrate on the effect of 
106: environmental processes on the colours of galaxies, particularly 
107: on the role that the ram pressure stripping of the hot gas from 
108: galaxy haloes plays.  At present, the semi-analytic models 
109: outlined above treat the stripping of the hot gas from the 
110: haloes of galaxies in a crude fashion, by assuming that the 
111: gaseous halo is completely and instantaneously stripped as the 
112: system crosses the virial radius of the (more massive) parent 
113: system and becomes a satellite.  
114: A direct consequence of this stripping is that the only fuel that 
115: is available for star formation once the galaxy becomes a 
116: satellite is that which resided in a cold disk when the galaxy 
117: first fell into the halo.  As a result, the satellite galaxy experiences a 
118: sharp decline in its star formation rate and its stellar 
119: population becomes red over time.  This process of cutting off 
120: the supply of hot gas that would otherwise cool and replenish 
121: the reservoir of cold, star-forming gas, is sometimes referred 
122: to as ``strangulation'' or ``starvation''  
123: (\citealt{larson80,balogh00}).  It is clear, however, that the 
124: maximally-efficient stripping assumed by most current semi-analytic 
125: models is not realistic, especially in cases where the mass of 
126: the satellite is comparable to that of the parent system \citep{wang07}. 
127: 
128: In a very recent study, \citet{kang08} investigate the effect of modelling the
129: stripping of satellites by an exponential decay. They adopt a stripping
130: factor that is independent of the properties and orbit of the parent and
131: satellite haloes and vary this to achieve the best match to the
132: observations. However, although this empirical approach is a step
133: forward, one expects the efficiency of the stripping to depend on the
134: orbit of the satellite and on the structural properties of both the
135: satellite and parent systems. In contrast, the model we present here aims 
136: to implement a physically motivated description for the stripping
137: effect, taking into account the relevant properties of the system.
138: 
139: Recently, \citet{mccarthy08} carried out a large suite of high 
140: resolution hydrodynamic simulations of the stripping of the 
141: hot gaseous haloes of galaxies and presented a simple, 
142: physically-motivated model that describes the 
143: simulation results remarkably well. These authors concluded 
144: that, typically, galaxies are able to retain a significant 
145: fraction of their hot haloes for long periods of time 
146: following virial crossing.  The results of these 
147: simulations are in qualitative agreement with a recent 
148: {\it Chandra} X-ray survey of massive 
149: satellite galaxies in hot clusters by Sun et al.\ (2007), who 
150: found that most massive satellites do indeed have detectable 
151: hot haloes (see also \citealt{jeltema08}).  The implication of 
152: these results is that the stripping of the hot gaseous haloes is 
153: much less efficient than previously assumed.  Consequently, some 
154: replenishment of the cold star-forming reservoir (via cooling of 
155: the hot halo) in the satellite galaxies is expected to take place, 
156: this prolonging star formation and resulting in bluer satellite galaxies.
157: 
158: In the present study, we incorporate the prescription for ram 
159: pressure stripping of hot haloes of \citet{mccarthy08} in the 
160: Durham semi-analytic code for galaxy formation, {\small GALFORM}, 
161: and show that this improvement in environmental physics brings 
162: the colours of satellite galaxies into much better agreement with the 
163: observational data.
164: 
165: The layout of the paper is as follows:  In Section  
166: \ref{sec:model} we present the details of the implementation of 
167: the new ram pressure model in {\small GALFORM}. In Section 
168: \ref{sec:results} we present results for the fraction of blue 
169: galaxies in the new model (\S \ref{sec:bluefrac}), discuss 
170: the environmental signatures in the colours of galaxies 
171: (\S \ref{sec:environ}) and present predictions for the red and blue
172: luminosity functions at redshift $z=0.1$ for galaxies of different type 
173: (\S \ref{sec:lumfunc}). In Section \S \ref{sec:discuss} we summarize 
174: and discuss our findings.
175: 
176: \section[]{A Semi-analytic Model of Galaxy Formation with Ram 
177: Pressure Stripping of Hot Gaseous Haloes}
178: \label{sec:model}
179: 
180: \subsection{The B06 version of {\small GALFORM}}
181: \label{sec:B06}
182: 
183: Apart from the minor changes described below, we use the 
184: version of the {\small GALFORM} semi-analytic code described in 
185: B06.  This version of {\small GALFORM} makes use of halo 
186: merger histories extracted from the {\it Millennium Simulation} 
187: with the techniques of \citet{helly03} (see also 
188: \citealt{harker06}). The {\it Millennium Simulation} 
189: \citep{springel_etal05} was carried out by the Virgo Consortium 
190: and it is one of the largest simulations of the growth of 
191: structure in the $\Lambda$CDM cosmology to date, containing 
192: approximately 10 billion dark matter particles in a cubic 
193: volume of $(500 h^{-1}$ Mpc)$^3$ with a particle mass of $8.6 
194: \times 10^{8} \, M_{\odot}/h$. Throughout the paper we adopt the 
195: same cosmological model assumed in the Millennium simulation (and 
196: in the B06 model) and quote our results in terms of the Hubble variable 
197: $h = H_0/100$~km$^{_1}$ Mpc$^{-1}$.
198: 
199: In terms of baryonic physics, 
200: the version of {\small GALFORM} developed by B06 has the same basic 
201: structure as in \citet{cole00}, but with the addition of an 
202: improved treatment of gas cooling and a new scheme for AGN feedback.  
203: In the present study, most of the basic parameters of the code 
204: (including, e.g., the efficiencies of supernovae and AGN feedback 
205: and the timescales for star formation and dynamical friction) are the 
206: same as those adopted in B06. However, in order to achieve better 
207: agreement with the zero-point colours of the observed red and blue 
208: sequences, we increase the value of the yield to $p=0.04$, which is 
209: a factor of two higher than the ``standard'' solar value adopted by
210: B06. With this change, the $^{0.1}(g-r)$ colours are redder by
211: $0.1$~mag compared with those in the B06 model (for galaxies on the 
212: red sequence). It also improves the metallicity of the intracluster
213: medium (ICM): whereas in the standard B06 model the metallicity of the
214: ICM was too low, $Z_{ICM} \simeq 0.15 Z_{\odot}$, in the model with 
215: double the yields $Z_{ICM} \simeq 0.3 Z_{\odot}$ (Bower et al. 2008), 
216: a value which is in better agreement with the X-ray measurements 
217: (e.g., \citealt{baum05}).  We note that large
218: yields such as these have been used in semi-analytical models in the
219: past in order to get better agreement between the model predictions
220: for galaxy colours and the observations
221: (see \citealt{kauffmann98,delucia04}; but see also the discussion of
222: \citealt{cole00} about increasing yields and consistency with 
223: stellar evolution models).  
224:  
225: One further minor change from the version of B06 concerns the  
226: division between haloes in the rapid cooling and hydrostatic regimes. 
227: B06 adopted a sharp transition between these two regimes.  In particular, 
228: the BO6 model assumes that radio-mode feedback is only effective in 
229: hydrostatic haloes with
230: 
231: \begin{equation}
232: t_{\rm cool}(r_{\rm cool}) > \alpha_{\rm cool}^{-1} t_{\rm 
233: ff}(r_{\rm cool}),
234: \label{eq:acool}
235: \end{equation}
236: 
237: \noindent where $t_{\rm cool}$ and $r_{\rm cool}$ are the cooling time 
238: and radius, $t_{\rm ff}$ is the free-fall time (as defined by
239: \citealt{cole00}) and $\alpha_{\rm cool}$ is an adjustable parameter 
240: (set in this model to $0.7$) that rescales the freefall time of the
241:  halo and has the effect of controlling the position of the break in the 
242: luminosity function\footnote{Equation (2) of B06 is incorrect;  
243: $\alpha_{\rm cool}$ should be replaced by $\alpha_{\rm 
244: cool}^{-1}$.}.  If the above condition is satisfied, it is then 
245: determined whether or not the central AGN is able to inject 
246: sufficient power to offset the energy being radiated away in 
247: the cooling flow and, if the available AGN power is greater 
248: than the cooling luminosity, the cooling flow is assumed to 
249: be completely quenched.
250: 
251: This strict dichotomy in halo properties has the undesirable
252:  effect that small changes in, for example, the gas mass
253:  fraction of haloes near the transition can greatly
254:  affect the star formation rate in their central galaxy. This
255:  leads to a small population of rapidly forming galaxies at
256:  the bright end of the blue sequence.  In the present paper,
257:  we have refined the criterion to make the transition more gradual.
258:  To do this, we reduce the cooling rate in haloes for which
259:  the cooling radius and effective freefall radius [$r_{\rm cool}(t)$
260:  and $r_{\rm ff,eff}(t) \equiv r_{\rm ff}(\alpha_{\rm cool} t)$] are close
261:  to each other. Specifically, if
262: 
263: \begin{equation}
264:  |r_{\rm cool}-r_{\rm ff,eff}| \le 0.5 \epsilon_{\rm cool}^{-1}r_{\rm
265:    cool},
266: \end{equation}
267: 
268: \noindent the net cooling rate is reduced:
269: 
270: \begin{equation} 
271: \dot{m}_{\rm cool,eff} = \dot{m}_{\rm cool} \left( 0.5 +
272:  \epsilon_{\rm cool} [1-r_{\rm ff}/r_{\rm cool}] \right), 
273: \end{equation}
274: 
275: \noindent where $\dot{m}_{cool}$ is the cooling rate in the absence of AGN
276: feedback. This function leaves the cooling rate unchanged if the
277: inequality above is not satisfied. For large values of $\epsilon_{\rm cool}$ 
278: the inequality applies for only a narrow range of 
279: $(r_{\rm cool}-r_{\rm ff,eff})$, and the behaviour of the B06 model is 
280: maintained. Here we adopt $\epsilon_{\rm cool}=10$, so that the
281: transition is still quite sharp, and the luminosity function is little 
282: affected.  The effect on the colours of galaxies at the bright tip of the blue
283:  sequence is, however, noticeable: by suppressing gas cooling in
284:  these objects, the tip of the blue sequence becomes redder,
285:  tending to curl up towards the red sequence.  This provides a
286:  slightly better match to observational data in this part of the
287:  colour-magnitude diagram (see \S \ref{sec:bluefrac}).
288: 
289: With these basic parameters, we run a model with a complete and 
290: instantaneous ram pressure stripping of the hot gaseous haloes 
291: of satellites (the ``default'' model) and a model where the ram 
292: pressure stripping of the hot haloes of satellites is calculated 
293: using the prescription of  \citet{mccarthy08} (henceforth 
294: called the ``hot ram pressure model'').  
295: 
296: \subsection{Implementation of Ram Pressure Stripping}
297: \label{sec:ram}
298: 
299: Below we give a brief description of the \citet{mccarthy08} 
300: ram pressure stripping model and how it is incorporated into 
301: {\small GALFORM}. The McCarthy et al.\ model is analogous to the 
302: original formulation of \citet{gunn72} for the stripping of a 
303: face-on cold disk, except that it applies to a 
304: spherical distribution of hot gas.  Specifically, the hot 
305: gaseous halo of the satellite will be stripped if the ram 
306: pressure ($P_{\rm ram}$) exceeds the satellite's gravitational 
307: restoring force per unit area ($P_{\rm grav}$):
308: 
309: \begin{equation}
310: \label{eq:ram}
311: P_{\rm ram} \equiv \rho_{\rm gas,p} v_{\rm sat}^2 > P_{\rm grav} 
312: \equiv \alpha_{\rm rp} \frac{G M_{\rm tot,sat}(r) \rho_{\rm 
313: gas,sat}(r)}{r} \, , 
314: \end{equation}
315: 
316: \noindent where $\rho_{\rm gas,p}$ is the gas density of the 
317: parent halo, $v_{\rm sat}$ is the velocity of the satellite with 
318: respect to this medium, $M_{\rm tot,sat}(r)$ is the total mass of 
319: the satellite within radius $r$ and $\rho_{\rm gas,sat}(r)$ is 
320: the density of the satellite's hot halo at this radius.  The
321: coefficient $\alpha_{\rm rp}$ is a geometric constant of order 
322: unity; McCarthy et al.\ find that $\alpha_{\rm rp} 
323: \approx 2$ gives the best fit to their hydrodynamic simulations.
324: Note that equation (\ref{eq:ram}) has not introduced any new free 
325: parameters into the {\small GALFORM} semi-analytical model, as 
326: $\alpha_{\rm rp}$ has been tuned to match the results of hydrodynamic 
327: simulations.
328: 
329: The satellite-centric radius where $P_{\rm ram} = P_{\rm grav}$ 
330: is referred to as the stripping radius and McCarthy et al.\  
331: assume that within this radius the satellite's gaseous halo 
332: remains intact while all gas exterior to this radius is 
333: stripped on approximately a sound crossing time.  This 
334: physically simple model has been shown to match the results of 
335: their high resolution hydrodynamic simulations\footnote{This 
336: includes both Lagrangian smooth particle hydrodynamic 
337: simulations with the GADGET-2 code \citep{springel05} and 
338: Eulerian adaptive mesh refinement simulations with the FLASH code 
339: \citep{fryxell00}.} remarkably well for a wide range of orbits, 
340: mass ratios, and structural properties.
341: 
342: In our implementation of the hot ram pressure model we fix the 
343: stripping radius by setting the ram pressure to its maximum 
344: value, which occurs at the pericentre of the satellite's orbit, 
345: and the gas is stripped at the instant it crosses the virial radius. 
346: We adopt this simplification as, by default, {\small GALFORM} does 
347: not track the full orbital evolution of the satellite galaxies 
348: (but see \citealt{benson03})\footnote{The Millennium simulation
349:   provides full dynamical information for subhaloes until they are 
350: tidally disrupted. In view of the limited resolution of the simulation, 
351: however, we choose to calculate the merging timescale of subhaloes
352: using the standard \citet{chandrasekhar43} formula rather than
353: following the subhalo orbits explicitly, as explained in BO6.
354: Since we are interested mainly in the broad statistical distribution of galaxy
355: colours, our results should not be affected by
356: the loss of information about individual orbits.}. In reality, the physical size of 
357: the stripping radius is a function of time, as the magnitude of 
358: the ram pressure varies along a non-circular orbit.  By setting 
359: the ram pressure to its maximum value, we overestimate the amount 
360: of stripping that occurs between the time when the satellite 
361: crosses the virial radius of the parent halo and when it reaches 
362: pericentre for the first time. However, for typical orbits this is not 
363: unreasonable, since this timescale is generally a small 
364: fraction of the total amount of time that the satellite spends 
365: in orbit about the parent halo.  In addition, since the full 
366: orbital evolution of the satellites is not followed, we neglect 
367: the effects of tidal heating and tidal stripping of the satellite 
368: system.  In terms of the removal of hot halo gas, however,
369: ram pressure stripping is a more efficient mechanism than tidal 
370: stripping for the vast majority of satellites (i.e., the 
371: stripping radius is typically smaller than the satellite's tidal 
372: radius; see \citealt{mccarthy08}).
373: 
374: The pericentres and the velocities at pericentre (both of which 
375: are required to compute the maximum ram pressure along the 
376: orbit) of the satellites are calculated by assuming the 
377: satellites have the same 2-D joint radial ($v_r$) and 
378: tangential ($v_\theta$) velocity distribution of infalling 
379: substructure as measured by \citet{benson05} from a large suite 
380: of Virgo Consortium cosmological simulations. \citet{benson05} finds that the following functional form describes the 2-D 
381: distribution well:
382: 
383: \begin{equation}
384: f(v_{r},v_{\theta})=a_{1} v_{\theta}
385: \exp{[-a_{2}(v_{\theta}-a_{9})^{2}-b_{1}(v_{\theta})(v_{r}-b_{2}(v_{\theta}))^{2}]}
386: \, ,
387: \end{equation}
388: 
389: \noindent where $a_{1}$, $a_{2}$, $a_{9}$, $b_{1}(v_{\theta})$ and
390: $b_{2}(v_{\theta})$ are coefficients and functions tabulated in
391: \citet{benson05}. We assume that this distribution is independent 
392: of halo mass and redshift. For each satellite we randomly sample this 
393: distribution, extracting a radial and tangential velocity 
394: pair which, in turn, allows us to calculate the energy and 
395: angular momentum (per unit mass) of the orbit.  We compute 
396: the pericentre radius and velocity of the satellite at 
397: pericentre by assuming that the orbital energy and angular 
398: momentum are conserved and by treating the satellite as a 
399: point mass orbiting within a Navarro-Frenk-White (NFW) 
400: \citep{navarro96,navarro97} potential with the same total 
401: mass and concentration as the parent halo (see, e.g.,
402: \citealt{binney87}).  By assuming that the orbital energy and angular 
403: momentum are conserved, we are ignoring the dynamical 
404: friction force acting on the satellites.  However, in cases 
405: where the dynamical friction force is strong, the satellite will 
406: quickly sink to the centre of the parent halo and, in any case, 
407: this process is important for the most massive satellites which are 
408: less affected by ram pressure stripping.
409: 
410: By default, the version of {\small GALFORM} presented in B06 
411: assumes that the gas density profiles of the hot haloes of all systems 
412: (including satellite systems) follow a $\beta$-model 
413: \citep{cavaliere76,cavaliere78}, with $\beta = 2/3$ (which 
414: provides a reasonable match to the X-ray surface brightness 
415: profiles of massive groups and clusters; Jones \& Forman 1984), a 
416: fixed core radius $r_c = 0.1 R_{\rm vir}$, and a normalization 
417: $\rho_0$ that is set to yield the correct mass of hot gas 
418: within the virial radius.  We adopt the same distribution in 
419: the present study.  We note, however, that we have also 
420: experimented with NFW density profiles for the hot gas 
421: (for both the satellite and parent systems), and find that 
422: the resulting fraction of satellite galaxies that are blue is 
423: quite similar to the case of the $\beta$-model. This is likely 
424: to stem from the fact that, typically, the hot gas is only stripped down 
425: to intermediate radii, where the $\beta$-model and the NFW profile 
426: are similar.
427: 
428: In addition to a hot tenuous atmosphere, each halo has a
429: reservoir of cold gas that developed from the cooling of
430: the hot atmosphere and which can potentially form stars.
431: As stars are formed, some of this cold gas is re-heated by feedback from,
432: for example, supernovae winds.  Prior to stripping, the re-heated gas 
433: is assumed to follow the same spatial and thermodynamic distribution as the hot 
434: gas in the halo (as in B06). We therefore treat the initial stripping
435: of re-heated gas in the same fashion as the hot gas, and transfer the
436: same fraction (as computed with the ram pressure stripping algorithm 
437: described above) from the re-heated gas of the satellite to the
438: re-heated gas reservoir of the parent halo. We record the total mass
439: stripped from the halo as $M_{\rm strip}$ and the fraction of the hot halo
440: that is stripped as $f_{\rm strip}$. Although this procedure is simple,
441: it is worth noting that in reality the spatial and thermodynamic
442: properties of the re-heated gas could be quite complicated (e.g.,
443: multiphase) and that it need not be distributed in the same way as the 
444: hot gas phase.  It would be worth revisiting this issue once a better 
445: treatment of the re-heated gas is incorporated into the semi-analytic 
446: model (Benson et al., in prep.).
447: 
448: The cooling rate of the remaining, unstripped gas
449: is calculated by cooling only the gas within the stripping radius and
450: assuming that stripping does not alter the mean density of gas within this
451: radius. We implement this by giving the satellite a nominal hot gas 
452: mass $M_{\rm hot}' = M_{\rm hot} + M_{strip}$ (where $M_{\rm hot}$ is the
453: true hot gas content of the halo) and applying the same cooling algorithm
454: as that used for central galaxies (except limiting the maximum cooling radius to
455: $r_{\rm strip}$ rather than $R_{\rm vir}$). This step ensures self-consistency in 
456: the treatment of the gas cooling between stripped and unstripped
457: galaxies, and therefore that the colours of satellites are predicted correctly.
458: 
459:    So far, we have only been considering the stripping of gas which
460:     was in the hot or re-heated phases at the time the satellite halo
461:     reached pericentre in the parent halo for the first
462:     time. However, as long as star formation continues in the
463:     satellite galaxy, supernova feedback will continue to re-heat gas
464:     and eject it from the satellite galaxy. The colours of satellite
465:     galaxies turn out to be sensitive to how much of this ``secondary'' 
466:     re-heated gas is stripped. The numerical simulations of
467:     \citep{mccarthy08} do not provide any direct information about
468:     the stripping of this re-heated gas, since they only treat the
469:     stripping of the initial hot halo. One can imagine
470:     several possible scenarios for the fate of the re-heated gas: 
471:     we adopt a picture
472:     in which the the re-heated gas is ballistically ejected as relatively 
473:     cold material that subsequently mixes or evaporates to become part 
474:     of the hot halo. In isolated galaxies, the re-heated gas is put 
475:     back with the same radial distribution as for the original
476:     hot gas halo. In satellite galaxies, we can consider 
477:     two extreme cases for the treatment of the secondary re-heated gas. 
478:     Continuing to apply the initial stripping criterion to the satellite
479:     galaxy as it orbits suggests that the re-heated gas should be
480:     stripped by the same factor $f_{\rm strip}$ as it is ejected from the
481:     galaxy. However, applying this at every
482:     timestep is too extreme: most of the re-heating occurs during the 
483:     outer part of the galaxy's orbit where the ram-pressure force 
484:     is small. It might therefore be more appropriate to consider a
485:     second case
486:     where little of this re-heated gas is stripped by ram-pressure
487:     effects. Further numerical simulations are required to elucidate
488:     which of these scenarios is physically more realistic and to
489:     determine a suitable parameterization of the time averaged stripping
490:     rate. At this point, the second (minimal stripping) case seems 
491:     more appropriate since the
492:     typical orbital timescale exceeds several Gyr, and is comparable
493:     to the timescale for the mass-growth of the parent halo 
494:     (see below).
495: 
496:     To allow for these uncertainties, we adopt a partially 
497:     empirical approach, and model the time-dependence of the hot gas
498:     mass in the satellite halo after its first pericentre passage as:
499: 
500:     \begin{equation}
501:     \dot{M}_{\rm hot} = (1 - \epsilon_{\rm strip}f_{\rm strip})
502:     \frac{M_{\rm reheat}}{\tau_{\rm reheat}} - \dot{M}_{\rm cool}
503:     \end{equation}
504: 
505:     Here ${M}_{\rm hot}$ is the mass of hot gas
506:     available for cooling and $\dot{M}_{\rm cool}$ is the rate at
507:     which it cools.  $M_{\rm reheat}$ is the mass of gas which has
508:     been re-heated by supernovae but not yet returned to the hot
509:     phase. In the absence of ram-pressure stripping, this re-heated
510:     gas is assumed to return to the hot phase on a timescale
511:     $\tau_{\rm reheat} = t_{\rm dyn}/\alpha_{\rm reheat}$, as in B06,
512:     where the value $\alpha_{\rm reheat}=1.26$ was chosen to match
513:     the observed galaxy luminosity function. The effect of
514:     ram-pressure stripping is described by the the term
515:     $\epsilon_{\rm strip}f_{\rm strip}$, where $f_{\rm strip}$ is the
516:     stripping factor already calculated for the initial pericentre
517:     using equation~(4), and
518:     $\epsilon_{\rm strip}$ is a new parameter (representing the time
519:     averaged stripping rate after the initial pericentre) which we
520:     adjust to fit
521:     the observations. The first (maximal) stripping case corresponds to
522:     $\epsilon_{\rm strip}=1$ if the orbital timescale is much
523:     shorter than the re-heating timescale, and the
524:     second case (minimal stripping) to $\epsilon_{\rm strip}=0$. The
525:     gas which is stripped from each satellite is added to the hot gas
526:     component of the parent halo.
527: 
528:     We find that if we take $\epsilon_{\rm strip}=1$, then the
529:     colour distribution of satellite galaxies looks very similar to
530:     the default model in which there is complete stripping of the
531:     initial hot gas and of the re-heated gas, with all satellites lying 
532:     on the red
533:     sequence. For $0.2 \la \epsilon_{\rm strip}<1$, the satellite
534:     colour distribution looks similar to the case $\epsilon_{\rm
535:     strip}=1$. For $\epsilon_{\rm strip} \la 0.2$, a blue sequence
536:     appears for the satellites, which results in better agreement
537:     with the observed colour-magnitude distribution. Values in the
538:     range $0 \leq \epsilon_{\rm strip} \la 0.2$ result in similar
539:     colour-magnitude distributions. For this paper, we adopt the
540:     value $\epsilon_{\rm strip} =0.1$.
541: 
542: The stripping of satellites is also affected by the growth of the halo in
543: which the satellite is orbiting. If we did not allow for the effect of
544: halo growth, a small satellite accreted at high redshift would not feel
545: the increasing ram-pressure effect as the parent halo grows in mass,
546: perhaps becoming a galaxy cluster by the present-day. In order to take
547: this effect into account,  we assign the satellite galaxy new orbital 
548: parameters and derive a new stripping factor every time the halo doubles 
549: in mass compared to the initial stripping event. If this factor exceeds
550: that applied previously, additional hot and re-heated material is
551: removed from the galaxy. This process tends additionally to
552: suppress on-going star formation in the satellites of massive haloes.
553: 
554: An additional physical effect that may be relevant but that is not 
555: accounted for in our model is that the (ram + thermal) pressure force 
556: exerted on the hot halo of the satellite could, in turn, raise the 
557: pressure force exerted on the cold gas and potentially stimulate 
558: additional star formation \citep{bekki03}. This would tend to result in bluer 
559: colours for the satellite galaxies.  The magnitude of this effect is 
560: presently unclear, however, and needs to be quantified 
561: either observationally or with the aid of hydrodynamic 
562: simulations that accurately include the effects of cooling, star 
563: formation, and feedback.
564: 
565: 
566: \section{Results}
567: \label{sec:results}
568: 
569: \begin{figure}
570: \includegraphics[width=9.cm]{fig1.ps}
571: \caption{\label{fig:lf}{Comparison of the predicted galaxy
572: luminosity functions of the default model with the hot ram
573: pressure stripping model at redshift $z=0$.  The data points represent 
574: observations from the 2dF Galaxy Redshift Survey \citep{norberg02}.
575: Magnitudes are $b_J$ (Vega) in both the models and the 
576: observations. Models use volume limited samples and the smallest 
577: resolved haloes have luminosities $\simeq 0.03 L_*$ (equivalent to 
578: $M_{b_J}-5\log h = -15.7$). Thick lines correspond to all galaxies and 
579: thin lines to satellite galaxies only.}}
580: \end{figure}
581: 
582: Since it is plausible that including a better treatment of the 
583: ram pressure stripping  may have a fairly significant impact on 
584: the observable properties of the galaxies, we first check if the 
585: predicted galaxy luminosity function is significantly altered.
586: 
587: In Fig.\ \ref{fig:lf} we compare the predicted galaxy luminosity 
588: function for the default\footnote{Our default model luminosity function is
589: essentially the same as that of B06, even after the minor changes 
590: outlined in \S \ref{sec:ram}} and hot ram pressure stripping models at
591: redshift $z=0$. The figure shows that our improved treatment of the 
592: ram pressure does not significantly alter the luminosity 
593: function.  This is primarily because ram pressure stripping 
594: preferentially affects satellite galaxies, whereas the total 
595: luminosity function (satellites + centrals) is dominated by 
596: central galaxies (see \S \ref{sec:lumfunc}).  In fact, even the 
597: luminosity function of satellite galaxies alone is only affected by 
598: a small amount (see thin lines in Fig. \ \ref{fig:lf}), however,  as we show 
599: below, the distribution of satellite colours at fixed luminosity 
600: is significantly altered.
601: 
602: \subsection{The blue fraction of galaxies}
603: \label{sec:bluefrac}
604: 
605: \begin{figure*}
606: \includegraphics[width=15.cm]{fig2.ps}
607: \caption{\label{fig:cmd}{Colour-magnitude diagram (CMD) of 
608: $z =0.1$ galaxies in the default model (upper panels) and 
609: in the hot ram pressure stripping model (lower panels). The 
610: three panels in each row, from left to right, represent
611: the CMDs of all galaxies, satellites and
612: central galaxies, respectively. Magnitudes are SDSS (AB 
613: system) $^{0.1}g$ and $^{0.1}r$ at redshift $z=0.1$. The 
614: contours are spaced linearly in galaxy number density, starting 
615: from 500 per ($500 \ h^{-1}$ Mpc)$^3$ and increasing in levels 
616: of 500 per ($500 \ h^{-1}$ Mpc)$^3$. The solid line represents 
617: the colour cut adopted by \citet{weinmann06a} in order to separate blue and 
618: red galaxies in the SDSS. The dashed line represents an 
619: alternative colour cut (see text).}}
620: \end{figure*}
621: 
622: Fig.\ \ref{fig:cmd} shows the predicted colour-magnitude 
623: diagram (CMD) of present-day galaxies for the default model 
624: (upper panels) and for the hot ram pressure stripping model 
625: (lower panels).  The three panels in each row, from left to 
626: right, represent the CMDs of all galaxies, 
627: satellites only, and central galaxies only.  For a more 
628: convenient comparison with the SDSS data, the model predictions 
629: are output at redshift $z=0.1$ (which corresponds approximately  
630: to the median redshift of the SDSS sample) and the filters chosen
631: for analysis are $^{0.1}g$ and $^{0.1}r$ in the AB system. 
632: 
633: Fig.\ \ref{fig:cmd} shows, as expected, that changing the 
634: treatment of ram pressure stripping primarily affects the 
635: colours of satellite galaxies.  In particular, the 
636: distribution of $^{0.1}(g-r)$ colours at a fixed magnitude is 
637: broader for the hot ram pressure stripping model and a larger 
638: fraction of the satellites now have bluer colours.  The physical 
639: reason for this behaviour is simply that the retention of some 
640: of the hot gas in the halo of the satellite allows for the 
641: replenishment of the cold gas reservoir which, in turn, 
642: prolongs star formation in the satellite after its accretion 
643: onto the parent halo.  
644: Although the satellite galaxies eventually stop forming new
645: stars as they consume their gas reservoir, this process is much more 
646: protracted in the hot ram pressure model.  
647: As a result, the satellites cover a wider range of colours, filling 
648: in the region between the two main colour sequences.  The red sequence 
649: is also much less pronounced in the new model.
650: 
651: To quantify this change in the colours, we classify galaxies as 
652: being either `blue' or `red' by adopting the colour cut proposed 
653: by \citet{weinmann06a} for galaxies in the SDSS i.e.,
654: 
655: \begin{equation}
656: ^{0.1}(g - r) = 0.7 - 0.032 \ (^{0.1}M_r - 5 \log h + 16.5) \, ,
657: \label{eq:colorcut}
658: \end{equation}
659: 
660: This cut, which is represented in the panels of Fig.\ 
661: \ref{fig:cmd} by solid lines, isolates reasonably well the red 
662: and the blue sequences in both the default and hot ram pressure 
663: models (see the top left and bottom left panels). 
664: This is similar to what is observed in the colour-magnitude
665: diagram of SDSS galaxies (see Figure 1 of \citealt{weinmann06a}; also Figure 1 
666: of \citealt{weinmann06b}).  For the SDSS data, the cut adopted by 
667: Weinmann et al. tends to isolate the red sequence at its base defined 
668: at the high luminosity end (as opposed to following the minimum between the red 
669: and blue sequences).  This effect is well reproduced by the solid line 
670: in Fig.\ \ref{fig:cmd}.
671: 
672: We note, however, that adopting exactly the same colour cut in our 
673: models as in the observations is not necessarily the best choice. 
674: A drawback of using the Weinmann et al. colour cut to define blue and
675: red fractions is that the results are quite sensitive to the precise position
676: of the red sequence. For this reason, we also employ a second cut 
677: (indicated by the dashed line in Fig.\ \ref{fig:cmd}), which intersects
678: the model's bimodal contours roughly at their minimum. In the
679: following, we will use by default the \citet{weinmann06a} cut and,
680: for illustrative purposes, show also results with the alternative cut.
681: 
682: The left hand and middle panels of Fig.\ \ref{fig:fblue} show, 
683: for the default and hot ram pressure models respectively, the 
684: fraction of blue galaxies per $^{0.1}M_r$ magnitude for 
685: satellites (filled red circles), centrals (filled blue 
686: squares) and all types (filled green triangles) using 
687: the \citet{weinmann06a} cut.  Also shown is the fraction 
688: of blue satellites (open red circles) using the alternative 
689: colour cut.
690: 
691: Fig.\ \ref{fig:fblue} shows that the fraction of satellites that 
692: are blue in the hot ram pressure model is about twice as high as 
693: in the default model.  As discussed above, the actual fraction of blue 
694: galaxies depends on the colour cut adopted; however the relative 
695: difference in the blue fractions between the two models is largely 
696: independent of the cut. In particular, we find that using 
697: either the \citet{weinmann06a} or the alternative cut (or any 
698: in between these two), the fraction of satellite galaxies that are 
699: blue in the hot ram pressure model is approximately $2-2.5$ times 
700: larger than in the default model.
701: 
702: \begin{figure*}
703: \centering
704: \includegraphics[width=18.cm]{fig3.ps}
705: \caption{\label{fig:fblue}{The fraction of blue galaxies per
706: $^{0.1}M_r$  magnitude. The left panel shows the 
707: default model, the middle panel shows the hot ram pressure 
708: stripping model. For comparison, we plot in the right panel 
709: the fraction of blue galaxies in the SDSS data as derived by 
710: \citet{weinmann06b}. The blue fractions are shown separately 
711: for satellites (red circles), central galaxies (blue squares) 
712: and all types (green triangles). The empty symbols 
713: represent the fraction of blue satellites with the 
714: alternative colour cut (see the dashed line in Fig.\ \ref{fig:cmd}). 
715: For better comparison, the black dotted line in the middle panel 
716: reproduces the blue fraction of SDSS satellites from the right panel.
717: All magnitudes are k-corrected to redshift $0.1$.}}
718: \end{figure*}
719: 
720: \begin{figure*}
721: \centering
722: \includegraphics[width=15.cm]{fig4.ps}
723: \caption{\label{fig:fblue_lim}{The fraction of blue 
724: satellites per $^{0.1}M_r$ magnitude calculated using the Weinmann et al 
725: colour cut (eq. \ref{eq:colorcut}), in parent haloes of different mass.  
726: The hot ram pressure stripping model is shown in the 
727: left panel. Data from the SDSS group catalogue of \citet{weinmann06b} 
728: are shown in the right panel. In both panels magnitudes have been 
729: k-corrected out to redshift $0.1$.}}
730: \end{figure*}
731: 
732: How do the models fare in comparison to the observational data?  
733: In the right hand panel, we show the blue fraction of galaxies 
734: measured by \citet{weinmann06b} using an SDSS group catalogue 
735: (kindly provided by S. Weinmann). The fraction of satellite galaxies 
736: that are blue in the observations increases up to approximately $60\%$ 
737: at the faint magnitude end. Encouragingly, the newly implemented 
738: ram pressure model yields a  fraction of blue satellites that is in 
739: strikingly good agreement with the SDSS results.  The satellite 
740: blue fraction rises also to about $60$\% at the faint end and the 
741: overall trend in the blue fraction with luminosity is also 
742: close to that observed.  In contrast, the default model yields a 
743: fraction of blue satellite galaxies at most $\approx 30$\% 
744: (at faint magnitudes). This is similar to the fraction of blue 
745: satellites ($\sim 20$\%) reported for the \citet{croton06} 
746: semi-analytic model by \citet{weinmann06b} (see the bottom right 
747: hand panel of their figure 4). 
748: 
749: We also note that the fraction of bright central galaxies 
750: that are blue is in broad agreement with the observations. This 
751: is largely a consequence of including a prescription for 
752: AGN feedback into {\small GALFORM} which tends to halt star 
753: formation in the most massive galaxies.  Finally, we note that 
754: the semi-analytic model predicts many more faint central 
755: galaxies than observed in the SDSS group catalogue (independently of how the 
756: stripping of the hot gaseous haloes is treated). However, the lack of faint 
757: central galaxies is a selection effect of the group sample in the
758: SDSS: the source catalogue excludes groups for which no member is
759: brighter than $^{0.1}M_r= -19.5 +5\log h$, where $h=0.73$ 
760: \citep{weinmann06a}. Although we do not attempt to reproduce the 
761: mock groups catalogue in detail here, including this selection criterion 
762: tends to makes the satellite blue fraction shown in Fig. \ref{fig:fblue} 
763: greater by 0.1, further improving the match to the observations.
764: 
765: 
766: \subsection{Environmental dependence of colours}
767: \label{sec:environ}
768: 
769: Ram pressure stripping is an environmentally-driven process, 
770: hence one should expect variations in the fraction of blue 
771: satellites as a function of parent halo mass.  The left hand 
772: panel of Fig.\ \ref{fig:fblue_lim} shows the fraction of 
773: blue satellites in the hot ram pressure model per $^{0.1}M_r$ 
774: bin, calculated using the Weinmann et al. colour cut
775: (eq. \ref{eq:colorcut}), for different parent haloes mass 
776: ranges.  The lower masses, $10^{12} < M_{\rm tot,parent} < 10^{14} \, M_{\odot}$, 
777: correspond to group- type environments, whereas the larger 
778: masses,  $M_{\rm tot,parent}>10^{14} \, M_{\odot}$ correspond 
779: to massive clusters. For comparison, in the right 
780: hand panel we plot the blue fractions of satellites in the SDSS 
781: group catalogue of \citet{weinmann06b} split into the same 
782: parent halo mass bins.  
783: 
784: Reasonably good agreement with the data is obtained.  The new model
785: reproduces the overall trend that satellites of a given magnitude are
786: bluer if they reside in less massive systems.  The physical explanation
787: for this behaviour is due to the slighthy higher density of the hot gas in
788: more massive systems (because lower mass systems, on average, have
789: converted a larger fraction of their baryons into stars) and to the higher
790: orbital velocities of satellites in more massive host haloes.  The
791: latter effect is the dominant one, especially since the ram pressure
792: scales with the square of the satellite's velocity.  In particular,
793: if we ignore the weak dependence of hot gas density on halo mass, the
794: ram pressure scales simply as $P_{\rm ram} \propto v_{\rm sat}^2$.
795: Typically, $v_{\rm sat}$ is of order the virial circular velocity of the
796: parent halo, implying the ram pressure will scale roughly as $P_{\rm ram}
797: \propto M_{tot,p}^{2/3}$.  As a result, stripping is more extensive in
798: more massive parent systems, reducing the fuel supply for star formation
799: in satellites.
800: 
801: 
802: The model shows a slight increase in the fraction of 
803: blue satellite galaxies towards the bright magnitude end, i.e. the ``bump''
804: at $ -19 >^{0.1}M_{r}> -21$.  As Fig. \ \ref{fig:fblue_lim} shows, this 
805: effect is more pronounced for bright galaxies residing in low mass groups, 
806: $10^{12} < M_{\rm tot,parent} < 10^{13} \, M_{\odot}$, and likely reflects
807: a limitation of the way in which the transition between rapid cooling and
808: hydrostatic regimes is currently treated in the code (see \S \ref{sec:B06}).
809: 
810: The direct comparison between the halo masses in the
811: models and those inferred from the observations should be treated 
812: with some caution.  In the case of a theoretical model, the true mass
813: of the parent halo is known precisely, however in the case of the 
814: observational data one must make use of a mass proxy (e.g.,
815: \citealt{eke04}). In the particular case of the SDSS data, \citet{weinmann06b} 
816: use an empirical relationship between the total optical luminosity 
817: of a system and its mass.  The semi-analytic models, however, show 
818: considerable scatter in the relationship between these quantities.  
819: Ultimately, one would like to construct a mock survey from the theoretical 
820: models with similar characteristics to the SDSS catalogue, and
821: analyze the mock data the same way as the real data.  This is beyond 
822: the scope of the present paper and will be addressed in a future 
823: study.
824: 
825: We also note that similar results have been obtained from studies 
826: that attempt to distinguish low mass from high mass environments 
827: using other mass proxies. For example, \citet{hogg04} find that 
828: bluer galaxies in the local Universe typically reside in low galaxy 
829: density environments, whereas redder galaxies tend to live in high 
830: galaxy density environments.  This is fully consistent with the results of 
831: \citet{weinmann06b} and our own model predictions if one makes the 
832: reasonable assumption that the projected surface density of 
833: galaxies increases with increasing parent halo mass (e.g., 
834: \citealt{gladders07}).
835: 
836: Lastly, the colours are expected to depend not only on the mass 
837: of the parent halo but also on the intrinsic properties of the 
838: satellite.  To help disentangle these two factors, we plot in 
839: Fig.\ \ref{fig:hist}  colour  histograms (in $^{0.1}(g-r)$) for 
840: satellite galaxies in both the default (dashed lines) and 
841: the hot ram pressure (solid lines) models, divided into 
842: different $^{0.1}M_r$ magnitude and parent halo mass bins.  
843: In the new ram pressure model, satellites brighter than 
844: $^{0.1}M_r \sim -20$ tend to be red, independent of the parent halo
845: mass (i.e., environment). This is because internal processes, 
846: specifically AGN feedback, are the dominant mechanism for quenching the 
847: star formation in these systems. Meanwhile, as expected, 
848: low mass satellites in the same model (corresponding to magnitudes 
849: fainter than $^{0.1}M_r \sim -20$) tend to be bluer in low mass parent 
850: haloes and redder in massive clusters. The shapes of the histograms and
851: the overall trends are qualitatively similar to observational results
852: \citep{balogh04,weinmann06a} $-$ but note that the observations include
853: both satellites and centrals. For example, the highest luminosity galaxies 
854: are always dominated by the narrow red peak, regardless of environment,
855: a weak tail of blue galaxies only becoming visible in the lowest density
856: regions. Since we plot only the satellite galaxies here, a distinct blue
857: sequence is not seen; rather, the satellite galaxies increasingly occupy
858: transition colours at low halo masses where star formation
859: has been partially suppressed but not completely extinguished.
860: The results of our model suggest that, at the low halo mass end,
861: environmental processes are as important as the intrinsic physical 
862: processes in determining the colour of satellite galaxies.
863: 
864: \begin{figure*}
865: \includegraphics[width=12.cm]{fig5.ps}
866: \caption{\label{fig:hist}{Fraction of satellite galaxies 
867: ($f[^{0.1}(g-r),M_{tot}]$) per $^{0.1}(g-r)$ colour bin, divided into 
868: different $^{0.1}M_r$ magnitude ranges and residing in parent haloes of 
869: different total mass. Results for the hot ram pressure
870: model are plotted with solid lines and for the default model with
871: dot-dashed lines. Magnitudes increase from left to right and parent halo 
872: mass increase from top towards bottom. Vertical lines correspond to the 
873: \citet{weinmann06a} colour cut (solid lines) and to the alternative
874: colour cut described in \S \ref{sec:bluefrac}(dotted lines).}}
875: \end{figure*}
876: 
877: \subsection{Luminosity functions of red and blue galaxies at $z=0.1$}
878: \label{sec:lumfunc}
879: 
880: 
881: \begin{figure*}
882: \centering
883: \includegraphics[width=18.cm]{fig6.ps}
884: \caption{\label{fig:lumfunc}{Luminosity functions in the $^{0.1}M_r$ band
885:     at redshift $z=0.1$. Panels include all galaxies (left), satellites
886:     (middle) and central galaxies (right). Solid lines correspond to the hot
887:     ram pressure model and dashed lines to the default model. 
888:     Black lines denote all galaxies in the (sub)sample, while red and 
889:     blue lines denote galaxies that are above and below the
890:     $^{0.1}(g-r)$ colour cut in equation \ref{eq:colorcut},
891:     respectively.  The green dotted lines in the middle and right hand 
892:    panels reproduce the total luminosity function of all galaxies for
893:    the hot ram pressure model (solid black line in left panel).
894:   }}
895: \label{fig:lumfunc}
896: 
897: \end{figure*}
898: 
899: As a starting point for future comparisons with SDSS and other data, we 
900: plot colour luminosity functions (LF) for the hot ram pressure model. 
901: Fig.~\ref{fig:lumfunc} shows the red and blue LFs in the $^{0.1}M_r$
902: band at redshift $z=0.1$, where colours are separated using the Weinmann et
903: al. cut.  Panels include all galaxies (left), satellites (middle) and 
904: central galaxies (right). Dashed lines represent the default model.
905: Current observational results for the luminosity functions are not
906: able to separate contributions of satellite and central galaxies
907: accurately over this range of luminosity due to the 
908: difficulty of robustly identifying faint central galaxies (However, see the
909:   recent HOD analysis of \citealt{brown08}). The comparison in 
910: Fig.~\ref{fig:lumfunc} highlights the origin of the variations of the 
911: blue fraction seen in Fig.~\ref{fig:fblue_lim}.
912: 
913: With the exception of the blue/red LF of satellites, the differences 
914: between the default and the hot ram pressure model are minor. Since the
915: centrals dominate in total number density, total LFs are similar for
916: both models (cf., discussion of Fig. \ref{fig:lf}).
917: When all galaxies are combined, the blue galaxy LF has a steeper 
918: faint end slope than the red counterpart, in agreement with observational
919: data (e.g., \citealt{baldry06}) \footnote{
920:     However, a study that appeared after our paper was submitted indicates that 
921:     the faint end slopes of the red and blue luminosity functions in
922:     the SDSS DR6 data are shallower than predicted by our model 
923:     \citep{montero08}. If confirmed, this may suggest that other
924:     physical processes unaccounted for in our model, 
925:     e.g. tidal stripping of stars \citep{henriques08}, may be
926:     responsible for the further flattening of the faint end slopes.}. By splitting the luminosity
927: function into central and satellite galaxies, we see that this is
928: driven by the rapidly increasing preponderance of blue central 
929: galaxies at faint magnitudes.  In contrast, the luminosity function of
930: central red galaxies drops by about three orders of magnitude between
931: $^{0.1}M_r$ of $-20$ and $-17$. This effect is driven by the AGN 
932: feedback in the model, and is independent of the stripping model.
933: 
934: The properties of satellites in the new model are much more dependent 
935: on the environmental physics included in the model. In the new model, 
936: the blue and red LFs reach similar values at the faint end 
937: (consistent with the results in Figure \ref{fig:fblue}
938: showing that at the faint end there are roughly similar numbers of red
939: and blue satellites).  The dramatic increase in the fraction of faint 
940: blue satellites can be seen by comparing the solid and dashed blue lines.
941: This is an alternative way of presenting the information in Fig.~\ref{fig:fblue_lim},
942: and underscores the importance of studying the environmental dependence of
943: galaxy properties in order to obtain a complete picture of galaxy formation
944: and evolution.
945: 
946: \section{Summary \& Discussion}
947: \label{sec:discuss}
948: 
949: Until now, most current semi-analytic models of galaxy 
950: formation have adopted a crude modelling of the ram pressure 
951: stripping of the hot gaseous halos of satellite galaxies.  In 
952: particular, they typically assume complete and instantaneous 
953: stripping of the hot gas halo when the galaxy first falls in, 
954: without regard for the galaxy's mass, orbit, or structural properties. 
955: This is at odds with results of hydrodynamic simulations, and also 
956: with recent X-ray observations of galaxies in massive clusters.  In 
957: the present study, we have improved the treatment of 
958: stripping by implementing the model of \citet{mccarthy08} 
959: (which has been shown to match simulations of ram pressure 
960: stripping to high accuracy) into the {\small GALFORM} 
961: semi-analytic model for galaxy formation. Although the initial
962: stripping event does not require us to add additional parameters 
963: to the model, subsequent stripping of the gas re-heated from the disk 
964: requires additional parameterization. We parameterize this process 
965: by assuming that most of the ejecta are retained in
966: the galaxy after the first pericentre passage.
967: 
968: We find that the newly implemented treatment of stripping leads to 
969: a significant improvement in the ability of the model to match 
970: the colours of satellite galaxies in the SDSS. The new model 
971: is also able to account for the environmental dependence of the 
972: colours of satellite galaxies.  Our results suggest that for the 
973: majority of satellite galaxies, this environmental 
974: process can be as important in modifying the galaxy colours as 
975: intrinsic processes, such as AGN or supernovae feedback, which 
976: operate within the satellites themselves. This finding is 
977: in broad agreement with previous studies that found that 
978: internal  processes that quench star formation do not seem 
979: capable of explaining the full range of colour and morphology 
980: data (e.g., \citealt{weinmann06a,baldry06}). 
981: 
982: Although we have only focused on the stripping of hot gaseous 
983: halos in the present study, other environmental processes may be 
984: relevant as well.  We now briefly review some of these and 
985: conclude that none of them appear to be as important as the ram pressure 
986: stripping of the hot haloes and the feedback already 
987: incorporated into our present model.
988: 
989: In cases where the ram pressure stripping of the hot halo is 
990: complete, some stripping of the cold gaseous disks may also 
991: occur \citep{gunn72,abadi99,quilis00}.  
992: However, \citet{okamoto03} and \citet{lanzoni05} have 
993: explored ram pressure stripping of disks in semi-analytic 
994: models and have concluded that the effect on the colours and star 
995: formation rates of satellite galaxies are minimal.  This most 
996: likely stems from the fact that disk stripping is only expected 
997: to be relevant for a minority of satellite galaxies whose 
998: orbits take them into the very centres of massive systems 
999: (e.g., \citealt{bruggen08}).  
1000: 
1001: Thermal evaporation of the hot gaseous haloes (and/or disks) 
1002: could also be relevant \citep{cowie77}, but observational studies of 
1003: bubbles and cold fronts in X-ray groups and clusters have placed strong 
1004: constraints on the efficiency of conduction, concluding that it 
1005: is strongly suppressed \citep{markevitch07,mcnamara07}.  
1006: Turbulent stripping via the generation of Kelvin-Helmholtz and 
1007: Rayleigh-Taylor instabilities at the interface between the hot 
1008: gaseous halo of the satellite and the parent system is possible, 
1009: but the timescale for this type of stripping is generally quite 
1010: long (see \citealt{mccarthy08}).  
1011: 
1012: Other possibly relevant processes include viscous stripping 
1013: (unfortunately at present the viscosity of the hot gas in groups 
1014: and clusters is poorly constrained; \citealt{mcnamara07}), tidal 
1015: effects on the gas as the result of the interaction with the gravitational 
1016: potential of the parent halo \citep{byrd90,merritt83} or with 
1017: the other satellites (i.e., mergers \citep{mihos95} 
1018: or harassment \citep{moore96}), and shock heating of the 
1019: satellite's hot gas as it falls in at transonic velocities.
1020: However, \citet{mccarthy08} have argued that ram pressure 
1021: stripping of the hot gaseous haloes is always more efficient 
1022: than tidal stripping by the parent halo's potential or 
1023: shock heating in cases where the satellite mass is less than 
1024: about 10\% of the parent halo's mass. 
1025: 
1026: In this paper we have concentrated only on comparisons with data
1027: at low redshifts ($z<0.1$). In a future study, we intend to 
1028: test the model at other redshifts and compare the redshift evolution of galaxy 
1029: colours with data from current deep surveys. Another important application of
1030: the model is the study of the clustering of galaxies as a function of different 
1031: physical quantities, such as colour. In addition to the large scale
1032: dependence driven by the relative importance of AGN feedback and
1033: ram pressure stripping as a function of halo mass, our model also predicts a 
1034: radial dependence within a halo driven by the variation in the strength
1035: of stripping with the galaxy orbit. The colour dependence of
1036: small-scale clustering will constrain the model and perhaps suggest a
1037: way to improve our treatment of the interaction between galaxies and their
1038: environment.
1039: 
1040: \vskip 6pt
1041: {\sf Galaxy catalogues for this model are available for download from this http URL: http://galaxy-catalogue.dur.ac.uk:8080/Millennium/}
1042: 
1043: \section*{Acknowledgments}
1044: 
1045: We thank Simone Weinmann for providing us the SDSS blue 
1046: fraction data in electronic format. We are grateful to 
1047: Simon White for
1048: a careful reading of the manuscript and for useful suggestions. We also
1049: acknowledge Michael Balogh, Michael Brown and David Wake for useful 
1050: discussions. ASF is supported by a STFC Fellowship at the Institute for
1051: Computational Cosmology in Durham. 
1052: RGB acknowledges the support of a STFC senior fellowship.  
1053: IGM acknowledges support from a postdoctoral 
1054: fellowship from the Natural Sciences and Engineering Research Council 
1055: (NSERC) of Canada. AJB acknowledges the support of the Gordon and Betty Moore 
1056: Foundation.  This work was supported in part by a STFC rolling 
1057: grant to Durham University.
1058: 
1059: 
1060: \begin{thebibliography}{99}
1061: 
1062: \bibitem[Abadi et al.(1999)]{abadi99} Abadi, M.~G., Moore, B., \& Bower, R.~G. \ 1999, MNRAS, 308, 947 
1063: 
1064: \bibitem[Baldry et al.(2004)]{baldry04} Baldry, I.~K., 
1065: Glazebrook, K., Brinkmann, J., Ivezi{\'c}, {\v Z}., Lupton, R.~H., Nichol, 
1066: R.~C., Szalay, A.~S.\ 2004, ApJ, 600, 681 
1067: 
1068: \bibitem[\protect\citeauthoryear{Baldry et al.}{2006}]{baldry06}
1069:   Baldry,~I.~K., Balogh,~M.~L., Bower,~R.~G., Glazebrook,~K.,
1070:   Nichol,~R.~C., Bamford,~S.~P., Budavari,~T. 2006, MNRAS, 373, 469
1071: 
1072: \bibitem[\protect\citeauthoryear{Balogh, Navarro \&
1073: 	 Morris}{2000}]{balogh00} Balogh,~M.~L., Navarro,~J.~F. \&
1074:   Morris,~S. 2000, ApJ, 540, 113
1075: 
1076: \bibitem[\protect\citeauthoryear{Balogh et al.}{2004}]{balogh04}
1077:   Balogh,~M.~L., Baldry, I.~K., Nichol,~R., Miller,~C., Bower, R., 
1078:   Glazebrook, K. 2004, ApJ, 615, 101
1079: 
1080: \bibitem[\protect\citeauthoryear{Baugh et al.}{2005}]{baugh05}
1081:   Baugh,~C.~M., Lacey,~C.~G., Frenk,~C.~S., Granato,~G.~L., Silva,~L.,
1082:   Bressan,~A., Benson,~A.~J., Cole,~S., 2005, MNRAS, 356, 1191
1083: 
1084: \bibitem[Baugh(2006)]{baugh06} Baugh, C.~M.\ 2006, Reports of 
1085: Progress in Physics, 69, 3101 
1086: 
1087: \bibitem[\protect\citeauthoryear{Baumgartner et al.}{2005}]{baum05} 
1088:   Baumgartner, W.~H., Loewenstein, M., Horner, D.~J., Mushotzky,
1089:   R.~F. \ 2005, ApJ, 620, 680
1090: 
1091: \bibitem[\protect\citeauthoryear{Bekki \& Couch}{2003}]{bekki03}
1092:   Bekki, K. \& Couch, W.~J. \ 2003, ApJL, 596, L13
1093: 
1094: \bibitem[\protect\citeauthoryear{Benson et al.}{2002}]{benson02}
1095:  Benson,~A.~J., Lacey,~C.~G., Baugh,~C.~M., Cole,~S., Frenk,~C.~S. 2002, MNRAS, 333,  156
1096: 
1097: \bibitem[\protect\citeauthoryear{Benson et al.}{2003}]{benson03} 
1098: Benson,~A.~J., Bower,~R.~G., Frenk,~C.~S., Lacey,~C.~G., Baugh,~C.~M., Cole,~S. 2003, MNRAS, 599, 38
1099: 
1100: \bibitem[\protect\citeauthoryear{Benson}{2005}]{benson05}
1101:   Benson,~A.~J. 2005, MNRAS, 358, 551 
1102: 
1103: \bibitem[\protect\citeauthoryear{Binney \& Tremaine}{1987}]{binney87}
1104:   Binney, J. \& Tremaine, S. 1987, ``Galactic Dynamics'', Princeton
1105:   University Press
1106: 
1107: \bibitem[\protect\citeauthoryear{Blanton et al.}{2003}]{blanton03}
1108:   Blanton,~M.~R., et al., 2003, ApJ, 592, 819
1109: 
1110: \bibitem[Blanton et al.(2005)]{blanton05} Blanton, M.~R., Lupton,
1111:   R.~H., Schlegel, D.~J., Strauss, M.~A., Brinkmann, J., Fukugita, M., Loveday, J.\ 2005, ApJ, 631, 208 
1112: 
1113: \bibitem[\protect\citeauthoryear{Bower et al.}{2006}]{bower06}
1114:   Bower,~R.~G., Benson,~A.~J., Malbon,~R., Helly,~J.~C., Frenk,~C.~S.,
1115:   Baugh,~C.~M., Cole,~S., Lacey,~C.~G., 2006, MNRAS, 370, 645
1116: 
1117: \bibitem[\protect\citeauthoryear{Bower et al.}{2008}]{bower08}
1118:   Bower,~R.~G., McCarthy, I.~G. \& Benson,~A.~J., 2008, MNRAS, submitted
1119: 
1120: \bibitem[\protect\citeauthoryear{Brinchmann et
1121: 	 al.}{2004}]{brinchmann04} Brinchmann,~J., Charlot, S., White, S. D. M., Tremonti, C., Kauffmann, G., Heckman, T., Brinkmann, J. 2004, MNRAS, 351, 1151
1122: 
1123: 
1124: \bibitem[\protect\citeauthoryear{Brown et al.}{2008}]{brown08}
1125:   Brown,~M.~J.~I. et al., ApJ, in press, ArXiv e-prints, arXiv:0804.2293v1
1126: 
1127: 
1128: \bibitem[\protect\citeauthoryear{Br{\"u}ggen \& De Lucia}{2008}]{bruggen08}
1129:   Br{\"u}ggen, M., \&  De Lucia, G.\ 2008, MNRAS, 383, 1336
1130: 
1131: \bibitem[Byrd \& Valtonen(1990)]{byrd90} Byrd, G., \& 
1132: Valtonen, M.\ 1990, ApJ, 350, 89 
1133: 
1134: \bibitem[\protect\citeauthoryear{Cattaneo et al.}{2006}]{cattaneo06}
1135:   Cattaneo,~A., Dekel,~A., Devriendt,~J., Guiderdoni,~B., Blaizot,~J.,
1136:   2006, MNRAS, 370, 1651
1137: 
1138: \bibitem[\protect\citeauthoryear{Chandrasekhar}{1943}]{chandrasekhar43}
1139:   Chandrasekhar  S., 1943, ApJ, 97, 255
1140: 
1141: \bibitem[\protect\citeauthoryear{Coil et al.}{2008}]{coil08}
1142:   Coil,~A.~L., et al. 2008, ApJ, 672, 153
1143: 
1144: \bibitem[\protect\citeauthoryear{Cole et al.}{1994}]{cole94} Cole,~S.,
1145:   Aragon-Salamanca, A., Frenk, C.~S., Navarro, J.~F. \& Zepf, S.~E. \ 1994, MNRAS, 271, 781
1146: 
1147: \bibitem[\protect\citeauthoryear{Cole et al.}{2000}]{cole00} Cole,~S.,
1148:   Lacey,~C.~G., Baugh,~C.~M., Frenk,~C.~S., 2000, MNRAS, 319, 168
1149: 
1150: \bibitem[Cowie \& Songaila(1977)]{cowie77} Cowie, L.~L., \& 
1151: Songaila, A.\ 1977, Nature, 266, 501 
1152: 
1153: \bibitem[\protect\citeauthoryear{Croton et al.}{2006}]{croton06}
1154:   Croton,~D.~J., Springel,~V., White,~S.~D.~M., De Lucia,~G.,
1155:   Frenk,~C.~S., Gao,~L., Jenkins,~A., Kauffmann,~G., Navarro,~J.~F.,
1156:   Yoshida,~N., 2006, MNRAS, 365, 11
1157: 
1158: \bibitem[\protect\citeauthoryear{Cavaliere \& Fusco-Femiano}{1976}]{cavaliere76}Cavaliere, A., \& Fusco-Femiano, R.\ 1976, Astronomy \& Astrophysics, 49, 137 
1159: 
1160: \bibitem[\protect\citeauthoryear{Cavaliere \& Fusco-Femiano}{1978}]{cavaliere78} Cavaliere, A., \& Fusco-Femiano, R.\ 1978, Astronomy \& Astrophysics, 70, 677
1161: 
1162: \bibitem[\protect\citeauthoryear{de Lucia et al.}{2004}]{delucia04} de
1163: Lucia,~G., Kauffmann, G. \& White, S.~D.~M. \ 2004, MNRAS, 349, 1101
1164: 
1165: \bibitem[\protect\citeauthoryear{Eke et al.}{2004}]{eke04} Eke,
1166:   V.~R. et al., 2004, MNRAS, 348, 866
1167: 
1168: \bibitem[Frei \& Gunn(1994)]{frei94} Frei, Z., \& Gunn, J.~E.\ 
1169: 1994, AJ, 108, 1476 
1170: 
1171: \bibitem[\protect\citeauthoryear{Fryxell et al.}{2000}]{fryxell00} Fryxell, B., et al.\ 
1172: 2000, ApJS, 131, 273 
1173: 
1174: \bibitem[\protect\citeauthoryear{Gladders et al.}{2007}]{gladders07}
1175:   Gladders, M.~D., Yee, H.~K.~C., Majumdar, S., Barrientos, 
1176:   L.~F., Hoekstra, H., Hall, P.~B., Infante, L.\ 2007, ApJ, 
1177:   655, 128 
1178: 
1179: \bibitem[Granato et al.(2004)]{granato04} Granato, G.~L., De 
1180: Zotti, G., Silva, L., Bressan, A., \& Danese, L.\ 2004, ApJ, 600, 580 
1181: 
1182: \bibitem[\protect\citeauthoryear{Gunn \& Gott}{1972}]{gunn72}
1183:   Gunn,~J.~E. \& Gott,~J.~R.~III, 1972, ApJ, 176, 1
1184: 
1185: \bibitem[\protect\citeauthoryear{Harker et al.}{2006}]{harker06}
1186: Harker, G., Cole, S., Helly, J., Frenk, C., Jenkins, A.\ 
1187: 2006, MNRAS, 367, 1039 
1188: 
1189: \bibitem[\protect\citeauthoryear{Helly et al.}{2003}]{helly03}
1190:   Helly,~J.~C., Cole,~S., Frenk,~C.~S., Baugh,~C.~M., Benson,~A.~J.,
1191:   Lacey,~C. 2003, MNRAS, 338, 903
1192: 
1193: \bibitem[\protect\citeauthoryear{Henriques et al.}{2008}]{henriques08}
1194:   Henriques, B.~M., Bertone, S., \& Thomas, P. 2008, MNRAS, 383, 1649
1195:  
1196: \bibitem[\protect\citeauthoryear{Hogg et al.}{2002}]{hogg02}
1197:   Hogg,~D.~W. et al., 2002, AJ, 124, 646
1198: 
1199: \bibitem[\protect\citeauthoryear{Hogg et al.}{2004}]{hogg04}
1200:   Hogg,~D.~W. et al., 2004, ApJ, 601, 29
1201: 
1202: \bibitem[\protect\citeauthoryear{Jeltema et al.}{2008}]{jeltema08}
1203:   Jeltema, T.~E, Binder, B., Mulchaey, J.~S. 2008, ApJ, 679, 1162
1204: 
1205: \bibitem[\protect\citeauthoryear{Jones \& Forman}{1984}]{jones84} Jones,~C. \& Forman,~W. 1984, ApJ, 276, 38
1206: 
1207: \bibitem[\protect\citeauthoryear{Kang et al.}{2006}]{kang06}
1208:   Kang, X., Jing, Y.~P., \& Silk, J.\ 2006, ApJ, 648, 820
1209:  
1210: \bibitem[\protect\citeauthoryear{Kang \& van den Bosch}{2008}]{kang08}
1211:   Kang, X. \& van den Bosch, F. ~C., \ 2008, ApJL, 676, L101
1212:  
1213: \bibitem[\protect\citeauthoryear{Kannappan}{2004}]{kannappan04}
1214:   Kannappan,~S.~J. 2004, ApJL, 611, L89
1215: 
1216: \bibitem[\protect\citeauthoryear{Kauffmann et al.}{1993}]{kauffmann93}
1217: Kauffmann, G., White, S.~D.~M. \& Guiderdoni, B.\ 1993, MNRAS, 264, 201
1218: 
1219: \bibitem[\protect\citeauthoryear{Kauffmann \&
1220:     Charlot}{1998}]{kauffmann98} 
1221: Kauffmann, G. \& {Charlot}, S. \ 1998, MNRAS, 294, 705
1222: 
1223: \bibitem[\protect\citeauthoryear{Kauffmann et al.}{1999}]{kauffmann99}
1224: Kauffmann, G., Colberg, J.~M., Diaferio, A., \& White, S.~D.~M.\ 1999, MNRAS, 303, 188 
1225: 
1226: \bibitem[\protect\citeauthoryear{Kauffmann et al.}{2003}]{kauffmann03}
1227:   Kauffmann,~G. et al. 2003, MNRAS, 341, 54
1228: 
1229: \bibitem[\protect\citeauthoryear{Kauffmann et al.}{2004}]{kauffmann04}
1230:   Kauffmann,~G., White,~S.~D.~M, Heckman,~T.~M., Menard,~B., Brinchmann,~J., Charlot,~S., Tremonti,~C., Brinkman,~J. 2004, MNRAS, 353, 713
1231: 
1232: \bibitem[\protect\citeauthoryear{Lacey et al.}{1993}]{lacey93}
1233:   Lacey, C., Guiderdoni, B., Rocca-Volmerange, B. \& Silk, J. \ 1993,
1234:   ApJ, 402, 15
1235: 
1236: \bibitem[\protect\citeauthoryear{Lanzoni et al.}{2005}]{lanzoni05}
1237:   Lanzoni, B., Guiderdoni, B., Mamon, G.~A., Devriendt, J., \& Hatton, S.\ 2005, MNRAS, 361, 369
1238: 
1239: \bibitem[\protect\citeauthoryear{Larson, Tinsley \& Caldwell}{1980}]{larson80}
1240:   Larson,~R.~B., Tinsley,~B.~M., Caldwell,~C.~N. 1980, ApJ, 237, 692
1241: 
1242: \bibitem[\protect\citeauthoryear{Markevitch \& Vikhlinin}{2007}]{markevitch07}
1243:   Markevitch, M., \& Vikhlinin, A.\ 2007, Physics Reports, 443, 1 
1244: 
1245: \bibitem[\protect\citeauthoryear{McCarthy et al.}{2008}]{mccarthy08}
1246:   McCarthy I.~G., Frenk,~C.~S., Font,~A.~S., Lacey,~C., Bower,~R.~G.,
1247:   Mitchell,~N.~L., Balogh,~M.~L., Theuns,~T. MNRAS, 2008, 383, 593
1248: 
1249: \bibitem[\protect\citeauthoryear{McNamara \& Nulsen}{2007}]{mcnamara07}
1250:   McNamara, B.~R., \& Nulsen, P.~E.~J.\ 2007, ARA\&A, 45, 117 
1251: 
1252: \bibitem[Menci et al.(2006)]{menci06} Menci, N., Fontana, A., Giallongo, E., Grazian, A., Salimbeni, S.\ 2006, ApJ, 647, 753 
1253: 
1254: \bibitem[Merritt(1983)]{merritt83} Merritt, D.\ 1983, ApJ, 264, 
1255: 24 
1256: 
1257: \bibitem[Mihos(1995)]{mihos95} Mihos, J.~C.\ 1995, ApJL, 438, 
1258: L75 
1259: 
1260: \bibitem[\protect\citeauthoryear{Moore et al.}{1996}]{moore96}
1261:   Moore,~B., Katz, N., Lake, G., Dressler, A., Oemler, A. 1996, Nature, 379, 613
1262: 
1263: \bibitem[\protect\citeauthoryear{Montero-Dorta \&
1264:     Prada}{2008}]{montero08} Montero-Dorta, A.~D., \& Prada, F. 2008, ArXiv e-prints, arXiv:0806.4930v1
1265: 
1266: \bibitem[Navarro et al.(1996)]{navarro96} Navarro, J.~F., Frenk, C.~S., \& White, S.~D.~M.\ 1996, ApJ, 462, 563 
1267: 
1268: \bibitem[Navarro et al.(1997)]{navarro97} Navarro, J.~F., Frenk, 
1269: C.~S., \& White, S.~D.~M.\ 1997, ApJ, 490, 493 
1270: 
1271: \bibitem[\protect\citeauthoryear{Norberg et al.}{2002}]{norberg02}
1272:   Norberg,~P. et~al., 2002, MNRAS, 336, 907
1273: 
1274: \bibitem[Nulsen(1982)]{nulsen82} Nulsen, P.~E.~J.\ 1982, MNRAS, 
1275: 198, 1007 
1276: 
1277: \bibitem[\protect\citeauthoryear{Okamoto \& Nagashima}{2003}]{okamoto03}
1278:   Okamoto, T., \& Nagashima, M.\ 2003, ApJ, 587, 500 
1279: 
1280: \bibitem[\protect\citeauthoryear{Quilis et al.}{2000}]{quilis00}
1281:   Quilis, V., Moore, B., \& Bower, R.\ 2000, Science, 288, 1617 
1282: 
1283: \bibitem[\protect\citeauthoryear{Somerville \& Primack}{1999}]{somerville99} 
1284:   Somerville, R.~S., \& Primack, J.~R.\ 1999, MNRAS, 310, 1087 
1285: 
1286: \bibitem[\protect\citeauthoryear{Springel}{2005}]{springel05}
1287:   Springel,~V. 2005, MNRAS, 364, 1105
1288: 
1289: \bibitem[\protect\citeauthoryear{Springel et al.}{2005}]{springel_etal05}
1290:   Springel,~V. et~al., 2005, Nature, 435, 629
1291: 
1292: \bibitem[\protect\citeauthoryear{Strateva et al.}{2001}]{strateva01}
1293:   Strateva,~I. et al., 2001, AJ, 122, 1861
1294: 
1295: \bibitem[\protect\citeauthoryear{Sun et al}{2007}]{sun07} Sun,~M.,
1296:   Jones,~C., Forman,~W, Vikhlinin,~A., Donahue,~M., Voit,~M. 2007, ApJ,
1297:   657, 197
1298: 
1299: \bibitem[Treu et al.(2003)]{treu03} Treu, T., Ellis, R.~S., 
1300: Kneib, J.-P., Dressler, A., Smail, I., Czoske, O., Oemler, A., Natarajan, P.\ 2003, ApJ, 591, 53 
1301: 
1302: \bibitem[\protect\citeauthoryear{Wang et al.}{2007}]{wang07} Wang, L., Li, C., Kauffmann, G.,
1303: \& De Lucia, G., 2007, MNRAS, 377, 1419
1304: 
1305: \bibitem[\protect\citeauthoryear{Weinmann et al.}{2006a}]{weinmann06a}
1306:   Weinmann~S.~M., van den Bosch,~F.~C., Yang,~X., Mo,~H.~J., 2006a,
1307:   MNRAS, 366, 2
1308: 
1309: \bibitem[\protect\citeauthoryear{Weinmann et al.}{2006b}]{weinmann06b}
1310:   Weinmann~S.~M., van den Bosch,~F.~C., Yang,~X., Mo,~H.~J.,
1311:   Croton,~D.~J., Moore,~B., 2006b, 372, 1161
1312: 
1313: \bibitem[\protect\citeauthoryear{York et al.}{2000}]{york00} York, D.~G. et al. \ 2000, AJ, 120, 1579
1314: 
1315: \end{thebibliography}
1316: 
1317: \end{document}
1318: