astro-ph0403629/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2: \shorttitle{Ellipticals from Self-Consistent Simulations}
3: \shortauthors{S\'aiz et al.}
4: 
5: %\documentstyle[12pt,aasms4,epsfig]{article} 
6: \received{2003 September 5}
7: \begin{document}
8:  
9: \title{Ellipticals at $z=0$ from Self-Consistent Hydrodynamical
10: Simulations: Comparison with SDSS Structural and Kinematical Data
11: }
12:  
13: \author{A. S\'aiz$^1$,   R. Dom\'{\i}nguez-Tenreiro
14: \footnote{ Dpt. F\'{\i}sica Te\'orica C-XI,
15: Universidad Aut\'onoma de Madrid, 
16: E-28049 Cantoblanco,
17: Madrid,
18: Spain;
19: e-mail: alejandro.saiz@uam.es, rosa.dominguez@uam.es}\hspace{0.3cm}
20: and
21: A. Serna
22: \footnote{Dpt. F\'{\i}sica y A.C., Universidad Miguel Hern\'andez,
23: E-03206 Elche, Alicante, Spain;
24: e-mail: arturo.serna@umh.es}}
25: 
26: 
27: \begin{abstract}
28:    We present   results of an  analysis
29:   of the structural and kinematical
30: properties of a sample of elliptical-like objects (ELOs)
31:  identified  in four hydrodynamical,   self-consistent
32:       simulations run with the DEVA code (Serna et al. 2003).
33: Star formation has been implemented in the code through a simple
34: phenomenological parameterization, that takes into account
35: stellar physics processes only implicitly through the values 
36: of a threshold gas density, $\rho_{\rm g,thres}$,
37: and an efficiency parameter, $c_*$.
38:   The four simulations operate in the context of a
39: $\Lambda$CDM cosmological model consistent with observations,
40: resolve ELO mass assembly at scales up to $\simeq$ 2 kpc,
41: and differ in the values of their star formation parameters.
42:  Stellar masses,
43: projected half-mass radii and central l.o.s. velocity dispersions,
44: $\sigma_{\rm los, 0}$, have been measured on the ELO sample
45: and their  values compared with data 
46: from the Sloan digital sky survey.
47: For the first time in self-consistent simulations,
48: a good degree of agreement has been shown, including
49: the Faber-Jackson  and  the $D_n - \sigma_{\rm los, 0}$ relations,
50: among others, but only 
51: when particular values of the  $\rho_{\rm g,thres}$ and $c_*$ parameters
52: are used. This demostrates the effect that the star formation parameterization
53: has on  the ELO mass distribution.
54: Additionally, our results suggest that  it is not strictly necessary,
55: at the scales resolved in this work,
56: to appeal to energy sources other than gravitational
57: (as for example   supernovae  feedback effects) to
58: account for the structure and kinematics of large ellipticals.
59:            
60: 
61: 
62: \end{abstract}
63: 
64: \keywords{ cosmology: theory - galaxies: ellipticals - galaxies: formation - galaxies: fundamental parameters - hydrodynamics
65: }
66: 
67: \section{INTRODUCTION}
68: 
69: One of the most challenging open problems in modern cosmology
70: is the origin of
71: the local galaxies of different Hubble types  we observe to-day.
72: Among them,  ellipticals
73: are the easiest to study.   
74: They form the  most homogeneous 
75: family and show the most precise regularities in the 
76: form of correlations among some pairs of their observable parameters.
77: The Sloan digital sky survey (SDSS, see York et al. 2000)
78: sample of early-type galaxies,
79: containing to date 9000 galaxies from different environments,
80: provides a new standard of 
81: reference for nearby elliptical  galaxies.
82: Its analysis confirmed correlations previously established,
83: such as those involving  structural and kinematical parameters
84: (the $L - \sigma_{\rm los, 0}$ or Faber-Jackson relation,
85: 1976; the surface-brightness - effective radius relation,
86: Kormendy 1977; the $D_n - \sigma_{\rm los, 0}$ relation, Dressler et al. 1987;
87: among others, see Bernardi et al. 2003a, 2003b, 2003c, 2003d,
88:   and references quoted
89: therein).
90: These  correlations, as well as the [$\alpha/$Fe] ratio trend 
91: with $\sigma_{\rm los, 0}$ (Jorgensen 1999), 
92: demand  short formation time-scales and old formation
93: ages for the bulk of the stellar populations of ellipticals.
94: These requirements are naturally met by one of the scenarios
95: proposed so-far to explain galaxy formation and evolution:
96: the so-called {\it monolithic collapse} scenario (ellipticals
97: would form  at high $z$ in a single burst of star formation,
98: and would passively evolve since then; Patridge \& Peebles
99: 1967; Larson 1974). 
100: The competing {\it merging  scenario} 
101: (galaxy mass assembly takes place gradually through repeated
102: mergers of smaller subunits; Toomre 1977; Kauffmann 1996)
103: meets some difficulties at
104: explaining the correlations above as well as other observations
105: on ellipticals, see Peebles 2002
106:  and Matteucci 2003 for  details and discussions.
107:  But the monolithic collapse scenario
108: does not recover all the currently available observations on ellipticals 
109: either.
110:  Such are, for example,  the range in ages their
111:  stellar populations
112: span in some cases or  their kinematical and dynamical peculiarities
113:  (Trager et al. 2000;
114: Menanteau, Abraham \& Ellis 2001), indicating
115: that
116:  an important fraction of
117: present-day ellipticals have recently experienced merger
118: events.
119: 
120: In order to reconcile all this observational background
121: within a formation scenario, it is preferable to study
122: galaxy assembly from simple physical principles and in
123: connection with the global cosmological model.
124: Self-consistent gravo-hydrodynamical simulations are a very
125: convenient tool to work out this problem 
126: (Navarro \& White 1994,
127: Tissera, Lambas \& Abadi 1997, Thacker \& Couchman 2000).
128: The method works as follows:
129: initial conditions are set at high
130: $z$ as a Montecarlo
131: realization of the field of primordial fluctuations
132: to a given cosmological model in a periodic, homogeneously sampled box;
133:  then the evolution of these fluctuations is numerically followed
134: up to $z =0$ by means of a computing code that solves the
135: N-body plus hydrodynamical
136: evolution equations.
137: In this way, the uncertainties resulting from prescriptions on
138: dynamics and gas  cooling and heating, present in other
139: methods such as semi-analytical ones (Kauffmann et al. 1999,
140: Mathis et al. 2002) can be removed, only star formation needs
141: further modelling. 
142: Individual galaxy-like objects (GLOs) naturally appear as an output
143: of the simulations,  
144:  no  prescriptions  are needed
145: as far as their mass  assembly processes 
146: are concerned. Moreover,
147: self-consistent simulations directly provide detailed information
148: on each individual GLO  at each
149: $ z$, namely  its  six dimensional phase space structure,
150:  as well as  the  temperature and age distributions of
151: its gaseous and stellar components, respectively.
152:  From this information, the 
153:  parameters  characterizing  each GLO
154:   can be estimated and compared
155: with observations (see, for example, S\'aiz et al. 2001,
156: concerning disk galaxies).
157: The first step in the program of studying the origin of galaxies
158: through self-consistent simulations, is to make sure that they
159: form GLO samples of different Hubble types that have counterparts
160: in the real local Universe. 
161: In particular, the possible simple correlations  involving  structural
162: and kinematical  parameters must be recovered.
163:  A detailed  analysis of this kind was
164: not yet available for ellipticals 
165:  (see previous work in Kobayashi 2002;
166: Sommer-Larsen, Gotz,  \& Portinari 2002;
167: Meza et al. 2003). 
168: 
169: We present in this paper the results of an analysis of
170: the structure and kinematics of a sample of
171:  elliptical-like objects (ELOs)
172: identified in four  self-consistent hydrodynamical simulations
173: run in the framework of a flat $\Lambda$CDM model consistent
174: with observations.
175: We have used 
176: DEVA,
177: an  AP3M-SPH code particularly designed to study
178: galaxy assembly in a cosmological context.
179: In DEVA, special attention has been paid that
180: the implementation of conservation laws (energy, entropy
181: and angular momentum) is as accurate as possible 
182:  (see Serna, Dom{\'{\i}}nguez-Tenreiro, \& S\'aiz, 2003 for details,
183: in particular for  a discussion on the observational implications of
184: violating some conservation laws).
185: Star formation (SF) processes have been implemented in the code
186: through a simple  parameterization,
187: similar to that first used by Katz (1992),  
188: that includes a  threshold gas density,
189: $\rho_{\rm g,thres}$ and an efficiency parameter, $c_{*}$, 
190: determining  the SF timescales
191:  according with a
192: Kennicutt-Schmidt law\footnote{See Elmegreen (2003) for a discussion
193: on the possibility that this law can be explained as a result of SF 
194: processes at the scale of molecular cloud cores, through an
195: interstellar medium  (ISM) gas structure whose density,
196: prior to SF, can be 
197: described by a log-normal probability distribution, as Wada \& Norman
198: (2001) found in their simulations
199: } (Kennicutt 1998).
200: 
201: Galaxy-like objects of different morphologies have been
202:  identified in the simulations 
203: (S\'aiz, Dom\'{\i}nguez-Tenreiro \& Serna 2002; S\'aiz 2003). 
204: The aim of this paper is to show that some of  those identified as 
205:  ELOs, have, at a structural and kinematical level,
206: counterparts in the local Universe, including parameter  correlations.
207: Data have been taken from the SDSS
208:  as analyzed by
209: Bernardi  et al.\ (2003a, 2003b),
210: Kauffmann et al. (2003a, 2003b) and
211:  Shen et al., 2003.
212: A brief account on ELO assembly is as follows: it mainly  occurs through
213: a multiclump  collapse at rather high $z$ 
214:   involving
215: many clumps; collapse 
216:  takes the clumps closer and closer along
217: filaments causing them to merge at very low relative angular
218: momentum and, consequently, at short timescales. This results
219: into  fast
220: SF bursts at high $z$ that transform most of the available gas
221: into stars.
222: The frequency of  head-on  mergers
223:  decreases with
224: $z$. ELO stellar populations are mostly old, and a trend exists 
225: with $\sigma_{\rm los, 0}$, as suggested by 
226: some observations (Thomas, Maraston \& Bender, 2002).     
227: 
228: \section{OBJECTS AT $z=0$}
229: 
230: We have analyzed ELOs identified in four different simulations,
231: namely S14, S16, S17 and S26, run within the same
232: global flat $\Lambda$ cosmological model, with $h=0.65$,
233: $\Omega_{\rm m} = 0.35$, $\Omega_{\rm b} = 0.06$.
234: The normalization parameter has been taken slightly high,
235: $\sigma_8 = 1.18$, as compared with the average fluctuations 
236: of 2dFGRS or SDSS  galaxies
237: (Lahav et al. 2002, Tegmark et al. 2003)
238: to mimic an active region of the Universe
239: (Evrard, Silk \& Szalay 1990).
240: The gravitational resolution is $\epsilon_{\rm g}$= 2.3 kpc.
241: Feedback effects from  stellar processes have  not been
242: explicitly included in the simulations, but the values of the
243: SF parameters we use mimic them in a sense
244: \footnote{Note that the role of stellar processes at kpc scales
245: is not yet clear,
246: as some authors argue that their  energy release drives
247: the structure of the ISM locally
248: at sub-kpc scales, while gas compression processes other
249: than stellar pressures, for example gravitational instabilities,
250: structure the gas prior to SF at scales $\ge$ kpc
251: (Elmegreen, 2003). Our ignorance of  sub-kpc (that is, subresolution)
252: scale processes, including SNe feedback effects, is circumvented 
253: by taking them implicitely  into
254: account  through the SF parameterization}.
255: S14, S16, and S26 share the same initial conditions 
256: (their  assembly histories are identical at halo scales)
257:  and differ only in the SF parameters
258: ($c_*$ = 0.1, 0.03 and 0.01, and $\rho_{\rm g,thres}$ = $6 \times 10^{-25},
259: 1.8 \times 10^{-24},6 \times 10^{-24}$ gr cm$^{-3}$ 
260: for S14, S16 and S26, respectively,
261: that is, SF becomes increasingly more difficult from S14 to S26), to test
262: the effects of SF parameterization to shape ELOs.
263: S17 is identical to S16, except that their initial conditions
264: differ, to test cosmic variance.
265: A standard cooling function has been used.
266: Each  of the four simulations started at a redshift $z_{\rm in} = 20$.
267: In any run, 64$^3$ dark matter and 64$^3$ baryon  particles,
268: with a mass of $1.29 \times 
269: 10^8$ and $2.67 \times 10^7 $M$_{\odot}$, respectively, have been used
270: to  homogeneously sample the density field in a periodic box of 10 Mpc side.
271: The mass function of galaxy-like objects formed in this box is consistent
272: with that of a small group environment
273: (Cuesta-Bolao \& Serna, private communication),
274:  where the efficiency of galaxy
275: formation per volume unit is higher than average in the universe, 
276: and, so, the fraction of cold baryons (i.e., cold gas and stars)
277: in the box is also higher than average.
278: ELOs have been identified as those objects having a prominent stellar
279: spheroidal component
280:  with hardly disks at all
281: (S\'aiz 2003; S\'aiz et al., in preparation). 
282: The 8 more massive objects identified at $z=0$
283: in S14, S16 and S17, and the 4 more massive in S26, fulfill this 
284: condition. The most (least) massive object in the sample
285: has, at $z=0$,
286:  59,300 (4,130) dark and 29,600 (2,610) baryon particles
287: within its virial radius. The   spin parameters of the ELO sample have 
288: an average value of $\bar{\lambda} = 0.033$. 
289: Their stellar components have ellipsoidal shapes, and, in most cases,
290: they are  dynamically relaxed systems.
291: 
292: 
293: The 3D stellar mass density profiles are steeper than those of dark matter (DM)
294: as a result
295: of  energy losses by gas particles as they cool and fall onto
296: the center of the configuration.
297: The size scales are closely correlated to the halo mass scale, 
298: $M_{\rm vir}$; they
299: also depend on the SF parameters,  because the easier it is
300: for a gas particle in a given proto-ELO  to be transformed into a
301: stellar particle, the lower the amount
302:  of energy it will radiate before being 
303: transformed, and, so, the lower the total dissipation 
304: ensuing the ELO formation.
305: Concerning ELO velocity distributions, their 3D structure
306: is such that stars and dark matter do not exhibit a global rotation;
307: the stellar velocity dispersion profiles are systematically lower 
308: than DM ones (as found by Loewenstein 2000 on theoretical grounds)
309: and they slightly decrease outwards;
310: the anisotropy parameter 
311:  is always positive,
312: that is, disordered energy lies preferentially on radial motions. 
313: Projected  mass and velocity distributions  
314: can be characterized by global
315: mass (i.e., luminosity), size and velocity dispersion
316: parameters that have been extensively observed in real 
317: ellipticals  and can be easily measured in simulated ELOs,
318: making the comparison  with observational data possible.
319: Physically, the mass parameter at the ELO  scale
320: is $M_{\rm bo}^{\rm cb}$, 
321: the total amount of cold baryons that have reached
322: the central $\sim $ few kpc volume of the halos,
323: forming an ELO.    
324: A fraction of these cold baryons
325: have turned into stars, depending on
326: the strength of the dynamical  activity in the
327: volume surrounding the  proto-ELO
328: at high $z$, and, also, on the efficiency of the SF algorithm. 
329: $M_{\rm bo}^{\rm star}$ is the stellar mass. 
330: Hereafter we will only refer to $M_{\rm bo}^{\rm star}$ as mass
331: scale, as it can be estimated from    
332:  luminosity data  through modelling (Kauffmann et al.
333: 2003a, for example).
334: Note that S14 objects have a slightly higher
335: stellar content 
336: than S26 objects, with those from S16 and S17 at intermediate
337: positions, as expected due to the values of their  SF parameters.
338: The projected  effective \emph{mass} radius,
339: $R_{\rm e, bo}^{\rm star}$,  is the radius
340:   enclosing a mass
341:  equal to  $M_{\rm bo}^{\rm star}/2$ as seen in projection.
342: Observationally,
343: a useful characterization of the
344: velocity dispersion of an elliptical galaxy is
345: provided by its central
346: line-of-sight velocity dispersion, $\sigma_{\rm los, 0}$. It
347:  has been measured in the ELO sample and found to be tightly
348: correlated with $M_{\rm vir}$.
349: 
350: Our next concern is related with the extent to which
351: ELOs in the sample as described by 
352: $M_{\rm bo}^{\rm star}$ as mass parameter,
353:  $R_{\rm e,bo}^{\rm star}$ as
354: projected size parameter, and $\sigma_{\rm los, 0}$
355: as velocity dispersion parameter, have observational
356: counterparts.      
357: On the observational side, Bernardi et al. (2003b)
358: provide maximum-likelihood estimates of the
359: parameters characterizing the joint probability distribution
360: (a trivariate gaussian) for  absolute luminosities, $M$,
361: (the logarithms of) intrinsic sizes, $R$, and central velocity
362: dispersions, $V$, namely, their mean values and the covariance
363: matrix.
364: Stellar masses of a sample of 10$^5$ SDSS galaxies
365: of different morphological types have been estimated by
366: Kauffmann et al. (2003a). Their results indicate that the
367:  stellar-mass-to-light ratio, $S$, can be taken to be constant in the
368: range of absolute luminosities $M < -21$, that is, for the more massive
369: or, equivalently, early-type galaxies in their sample
370: (see Kauffmann et al. 2003b).
371: The values of the logarithm 
372: of this ratio are  $S \simeq 0.53$ and $S \simeq 0.25$,
373: with dispersions $\sigma_S < 0.15$ and 0.1,  in the
374: $r$ and $z$ SDSS bands, respectively.
375: The independence of the value of $S$ with $M$ implies that
376: the pairwise correlation coefficients, $\rho_{\rm SM}$, can 
377: be taken $\simeq 0$ to a good accuracy.
378: Assuming, moreover, that the pairwise 
379: correlation coefficients $\rho_{\rm SR}$ and $   \rho_{\rm SV}$,
380: are also $\simeq 0$,
381: the pairwise concentration ellipses (and regression lines) can be drawn
382: for  the same data variables as those we have measured
383: on simulated ELOs (that is, stellar masses instead of absolute
384: luminosity, $R$ and $V$). This allows us to do  a more meaningful comparison.
385: 
386: In Figure~\ref{Fig1ObsLetter},  the consistency 
387: between SDSS data and ELOs as seen in the
388:  $\sigma_{\rm los, 0}$ versus $R_{\rm e, bo}^{\rm star}$ plot
389: is addressed.
390: Points are ELO measured values, with different symbols for
391: different simulations. The ellipse is the 2$\sigma$
392: concentration ellipse and the lines are its major
393: and minor axes for
394: the SDSS early-type galaxy  sample in the $z$ band (the other SDSS bands
395: lead to almost identical data diagrams).
396: We recall that the major axis corresponds to  the orthogonal mean square
397: regression line, o.m.s.r.l.,
398: for the two variables in the Figure, and that
399: the 1$\sigma$ (2$\sigma$) region from the o.m.s.r.l. can be drawn
400: by tracing its parallel lines through the middle (end) points of the
401: minor semiaxis; they have not been drawn in the Figure for the sake
402: of clarity.
403: We see that, except for 5 of the S14  ELOs,
404: the whole ELO sample lies within the 2$\sigma$ concentration
405: ellipse.
406: The effects of the different SF parameterizations are apparent
407: in this plot. We see that while those ELOs formed in simulations
408: where SF is easy (S14), tend to be too large for their
409: $\sigma_{\rm los, 0}$ and outside the 1$\sigma$ 
410: (or even the 2$\sigma$) o.m.s.r.l. region,
411: those formed in simulations where SF is difficult (S26), tend to be 
412: smaller for their velocity dispersion.
413: ELOs formed in S16 and S17 have intermediate sizes and agree with
414: SDSS data at 1$\sigma$ level, showing a power-law correlation
415: (the $D_n - \sigma_{\rm los, 0}$ relation). 
416: This behaviour with the SF parameterization is consistent with
417: the interpretation that the easier to form stars, the lower
418: the amount of energy lost by the gas before being turned into stars.
419: 
420: In Figure~\ref{Fig2ObsLetter} we plot the stellar half-mass radii,
421: $R_{\rm e, bo}^{\rm star}$, versus the stellar masses,
422: $M_{\rm bo}^{\rm star}$.
423: Data are the median values (points) and
424: 1 sigma dispersions (error bars) for the distribution of 
425:  S\'ersic (1968)  half-light radius in the $z$-band
426: as a function of stellar mass for SDSS early-type galaxies
427: (Shen et al. 2003, their Figure 11; data in the $r$-band do not
428: differ substantially).
429: We have also plot the 2$\sigma$ concentration ellipse drawn from
430:  Bernardi et al.'s
431: sample  as explained above.
432: The vertical line stands for
433: the  lower bound  of  early-type galaxy stellar masses,
434:   as found by Kauffmann
435: et al. 2003b.
436: We first note  that, when, say, 
437: $M_{\rm bo}^{\rm star} > 6 \times 10^{10} M_{\odot}$,
438: the consistency of data analysis by  Shen et al. 2003
439: and Bernardi et al. 2003 is remarkable.
440: Comparing Shen et al. 2003 results and our simulations
441: in this mass range, we see
442: that the agreement  
443: is good at a 1$\sigma$
444: level for S14 ELOs (even if they are larger than median 
445:  data values), as well as for the
446:   S17 and S16 ELO sample (smaller than median data values),
447:  except for three of them, that are  within  2$\sigma$ of Shen et al.'s
448:  data.
449: In this range of stellar mass, $R_{\rm e,bo}^{\rm star}$
450: and $M_{\rm bo}^{\rm star}$ show a power law correlation 
451: with a low dispersion and an exponent and
452: zero point consistent with
453: data. 
454: For lower masses, S17 and S16 ELOs are too small for their stellar
455: masses at
456: 1$\sigma$, as are S26 ELOs in any range of stellar mass. 
457: The effects of the SF parameterization on this plot are similar
458: to those on Figure~\ref{Fig1ObsLetter} we have just discussed.
459: 
460: Finally, we analyze the consistency between
461: simulations and data from the
462: $\sigma_{\rm los, 0}$ versus $M_{\rm bo}^{\rm star}$
463: plot (Figure~\ref{Fig3ObsLetter}).
464: We draw points for results of ELO measurements and
465: the 2$\sigma$ concentration ellipse for SDSS data. 
466: We see that most of the ELO sample lies within this  
467: ellipse, even if ELOs, and particularly so the less massive
468: ones, tend to have rather high stellar masses for their
469: $\sigma_{\rm los, 0}$ 
470: as compared with SDSS galaxies in the same $\sigma_{\rm los, 0}$
471: range.
472: Most ELOs with  $M_{\rm bo}^{\rm star} > 10^{11} M_{\odot}$,
473:  are inside the 1$\sigma$ o.m.s.r.l. region for SDSS data,
474:  and they show a power-law relation with slope
475: and zero point consistent with data and with low dispersion
476: (note that as $M^{\rm star}/L$ is constant, 
477: this is just the Faber-Jackson relation).
478: The dispersion is even lower for S17 and S16 ELOs with, say,
479:  $M_{\rm bo}^{\rm star} > 6 \times 10^{10} M_{\odot}$ when taken
480:  alone.
481:  Note that, even if most S14 objects
482:  have higher values of $M_{\rm bo}^{\rm star}$ than their
483:  S26 counterparts, with S16  and S17 objects at an intermediate position,
484:  all the
485:  objects follow the same relation, and particularly so the most massive ones,
486:   so that the different SF parameterizations
487:   have no important effects on this plot as far as the correlation
488:   is concerned.
489:   
490: 
491: 
492: \section{Discussion}
493: 
494: 
495: The degree of consistency reported here between sizes,
496: velocity dispersions and stellar masses of simulated
497: ELOs and SDSS data is very good. The agreement is
498: particularly outstanding when the simplicity of 
499: our working scheme is recalled: ELO assembly has been
500: simulated in the context of a cosmological model roughly
501: consistent with observations; Newton laws and hydrodynamical
502: equations have been integrated in this context,
503: with a standard cooling algorithm and a SF parameterization
504: through a Kennicutt-Schmidt-like law, 
505: containing our ignorance about its details at sub-kpc scales.
506: No further hypotheses to model  the assembly
507:  processes have been made.
508: Our results suggest that, at least  for the more massive objects,
509: say $\ge 6 \times 10^{10}$ M$_{\odot}$, it is not strictly
510: necessary to appeal to energy sources, {\it at the scales
511: resolved in this work}, that is, up to $\sim 2$ kpc, other than
512: gravitational to account for their sizes, velocity dispersions,
513: stellar masses and their correlations. 
514: This result is consistent with the idea that SNe explosions
515: are not relevant at these scales, but only at smaller scales,
516: to make shells and trigger star formation in molecular cloud cores.
517: Their effect would be accounted for in the SF parameters
518: (Elmegreen 2003).
519: 
520: This work strongly suggests that the SF parameterization
521: is a key ingredient to determine the 
522: compactness of elliptical galaxies.  
523: A good agreement has been found in the correlations
524: addressed here with SDSS data, but 
525: in the case of those involving sizes, only when  particular  values
526: of the SF  parameters are used. Our results 
527:  push the problem of elliptical 
528: galaxy formation from understanding 
529: their mass  assembly  at scales $>$  kpc in 
530: a cosmological context (Dom\'{\i}nguez-Tenreiro et al. 2003, in preparation), 
531: to understanding how SF was regulated
532: at high $z$ as the gas falls within
533: collapsing volumes, or other shock locations, so as to proceed 
534:  just with  the   efficiency necessary to  produce
535: the sizes observed in  to-day ellipticals.   
536: 
537: It is a pleasure to thank J. Silk and J. Sommer-Larsen for
538: useful information on the topics addressed in this paper.
539: This work was partially supported by the MCyT (Spain) through grant
540: AYA-0973 from the Programa Nacional de Astronom\'{\i}a 
541: y Astrof\'{\i}sica. We also thank the Centro de Computaci\'on
542: Cient\'{\i}fica (UAM, Spain) for computing facilities.
543: 
544: 
545: \clearpage
546: 
547: \begin{thebibliography}{}
548: 
549: 
550: \bibitem[]{575} Bernardi, M., et al. 2003a, AJ, 125, 1817; 2003b, AJ, 125, 1849; 2003c, AJ, 125, 1866
551: 
552: \bibitem[]{577} Bernardi, M., et al. 2003d, AJ, 125, 1882 
553: 
554: \bibitem[]{579}
555: Dressler, A., Lynden-Bell, D., Burstein, D., Davies, R.~L., Faber,  S.~M., Terlevich, R., \& Wegner, G. 1987, ApJ, 313, 42
556:   
557: \bibitem[]{582} Elmegreen, B. 2003, Ap\&SS, 284, 819
558: 
559: \bibitem[]{584} Evrard, A., Silk, J., \& Szalay, A.S. 1990, ApJ, 365, 13
560: 
561: \bibitem[]{586}
562: Faber, S.~M., \& Jackson, R.~E. 1976, \apj, 204, 668    
563: 
564: \bibitem[]{589} Jorgensen, I. 1999, MNRAS, 306, 607
565: 
566: \bibitem[]{591} Katz, N. 1992, ApJ, 391, 502
567: 
568: \bibitem[]{593} Kauffmann, G. 1996, MNRAS, 281, 475
569: 
570: \bibitem[]{595}
571: Kauffmann, G., Colberg, J.M., Diaferio, A., \& White, S.D.M., 1999, MNRAS, 303, 188
572: 
573: \bibitem[]{598}
574: Kauffmann, G., et al., 2003a, MNRAS, 341, 33
575:  
576: \bibitem[]{601}
577: Kauffmann, G., et al., 2003b, MNRAS, 341, 54
578: 
579: \bibitem[]{604} Kennicutt, R. 1998, ApJ, 498, 541
580: 
581: \bibitem[]{606} Kobayashi, C. 2002, Ap\&SS, 281, 301
582: 
583: \bibitem[]{608}
584: Kormendy, J. 1977, \apj, 218, 333        
585: 
586: \bibitem[]{611} Lahav, O., et al.  2002, MNRAS, 333, 961L 
587: 
588: \bibitem[]{613}
589: Larson, R.~B. 1975, \mnras, 164, 585 
590: 
591: \bibitem[]{616}
592: Loewenstein, M. 2000, \apj, 532, 17   
593: 
594: \bibitem[]{619}
595: Mathis, H., Lemson, G., Springel, V., Kauffmann, G., White, S.D.M.,
596: Eldar, A., \& Dekel, A. 2002, \mnras, 333, 739 
597: 
598: \bibitem[]{623} Matteucci, F. 2003, Ap\&SS, 284, 539
599: 
600: 
601: \bibitem[]{626} Menanteau, F., Abraham, R.G., \& Ellis, R.S. 2001, MNRAS, 322, 1
602: 
603: \bibitem[]{628} Meza, A., Navarro, J., Steinmetz, M., \& Eke, V.R., 2003,
604: ApJ, 590, 619 
605: 
606: \bibitem[]{631} Navarro, J.F., \& White, S.D.M., 1994, \mnras, 267, 401
607: 
608: \bibitem[]{633} Partridge, R.B., \& Peebles, P.J.E. 1967, ApJ, 147, 868
609: 
610: \bibitem[]{635} Peebles, P.J.E. 2002, in A New Era in Cosmology, ASP Conf., eds. N. Metcalf and T. Shanks  
611: 
612: \bibitem[]{637} S\'aiz, A., Dom{\'{\i}}nguez-Tenreiro, R., Tissera, P.B., \& Courteau, S. 2001, MNRAS, 325, 119
613: 
614: \bibitem[]{639} S\'aiz, A., Dom{\'{\i}}nguez-Tenreiro, R., \& Serna, A. 2002, Ap\&SS 281, 309
615: 
616: \bibitem[]{641} S\'aiz, A. 2003, PhD, Universidad Aut\'onoma de Madrid
617: 
618: \bibitem[]{643} Serna, A., Dom{\'{\i}}nguez-Tenreiro, R., \& S\'aiz, A. 2003, ApJ  in press, astro-ph/0307312 preprint
619: 
620: \bibitem[]{645} S\'ersic, J.L. 1968, Atlas de galaxias australes (C\'ordoba,
621: Argentina: Observatorio Astron\'omico)
622: 
623: \bibitem[]{648} Shen, S., et al. 2003, MNRAS, 343, 978
624: 
625: \bibitem[]{650} Sommer-Larsen, J., Gotz, M., \& Portinari, L. 2002, astro-ph/0204366 preprint
626: 
627: \bibitem[]{652} Tegmark, M.,  et al. 2003, astro-ph/0310723 preprint
628: 
629: \bibitem[]{654} Thacker, R.J., \& Couchman, H.M.P. 2000, ApJ, 545, 728
630: 
631: \bibitem[]{656} Thomas, D.,  Maraston, C., \& Bender, R. 2002, R.E. Schielicke (ed.), Reviews in Modern Astronomy, Vol. 15, Astronomische Gesellschaft, astro-ph/0202166 preprint  
632: 
633: \bibitem[]{658} Tissera, P.B., Lambas, D.G., \& Abadi, M.C. 19997, \mnras, 286, 384
634: 
635: \bibitem[]{660} Toomre, A. 1977, in The Evolution of Galaxies and Stellar Populations, eds. B. Tinsley \& R. Larson (New Have, CN: Yale Univ. Press)
636: 
637: \bibitem[]{662} Trager, S.C., Faber, S.M., Worthey, G., \& Gonz\'alez, J.J. 2000, AJ, 119, 1645
638: 
639: \bibitem[]{664} 
640: Wada, K., \& Norman, C.A., 2001, ApJ, 574, 172
641: 
642: \bibitem[]{}
643: York D.G., et al., 2000, AJ, 120, 1579
644: 
645: \end{thebibliography}
646: 
647: \clearpage
648: 
649: \begin{figure}
650: \centerline{\includegraphics[height=15cm]{f1.eps}}
651: \caption[]{
652: The central l.o.s. velocity dispersions of  ELOs in the   sample versus
653: their  stellar  half-mass radii. Different symbols stand for different
654: simulations.
655: We also draw the concentration ellipse (with their major and minor
656: axes) for the SDSS early-type galaxy sample from Bernardi et al. (2003b) in the $z$-band.
657: See text for more details 
658: }
659: \label{Fig1ObsLetter}
660: \end{figure}
661: 				
662: 
663: 
664: 
665: 
666: \clearpage
667: \begin{figure}
668:   \centerline{\includegraphics[height=15cm]{f2.eps}}
669:     \caption
670:       {
671:       The projected  stellar 
672:       half-mass radii versus 
673:       stellar masses.
674:     Filled diamonds with error bars correspond to
675:     median values of the S\'ersic half-light radii in the $z$-band
676:     and their 1 sigma dispersions (Shen et al. 2003).
677: We also draw the concentration ellipse 
678: for Bernardi et al.'s (2003b) sample in the $z$ band
679: and the lower limit for early-type-like galaxy stellar masses 
680: in the SDSS sample as found by
681: Kauffmann et al. (2003b, dotted vertical line).
682: See text for more details
683:       }
684:    \label{Fig2ObsLetter}
685: \end{figure}
686: 
687: 
688: \clearpage
689: 
690: \begin{figure}
691: \centerline{\includegraphics[height=15cm]{f3.eps}}
692: \caption{
693:    Same as  Figure 1 for the
694:    central l.o.s.\  velocity dispersions versus
695:    stellar masses. Vertical line is as in Figure 2
696:     }
697: \label{Fig3ObsLetter}
698: \end{figure}
699: 			       
700: 				
701: %\clearpage
702: 
703: 
704: 			       
705: 
706: 
707: \end{document}
708: 
709: 
710: 
711: 
712: