1: \documentclass[12pt,preprint]{aastex}

2: %\documentstyle[epsfig]{aastex}

3: %\received{}

4: %\revised{}

5: %\accepted{}

6: 

7: %\documentstyle[epsfig, emulateapj5]{aastex} \topmargin=0.5in

8: 

9: \def\msun{{\rm\,M_\odot}} 

10: \def\lsun{{\rm\,L_\odot}}

11: \def\zsun{{\rm\,Z_\odot}}

12: \newcommand{\etal}{et al.\ }

13: \newcommand{\kms}{\, {\rm km\, s}^{-1}}

14: \newcommand{\ikms}{(\kms)^{-1}}

15: \newcommand{\mpc}{\, {\rm Mpc}}

16: \newcommand{\kpc}{\, {\rm kpc}}

17: %\newcommand{\hmpc}{\, h^{-1} \mpc}

18: %\newcommand{\ihmpc}{(\hmpc)^{-1}}

19: %\newcommand{\hkpc}{\, h^{-1} \kpc}

20: \newcommand{\lya}{Ly$\alpha$ }

21: \newcommand{\lyaf}{Ly$\alpha$ forest}

22: \newcommand{\ch}{\bf change}

23: \newcommand{\gmo}{{\gamma-1}}

24: \newcommand{\bF}{\bar{F}}

25: \newcommand{\hi}{\mbox{H\,{\scriptsize I}\ }}

26: \newcommand{\heii}{\mbox{He\,{\scriptsize II}\ }}

27: \newcommand{\kpa}{k_\parallel}

28: \newcommand{\vk}{{\mathbf k}}

29: \newcommand{\df}{\delta_F}

30: \newcommand{\sF}{{F_s}}

31: \newcommand{\sdelta}{{\delta_s}}

32: \newcommand{\seta}{{\eta_s}}

33: \newcommand{\dt}{\Delta \theta}

34: \newcommand{\dv}{\Delta v}

35: \newcommand{\pa}{\parallel}

36: \newcommand{\pe}{\perp}

37: \newcommand{\dz}{\Delta z}

38: \def\h2{${\rm\,H_2}$}

39: \def\muG{\rm \mu{G}}

40: 

41: 

42: %%%%% apj5 emulate stuff %%%%%

43: \makeatletter \newenvironment{tablehere} {\def\@captype{table}} {}

44: \newenvironment{figurehere} {\def\@captype{figure}} {}

45: %%%%% apj5 emulate stuff %%%%%

46: 

47: \def\etal   {{et~al.}\ }

48: \def\ni{\noindent}        %No Indent%

49: \def\ub{\underbar}

50: \def\hi{\noindent \hangindent=2.5em}

51: \def\et{{\it et\thinspace al.}}    %et al.%

52: \def\pc{{\rm\,pc}}

53: \def\kpc{{\rm\,kpc}}

54: \def\Mpc{{\rm\,Mpc}}

55: \def\mpc{{\rm\,Mpc}}

56: \def\hpc{{~h_{73}^{-1}\rm\,pc}}

57: \def\hkpc{{~h_{73}^{-1}\rm\,kpc}}

58: \def\hMpc{{~h_{73}^{-1}\rm\,Mpc}}

59: \def\hnot{{\rm\,km/s/Mpc}}

60: \def\kmsec{{\rm\,km/s}}

61: \def\kms{{\rm\,km/s}}

62: \def\msun{{\rm\,M_\odot}}

63: \def\Zsun{{\rm\,Z_\odot}}

64: \def\lsun{{\rm\,L_\odot}}

65: \def\mdot{{\rm\,M_\odot}}

66: \def\vol#1  {{{#1}{\rm,}\ }}

67: \def\lya{{\rm Ly}\alpha}

68: \def\mytau{\tau_{{\rm Ly}\alpha}}

69: \def\hi{{H\thinspace I}\ }

70: \def\hii{{H\thinspace II}\ }

71: \def\hei{{He\thinspace I}\ }

72: \def\heii{{He\thinspace II}\ }

73: \def\heiii{{He\thinspace III}\ }

74: \def\nhi{N_{HI}}

75: \def\nheii{N_{\heii}}

76: \def\jhi{J_{\hi}}

77: \def\jhiu{J_{-21}}

78: \def\jheii{J_{\heii}}

79: \def\shi{\sigma_{\hi}(\nu)}

80: \def\shia{\bar\sigma_{\hi}}

81: \def\etal{et al.\ }

82: \def\Sec{\S }

83: \def\eqs{eqs.\ }

84: \def\cf{{cf.}\ }

85: \def\refset{\parindent=0pt\hangafter=1\hangindent=1em}

86: \newcount\refno

87: \refno = 1

88: \newcount\rfno

89: \rfno = 1

90: \def\eq{$^{\the\refno\ }$\advance\refno by 1}

91: \def\ad{\advance\rfno by 1}

92: 

93: \def\clock{\count0=\time \divide\count0 by 60

94:      \count1=\count0 \multiply\count1 by -60 \advance\count1 by \time

95:      \number\count0:\ifnum\count1<10{0\number\count1}\else\number\count1\fi}

96: \def\draft{

97:    \rightline{DRAFT: \today \quad\quad \clock}}

98: \def\myputfigure#1#2#3#4#5%

99: {\hskip0.03\textwidth\vskip#5pt%\hfill

100: \makebox[0pt]{\hskip#2in

101: \includegraphics[width=#3\textwidth]{#1}}\vskip#4pt\hfill}

102: 

103: \begin{document}

104: %\title{Galaxy Size Problem at $z=3$ in the Standard Cosmological Model}

105: \title{Galaxy Size Problem at \lowercase{$z=3$}: Simulated Galaxies Are Too Small}

106: 

107: \author{M.~Ryan~Joung\altaffilmark{1}, Renyue~Cen\altaffilmark{1}, \& Greg~L.~Bryan\altaffilmark{2}}  

108: 

109: \begin{abstract}

110: Using state-of-the-art adaptive mesh refinement cosmological hydrodynamic

111: simulations with a spatial resolution of proper $0.21\hkpc$ 

112: in refined subregions embedded within a comoving cosmological 

113: volume ($27.4\hMpc$)$^3$,

114: we investigate the sizes of galaxies at $z=3$ in the standard cold dark matter model

115: where reionization is assumed to complete at $z_{ri}\sim 6$.

116: Our simulated galaxies 

117: are found to be significantly smaller than the observed ones:

118: while more than one half of the galaxies 

119: observed by HST and VLT ranging from rest-frame UV to optical bands

120: with stellar masses larger than $2\times 10^{10}\msun$ have  

121: half-light radii larger than $\sim$$2\hkpc$, 

122: none of the simulated massive galaxies in the same mass range have 

123: half-light radii larger than $\sim$$2\hkpc$,

124: after taking into account dust extinction.

125: Corroborative evidence is provided by the rotation curves of the simulated galaxies 

126: with total masses of $10^{11}$-$10^{12}\msun$,

127: which display values ($300$-$1000~$km s$^{-1}$) 

128: at small radii ($\sim$$0.5\hkpc$) due to high stellar 

129: concentration in the central regions, 

130: larger than those of any well observed galaxies. 

131: Possible physical mechanisms 

132: to resolve this serious problem 

133: include: 

134: (1) an early reionization at $z_{ri}\gg 6$

135: to suppress gas condensation hence star formation,

136: (2) a strong, internal energetic feedback from stars or central black holes

137: to reduce the overall star formation efficiency,

138: %(3) a significantly reduced global gas density at the epoch of early galaxy formation ($z=6$-$20$)

139: %compared to what is seen at $z \approx 1100$ by WMAP,

140: or (3) a substantial small-scale

141: cutoff in the matter power spectrum.

142: 

143: \end{abstract}

144: 

145: 

146: \keywords{hydrodynamics --- galaxies: formation --- galaxies: kinematics

147: and dynamics --- cosmology: theory --- methods: numerical --- ultraviolet:

148: galaxies}

149: 

150: \altaffiltext{1}{Department of Astrophysical Sciences, Princeton

151:   University, Peyton Hall, Ivy Lane, Princeton, NJ 08544}

152: 

153: \altaffiltext{2}{Department of Astronomy, Columbia 

154:   University, Pupin Physics Laboratory, New York, NY 10027}

155: 

156: 

157: \section{INTRODUCTION}

158: 

159: The standard cosmological model

160: has been remarkably successful in accounting for

161: observations on scales larger than galaxy sizes 

162: (Krauss \& Turner 1995; Ostriker \& Steinhardt 1995;

163: Bahcall \etal 1999;

164: Tegmark \etal 2004; Spergel \etal 2007).

165: We intend to test this same model  

166: with regard to galaxy formation and evolution,

167: a regime where astrophysical processes are important and

168: hence a detailed testing of the cosmological model becomes 

169: more intricate.

170: In this paper, the first of a series,

171: we focus on the sizes of galaxies at redshift $z=3$, 

172: including ``Lyman Break Galaxies'' (LBGs) (Steidel \etal 2003).

173: Previous works on this subject include those based on semi-analytic methods

174: (e.g., Mo \etal 1999; Somerville \etal 2001; Somerville \etal 2008);  

175: the observed size-mass and size-luminosity relations at $z=3$ were reproduced 

176: in these studies.

177: 

178: Here, we take a brute-force approach

179: using high-resolution adaptive mesh refinement (AMR)

180: cosmological simulations to minimize

181: the number of adjustable astrophysical parameters and thereby maximize

182: the predictability of the standard model.

183: A physical resolution of our simulations of $0.21\hkpc$ proper

184: at $z=3$ in refined subregions 

185: embedded within a comoving cosmological

186: volume ($27.4\hMpc$) permits, 

187: an accurate characterization of the sizes of galaxies at $z=3$.

188: More specifically, we compare the half-light radii of simulated and 

189: observed galaxies in terms of size-mass and size-luminosity relations.  

190: This comparison is motivated by a series of observations of $z \sim 3$ 

191: galaxies in the rest-frame UV (Giavalisco \etal 1996; Lowenthal \etal 1997; 

192: Ferguson \etal 2004; Giavalisco \etal 2008) 

193: and in the rest-frame optical (Trujillo \etal 2006; 

194: Toft \etal 2007; Zirm \etal 2007; Buitrago \etal 2008).

195: 

196: We find the simulated galaxies to be generally smaller than the 

197: observed galaxies at $z=3$ in the stellar mass range $\ge 10^{10.5}\msun$

198: where comparisons can be made.  

199: This problem may be related to the disk size problem 

200: (also called the ``angular momentum problem'') 

201: at $z=0$ (e.g, Navarro \etal 1995;

202: Navarro \& Steinmetz 1997;

203: Governato \etal 2004).

204: The galaxy size problem at $z=3$ found in the present work may also

205: be related to an apparent large excess 

206: of predicted but unobserved dwarf halos (Klypin \etal 1999; Moore \etal 1999)

207: and an over-concentration of dark matter in simulated %model

208: dwarf galaxies on the scale of $\sim$$1\hkpc$ 

209: (Moore 1994; Flores \& Primack 1994; Burkert 1995;

210: McGaugh \& de Blok 1998; 

211: Kravtsov \etal 1998;

212: Moore \etal 1999).

213: Physical processes at high redshift that may be responsible

214: for the eventual resolution of this problem

215: should be manifested more clearly

216: %may be less ``diluted" 

217: at $z=3$, as they are unaffected by additional complications that 

218: may occur at lower redshifts.

219: Therefore, one may be able to obtain important ``cleaner" clues

220: to the nature of the dark matter and/or important astrophysical 

221: processes at high redshifts 

222: by studying 

223: %using 

224: $z=3$ galaxies.

225: Moreover, combining observations at both $z=3$ and $z=0$

226: may provide still more powerful constraints.

227: 

228: 

229: \section{SIMULATION AND ANALYSIS METHODS}

230: 

231: We perform cosmological simulations with the adaptive mesh refinement (AMR) 

232: Eulerian hydro code, Enzo (Bryan 1999; Norman \& Bryan 1999; O'Shea et al. 2004).  

233: First, we ran a low resolution simulation with a periodic box of $27.4\hMpc$ comoving on a side 

234: in a $\Lambda$CDM universe with cosmological parameters consistent with the WMAP3 results: 

235: ($\Omega_m$, $\Omega_{\Lambda}$, $\Omega_b$, $h$, $\sigma_8$, $n_s$) = (0.24, 0.76, 0.042, 0.73, 0.74, 0.95).  

236: We identified virialized dark matter halos in this simulation at $z=3$ 

237: and resimulated 11 of the most massive 20 halos 

238: in a suite of five high resolution simulations embedded within the same ($27.4\hMpc$)$^3$ comoving box 

239: using the multimass initialization technique.

240: The five high resolution subregions have comoving volumes ranging

241: from $\sim$($3.4\hMpc$)$^3$ to $\sim$($8.8\hMpc$)$^3$. 

242: Within the high-resolution regions, the cell size of the root grid is $53.5\hkpc$ 

243: and additional grid refinements are allowed to reach 

244: a maximum level of $l_{max}=6$, resulting in the maximum spatial resolution 

245: of $0.84\hkpc$ (comoving) or $0.21\hkpc$ (proper) at $z=3$,  

246: while the rest of the box is evolved at a lower resolution of $214\hkpc$. 

247: The dark matter particle mass in the high-resolution region is $4.6 \times 10^6$ M$_{\odot}$.

248: The simulations include a metagalactic UV background (Haardt \& Madau 1996), 

249: a diffuse form of photoelectric and photoionization heating 

250: (Abbott 1982; Joung \& Mac Low 2006)

251: and shielding of UV radiation by neutral hydrogen (Cen \etal 2005). 

252: They also include cooling due to molecular hydrogen (Abel et al. 1997)

253: and  metallicity-dependent radiative cooling 

254: (Cen \etal 1995) extended down to $10~$K (Dalgarno \& McCray 1972). 

255: Star particles are created in cells that satisfy a set of criteria for 

256: star formation proposed by Cen \& Ostriker (1992).  

257: A stellar particle of mass

258: $m_{*}=c_{*} m_{\rm gas} \Delta t/t_{*}$ is created

259: (the same amount is removed from the gas mass in the cell),

260: if the gas in a cell at any time meets

261: the following three conditions simultaneously:

262: (1) contracting flow, (2) cooling time less than dynamic time, and  (3)

263: Jeans unstable,

264: where $\Delta t$ is the time step, $t_{*}={\rm max}(t_{\rm dyn}, 10^5$yrs),

265: $t_{dyn}=\sqrt{3\pi/(32G\rho_{tot})}$ is the dynamical time of the cell,

266: $m_{\rm gas}$ is the baryonic gas mass in the cell and

267: $c_*=0.03$ is star formation efficiency.

268: Each star particle is tagged with its initial mass, creation time, and metallicity; 

269: star particles typically have masses of $\sim$$10^5\msun$.

270: Star formation and supernovae feedback are modeled following Cen \etal (2005)

271: with $e_{SN}=3\times 10^{-6}$.

272: Feedback energy and ejected metals are distributed into 

273: 27 local gas cells centered at the star particle in question, weighted by the specific volume of each cell.

274: The temporal release of metal-enriched gas and thermal energy at time $t$

275: has the following form: 

276: $f(t,t_i,t_{dyn}) \equiv (1/ t_{dyn})

277: [(t-t_i)/t_{dyn}]\exp[-(t-t_i)/t_{dyn}]$,

278: where $t_i$ is the formation time of a given star particle.

279: The metal enrichment inside galaxies and in the intergalatic medium 

280: is followed self-consistently in a spatially resolved fashion (Cen \etal 2005).

281: 

282: We identify virialized objects in our high resolution simulations using the HOP algorithm (Eisenstein \& Hut 1998),

283: which is tested to be robust.

284: % with $\delta_{outer}$, the minimum overdensity parameter to be part

285: % of a group, of 200.  To select out individual galaxies, we use a 

286: %more stringent threshold of $\delta_{outer}=10^4$ while the other two

287: %parameters, $\delta_{peak}$ and $\delta_{saddle}$, that control 

288: %minimum central overdensity for groups and minimum overdensity to 

289: %merge groups have values $3.0 \times$ and $2.5 \times \delta_{outer}$.

290: We find 49 halos with virial masses $>5 \times 10^{10}$ M$_{\odot}$ 

291: to compare with observations.

292: 

293: The light distribution is computed from the star particles using the GISSEL stellar

294: synthesis code (Bruzual \& Charlot 2003).  

295: We calculate the luminosities of the simulated galaxies in various 

296: rest-frame UV and optical bands where observations are available for 

297: $z \sim 3$ galxies with measured half-light radii (see Fig. 2).  To 

298: obtain half-light radii in the right wavelength range, we placed the same 

299: filter (blueshited to z=3) as used in each sample of observed galaxies;  

300: this was necessary because the sizes of simulated galaxies vary depending 

301: on the observed band, as shown in the size-luminosity plot.  

302: We followed the procedure described in Appendix B of Rudnick et al. (2003) 

303: to compute luminosities of the model galaxies in each given 

304: band.\footnote{For photon counting devices, the filter transmission curves 

305: $T(\lambda)$ in Rudnick et al. (2003) should be replaced by $T(\lambda) \lambda$.}  The stellar 

306: mass of each simulated galaxy is equal to the sum of the masses of the 

307: star particles located 

308: within 15\% of the virial radius of the galaxy at z=3.  For the observed 

309: galaxies, we adopt the stellar masses reported in the papers referenced 

310: in Figure 2 and, where appropriate, convert apparent magnitudes to 

311: luminosities in the wavelength range of the given filter blueshifted to 

312: z=3.

313: %We focus on the luminosity in the SDSS $g'$-band 

314: %[note that the $G$ filter in Steidel \& Hamilton (1993) is nearly identical to the $g'$-band], corresponding to

315: %restframe $\lambda_{eff} \approx 1200~\AA$.  

316: %The spectrum is nearly flat in the UV, so the two bands (with

317: %effective restframe wavelengths of $1200\AA$ and $1500\AA$) give

318: %practically the same results.  

319: We adopt an approximate model for dust extinction following Binney \& Merrifield (1998) but assume that dust attenuation is proportional to the metal column density rather than the total hydrogen density and correct for depletion of refractory elements (Zn) onto dust 

320: grains parametrized by $f_{Fe}$, fraction of iron in dust (Vladilo \& Peroux 2005):

321: $A_V = \Sigma_Z~f_{Fe}/(4\times 10^{19}~m_p F$ cm$^{-2})$ mag, 

322: where $m_p$ is the proton mass, $\Sigma_Z$ is the mass column 

323: density of metals in front of a given star particle. 

324: The factor, $F=3.7$, is chosen so that $E(B-V) = 0.3$ (Pettini \etal 1998; 

325: Shapley \etal 2003; Steidel \etal 2003), corresponding to $A_{1500\AA} 

326: \approx 1.5$ and an escape fraction in the rest-frame $1500\AA$ is $\sim$26\%. 

327: We choose a relatively high value of $E(B-V)$ to strengthen our conclusion 

328: that, even though dust extinction acts to increase half-light radii of the 

329: galaxies, our simulated galaxies are still generally smaller than the 

330: observed ones.  

331: The conversions from $A_V$ to $A$ at other bands are based on 

332: the dust extinction law proposed by Calzetti \etal (2000). 

333: All half-light radii are measured directly using projected 2D maps.  

334: 

335: 

336: \section{RESULTS}

337: 

338: 

339: 

340: 

341: 

342: Figure 1 shows the projected 

343: stellar mass density of a region of comoving size

344: ($1.4\hMpc$)$^2$ with a depth of comoving $3.3\hMpc$ at $z=3$,

345: cut out from our largest high resolution simulation sub-volume of size $\sim$($8.8\hMpc$)$^3$.

346: The insets show magnified images of the

347: four most massive halos in the displayed region 

348: in their projected luminosity density distributions. 

349: For each galaxy, the

350: images show, from left to right, intrinsic (before dust attenuation is

351: applied), apparent (after dust attenuation), and smoothed (after dust

352: attenuation and PSF filtering) luminosity densities, respectively, all in

353: the observed $I$-band.\footnote{We use the response curve for the F814W filter on HST's WFPC2.}

354: %$g'$-band. 

355: On the one hand, we see that our high resolution permits formation of extremely dense

356: structures.

357: On the other hand, many rich and salient features produced by cosmological processes,

358: such as mergers and tidal tails, are clearly visible and striking.

359: Comparing the leftmost and middle pictures of each row of the insets 

360: suggests that dust extinction significantly affects 

361: the observed luminosity density distribution 

362: in the observed $I$-band. 

363: %$g'$-band. 

364: This can be understood simply as a consequence of the empirical Schmidt-Kennicutt relation (Kennicutt 1998): 

365: gas surface density increases monotonically 

366: as star formation rate density increases; and we assumed that 

367: dust extinction is proportional to the metal surface density 

368: hence gas surface density (for a constant metallicity).

369: 

370: 

371: 

372: 

373: Figure 2 shows the size-mass (top) and size-luminosity relations (bottom)

374: of the simulated and observed galaxies.  The ``intrinsic" half-light radii of the 

375: simulated galaxies are displayed as ``x", while dust extinction-applied 

376: half-light radii are shown as filled circles where the shade in grey scale 

377: indicates the amount of dust extinction in the rest-frame $V$-band (darker 

378: for higher extinction; $0.35 \lesssim A_V \lesssim 2.5$).  Each sample of 

379: observed galaxies is shown by distinct symbols as indicated.

380: The observed $z \sim 3$ galaxies include the rest-frame UV samples from 

381: Lowenthal \etal (1997) and Giavalisco \etal (2008) as well as rest-frame 

382: optical samples from Trujillo \etal (2006), Toft \etal (2007), Zirm 

383: \etal (2007) and Buitrago \etal (2008).

384: In all panels, the intrinsic sizes of the simulated galaxies lie mostly between 0.2 and 

385: $0.5 \hkpc$, while the dust extinction-applied sizes can be significantly larger.  

386: This increase in apparent sizes is, on average, 

387: greater at shorter wavelengths and for more massive and luminous galaxies, 

388: and may be partly responsible for the observed relation between star formation 

389: and size at $z \sim 2.5$ (Toft et al. 2007; Zirm et al. 2007).  

390: Remarkably, this trend is also seen in the sizes of the simulated galaxies 

391: measured in the rest-frame optical bands. 

392: %, where dust extinction is less important

393: 

394: In the top display of Figure 2, 

395: the observational data are only available for galaxies with 

396: stellar masses of $\ge 2\times 10^{10}\msun$.

397: In this mass range, there are more than one half of the observed galaxies

398: having the half-light radii (all measured in the rest-frame optical) 

399: greater than $2 \hkpc$,

400: whereas all of the simulated galaxies in the same mass range

401: have the half-light radii smaller than $2 \hkpc$.

402: A significant fraction of these relatively massive galaxies 

403: that are actively star-forming (SFR $> 10$ M$_{\odot}$ yr$^{-1}$) 

404: may be identified with the LBGs (Steidel \etal 1996).

405: 

406: This discrepancy in half-light radii between simulated 

407: and observed galaxies is also seen in the size-luminosity relation

408: at the bottom display of Figure 2, where luminosities at several different 

409: observed bands are shown in four separate panels.

410: Here, the half-light radii of the simulated 

411: galaxies again occupy the low-end of the distributions compared to 

412: galaxies observed in the HST WFPC2 F814W filter 

413: (Lowenthal \etal 1997),

414: in the VLT ISAAC K$_s$ band (Trujillo \etal 2006;

415: Toft \etal 2007).

416: However, the situation here is slightly more complicated 

417: in that we find apparent agreement between simulated galaxies

418: and observations of Giavalisco \etal (2008) 

419: in the HST ACS F850LP filter 

420: and, to some degree, of Buitrago \etal (2008) in the VLT ISAAC K$_s$ band.

421: Whether the 

422: discrepancy among observed sizes is real or an artifact of the different 

423: source extraction and fitting algorithms employed is an open 

424: question.\footnote{We note, however, that the measured half-light radii 

425: in Giavalisco et al. (2008) 

426: -- based on the SExtractor program -- may be biased low compared to size 

427: measurements derived from fits to Sersic profiles, for the faint galaxies in 

428: their sample (M. Giavalisco, private comm.).}

429: 

430: Overall, we conclude that there is fairly strong indication

431: that the simulated massive galaxies are too small 

432: compared to their observed counterparts at $z \sim 3$,

433: from rest-frame UV to rest-frame optical bands.

434: This suggests that the concurrent star formation activities and the overall 

435: stellar mass distributions in the simulated 

436: galaxies at $z=3$ are both too concentrated near the galactic centers.

437: Concentration of stellar mass may 

438: be shown in an alternative way using rotation curves.

439: Figure 3 shows rotation velocity curves for the three top galaxies 

440: of total mass of $10^{11}-10^{12}\msun$.

441: These curves seem to peak at too high a value ($300$-$1000~$km s$^{-1}$)

442: at small radii ($\sim$$0.5\hkpc$).

443: 

444: It is prudent to ask if our results depend on limited resolution.

445: To address this point, we ran higher resolution simulations (with the 

446: smallest cell size of  0.11 proper$\hkpc$ or twice the linear 

447: resolution as in our fiducial models) of 2 of the 5 subvolumes,

448: or 8 of the 49 galaxies discussed in this Letter.  In all of these cases,

449: we find that both the intrinsic half-mass and intrinsic half-light radii

450: are significantly smaller (by nearly a factor of 2), indicating that our

451: results have not yet converged in terms of obtaining the absolute sizes

452: of the high-z galaxies.

453: We find that the dust attenuated half-light

454: radii with the higher resolution simulations are slightly smaller (10-30\%)

455: than those from lower resolution simulation. 

456: While the absolute sizes have not converged,

457: these tests support and strengthen our

458: conclusion that the simulated galaxies at $z=3$ are too small

459: compared to their observed counterparts.

460: 

461: 

462: 

463: 

464: 

465: \section{DISCUSSION AND CONCLUSIONS}

466: 

467: 

468: The standard cold dark matter cosmological model is in good agreement 

469: with a rich set of observations on large scales.

470: Our main purpose is to systematically examine 

471: it, through a series of papers,

472: in the context of galaxy formation and evolution,

473: a regime where, relatively speaking, 

474: it has not been seriously contested.

475: Because astrophysical processes tend to play a progressively 

476: more important role at smaller scales,

477: in particular, on galactic scales and smaller,

478: it is vital to employ as few 

479: adjustable astrophysical parameters as possible,

480: to have a true test of the cosmological model.

481: %Therefore, we take a brute-force approach

482: %using high-resolution adaptive mesh refinement

483: %cosmological simulations to maximize

484: %the predictability and the falsifiability of the standard cosmological model

485: %with respect to galaxy formation and evolution.

486: 

487: In this paper we focus on the sizes of galaxies, including

488: LBGs (Steidel \etal 2003) at redshift $z=3$,

489: using state-of-the-art AMR cosmological hydrodynamic

490: simulations with a spatial resolution of proper $0.21\hkpc$ 

491: in five refined subregions embedded within a comoving cosmological volume ($27.4 \hMpc$)$^3$.

492: We find that, taking into account dust extinction, 

493: the computed distribution of half-light radii of simulated galaxies 

494: in the rest-frame $I$- and $V$-bands occupy only the low-end of the 

495: observed size distributions.  

496: While none of the simulated massive galaxies have 

497: half-light radii larger than $\sim$$2\hkpc$, 

498: more than one half of the observed galaxies have  

499: half-light radii exceeding that value. 

500: %$\sim$$2\hkpc$. 

501: We note that the intrinsic 

502: (i.e., in the absence of dust extinction and finite instrumental resolution)

503: half-light (and half-stellar-mass) radii are smaller at $\sim 0.3\hkpc$

504: and our resolution tests indicate that they become still smaller

505: with higher resolution, hence strengthening our conclusions. 

506: Consistent with this apparent discrepancy between simulated and observed

507: LBGs, the rotation curves of the simulated galaxies

508: of total masses of $10^{11}$-$10^{12}\msun$

509: have unusually high values ($300$-$1000~$km s$^{-1}$) 

510: at small radii ($\sim$$0.5\hkpc$).

511: Such rotation curves are not seen in local galaxies for which accurate 

512: measurements are available,

513: although no direct comparison with 

514: observed high-redshift galaxies 

515: can be properly made at this time.

516: 

517: These discrepancies appear to originate from 

518: %The origin appears that

519: stellar masses that are too highly concentrated in small ($< 1\hkpc$) central regions

520: of the simulated galaxies.

521: This may be caused by an over-abundance of smaller galaxies

522: that formed at high redshifts and subsequently sank to the centers

523: via dynamical friction {\it or} by vigorous {\it in situ} star formation in the central regions.

524: Likely, 

525: any potential viable solution to this apparent problem would have to reduce the amount 

526: of stars that formed.

527: Possible physical mechanisms include: 

528: (1) an early reionization with $z_{ri}\gg 6$

529: to suppress gas condensation that will reduce earlier star formation

530: (e.g., Bullock, Kravtsov, \& Weinberg 2000),

531: (2) a strong, internal energetic feedback from stars or central black holes

532: to reduce the overall star formation efficiency (e.g., Sommer-Larson \etal 2003;

533: Governato \etal 2007),

534: or 

535: (3) a substantial small-scale

536: cutoff in the matter power spectrum,

537: for example, if the dark matter particles are warm rather than cold

538: (e.g., Hogan \& Dalcanton 2000; Sommer-Larsen \& Dolgov 2001;

539: Bode, Ostriker \& Turok 2001)

540: 

541: Since the age of the universe at $z=3$ is only about $1/6$ of the present age, 

542: a successful resolution to the galaxy size problem at $z\sim 3$

543: may provide important ``cleaner" clues

544: to the nature of the dark matter and/or important astrophysical 

545: processes at high redshifts.

546: Moreover, combining observations at both $z=3$ and $z=0$

547: may provide still more powerful constraints.

548: 

549: 

550: %\smallskip

551: \acknowledgements{

552: We thank A. Burkert, Y.-T. Lin, M.-M. Mac Low, 

553: L. Mayer, B. Moore, R. Overzier, A. Shapley, and P. van Dokkum 

554: for helpful discussions and comments; 

555: F. Buitrago, M. Giavalisco, M. Kriek, G. Rudnick, and I. Trujillo 

556: for generous help with observational data; and 

557: M.-S. Shin and K. Nagamine 

558: for help with the galaxy spectrum code.

559: We are grateful to M. Giavalisco and his colleagues for

560: making their new GOODS-S sample available to us prior to publication.

561: We are indebted to an anonymous referee for many useful suggestions

562: that help improve the paper.

563: We thank M. Norman and his team 

564: in University of California at San Diego

565: for useful assistance in code development.

566: We gratefully acknowledge financial support by

567: grants AST-0507521, NNG05GK10G and NNX08AH31G.

568: The simulations were performed at NCSA with computing time provided by

569: LRAC allocation TG-MCA04N012.

570: }

571: 

572: \begin{thebibliography}{99}

573: 

574: \bibitem[Abbott(1982)]{abb82} Abbott, D. C. 1982, ApJ, 263, 723

575: 

576: \bibitem[Abel \etal(1997)]{abe97} Abel, T., Anninos, P., Zhang, Y., \& Norman, M. L. 1997, New Astronomy, 2, 181

577: 

578: \bibitem[Bahcall \etal 1999]{b99} Bahcall, N.A., Ostriker, J.P., Perlmutter, S., \& Steinhardt, P. 1999, Science, 284, 1481

579: 

580: \bibitem[Binney \& Merrifield(1998)]{bin98} Binney, J., \& Merrifield, M. 1998, Galactic Astronomy (Princeton: Princeton Univ. Press)

581: 

582: \bibitem[Bode \etal (2001)]{bod01} Bode, P., Ostriker, J.P., \& Turok, N. 2001, ApJ, 556, 93

583: 

584: \bibitem[Bruzual \& Charlot(2003)]{bru03} Bruzual, G., \& Charlot, S. 2003, MNRAS, 344, 1000

585: 

586: \bibitem[Bryan(1999)]{bry99} Bryan, G. L., 1999, Comp. Sci. Eng., 1, 46

587: 

588: \bibitem[Buitrago \etal (2008)]{bui08} Buitrago, F., Trujillo, I., Conselice, C. et al., 2008, ApJL, in press (arXiv:0807.4141v2 [astro-ph])

589: 

590: \bibitem[Bullock \etal (2000)]{bkw00} Bullock, J.S., Kravtsov, A.V., \& Weinberg, D.H. 2000, 539, 517

591: 

592: \bibitem[Burkert 1995]{b95} Burkert, A. 1995, ApJ, 447, L25

593: 

594: %\bibitem[Calzetti et al.(2000)]{cal00} Calzetti, D., Armus, L., Bohlin, R. C., Kinney, A. L., Koornneef, J., \& Storchi-Bergmann, T. 2000, ApJ, 533, 682

595: \bibitem[Calzetti et al.(2000)]{cal00} Calzetti, D., Armus, L., Bohlin, R. C. \etal 2000, ApJ, 533, 682

596: 

597: \bibitem[Cen \& Ostriker 1992]{co92} Cen, R., \& Ostriker, J.P. 1992, ApJ, 399, L113

598: 

599: \bibitem[Cen \etal 1995]{c95} Cen, R., Kang, H., Ostriker, J.P., \& Ryu, D. 1995, ApJ, 451, 436

600: 

601: \bibitem[Cen \etal 2005]{c05} Cen, R., Nagamine, K., \& Ostriker, J.P. 2005, ApJ, 635, 86

602: 

603: \bibitem[Dalgarno \& McCray(1972)]{dal72} Dalgarno, A., \& McCray, R.A. 1972, ARA\&A, 10, 375

604: 

605: \bibitem[Eisenstein \& Hut(1998)]{eis98} Eisenstein, D.J., \& Hut, P. 1998, ApJ, 498, 137

606: 

607: \bibitem[Ferguson et al.(2004)]{fur04} Ferguson, H., Dickinson, M., Giavalisco, M. \etal 2004, ApJ, 600, L107

608: 

609: \bibitem[Flores \& Primack (1995)]{fp95} Flores, R.A., \& Primack, J.R. 1995, ApJ, 457, L5

610: 

611: \bibitem[Giavalisco \etal 1996]{g96} Giavalisco, M., Steidel, C.C., \& Macchetto, F.D. 2004, ApJ, 607, 688

612: 

613: \bibitem[Giavalisco \etal 2008]{g08} Giavalisco, M. and the GOODS Team, 2008, in preparation

614: 

615: \bibitem[Governato \etal 2007]{g07} Governato, F., \etal 2004, ApJ, 607, 688

616: 

617: \bibitem[Governato \etal 2007]{g07} Governato, F., \etal 2007, MNRAS, 374, 1479

618: 

619: \bibitem[Haardt \& Madau(1996)]{haa96} Haardt, F., \& Madau, P. 1996, ApJ, 461, 20

620: 

621: \bibitem[Hogan \& Dalcanton (2000)]{h00} Hogan, C.J., \& Dalcanton, J.J. 2000, PhRvD, 62, 063511

622: 

623: \bibitem[Joung \& Mac Low(2006)]{jou06} Joung, M.K.R., \& Mac Low, M.-M. 2006, ApJ, 653, 1266

624: 

625: \bibitem[Kennicutt (1998)]{ken98} Kennicutt, R.C., Jr. 1998, ARA\&A, 36, 189

626: 

627: \bibitem[Klypin \etal (1999)]{k99} Klypin, A.A., Kravtsov, A.V., Valenzuela, O., \& Prada, F. 1999, ApJ, 522, 82

628: 

629: %\bibitem[Klypin \etal (2001)]{k01} Klypin, A.A., Kravtsov, A.V., Bullock, J., \& Primack, J. 2001, 554, 903

630: 

631: \bibitem[Krauss \& Turner 1995]{kt95} Krauss, L., \& Turner, M.S. 1995, Gen. Rel. Grav., 27, 1137

632: 

633: \bibitem[Kravtsov \etal 1998]{kk98} Kravtsov, A.V., Klypin, A.A., Bullock, J.S., \& Primack, J.R. 1998, ApJ, 502, 48

634: 

635: \bibitem[Lowenthal \etal 1997]{l97} Lowenthal, J.D., Koo, D.C., Guzman, R. \etal 1997, ApJ, 481, 673

636: 

637: \bibitem[McGaugh \& de Blok (1998)]{MdB98} McGaugh, S.S., de Blok, W.J.G. 1998, ApJ, 499, 41

638: 

639: %\bibitem[Meurer \etal 1997]{m97} Meurer, G.R., Heckman, T.M., Lehnert, M.D., Leitherer, C., \& Lowenthal, J. AJ, 114, 54

640: 

641: \bibitem[Mo, Mao, \& White (1999)]{MMW99} Mo, H.J., Mao, S., \& White, S.D.M. 1999, MNRAS, 304, 175

642: 

643: \bibitem[Moore 1994]{m94} Moore, B. 1994, Nature, 370, 629

644: 

645: \bibitem[Moore 1999]{m99} Moore, B., Quinn, T., Governato, F. \etal 1999, MNRAS, 310, 1147

646: 

647: \bibitem[Navarro, Frenk, \& White 1995]{nfw95} Navarro, J.F., Frenk, C.S., \& White, S.D.M. 1995, MNRAS, 275, 56

648: 

649: \bibitem[Navarro \& Steinmetz 1997]{ns97} Navarro, J.F., \& Steinmetz, M. 1997, ApJ, 478, 13

650: 

651: \bibitem[Norman \& Bryan(1999)]{nor99} Norman, M. L., \& Bryan, G. L. 1999, ASSL Vol. 240: Numerical Astrophysics, 19

652: 

653: \bibitem[O'Shea et al.(2004)]{osh04} O'Shea, B. W., Bryan, G. L., Bordner, J., Norman, M. L. \etal 2004, ``Adaptive Mesh Refinement - Theory and Applications'', ed. T. Plewa, T. Linde \& V. G. Weirs, Springer Lecture Notes in Comp. Sci. Eng. (arXiv:astro-ph/0403044)

654: 

655: \bibitem[Ostriker \& Steinhardt 1995]{os95} Ostriker, J.P., \& Steinhardt, P. 1995, Nature, 377, 600

656: 

657: \bibitem[Pettini \etal 1998]{p98} Pettini, M., Kellogg, M., Steidel, C.C., Dickinson, M., Adelberger, K.L., \& Giavalisco, M. 1998, ApJ, 508, 539

658: 

659: \bibitem[Rudnick et al. (2003)]{rud03} Rudnick, G., Rix, H.-W., Franx, M. \etal 2003, ApJ, 599, 847

660: 

661: %\bibitem[Sawicki \& Yee (1998)]{sy98} Sawicki, M., \& Yee, H.K. 1998, AJ, 115, 1329

662: 

663: \bibitem[Shapley et al. (2003)]{sha03} Shapley, A.E., Steidel, C.C., Pettini, M., \& Adelberger, K.L. 2003, ApJ, 588, 65

664: 

665: \bibitem[Somerville et al. (2001)]{som01} Somerville, R.S., Primack, J.R., \& Faber, S.M. 2001, MNRAS, 320, 504

666: 

667: \bibitem[Somerville et al. (2008)]{som08} Somerville, R.S., Barden, M., Rix, H.-W. \etal 2008, ApJ, 672, 776

668: 

669: \bibitem[Sommer-Larsen, \& Dolgov (2001)]{sd00} Sommer-Larsen, J., \& Dolgov, A. 2001, ApJ, 551, 608

670: 

671: \bibitem[Sommer-Larsen \etal (2003)]{smp03} Sommer-Larsen, J., Gotz, M., \& Portinari, L. 2003, ApJ, 596, 47

672: 

673: \bibitem[Spergel \etal (2007)]{Spergel07} Spergel, D.N., \etal 2007, ApJS, 170, 377

674: 

675: %\bibitem[Steidel \& Hamilton(1993)]{ste93} Steidel, C. C., \& Hamilton, D. 1993, AJ, 105, 2017

676: 

677: \bibitem[Steidel \etal (1996)]{Steidel96} Steidel, C.C., Giavalisco, M., Pettini, M. \etal 1996, ApJ, 462, 17

678: 

679: \bibitem[Steidel \etal (2003)]{Steidel03} Steidel, C.C., Adelberger, K.L., Shapley, A.E. \etal 2003, ApJ, 592, 728

680: 

681: \bibitem[Tegmark \etal (2004)]{Tegmark04} Tegmark, M., \etal 2004, Phys. Rev. D69, 103501

682: 

683: \bibitem[Toft \etal (2007)]{Toft07} Toft, S., van Dokkum, P., Franx, M. \etal 2007, ApJ, 671, 285

684: 

685: \bibitem[Trujillo \etal (2006)]{Truj06} Trujillo, I., F\"orster Schreiber, N.M., Rudnick, G. \etal 2006, ApJ, 650, 18

686: 

687: \bibitem[Vladilo \& Peroux (2005)]{VladiloPerous05} Vladilo, G., Peroux, C., 2005, A\&A, 444, 461

688: 

689: \bibitem[Zirm \etal (2007)]{zirm07} Zirm, A.W., van der Wel, A., Franx, M. \etal 2007, ApJ, 656, 66

690: 

691: \end{thebibliography}

692: 

693: \clearpage

694: 

695: %\begin{figurehere}

696: \begin{figure*}

697: \begin{center}

698: \resizebox{7.5in}{!}{\includegraphics[angle=0]{f1.eps}} %for ApJ high res f1

699: %\resizebox{4.8in}{!}{\includegraphics[angle=0]{f1.eps}}

700: \end{center}

701: \caption{

702: The large image displays the projected 

703: stellar mass density of a region of comoving size

704: $(1.4\hMpc$)$^2$ with a depth of comoving $3.3\hMpc$ at $z=3$,

705: cut out from one of our simulation volumes.

706: The insets zoom in on the

707: four most massive galaxies in volume region

708: in terms of virial mass and show

709: their projected luminosity density distributions.  For each galaxy, the

710: images show, from left to right, intrinsic (before dust attenuation is

711: applied), apparent (after dust attenuation), and smoothed (after dust

712: attenuation and PSF filtering) luminosity densities, respectively, all in

713: the observed $I$-band.  %$g'$-band. 

714: A Gaussian filter with FWHM $=0''.125$ is applied for the PSF.

715: The bars indicate lengths in proper$\hkpc$.

716: }

717: \label{fig:1}

718: \vskip7pt

719: %\end{figurehere}

720: \end{figure*}

721: 

722: %\begin{figurehere}

723: \begin{figure*}

724: \begin{center}

725: \resizebox{3.7in}{!}{\includegraphics[angle=0]{f2a.eps}}

726: \resizebox{4.0in}{!}{\includegraphics[angle=0]{f2b.eps}}

727: %\resizebox{6.0in}{!}{\includegraphics[angle=0]{f2.eps}}

728: \end{center}

729: \caption{

730: Size-mass ({\it top}) and size-luminosity ({\it bottom}) relations 

731: of the simulated and observed galaxies.  The ``intrinsic" half-light radii 

732: of the simulate galaxies (along six orthogonal projection directions) are 

733: displayed as crosses, while dust extinction-applied 

734: half-light radii are shown as filled circles where the shade in grey scale 

735: indicates the amount of dust extinction in the rest-frame $V$-band (darker 

736: for higher extinction; $0.35 \lesssim A_V \lesssim 2.5$).  

737: In the top display the observed 

738: galaxies that appear in the size-mass relation are based on measurements 

739: in the rest-frame optical.  

740: In the bottom display 

741: the size-luminosity relations are shown for 

742: both rest-frame UV ({\it top panels}) and rest-frame optical ({\it 

743: bottom panels}). 

744: The observed $z \sim 3$ galaxies include the rest-frame UV samples from 

745: Lowenthal \etal (1997) using HST WFPC2 F814W 

746: filter ($\lambda_c=8203\AA$ and FWHM=$1758\AA$) 

747: and from Giavalisco \etal (2008) using HST ACS F850LP 

748: filter ($\lambda_c=8950\AA$ and FWHM=$900\AA$),

749: rest-frame optical samples from Trujillo \etal (2006), 

750: Toft \etal (2007) and Buitrago \etal (2008)

751: using VLT ISAAC K$_s$ band ($\lambda_c=21600\AA$ and FWHM=$2700\AA$)

752: and from Zirm \etal (2007) using HST WFPC2 F160W 

753: filter ($\lambda_c=16089\AA$ and FWHM=$4010\AA$).

754: }

755: \label{fig:2}

756: \vskip7pt

757: \end{figure*}

758: %\end{figurehere}

759: 

760: %\begin{figurehere}

761: \begin{figure*}

762: \begin{center}

763: \resizebox{6.0in}{!}{\includegraphics[angle=0]{f3.eps}}

764: %\resizebox{3.7in}{!}{\includegraphics[angle=0]{f3.eps}}

765: \end{center}

766: \caption{

767: Rotation velocity curves for the three galaxies labeled ``$1$'', ``$2$''

768: and ``$3$'' in Fig. 1.  The solid curves represent rotation velocities

769: due to all the matter within a given galactocentric radius,

770: while the dotted curves show those due to stellar mass only.  

771: The virial masses of the three galaxies

772: are 

773: $8\times 10^{11}$, 

774: $4\times 10^{11}$

775: and 

776: $1\times 10^{11}\msun$, respectively.

777: }

778: \label{fig:3}

779: \vskip7pt

780: %\end{figurehere}

781: \end{figure*}

782: 

783: \end{document}

784: 

785: