1: %% Beginning of file 'sample.tex'

2: %%

3: %% Modified 2005 December 5

4: %%

5: %% This is a sample manuscript marked up using the

6: %% AASTeX v5.x LaTeX 2e macros.

7: 

8: %% The first piece of markup in an AASTeX v5.x document

9: %% is the \documentclass command. LaTeX will ignore

10: %% any data that comes before this command.

11: 

12: %% The command below calls the preprint style

13: %% which will produce a one-column, single-spaced document.

14: %% Examples of commands for other substyles follow. Use

15: %% whichever is most appropriate for your purposes.

16: %%

17: 

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

19: 

20: %\usepackage{ulem}

21: %\usepackage{color}

22: 

23: %% manuscript produces a one-column, double-spaced document:

24: 

25: %\documentclass[manuscript]{aastex}

26: 

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

28: \usepackage{ulem}

29: \usepackage[usenames]{color}

30: 

31: %\documentclass{emulateapj}

32: 

33: %% preprint2 produces a double-column, single-spaced document:

34: 

35: %% Sometimes a paper's abstract is too long to fit on the

36: %% title page in preprint2 mode. When that is the case,

37: %% use the longabstract style option.

38:         

39: %% \documentclass[preprint2,longabstract]{aastex}

40:        

41: %% If you want to create your own macros, you can do so

42: %% using \newcommand. Your macros should appear before

43: %% the \begin{document} command.

44: %%

45: %% If you are submitting to a journal that translates manuscripts

46: %% into SGML, you need to follow certain guidelines when preparing

47: %% your macros. See the AASTeX v5.x Author Guide

48: %% for information.

49: 

50: %\newcommand{\vdag}{(v)^\dagger}

51: %\newcommand{\myemail}{skywalker@galaxy.far.far.away}

52: 

53: \newcommand{\BP}{Ballesteros-Paredes}

54: \newcommand{\cs}{c_{\rm s}}

55: \newcommand{\Eg}{E_{\rm g}}

56: \newcommand{\Ek}{E_{\rm k}}

57: \newcommand{\Eth}{E_{\rm th}}

58: \newcommand{\eq}{{\rm eq}}

59: \newcommand{\kms}{{\rm ~km~s}^{-1}}

60: \newcommand{\Lbox}{L_{\rm box}}

61: \newcommand{\Lj}{L_{\rm J}}

62: \newcommand{\Mcl}{M_{\rm cl}}

63: \newcommand{\Mj}{M_{\rm J}}

64: \newcommand{\Msun} {M_\sun}

65: \newcommand{\nmax}{n_{\rm max}}

66: \newcommand{\npk}{n_{\rm pk}}

67: \newcommand{\nsat}{n_{\rm sat}}

68: \newcommand{\nthr}{n_{\rm thr}}

69: \newcommand{\pcc}{{\rm ~cm}^{-3}}

70: \newcommand{\psc}{{\rm ~cm}^{-2}}

71: \newcommand{\ri}{r_{\rm inf}}

72: \newcommand{\racc}{r_{\rm acc}}

73: \newcommand{\tad}{\tau_{\rm AD}}

74: \newcommand{\tcol}{\tau_{\rm col}}

75: \newcommand{\tff}{\tau_{\rm ff}}

76: \newcommand{\tpre}{\tau_{\rm pre}}

77: \newcommand{\tyso}{\tau_{\rm yso}}

78: \newcommand{\VS}{V\'azquez-Semadeni}

79: 

80: % definiciones de javier

81: \def\alamenos#1{$^{-#1}$}

82: \def\ala#1{$^{#1}$}

83: 

84: \def\diezalamenos#1{10$^{-#1}$}

85: \def\diezala#1{10$^{#1}$}

86: 

87: 

88: %% You can insert a short comment on the title page using the command below.

89: 

90: \slugcomment{Submitted to The Astrophysical Journal}

91: 

92: %% If you wish, you may supply running head information, although

93: %% this information may be modified by the editorial offices.

94: %% The left head contains a list of authors,

95: %% usually a maximum of three (otherwise use et al.).  The right

96: %% head is a modified title of up to roughly 44 characters.

97: %% Running heads will not print in the manuscript style.

98: 

99: \shorttitle{CORE LIFETIMES IN NUMERICAL SIMULATIONS}

100: \shortauthors{GALV\'AN-MADRID ET AL.}

101: 

102: %% This is the end of the preamble.  Indicate the beginning of the

103: %% paper itself with \begin{document}.

104: 

105: \begin{document}

106: 

107: %% LaTeX will automatically break titles if they run longer than

108: %% one line. However, you may use \\ to force a line break if

109: %% you desire.

110: 

111: \title{Statistics of Core Lifetimes in Numerical Simulations of

112: Turbulent, Magnetically Supercritical Molecular Clouds}

113: 

114: %% Use \author, \affil, and the \and command to format

115: %% author and affiliation information.

116: %% Note that \email has replaced the old \authoremail command

117: %% from AASTeX v4.0. You can use \email to mark an email address

118: %% anywhere in the paper, not just in the front matter.

119: %% As in the title, use \\ to force line breaks.

120: 

121: \author{Roberto Galv\'an-Madrid\altaffilmark{1}, Enrique

122: \VS\altaffilmark{1} Jongsoo Kim\altaffilmark{2}, and Javier

123: \BP\altaffilmark{1}}

124: 

125: \altaffiltext{1}{Centro de Radioastronom\'ia y Astrof\'isica (CRyA),

126: Universidad Nacional Aut\'onoma de M\'exico,

127: Apdo. Postal 72-3 (Xangari), Morelia, Michoac\'an 58089, M\'exico}

128: \email{r.galvan, e.vazquez, j.ballesteros@astrosmo.unam.mx}

129: 

130: \altaffiltext{2}{Korea Astronomy and Space Science Institute, 61-1,

131: Hwaam-dong, Yuseong-gu, Daejon 305-764, Korea}

132: \email{jskim@kasi.re.kr}

133: 

134: %% Mark off your abstract in the ``abstract'' environment. In the manuscript

135: %% style, abstract will output a Received/Accepted line after the

136: %% title and affiliation information. No date will appear since the author

137: %% does not have this information. The dates will be filled in by the

138: %% editorial office after submission.

139: 

140: \begin{abstract}

141: We present measurements of the mean dense core lifetimes in numerical

142: simulations of magnetically supercritical, turbulent, isothermal

143: molecular clouds (MCs), in order to compare with observational

144: determinations. The mean ``prestellar'' lifetimes are given as a

145: function of the mean

146: density within the cores, which in turn is determined by the density

147: threshold $\nthr$ used to define them. The mean lifetimes are consistent

148: with observationally reported values, ranging from a few to several

149: free-fall times. We also present estimates of the

150: fraction of cores in the ``prestellar'',

151:  ``stellar'', and

152: ``failed'' stages as a function of $\nthr$.  Failed cores are defined as

153: those that do not manage to collapse, but rather re-disperse back into

154: the environment. Due to resolution limitations, the

155: number ratios are

156: measured indirectly in the simulations, as either lifetime ratios (for

157: the prestellar cores), or as time-weighted mass ratios (for the failed

158: cores). Our approach contains

159: one free parameter, the lifetime of a protostellar object $\tyso$ (Class

160: 0 + Class I stages), which is outside the realm of the

161: simulations. Assuming a value $\tau_\mathrm{yso} = 0.46$ Myr, we obtain

162: number ratios of starless to stellar cores ranging from 4--5 at $\nthr =

163: 1.5 \times 10^4 \pcc$ to $\sim 1$ at $\nthr = 1.2 \times 10^5 \pcc$,

164: again in good agreement with observational determinations. We also find

165: that the failed cores are generally difficult to

166: detect, although the

167: mass in these cores is comparable to that in stellar cores at

168: $\nthr = 1.5 \times 10^4 \pcc$. At $\nthr = 1.2

169: \times 10^5 \pcc$ the mass in failed cores is negligible, in agreement

170: with recent observational suggestions

171: that at the latter densities the cores are in general gravitationally

172: dominated. We conclude by noting that the timescale for core contraction

173: and collapse is virtually the same in the subcritical, ambipolar

174: diffusion-mediated model of star formation, in the model of star

175: formation in turbulent supercritical clouds, and in a model

176: intermediate between the previous two, suggesting a convergence of the

177: models at least at the level of the core lifetimes, for currently

178: accepted values of the clouds' magnetic criticality.

179: 

180: \end{abstract}

181: 

182: %% Keywords should appear after the \end{abstract} command. The uncommented

183: %% example has been keyed in ApJ style. See the instructions to authors

184: %% for the journal to which you are submitting your paper to determine

185: %% what keyword punctuation is appropriate.

186: 

187: \keywords{ISM: clouds --- ISM: evolution --- stars:

188: formation --- stars: pre-main sequence --- turbulence}

189: %\keywords{ISM: clouds --- Stars: formation--- Turbulence}

190: 

191: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

192: 

193: \section{Introduction} \label{sec:intro}

194: 

195: The evolution of the dense cores within molecular clouds (MCs) on their route

196: to forming stars, and the duration of the various stages of contraction

197: and collapse in particular, are key ingredients in our understanding of

198: the star formation process. The process of contraction and final

199: collapse of a core is generally divided in the so-called ``prestellar''

200: phase, which includes the span between the time when the core is

201: first detected and when a protostellar object forms in its deepest

202: regions, and the ``stellar'' phase, which includes the Class 0 and Class

203: I protostellar stages (from the appearance of the protostellar object to

204: the clearance of the surrounding material).

205: 

206: Observationally, several works have estimated the duration of the

207: prestellar stage through either statistical or chemical methods

208: \citep[see the discussion in][ and references therein]{BKMV07}. The

209: former rely on measuring the number ratio of prestellar to stellar

210: cores, and assuming that this ratio equals the ratio of durations of

211: these stages \citep[e.g.,][]{Beich86, LM99, Oni02, Hat06, Jor07}. A

212: summary of prestellar durations, or ``lifetimes'', from

213: several studies, has been recently presented by \citet{WT_etal07}.

214: 

215: Theoretically, there are two main competing models that describe the

216: evolution of the cloud cores.  The so-called ``standard model'' of

217: quasi-static contraction mediated by ambipolar diffusion \citep[AD;

218: e.g.,][]{Mou76, Mou91, SAL87} was the dominant paradigm until

219: recently. It postulated that the cores in low-mass star-forming regions

220: evolved quasi-statically in magnetically subcritical clouds, contracting

221: gravitationally at the rate allowed by AD, which causes a redistribution

222: of the magnetic flux, until the central parts of the cores become

223: magnetically supercritical and proceed to dynamical collapse. The

224: process was originally thought to be slow, because the clouds were

225: thought to be strongly magnetically subcritical, in which case the AD

226: timescale $\tad$ is roughly one order of magnitude larger than the

227: free-fall time.  However, it was later recognized that near-magnetically

228: critical initial conditions \citep{FM93,CB01} or turbulent conditions

229: \citep{FA02, Heitsch_etal04} cause the AD timescale to become

230: significantly shorter, only a few times longer than the free-fall time

231: \citep[][ hereafter Paper I]{CB01, Paper I}. This

232: realization was motivated in part by observational studies suggesting

233: core lifetimes not much longer than their free-fall times

234: \citep[e.g.,][]{LM99, JMA99}. A recent numerical study 

235: of core lifetimes dominated by AD

236: but with turbulent initial conditions has been presented recently by

237: \citet{NL05}, finding core lifetimes in the range 1.5--10 times the

238: free-fall time.

239: 

240: On the other hand, the so called ``turbulent'' model of star formation

241: \cite[e.g.,][]{BVS99, VS_etal00, MK04, BKMV07} takes into account the

242: fact that MCs are turbulent, and attributes a dual character to the

243: turbulence: on the one hand it provides support against generalized

244: cloud collapse, but on the other it produces strong local density

245: enhancements, which constitute the clumps and cores within the clouds.

246: Thus, their formation timescales are of the order of the turbulent

247: crossing time across the size scales of the fluid parcels that collide

248: to produce the clumps and cores \citep{BHV99, Elm00, PN02}. However,

249: Paper I showed that the lifetimes of randomly selected cores at

250: densities comparable to those of observed ammonia cores are nevertheless

251: of the order of a few times the free-fall time of the cores, because

252: their evolution while they are detectable includes part of the assembly

253: process and then the collapsing stage. 

254: Note that this is contrary to

255: frequent perceptions that the turbulent model implies core lifetimes of

256: approximately \emph{one} free-fall time \citep[e.g.,][]{WT_etal07}.

257: 

258: Within the context of the turbulent model, it is natural to expect the

259: existence of an additional class of MC clumps that end up redispersing

260: rather than collapsing; that is, turbulent density fluctuations that do

261: not manage to become locally gravitationally unstable (Elmegreen 1993;

262: Taylor et al.\ 1996; Paper I). We refer to these as ``failed''

263: clumps. Also, in this model, it is not crucial whether the clouds are

264: magnetically subcritical or supercritical, since much of the ``filtering'' of

265: mass that reaches the collapsed objects, and which regulates the star

266: formation efficiency, is provided by the dual role of the turbulence.

267: A stronger magnetic field simply appears to provide an extra reduction

268: factor for the star formation efficiency, whose precise magnitude

269: depends on whether the turbulence in the clouds is driven or decaying

270: \citep[see the discussion in][ and references therein]{VS07}.

271: 

272: In this paper we present a survey of the core lifetimes in a set of numerical

273: simulations of a turbulent, continuously driven, magnetically

274: supercritical, isothermal MC, with the aim of obtaining statistically

275: reliable data about them within this model, that can be 

276: compared to

277: observational determinations. We also discuss the number ratio of

278: ``starless'' to ``stellar'' (in the simulations, collapsed) cores. The

279: plan of the paper is as follows: In \S \ref{sec:simulation} we describe

280: the numerical simulations we analyzed; in \S

281: \ref{sec:procedure} we describe the method we used to measure the

282: lifetimes and the number ratios of the various kinds of objects; in  \S

283: \ref{sec:discussion} we compare them with existing observational and

284: theoretical work, and discuss their implications. In this section we also

285: discuss the limitations and accomplishments of the present work.

286: 

287: \section{The Numerical Simulations} \label{sec:simulation}

288: 

289: We have chosen the moderately supercritical simulation (named

290: M10J4$\beta$.1) reported in Paper I for analysis of the core

291: evolution. We refer to this run as ``R1'' in the rest of the paper,

292: and its selection was based on the fact that observations suggest that

293: MCs are nearly critical or moderately supercritical \citep{Crut99}.

294: The computational domain contained 64 Jeans masses, or 1860 $M_\odot$

295: \citep[][]{VKB05}. The mean number density is $n_0=500 \pcc$, the

296: simulation box size is 4 pc, and the (uniform) temperature is $T=11.4$

297: K, so that the sound speed is $\cs = 0.2 \kms$. The mean magnetic

298: field was $B_0 = 14.5~\mu$G, giving a mass-to-magnetic-flux ratio $\mu

299: =2.8$ times the critical value. These physical properties and

300: conditions of the simulation make it directly comparable to the most

301: massive ``supercores'' in the Perseus cloud, as defined by

302: \citet{Kirk06}. Indeed,

303: supercore No.\ 2 (IC 348) is reported by

304: those authors to have a mass $\sim 1940~\Msun$, and, at a distance of

305: 250 pc, its size is $\sim 5$ pc, very close to the corresponding

306: parameters of our simulation. Supercore No.\ 7 (NGC 1333) is reported to be

307: somewhat smaller and less massive, at $\sim 970~\Msun$.

308: 

309: The turbulence in this simulation was continuously driven in order to

310: maintain a turbulent Mach number $\sim 10$.  For the turbulence random

311: driver, we follow the method in \citet{Stone98}, i.e., we applied

312: a random Fourier driver operating at the largest scales with the

313: functional form $P(k) \propto k^6 \exp{(-8k/k_{\rm peak})} $,  

314: where 

315: $k=2\pi/l$ is the wavenumber corresponding to the

316: scale $l$, $k_{\rm peak} = 2(2\pi/L)$ is the wavenumber where 

317: the input power spectrum

318: peaks, and $L$ is the one-dimensional size of the computational box.

319: As discussed in Paper I, this is motivated by the fact that turbulence

320: is ubiquitous at all scales in molecular clouds \citep[e.g.,][]

321: {Larson81, Blitz93, Heyer_Brunt04}, including starless ones such as

322: the so-called Maddalena's cloud \citep{Maddalena_Thaddeus85}, which

323: suggests a universal origin and maintenance mechanism for the clouds'

324: turbulence, and that turbulence is driven at the

325: largest scales in the clouds \citep{Heyer_Brunt04, Heyer_Brunt07}.

326: 

327: Two more simulations (called R2 and R3) were performed 

328: with the

329: same physical parameters but different random 

330: seeds for the Fourier driver. R2 

331: was performed at a resolution of 256 cells per dimension, and R3 

332: was performed at 512 cells per dimension. We used a total

333: variation diminishing (TVD) scheme \citep{Kim_etal99}. Runs R1 and R2

334: were integrated over 10~Myr, before self-gravity was turned on (5

335: turbulent crossing times). This is a standard procedure aimed at allowing the

336: system to attain a

337: well-developed, stationary turbulent 

338: state by the time gravity is turned on, thus preventing it from

339: ``capturing'' features produced directly by the artificial random

340: driver. Note that, however, one turbulent crossing time is 

341: generally enough to 

342: achieve such a stationary state \citep{BP_etal06}. 

343: Thus, run R3 (the more expensive of

344: these simulations in terms of computational time) 

345: was run only for one dynamical time before self-gravity was turned on. 

346: 

347: Once self-gravity was turned on, runs R1, R2 and R3 were 

348: integrated over 13, 9.1, and 4.7 Myr respectively. 

349: In reality, however, as suggested by the estimated ages

350: of the stellar population in the Solar Neighborhood \citep{BHV99,

351: HBB01, BH07}, molecular clouds should 

352: live not much longer than 5 Myr,

353: since they are destroyed by stellar winds, bipolar outflows, SNe explotions, 

354: etc. For this reason, we only used the first 8 Myr of the evolution of runs 

355: R1 and R2 with self-gravity.

356: 

357: Note that run R1 was used to perform the analysis over all types of

358: cores (collapsed and failed).  However, since only two collapse events

359: were found in R1, we used runs R2 and R3 to verify that the timescales

360: for formation and collapse obtained in R1 were consistent, and

361: independent of the resolution.  Run R2 produced two collapse

362: events, while run R3 produced three.

363: 

364: 

365: \section{Measurement procedure} \label{sec:procedure}

366: 

367: \subsection{General considerations} \label{sec:gral_cons}

368: 

369: Just as with observations, estimating the core lifetimes in the

370: simulations is not a straightforward task. The numerical simulations

371: have both advantages and disadvantages compared to the

372: observations. First, in the simulations, 

373: the density, velocity

374: and magnetic fields defined over the three spatial and one temporal 

375: dimensions are given. 

376: However, the cores, defined as we describe below, are elusive entities 

377: that in general do

378: not preserve their identity over time; i.e., they do not involve the

379: same mass at different times. After all, defining a core amounts to

380: defining a certain region of space in what is really a fluid

381: continuum. We choose to define a core as the set of connected grid cells

382: surrounding a local density peak and having densities above a given

383: density threshold $\nthr$. This definition mimics the emission observed

384: when using tracers sensitive to different density regimes. We have

385: chosen not to use an algorithm such as CLUMPFIND 

386: \citep{Williams_etal94} 

387: because it can sometimes

388: introduce artificial divisions across mostly continuous objects. Our

389: chosen definition allows large-scale clumps to be

390: defined at low $\nthr$ even if the region contains more than one

391: local density maximum.

392: 

393: In general, the cores are more

394: poorly defined as lower values of $\nthr$ are considered, while instead this

395: problem disappears as one considers higher $\nthr$.

396: Once defined, the cores must be followed over

397: time. The animation of R1 associated to Fig.\ \ref{fig:core_evol}

398: in the electronic

399: edition shows the evolution of the cores when $\nthr = 30~n_0 =

400: 1.5 \times 10^4 \pcc$, illustrating how the failed cores appear and

401: disappear, and sometimes merge or split. It also illustrates a merger of

402: collapsing cores (cf.\ \S \ref{sec:collapsing}).

403: The measured core properties (e.g., volume $V$, mass $m$, and mean

404: density $\bar{n}$) depend on $\nthr$, and furthermore, at a fixed value

405: of $\nthr$, these properties vary in time for every core.

406: Note in particular that the cores are not generally composed of the same

407: fluid parcels throughout their history.

408: Peak densities and their associated coordinates within a core

409: do not depend on $\nthr$, which only defines the time at which these two

410: properties begin to be measured.

411: 

412: The cores in the simulations can either collapse or rebound (no

413: stable hydrostatic solutions are possible in the isothermal, magnetically

414: supercritical regime; see Paper I). Thus, there exist two classes of

415: starless cores: those that are on route to collapse, to which we refer

416: as ``prestellar'', and those that will eventually redisperse back to

417: the surrounding medium, to which we refer as ``failed'' cores. In

418: addition, we refer to cores that have already collapsed as ``stellar''.

419: With these considerations in mind, we see that the number ratio of

420: starless to stellar cores is given by

421: %

422: \begin{equation}

423: \frac{N_{\star\mathrm{less}}}{N_\star}=\frac{N_\mathrm{pre}} {N_\star}

424: +\frac{N_\mathrm{f}}{N_\star} ,

425: \label{eq:starless2stellar}

426: \end{equation}

427: %

428: where $N_{\star\mathrm{less}}$, $N_{\rm pre}$, $N_\star$ and $N_{\rm f}$

429: are respectively the numbers of starless, prestellar, stellar and failed

430: cores.

431: 

432: Our goal in the present paper is to determine the lifetimes of the cores

433: and their number ratios as a function of the threshold used to define

434: them or, almost equivalently, as a function of their mean densities. The

435: lifetimes that are reported in observational studies correspond

436: to the duration of the prestellar stage, and are 

437: normally estimated by 

438: measuring the number of cores in the prestellar and stellar stages, and

439: assuming that the ratio of their numbers equals the ratio of

440: their durations. However, it was shown in Paper I that the ratio of the

441: numbers is an upper limit to the ratio of durations if there exists a

442: population of failed cores that is not accounted for. 

443: Furthermore, as discussed in \citet{BH07}, potential 

444: source count incompleteness in several observational surveys also

445: contributes to make this an upper limit.

446: 

447: In principle, one 

448: could simply measure the numbers of the various types

449: of cores (failed, prestellar and stellar) in the simulations. However, a

450: number of limitations prevent us from doing so. We have found that the

451: collapsed, or stellar, cores in our simulations are clearly

452: under-resolved. Their masses are typically between 50 and 100 $\Msun$,

453: while the smallest and densest cores in, e.g., Perseus, 

454: have masses $\sim 1 \Msun$,

455: although they are clustered in groups of up to several tens of them

456: \citep[see, for example, Figs.\ 4 and 5 of][]{Kirk06}. 

457: Specifically, we have extracted the masses of the clustered submillimeter

458: (submm) cores in supercore No.\ 2 and supercore No.\ 7 from tables 1, 2

459: and 3 of \citet{Kirk06}. We find a total mass in these clustered submm

460: cores of  

461: $\sim 7~\Msun$ for supercore No.\ 2  and $\sim 32~\Msun$ for supercore

462: No.\ 7. In comparison, the masses at the time of peak density saturation

463: of each of the collapsed objects in R1 at $\nthr

464: = 1.2 \times 10^5 \pcc$ (emulating the density of the submm cores) are

465: $61~\Msun$ and $62~\Msun$ respectively. 

466: Thus, each of our

467: cores with collapsed objects corresponds most directly to a cluster of

468: protostellar cores.

469: 

470: Another important issue is that the number of objects fluctuates

471: statistically in time. This

472: suggests that we should consider appropriately weighted time averages in

473: order to obtain the most representative numbers. As mentioned, our analysis 

474: in R1 was 

475: performed over the first 8 Myr of the simulation after gravity was turned

476: on, except for the failed cores at the lowest $\nthr$, for which a

477: sufficiently large statistical sample was obtained considering only the

478: first 4 Myr (due to the large number of these objects). Analysis in R2 and

479: R3 consisted only of collapse time measurements taken over 

480: 8 Myr and 5 Myr (the duration of R3) after gravity is turned on respectively.

481: 

482: The time interval

483: between outputs in the simulations was 0.04 Myr, and we used four different 

484: values of the

485: density threshold: $\nthr = 1.5 \times 10^4, ~3 \times 10^4, ~6

486: \times 10^4,$ and $1.2 \times 10^5 \pcc$. Higher threshold densities were not

487: used because cores with $n > 256~n_0 = 1.28 \times 10^5 \pcc$ are not

488: well resolved in all the simulations according to the Jeans condition

489: \citep{True_etal97}, as

490: explained in Paper I.

491: Thresholds lower than $20~n_0 = 10^4 \pcc$ were not investigated either,

492: because at such low densities the cores become impossible to follow in

493: time due to their poorly defined identities (see, for example, the

494: animations accompanying Paper I). In the next subsections we describe

495: how we estimate the numbers and lifetimes of the various kinds of cores.

496: 

497: \subsection{Collapsing cores} \label{sec:collapsing}

498: 

499: \subsubsection{Time scales} \label{sec:coll_timescales}

500: 

501: When a core collapses in the simulations, in practice its peak

502: density $\npk$ does not increase to infinity, but rather reaches a

503: saturated value $\nsat \sim$ 2--3 $\times 10^6

504: \pcc$ at $256^3$ resolution or $\nsat \sim$ 1 $\times 10^7 \pcc$ in the $512^3$

505: simulation, beyond which the collapse cannot

506: be followed further by the code. 

507: In this state, the mass of the collapsed object

508: is spread out over a few pixels in each direction.

509: This process is

510: illustrated in Fig.\ \ref{fig:core_peak_dens_evol}, which shows the

511: evolution of the peak densities of the objects that collapse

512: in the three runs.

513:  Upper, middle and lower panels represent the peak density as a

514:  function of time for runs R1, R2, and R3, respectively. 

515: The saturated density of a collapsed

516: core in general oscillates in time. This is because the code contains no

517: prescription for ``capturing'' the mass in the collapsed objects (the

518: analogue of a ``sink particle'' in SPH codes, for example), and

519: therefore this mass still interacts with its surroundings. Also, $\nsat$

520: increases slowly on average, because 

521: the collapsed object continues to slowly accrete mass.

522: 

523: 

524: We define the \emph{prestellar} lifetime of a core, $\tpre$, as

525: the interval between the

526:  time when it is first detected at a given threshold $\nthr$, and the

527:  time when its peak density reaches the saturated value. This

528:  implicitly assumes that any further subfragmentation of the core

529:  occurs essentially simultaneously, so that all protostars form at roughly 

530: the same time, at least within the precision of our temporal resolution. This

531: definition of the prestellar lifetime neglects the

532: duration of the final stages of collapse (from

533: the time when the density saturates in the simulation to the time at

534: which actual protostellar densities would be reached). This

535: approximation should not introduce a large error, since the free-fall

536: time at the saturation densities is

537: %

538: \begin{equation}

539: \tff \equiv (3 \pi/32 G \rho)^{1/2} \sim 2 \times 10^{4} ~\hbox{yr},

540: \label{eq:free-fall_time}

541: \end{equation}

542: %

543: which is significantly

544: shorter than the temporal resolution of

545: $4 \times 10^{4}$ yr between snapshots in our simulation.

546: In any case, this slight underestimate of

547: the true prestellar lifetime acts in the opposite direction of any delay of

548: the collapse introduced by numerical diffusion, and thus the two effects

549: should tend to cancel each other out (see the 

550: discussion in \S \ref{sec:caveats}).

551: 

552: The measured prestellar lifetime obviously depends on $\nthr$, since this

553: quantity defines the starting time of a core's observed

554: ``life''. Empirically, we have found that the mean density in the cores

555: scales close to linearly with $\nthr$ \citep[e.g.,][]{VBR97}, and so

556: this means that the measured prestellar lifetime should be a function of

557: the mean density as well, in qualitative agreement with observational

558: data \citep{WT_etal07}.

559: 

560: We ignore mergers of collapsing cores in the measurement of

561: $\tpre$. That is, if two cores appear at times $t_1$ and $t_2 > t_1$

562: respectively, and they merge at a later time $t_3$ into a new core that

563: eventually collapses, we take the prestellar lifetime $\tpre$ as

564: the time elapsed between $t_1$ and the time of density saturation. 

565: We do this motivated by the fact that the peak densities 

566: $\npk$ of cores that merge into a new core which collapses later are

567: seen to rise rapidly even before the merging event, suggesting that

568: these cores were already on their own independent route to collapse,

569: rather than the latter being triggered by the merging event. For

570: instance, the first two cores in R1 (upper panel of Fig.\

571: \ref{fig:core_peak_dens_evol}, {\it solid} and {\it dashed} lines

572: respectively) are seen to appear at roughly the same time ($t \sim

573: 1.4$ Myr), and to merge shortly thereafter. Thus, we consider them

574: together as the single first collapse event in R1.  The third core is seen to

575: have very similar physical properties to those of the merger, and we

576: refer to its collapse as the \emph{second} collapse event in R1.  In

577: Fig.\ \ref{fig:coll_times} we report the prestellar lifetimes measured

578: for the collapse events in R1 ({\it diamonds}), R2 ({\it triangles})

579: and R3 ({\it squares}) as a function of their mean densities (which in

580: turn are $\sim \nthr$).

581: 

582: It is important to remark that the long durations of the collapsed

583: objects \emph{after} they have collapsed seen in Fig.\

584: \ref{fig:core_peak_dens_evol} does not imply that the stellar cores

585: last that long, because no modeling of stellar energy feedback (e.g.,

586: winds, bipolar outflows, ionizing radiation) is included in the

587: numerical model, and thus the cores cannot be dispersed after they

588: produce collapsed objects.

589: 

590: \subsubsection{Number ratios} \label{sec:number_coll}

591: 

592: As mentioned in \S \ref{sec:gral_cons}, we cannot simply count the

593: number of cores with collapsed objects because it is clear that at the

594: resolutions we

595: used their subfragmentation was not appropriately captured. Thus,

596: instead, we proceed inversely to the standard observational procedure,

597: and estimate the number ratio of prestellar to stellar cores simply as

598: the ratio of the prestellar lifetime to the lifetime of the embedded

599: protostellar phase:

600: %

601: \begin{equation}

602: \frac{N_\mathrm{pre}}{N_\star}=\frac{\overline{\tau_\mathrm{pre}}}

603: {\tau_\mathrm{yso}},

604: \label{eq:Npre/Nstar}

605: \end{equation}

606: %

607: where $\overline{\tau_\mathrm{pre}}$ is the average of the measured prestellar

608: lifetimes for the

609: seven collapse events in the simulations, and $\tyso$ is the duration of the

610: protostellar stage. We

611: emphasize that eq.\ (\ref{eq:Npre/Nstar}) allows us to estimate the

612: number \emph{ratio} of prestellar to stellar cores, but not the actual

613: numbers, and that this ratio is a function of the threshold density,

614: since $\tau_\mathrm{pre}$ is.

615: 

616: For us, $\tau_\mathrm{yso}$ is a free parameter because it is outside

617: the realm of our simulations. As mentioned in \S

618: \ref{sec:coll_timescales}, no modeling of the feedback from the

619: protostars such as bipolar outflows, ionization radiation or winds is

620: included, so there are no energy sources available for dispersing the

621: cores once they are formed. Thus, we have to consider it as a free

622: parameter in our study, whose value is to be set on the basis of external

623: information.

624: 

625: Observationally, $\tyso$ is normally estimated by obtaining the ratio of

626: the number of embedded objects to that of the T Tauri stars \citep[e.g.,

627: ][]{WLY89, Ken90, Greene_etal94, KH95, Hat06}. Values for

628: $\tau_\mathrm{yso}$ reported in the literature are $\sim 0.6\pm 0.3$ Myr

629: \citep[see review by][ and references therein]{WT_etal07}, the spread

630: being due to both observational constraints and real differences 

631: among the

632: observed regions.

633: 

634: We take $\tau_\mathrm{yso}=0.46$ Myr, the mean value

635: obtained for Perseus by \citet{Hat06}. We base this selection on

636: two considerations: First, their estimate of $\tau_\mathrm{yso}$

637: includes corrections for incompleteness of surveys toward very low

638: masses and multiplicity. Second, we have found that our simulations 

639: are reasonably

640: resemblant of a massive ``supercore'' in Perseus, as defined by

641: \citet{Kirk06} (see \S \ref{sec:simulation} and \S \ref{sec:gral_cons}).

642: 

643: \subsection{Failed cores} \label{sec:failed}

644: 

645: Failed cores in our simulations are less

646: dense than collapsing cores, and

647: therefore are not as strongly affected by the limited resolution of our

648: simulations. Nevertheless, we consider it safer to estimate their number

649: ratio with respect to the stellar cores through the ratio of masses in

650: these two kinds of objects. This estimator avoids the uncertainty

651: associated with the possible failure to capture further subfragmentation

652: of the cores. However, even this is still not completely straightforward

653: because we need to take into account temporal effects.

654: 

655: The failed cores ``appear'' at some time and ``disappear'' at a later

656: time at a given density threshold $\nthr$. Furthermore, their masses are

657: continuously varying over this time interval, first increasing, reaching

658: a maximum value, and then decreasing again. In order to obtain an

659: estimator that is most representative of the number ratio expected to be

660: found upon a single instantaneous observation, we 

661: take the ratio of

662: the temporal averages  of the 

663: masses of the failed cores  to those of the

664: stellar cores, weighted by the ratio of their typical durations

665: 

666: \begin{equation}

667: \frac{N_\mathrm{f}}{N_\star}= \frac{\sum_i \langle

668: m_i \rangle \tau_i} {\tau_\mathrm{yso}\sum_j \langle m_j \rangle},

669: \label{eq:failed2stellar}

670: \end{equation}

671: %

672: where the summations over $i$ and $j$ refer to the failed and stellar

673: cores respectively, $\langle m_i \rangle$ and $\tau_i$ are respectively

674: the time-averaged mass and the duration of the $i$-th failed core,

675: and $\tyso$ is the duration of the protostellar phase discussed in \S

676: \ref{sec:collapsing}. For the stellar cores, we take $\langle m_j \rangle$

677: as the average mass over a time interval $\tyso$ after the core has collapsed.

678: 

679: Equation (\ref{eq:failed2stellar}) assumes that 

680: actual failed cores have,

681: on average, the same mass as stellar cores. 

682: However, it is reasonable to expect failed cores to be less  

683: massive in general than individual stellar cores, 

684: and in this case the right-hand side of eq.  

685: (\ref{eq:failed2stellar}) is a lower limit for the failed-to-stellar core

686: ratio\footnote{See, however, the discussion in \S

687: \ref{sec:comparison}

688: concerning the possibility that failed cores 

689: are hard to detect even at low densities, an effect that would act in the

690: opposite direction, lowering the observed failed to stellar core ratio.}.

691: Furthermore, eq. \ref{eq:failed2stellar} also assumes that no 

692: fragment of a failed core ever ends up collapsing, and no fragment of a

693: stellar core ends up redispersing. This is a reasonable assumption,

694: since subfragmentation of the cores is mostly a gravitational 

695: phenomenon, which occurs after the onset of collapse. Thus, collapsing

696: cores are expected to fragment into further collapsing units, while

697: failed cores are not expected to fragment because they do not engage

698: into collapse. In fact, the failed cores in our simulations generally have

699: very small, sub-solar  masses (see \S \ref{sec:comparison}).

700: 

701: 

702: \subsection{The starless to stellar ratio} \label{sec:svsratio}

703: 

704: Equations (\ref{eq:Npre/Nstar}) and (\ref{eq:failed2stellar}) give us

705: the prescriptions to estimate the terms on the right-hand side of eq.\

706: (\ref{eq:starless2stellar}). We can then estimate number ratios with

707: respect to the total number of cores $N_\mathrm{tot}= N_\mathrm{pre}+

708: N_\mathrm{f}+N_\star$ in terms of the ratios derived in eqs.\

709: (\ref{eq:starless2stellar}), (\ref{eq:Npre/Nstar}), and

710: (\ref{eq:failed2stellar}) as follows:

711: %

712: \begin{eqnarray}

713: \frac{N_\mathrm{pre}}{N_\mathrm{tot}}&=&\biggl[ \frac{

714: N_{\star\mathrm{less}} /N_\star} {N_\mathrm{pre}/N_\star} +

715: \biggl(\frac{N_\mathrm{pre}}{N_\star}\biggr)^{-1}\biggr]^{-1}\\

716: \frac{N_\mathrm{f}}{N_\mathrm{tot}}&=&\biggl[\frac{N_{\star\mathrm{less}}

717: /N_\star}{N_\mathrm{f}/N_\star}+\biggl(\frac{N_\mathrm{f}}{N_\star}

718: \biggr)^{-1}\biggr]^{-1}\\

719: \frac{N_{\star\mathrm{less}}}{N_\mathrm{tot}}&=&\biggl[1 +

720: \biggl(\frac{N_{\star\mathrm{less}}}{N_\star}\biggr)^{-1}\biggr]^{-1}\\

721: \frac{N_\star}{N_\mathrm{tot}}&=&\biggl[1 +

722: \frac{N_{\star\mathrm{less}}} {N_\star}\biggr]^{-1}

723: \end{eqnarray}

724: %

725: These fractions are shown in Fig.\ \ref{fig:number_ratios} ({\it left

726: panel}) as a function of $\nthr$. The {\it right panel} of this figure

727: shows the same ratios but normalized to the number of \emph{stellar} cores.

728: 

729: \section{Discussion} \label{sec:discussion}

730: 

731: \subsection{Comparison with previous work} \label{sec:comparison}

732: 

733: In addition to the core lifetimes as a function of their mean densities,

734: Fig.\ \ref{fig:coll_times} shows two lines indicating the locus of

735: $t=\tff$ and $t=10 \tff$, with $\tff$ defined in eq.\

736: (\ref{eq:free-fall_time}). This figure can be compared with Figure 2 of

737: \citet{WT_etal07}. It is seen that, within the uncertainties,

738: the core lifetimes in our simulations are in good agreement with the

739: observational values, being in general a few to several times the

740: free-fall time. However, such a comparison should be taken with care

741: because in the determination of observational lifetimes, the assumed

742: values for $\tyso$ span one order of magnitude (from $10^5$ to $10^6$

743: yr). In any case, one can unambiguously conclude that \emph{the prestellar

744: core lifetimes in the turbulent scenario are a few times $\tff$}, and not

745: only $1 \tff$ as is often stated in literature \citep[e.g.,][]{JKirk05,

746: WT_etal07}.

747: 

748: These results can be understood as a consequence of several factors. 

749: First, the

750: cores take some time to be assembled by the external turbulent

751: compressions before they become gravitationally unstable

752: \citep{Gomez_etal07}, and even after that, some readjustments of the

753: mass and some oscillations are expected to occur before gravitational

754: collapse sets in. Moreover, as pointed out by \citet{Larson69}, the

755: duration of the actual collapse is longer than the free-fall time

756: because the thermal pressure gradient is never negligible. So, even though 

757: the cores in the simulation are formed dynamically, their lifetimes are

758: fully consistent with observational estimates.

759: 

760: The number ratio of prestellar to stellar cores in our simulations and

761: the ratio of durations of these stages at low values of $\nthr$ are $\sim

762: 3$ (Fig.\ \ref{fig:number_ratios}, {\it right panel}). The same value is

763: obtained by \citet{LM99}, who studied an

764: optically-selected sample of cores, characterized by relatively low mean

765: densities, $\lesssim 10^4

766: \pcc$. Note, however, that we obtain a number ratio of

767: \emph{total} starless (prestellar + failed) to stellar cores of $\sim

768: 5$, larger than the observation of \citet{LM99}. This suggests that

769: their sample consisted mostly of prestellar cores, possibly because the

770: failed cores tend to be smaller and therefore less easily

771: detected. Indeed, Fig.\ \ref{fig:size_hist} ({\it left panel}) shows a

772: size histogram of the failed cores at $\nthr = 30~n_0 = 1.5 \times 10^4

773: \pcc$ for R1, and it can be seen that virtually all of them have sizes $\ell <

774: 0.2$ pc, with $\ell \sim 0.07$ pc being a characteristic size; instead, 

775: \citet{LM99} report a mean size $\ell \simeq 0.2$ pc. Thus, real 

776: failed cores may be harder to detect observationally, and

777: therefore not very common in observational surveys, even of low-density

778: cores. For reference, a histogram of the failed core masses is shown in

779: the {\it right panel} of Fig.\

780: \ref{fig:size_hist} as well.

781: 

782: At high thresholds, $\nthr = 1.2\times 10^5 \pcc$, the ratio $N_{\rm

783: pre}/N_\star \simeq 1.4$. This is consistent with the observed value of $\sim 1$

784: for this ratio in recent surveys of dense, submm cores

785: \citep{Hat06,Jor07},

786: sensitive to $n \sim 10^5 \pcc$. Thus, with a single assumed value of

787: $\tyso$, our estimated number ratios are in good agreement with

788: observations at both low and high densities. Moreover, the number of

789: failed cores at the higher densities goes to zero, suggesting that indeed

790: submm surveys sample essentially all gravitationally bound

791: cores, thus tracing the final stages of core evolution prior to

792: protostar formation.

793: 

794: Our estimated core lifetimes are also in good agreement with the

795: predictions from the AD-mediated model applied to moderately subcritical

796: clouds, either in analytical \citep{CB01} or in numerical \citep{NL05}

797: form. The latter authors report lifetimes 1.5--10 times the local

798: free-fall time in the collapsing cores, while the lifetimes of our cores

799: are $\sim 6~\tff$ (see Fig.\ \ref{fig:coll_times}). The similarity between

800: the mean lifetimes in both cases occurs because, on the one hand,

801: $\tad$ is only a few times longer than $\tff$ at the densities and

802: observed values of the mass-to-magnetic flux ratio of molecular cloud

803: cores while, on the other, the cores in our simulations are not

804: necessarily in a direct route to collapse when first detected, but

805: instead undergo a period of 

806: build up and readjustment, as mentioned above. Therefore,

807: we conclude that both models give comparable

808: predictions for the lifetimes of the cores in their observable stages,

809: with AD not significantly delaying the final stages of contraction and

810: collapse.

811: 

812: \subsection{Caveats, limitations, and error estimates} \label{sec:caveats}

813: 

814: The study presented in this paper has faced a number of difficulties

815: comparable to those encountered by observational studies, and in this

816: sense, its results are only suggestive, rather than conclusive. The

817: main limitations were:

818: 

819: \medskip

820: \noindent

821: 1. {\it The limited resolution of the simulations}. This

822: limitation prevented us from 

823: correctly following the fragmentation of collapsing cores, and

824: forced us to estimate the number ratio of prestellar to stellar cores

825: through the ratio of their durations ($\tau_\mathrm{pre}/\tyso$) rather than

826: through

827: direct counting, with $\tyso$ being a free parameter of our study. 

828: Similarly, the

829: number ratio of failed to stellar cores had to be estimated in terms of

830: the ratio of their masses, again due to the inability to resolve

831: individual stellar cores.

832: 

833: The limited resolution also restricted the number of collapse events to

834: only $\sim 2$ in each of our runs, although each collapse event clearly

835: corresponds to the formation of a stellar cluster (i.e., to many collapse events

836: leading to

837: individual stars), since the masses of the 

838: collapsed objects are 50--100 $\Msun$. Unfortunately, we estimate that, in

839: order to adequately resolve these objects into solar-mass-like cores, an

840: increase of at least a factor of $50^{1/3}$--$100^{1/3}$, or $\sim$ 4 in

841: resolution would be required (i.e., resolutions $> 1024^3$), or the

842: usage of a Lagrangian-type of numerical algorithm (SPH or

843: adaptive-mesh-refinement), in both magnetic and non-magnetic regimes. We

844: hope to perform such a study in the near future.

845: 

846: The limited resolution might also possibly affect the collapse

847: times measured in the simulation (see \S

848: \ref{sec:coll_timescales}), both because of the saturation values of the

849: density reached by the collapsed objects as well as because of the possible

850: slowing down of the collapse due to numerical viscosity. To test for this, 

851: Figures \ref{fig:rhomax_conv_test} and \ref{fig:colltime_vs_resol} 

852: illustrate the dependence of collapse 

853: on resolution. Figure \ref{fig:rhomax_conv_test} shows the evolution of the

854: maximum density $\nmax$ in four simulations with the same global

855: parameters (turbulent Mach number $\sim 10$, Jeans number $J=4$) at

856: resolutions $64^3$ ({\it red line}), $128^3$ ({\it green line}), $256^3$

857: ({\it dark-blue line}), and $512^3$ ({\it clear-blue line}). The collapse

858: in each simulation consists of the sharp transition from a low

859: typical level of the maximum density ($\nmax \lesssim 100~n_0 = 5 \times

860: 10^4 \pcc$) to a high one ($\nmax \gtrsim 10^3~n_0 = 5 \times 10^5

861: \pcc$). We define the ``collapse time'' of the simulations

862: as the duration of the sharp rise of $\nmax$ over more than

863: one order of magnitude separating the two levels. This is well defined

864: for resolutions equal or larger than $128^3$. For economy reasons, the

865: simulation at $512^3$ (i.e., R3) is only run for a short time compared to the

866: other simulations. In all cases, there is seen to be a rebound from the 

867: maximum.

868: 

869: Figure \ref{fig:colltime_vs_resol} summarizes the collapse times as a

870: function of resolution. For the $128^3$ simulation the collapse time 

871: corresponds directly to the measured duration of the rise of $\nmax$ for

872: the first collapse, and the error bar denotes the full width of the

873: rebound around the saturated value. For the $256^3$ (R1) and $512^3$ (R3)

874: simulations, the collapse times correspond to the average of the durations

875: measured for the available collapse events. The error bars denote the range of

876: values found. This figure shows that the collapse time appears to be

877: independent of resolution for resolutions $128^3$ and above, meaning that

878: our longer-than-free-fall collapse times are not an artifact of the

879: resolution.

880: If anything, Figure \ref{fig:colltime_vs_resol} shows that there is a

881: slight trend

882: of the collapse time to {\it increase} with resolution, probably

883: suggesting that the effect of traversing a larger dynamic range in

884: density at higher resolution dominates over any slowing down of the

885: collapse by numerical viscosity at lower resolution.

886: 

887: \medskip

888: \noindent

889: 2.  {\it The simulations do not include ambipolar diffusion (AD)}. Even

890: though the simulations were supercritical, in the presence of AD there

891: would exist the possibility that some failed cores could be ``captured''

892: by AD if the main agent supporting them is the magnetic energy, since in

893: this case AD could cause a redistribution of the magnetic field, leaving

894: the centermost parts of the core with less support than in the case

895: without AD. This, however, does not appear to be the case. For example,

896: the longest-lasting failed core in R1, with a lifetime of

897: $\sim 1.9$ Myr defined at $\nthr = 3 \times 10^4 \pcc$, is clearly

898: supercritical ($\mu \approx 4.0 \pm 2.1$ times critical, where the

899: uncertainty comes from the estimation of its ``radius'', since the core

900: is far from spherical; see Paper I), indicating that its failure to

901: collapse was due to thermal+turbulent rather than magnetic support.

902: 

903: In any case, a worst-case estimate of the error committed by the lack of AD

904: can be made by assuming that all failed cores with lifetimes longer than

905: a representative AD timescale $\tad$ proceed to collapse rather than

906: rebound. A reasonable estimate for $\tad$ is $\sim 1.5$ Myr (see

907: the Appendix in Paper I). Being even more restrictive, we recalculated the

908: number ratios assuming that the failed cores in R1 with lifetimes $\tau > 0.9$

909: Myr are ``captured'' by AD (see Fig.\ \ref{fig:number_ratios_AD}). In this

910: case, we find that the estimated

911: number ratios of failed cores decrease by a factor of $\sim 2$, but this

912: does not significantly affect the starless to stellar core ratio,

913: because the failed cores have very low masses.

914: 

915: \subsection{Conclusions} \label{sec:concl}

916: 

917: In spite of its limitations, the results of the present study are

918: nevertheless encouraging, as the prestellar lifetimes measured directly from

919: the density evolution of the cores are in good agreement with

920: observational determinations. The same is true for the number ratios of

921: prestellar to stellar cores. This estimation depends on the free

922: parameter $\tyso$, but a single value for it, chosen as the mean of the

923: range reported by \citet{Hat06} for Perseus, produces number ratios that

924: are in good agreement with observational determinations at both high and

925: low values of the density threshold $\nthr$. The present study thus

926: suggests that the turbulent model of star formation is not inconsistent

927: with observational determinations of core lifetimes and the number

928: ratios of prestellar to stellar cores.

929: 

930: \acknowledgements

931: We thank Gilberto C. G\'omez for help with IDL visualization. 

932: We are also grateful for the helpful comments provided by an anonymous referee. 

933: This work has received partial financial support from CONACYT grant U47366-F

934: to E. V.-S. The work of J. Kim was supported by the Astrophysical Research

935: Center for

936: the Structure and Evolution of the Cosmos (ARCSEC) of Korea Science and

937: Engineering Foundation (KOSEF) through the Science Research Center (SRC)

938: program.

939: The numerical simulations were performed on the Linux cluster

940: at KAO, with funding from KAO and ARCSEC.

941: 

942: \begin{thebibliography}{}

943: 

944: \bibitem[\BP~et al.\ (1999a)]{BVS99} \BP, J., \VS, E., \& Scalo,

945: J. 1999a, ApJ, 515, 286

946: 

947: \bibitem[\BP~et al.\ (1999b)]{BHV99} \BP, J., Hartmann, L., \&

948: \VS, E. 1999b, ApJ, 527, 285

949: 

950: \bibitem[Ballesteros-Paredes et al.(2006)]{BP_etal06}

951: Ballesteros-Paredes, J., Gazol, A., Kim, J., Klessen, R.~S., Jappsen,

952: A.-K., \& Tejero, E.\ 2006, \apj, 637, 384

953: 

954: \bibitem[Ballesteros-Paredes \& Hartmann(2007)]{BH07}

955: Ballesteros-Paredes, J., \& Hartmann, L. \ 2007, Rev. Mex. AA, 43, 123

956: 

957: \bibitem[\BP\ et al.\ (2007)]{BKMV07} \BP, J., Klessen, R., Mac Low,

958: M.-M., \& \VS, E. 2007, in Protostars and Planets V, ed. B. Reipurth,

959: D. Jewitt, \& K. Keil (Tucson: Univ. of Arizona Press), 63  

960: 

961: \bibitem[Beichman et al.\ (1986)]{Beich86} Beichman, C. A., Myers,

962: P. C., Emerson, J. P., Harris, S., Mathieu, R., Benson, P. J., \&

963: Jennings, R. E. 1986, ApJ, 307, 337

964: 

965: \bibitem[Blitz (1993)]{Blitz93} Blitz, L., 1993, in Protostars and

966: Planets III, ed. E. H. Levy, \& J. I. Lunine (Tucson: Univ. of

967: Arizona Press), 125

968: 

969: \bibitem[Ciolek \& Basu (2001)]{CB01} Ciolek, G. E., \& Basu, S. 2001,

970: ApJ, 547, 272  

971: 

972: \bibitem[Crutcher (1999)]{Crut99} Crutcher, R. M. 1999 ApJ, 520, 706

973: 

974: \bibitem[Elmegreen (1993)]{Elm93} Elmegreen, B. G. 1993, ApJ 419, L29

975: 

976: \bibitem[Elmegreen (2000)]{Elm00} Elmegreen, B. G. 2000, ApJ, 530, 277

977: 

978: \bibitem[Fatuzzo \& Adams (2002)]{FA02}

979: Fatuzzo, M., \& Adams, F. C. 2002, ApJ, 570, 210

980: 

981: \bibitem[Fiedler \& Mouschovias (1993)]{FM93} Fiedler, R. A., \&

982: Mouschovias, T. 1993 ApJ, 415, 680

983: 

984: \bibitem[G\'omez et al.\ (2007)]{Gomez_etal07} G\'omez, G. C., \VS, E.,

985: Shadmehri, M., \& \BP, J. 2007, preprint (arXiv:0705.0559)

986: 

987: \bibitem[Greene et al.\ (1994)]{Greene_etal94} Greene, T. P.,

988: Wilking, B. A., Andre, P., Young, E. T., \& Lada, C. J. 1994, ApJ, 434, 614

989: 

990: \bibitem[Hartmann et al.\ (2001)]{HBB01} Hartmann, L.,

991: Ballesteros-Paredes, J., \& Bergin, E. A. 2001, ApJ, 562, 852

992: 

993: \bibitem[Hatchell et al.\ (2007)]{Hat06} Hatchell, J., Fuller, G. A.,

994: Richer, J. S., Harries, T. J., \& Ladd, E. F. 2007, A\&A, 468, 1009

995: 

996: \bibitem[Heitsch et al.\ (2004)]{Heitsch_etal04} Heitsch, F., Zweibel,

997: E. G., Slyz, A. D., \& Devriendt, J. E. G. 2004, ApJ, 603, 165

998: 

999: \bibitem[Heyer \& Brunt (2004)]{Heyer_Brunt04} Heyer, M.~H., \& Brunt,

1000: C.~M.\ 2004, \apjl, 615, L45

1001: 

1002: \bibitem[Heyer \& Brunt (2007)]{Heyer_Brunt07} Heyer, M.~H., \& Brunt,

1003: C.~M.\ 2007, in IAU Symposium 237, Triggered Star Formation in a

1004: Turbulent ISM, ed. B. Elmegreen \& J. Palous (Cambridge: Cambridge

1005: University Press), 9

1006: 

1007: \bibitem[Jijina et al.\ (1999)]{JMA99} Jijina, J., Myers, P. C., \&

1008: Adams, F. C. 1999, ApJS, 125, 161

1009: 

1010: \bibitem[J$\o$rgensen et al.\ (2007)]{Jor07} J\o rgensen, J. K.,

1011: Johnstone, D., Kirk, H., \& Myers, P. C. 2007, ApJ, 656, 293

1012: 

1013: \bibitem[Kenyon et al.\ (1990)]{Ken90}

1014: Kenyon, S. J., Hartmann, L. W., Strom, K. M., \& Strom, S. E.

1015: 1990, AJ, 99, 869

1016: 

1017: \bibitem[Kenyon \& Hartmann (1995)]{KH95} Kenyon, S. J., \& Hartmann,

1018: L. 1995, ApJS, 101, 117

1019: 

1020: \bibitem[Kim et al.\ (1999)]{Kim_etal99} Kim, J., Ryu, D., Jones, T. W.,

1021: \& Hong, S. S. 1999, ApJ, 514, 506

1022: 

1023: \bibitem[Kirk et al.\ (2006)]{Kirk06} Kirk, H., Johnstone, D., \& Di Francesco, J. 2006, ApJ, 646, 1009

1024: 

1025: \bibitem[Kirk et al.\ (2005)]{JKirk05} Kirk, J. M., Ward-Thompson, D.,

1026: \& Andr\'e, P. 2005, MNRAS, 360, 1506

1027: 

1028: \bibitem[Larson(1969)]{Larson69} Larson, R.~B.\ 1969, \mnras,

1029: 145, 271

1030: 

1031: \bibitem[Larson(1981)]{Larson81} Larson, R.~B.\ 1981, \mnras,

1032: 194, 809

1033: 

1034: \bibitem[Lee \& Myers (1999)]{LM99} Lee, C. W., \& Myers, P. C. 1999,

1035: ApJS, 123, 233

1036: 

1037: \bibitem[Mac Low \& Klessen (2004)]{MK04} Mac Low, M.-M., \& Klessen,

1038: R. S. 2004, Rev. Mod. Phys., 76, 125

1039: 

1040: \bibitem[Maddalena \& Thaddeus (1985)]{Maddalena_Thaddeus85}

1041: Maddalena, R. J., \& Thaddeus, P. 1985, ApJ, 294, 231

1042: 

1043: \bibitem[Mouschovias (1976)]{Mou76} Mouschovias, T. C. 1976, ApJ, 207, 141

1044: 

1045: \bibitem[Mouschovias (1991)]{Mou91} Mouschovias, T. Ch. 1991, in The

1046: Physics of Star Formation and Early Stellar Evolution, ed. C. J. Lada,

1047: \& N. D. Kylafis (Dordrecht: Kluwer), 449

1048: 

1049: \bibitem[Nakamura \& Li (2005)]{NL05} Nakamura, F., \& Li, Z.-Y. 2005,

1050: ApJ, 631, 411

1051: 

1052: \bibitem[Onishi et al.\ (2002)]{Oni02} Onishi, T., Mizuno, A., Kawamura,

1053: A., Tachihara, K., \& Fukui, Y. 2002, ApJ, 575, 950

1054: 

1055: \bibitem[Padoan \& Nordlund (2002)]{PN02}

1056: Padoan, P., \& Nordlund, \AA\ 2002, ApJ, 576, 870

1057: 

1058: \bibitem[Shu et al.\ (1987)]{SAL87} Shu, F. H., Adams, F. C., \& Lizano,

1059: S. 1987, ARA\&A, 25, 23

1060: 

1061: \bibitem[Stone et al.\ (1998)]{Stone98} 

1062: Stone, J. M., Ostriker, E. C., \& Gammie, C. F. 1998, ApJ, 508, L99

1063: 

1064: \bibitem[Taylor et al.\ (1996)]{TMW96} Taylor, S. D., Morata, O., \&

1065: Williams, D. A. 1996, A\&A, 313, 269

1066: 

1067: \bibitem[Truelove et al.\ (1997)]{True_etal97} Truelove, J. K., Klein,

1068: R. I., McKee, C. F., Hilliman, J. H. II., Howell, L. H., \& Greenough,

1069: J. A. 1997, ApJ, 489, L179

1070: 

1071: \bibitem[\VS\ et al.\ (1997)]{VBR97} \VS, E., \BP, J., \& Rodr\'iguez,

1072: L. F. 1997, ApJ, 474, 292

1073: 

1074: \bibitem[\VS\ et al.\ (2000)]{VS_etal00} V\'azquez-Semadeni, E.,

1075: Ostriker, E. C., Passot, T., Gammie, C., 

1076: \& Stone, J. 2000, in Protostars and Planets IV, ed. V. Mannings,

1077: A. Boss, \& S. Russell (Tucson: Univ.\ of Arizona Press), 3

1078: 

1079: \bibitem[\VS\ et al.\ (2005a)]{Paper I} \VS, E., Kim, J., Shadmehri, M., \& 

1080: \BP, J. 2005a, ApJ, 618, 344 (Paper I)

1081: 

1082: \bibitem[\VS\ et al.\ (2005b)]{VKB05} \VS, E., Kim, J., \&

1083: \BP, J. 2005, ApJ, 630, L49

1084: 

1085: \bibitem[\VS\ (2007)]{VS07} \VS\ 2007, in IAU Symp. 237, Triggered Star

1086: Formation in a Turbulent ISM, ed. B.G. Elmegreen, \& J. Palous

1087: (Dordrecht: Kluwer), 50

1088: 

1089: \bibitem[Ward-Thompson et al.\ (2007)]{WT_etal07} Ward-Thompson, D.,

1090: Andre, P., Crutcher, R., Johnstone, D., Onishi, T., \& Wilson, C. 2007, in

1091: Protostars and Planets V, ed. B. Reipurth, D. Jewitt, \& K. Keil

1092: (Tucson: Univ.\ of Arizona Press), 33

1093: 

1094: \bibitem[Wilking et al.\ (1989)]{WLY89} Wilking, B. A., Lada, C. J., \&

1095: Young, E. T. 1989, ApJ, 340, 823

1096: 

1097: \bibitem[Williams et al.(1994)]{Williams_etal94} Williams, J.~P., de

1098: Geus, E.~J., \& Blitz, L.\ 1994, \apj, 428, 693

1099: 

1100: 

1101: \end{thebibliography}

1102: 

1103: \begin{figure}

1104: \epsscale{1.0}

1105: \plotone{f1.eps}

1106: \caption{Cores defined at $\nthr = 30~n_0 = 1.5 \times 10^4 \pcc$, and at

1107: $t=1.96$ Myr for R1. This time marks the end of collapse for the first

1108: collapsing core (the biggest core seen near the center of the simulation

1109: box) in this run. At this threshold, failed cores become an important feature

1110: and

1111: many of them coexist at every temporal step. The two ``spots'' at

1112: opposite sides of the $y$-axis boundaries, and at $(x,z) = (203,197)$

1113: correspond to just one failed core crossing the periodic boundaries. An

1114: animation of the simulation for $0.32 \le t/\mathrm{Myr} \le 8$ is

1115: presented in the electronic version of the Journal, with the frames

1116: separated by $\Delta t = 0.04$ Myr. The first 9 frames are omitted,

1117: since there are no structures above $\nthr$ at those times. The figure

1118: shown in the printed version of the Journal corresponds to frame \#40 in

1119: the animation.}

1120: \label{fig:core_evol}

1121: \end{figure}

1122: 

1123: \begin{figure}

1124: \epsscale{0.6}

1125: \plotone{f2.eps}

1126: \caption{Peak density as function of time for collapsing cores in R1 

1127: ({\it upper panel}), R2 ({\it middle panel}) and R3 ({\it lower panel})

1128: after

1129: they have exceeded $n_\mathrm{pk}=6\times10^4\pcc$.  A merging event

1130: occurs among the first two collapsing cores of R1 ({\it solid} and {\it

1131: dashed} lines respectively), but their peak densities are seen to rise

1132: steeply before their merger (the time when the lines converge). 

1133: A merger also occurs in R2, but well after the individual collapse 

1134: of the cores.}

1135: \label{fig:core_peak_dens_evol}

1136: \end{figure}

1137: 

1138: \begin{figure}

1139: \epsscale{1.0}

1140: \plotone{f3.eps}

1141: \caption{Duration of the prestellar stage for the

1142: collapsing cores in the simulations as a function of their

1143: mean density at the time collapse starts being measured. {\it Diamonds},

1144: {\it triangles} and {\it squares} respectively

1145: correspond to the R1, R2 and R3 runs. The loci of

1146: $\tau_\mathrm{pre} = \tff$ and $\tau_\mathrm{pre} = 10~\tff$ is marked by the

1147: lower and upper {\it long-dashed} lines respectively. Also shown is the fit for

1148: the

1149: mean prestellar lifetime as a function of mean density ({\it dash-dotted} line),

1150: which is $\simeq 6~\tff$.  }

1151: \label{fig:coll_times}

1152: \end{figure}

1153: 

1154: \begin{figure}

1155: \epsscale{1.0}

1156: \plottwo{f4a.eps}{f4b.eps}

1157: \caption{{\it Left panel}: Number ratios of the stellar ({\it dashed

1158: line, diamonds}), total starless ({\it solid line, triangles}),

1159: prestellar ({\it dotted line, plus signs}) and failed ({\it dash-dotted

1160: line, crosses}) cores. {\it Right panel}: Number ratios of the total

1161: starless, prestellar and failed cores (same lines and symbols as in the

1162: left panel), normalized to the number of stellar cores.}

1163: \label{fig:number_ratios}

1164: \end{figure}

1165: 

1166: \begin{figure}

1167: %\epsscale{1.}

1168: \plottwo{f5a.eps}{f5b.eps}

1169: \caption{{\it Left:} Size histogram of the failed cores at $\nthr=30~n_0 = 1.5

1170: \times 10^4 \pcc$ for R1. {\it Right:} Mass histogram for the same cores.}

1171: \label{fig:size_hist}

1172: \end{figure}

1173: 

1174: \begin{figure}

1175: \epsscale{0.8}

1176: \plotone{f6.eps}

1177: \caption{Evolution of the maximum density $\nmax$ in four

1178: simulations with the same global parameters (turbulent Mach number $\sim

1179: 10$, Jeans number $J=4$) at resolutions $64^3$ ({\it red line}), $128^3$

1180: ({\it green line}), $256^3$ ({\it dark-blue line}), and $512^3$ ({\it light-blue

1181: line}). The time axis is given in units of the sound crossing time ({\it

1182: lower axis}), equal to 20 Myr, and of the quantity $\Lj/\cs$ ({\it upper

1183: axis}), which is close to the free-fall time. The density axis is in units of

1184: the mean density in simulations $n_0=500 \pcc$. The first collapse event

1185: in each run is denoted by the

1186: sharp transition from a low mean maximum density level to a high one. The

1187: ``collapse time'' of the simulation is defined as the duration of the sharp rise

1188: of $\nmax$ over more than one order of magnitude separating the two

1189: levels. This is well defined in all cases except at $64^3$.}

1190: \label{fig:rhomax_conv_test}

1191: \end{figure}

1192: 

1193: \begin{figure}

1194: \epsscale{1.0}

1195: \plotone{f7.eps}

1196: \caption{Collapse time {\it versus} resolution for the three

1197: highest-resolution runs shown in Fig.\ \ref{fig:rhomax_conv_test}. For

1198: the $128^3$ simulation, this time   

1199: corresponds to the measured duration of the rise of $\nmax$ for

1200: the first collapse, and the error bar denotes the full width of the

1201: rebound around the saturated value. For the $256^3$ and $512^3$

1202: simulations, collapse times correspond to the average of the durations

1203: measured for the collapse events shown in Fig.\

1204: \ref{fig:core_peak_dens_evol}. 

1205: The error bars denote the range of values found.}

1206: \label{fig:colltime_vs_resol}

1207: \end{figure}

1208: 

1209: \begin{figure}

1210: \epsscale{1.0}

1211: \plottwo{f8a.eps}{f8b.eps}

1212: \caption{Same as Fig.\ \ref{fig:number_ratios} but assuming that all

1213: failed cores with lifetimes $\tau > 0.9$ Myr are ``captured'' by AD and

1214: led to gravitational collapse. The fraction of failed cores is seen to

1215: decrease, but their masses are so small that the fraction of stellar cores

1216: remain essentially unchanged.}

1217: \label{fig:number_ratios_AD}

1218: \end{figure}

1219: 

1220: \end{document}

1221: 

1222: %%

1223: %% End of file `sample.tex'.