1: %%

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

3: %%

4: %% Modified 2005 December 5

5: %%

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

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

8: 

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

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

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

12: 

13: %% The command below calls the preprint style

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

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

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

17: %%

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

19:   

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

21:     

22: %%\documentclass[manuscript]{aastex}

23: 

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

25: 

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

27: 

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

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

30: %% use the longabstract style option.

31: 

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

33: 

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

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

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

37: %%

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

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

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

41: %% for information.

42: 

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

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

45: 

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

47: 

48: %%\slugcomment{Not to appear in Nonlearned J., 45.}

49: 

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

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

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

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

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

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

56: 

57: \shorttitle{Accretion of terrestrial planets}

58: \shortauthors{Morishima et al.}

59: 

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

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

62: 

63: \begin{document}

64: 

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

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

67: %% you desire.

68: 

69: \title{Formation and accretion history of terrestrial planets from runaway growth through to late time 

70: : implications for orbital eccentricity}

71: 

72: 

73: 

74: %\title{Accretion of terrestrial planets from a compact 

75: %planetesimal disk}

76: 

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

78: %% author and affiliation information.

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

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

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

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

83: 

84: \author{RYUJI MORISHIMA\altaffilmark{1}, MAX W. SCHMIDT}

85: \affil{Institute for Mineralogy and    

86: Petrography, Swiss Federal Institute of Technology,

87: Clausiusstrasse 25, 8092 Zurich, Switzerland}

88: \email{Ryuji.Morishima@erdw.ethz.ch}

89: 

90: \and

91: 

92: \author{JOACHIM STADEL, BEN MOORE}

93: \affil{Institute for theoretical physics, University of   

94:  Zurich, Winterthurerstrasse 190, 8057 Zurich,    

95:  Switzerland

96: }

97: 

98: 

99: %% Notice that each of these authors has alternate affiliations, which

100: %% are identified by the \altaffilmark after each name.  Specify alternate

101: %% affiliation information with \altaffiltext, with one command per each

102: %% affiliation.

103: 

104: \altaffiltext{1}{The office is at

105: Institute for theoretical physics, University of   

106: Zurich, Winterthurerstrasse 190, 8057 Zurich,    

107: Switzerland}

108: 

109: 

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

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

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

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

114: %% editorial office after submission.

115: 

116: \begin{abstract}

117: Remnant planetesimals might have played an important 

118: role in reducing the orbital eccentricities of the 

119: terrestrial planets after their formation via giant impacts. 

120: However, the population and the size distribution of 

121: remnant planetesimals during and after the giant impact stage

122: are unknown, because simulations of planetary 

123: accretion in the runaway growth and giant impact stages have been 

124: conducted independently.  

125: Here we report results of direct N-body simulations of the

126: formation of terrestrial planets beginning with a compact 

127: planetesimal disk. The initial planetesimal disk has 

128: a total mass and angular momentum as observed for the

129: terrestrial planets, and we vary the width (0.3 and 0.5AU) and 

130: the number of planetesimals (1000-5000). This initial configuration

131: generally gives rise to three final planets of similar size, and 

132: sometimes a fourth small planet forms near the location of Mars. 

133: Since a sufficient number of planetesimals remains, 

134: even after the giant impact phase, the final orbital

135: eccentricities are as small as those of the Earth and Venus.

136: %Final eccentricities are likely to increase with 

137: %the initial mass and surface density of planetesimals, 

138: %although the dependence on the initial mass is very weak.

139: 

140: \end{abstract}

141: 

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

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

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

145: %% what keyword punctuation is appropriate.

146: 

147: \keywords{Accretion, terrestrial planets}

148: 

149: %% From the front matter, we move on to the body of the paper.

150: %% In the first two sections, notice the use of the natbib \citep

151: %% and \citet commands to identify citations.  The citations are

152: %% tied to the reference list via symbolic KEYs. The KEY corresponds

153: %% to the KEY in the \bibitem in the reference list below. We have

154: %% chosen the first three characters of the first author's name plus

155: %% the last two numeral of the year of publication as our KEY for

156: %% each reference.

157: 

158: 

159: %% Authors who wish to have the most important objects in their paper

160: %% linked in the electronic edition to a data center may do so by tagging

161: %% their objects with \objectname{} or \object{}.  Each macro takes the

162: %% object name as its required argument. The optional, square-bracket 

163: %% argument should be used in cases where the data center identification

164: %% differs from what is to be printed in the paper.  The text appearing 

165: %% in curly braces is what will appear in print in the published paper. 

166: %% If the object name is recognized by the data centers, it will be linked

167: %% in the electronic edition to the object data available at the data centers  

168: %%

169: %% Note that for sources with brackets in their names, e.g. [WEG2004] 14h-090,

170: %% the brackets must be escaped with backslashes when used in the first

171: %% square-bracket argument, for instance, \object[\[WEG2004\] 14h-090]{90}).

172: %%  Otherwise, LaTeX will issue an error. 

173: 

174: 

175: 

176: 

177: 

178: 

179: 

180: %% Putting eqnarrays or equations inside the mathletters environment groups

181: %% the enclosed equations by letter. For instance, the eqnarray below, instead

182: %% of being numbered, say, (4) and (5), would be numbered (4a) and (4b).

183: %% LaTeX the paper and look at the output to see the results.

184: 

185: \section{Introduction}

186: 

187: The rocky planets are believed to have formed via the accretion of small planetesimals.

188: The formation mechanism of planetesimals and their initial mass and 

189: spatial distribution are however still controversial. 

190: The standard picture of accretion of terrestrial 

191: planets from planetesimals is as follows.

192: During the early stages of planetesimal accretion, larger planetesimals 

193: grow faster than smaller ones owing to their stronger gravity 

194: \citep{Gre78, WS89}. 

195: Through this runaway growth stage, 

196: a few tens of Mars-size protoplanets form with mutual separation of 

197: 10 Hill radii \citep{KI98, KI02}. 

198: The growth time scale of protoplanets 

199: is estimated to be  $\sim$ 0.1--1Myr around 1AU, and is longer 

200: with larger distance from the Sun 

201: \citep{WS93, KI02}. 

202: As long as some amount of nebular gas and/or planetesimals remain, 

203: its damping effect stabilizes the orbits of protoplanets, preventing 

204: mutual collisions 

205: \citep{Iwa02, KomI02}.

206: As the amount of remnant gas and/or planetesimals decreases,

207: the orbital eccentricities of protoplanets increases due to 

208: their mutual interactions. Eventually their orbits become chaotic  

209: and late time giant impacts occur.

210: During this giant impact stage, whose time scale  

211: is considered to be $\sim$ 100Myr, the current terrestrial planets form

212: \citep[e.g.][]{CW98, Agn99, Kok06}.

213: The orbital eccentricities of planets immediately after 

214: giant impacts are likely to be much larger than 

215: those of the current terrestrial planets. 

216: Therefore, interactions with remnant gas and/or planetesimals is expected to

217: reduce their eccentricities. 

218: Reviews for the processes described above 

219: are given by \citet{Cha04,Nag07}.

220: 

221: Whether remnant gas or remnant planetesimals is more 

222: important for reducing eccentricities primarily depends 

223: on the time scale of gas dissipation.   

224: If the time scale is long, the velocity dispersion of planetesimals 

225: is suppressed by the gas drag.

226: Hence, the gravitational focusing effect of protoplanets 

227: is enhanced, resulting in a fast clean up of remnant 

228: planetesimals and a lower eccentricity distribution of

229: the final planets

230: \citep{AW02, KomI02, KomI04, Nag05, Ogi07}.

231: On the other hand, if the time scale of gas dissipation 

232: is short, planetesimals remain unaccreted by protoplanets for a longer

233: period of time. In this case, remnant planetesimals may be required

234: to reduce

235: the eccentricities \citep{Cha01, Obr06, Ray06}.

236: In this paper, we examine the latter scenario. 

237: In other words, we ignore the effects of the gas drag and 

238: the tidal interaction between a gas disk and protoplanets.      

239: The effects of gas will be investigated in a future study.

240: 

241: There have been several attempts to examine the effect of remnant 

242: planetesimals based on direct $N$-body simulations \citep{Cha01, Obr06, Ray06},

243: and with simulations 

244: using a hybrid-code \citep{KB06}. Direct $N$-body simulations usually adopt

245: lunar to Mars size protoplanets surrounded 

246: by smaller planetesimals as initial conditions. 

247: Direct $N$-body simulations suggest that the eccentricities of final planets 

248: are further reduced as the total mass of planetesimals increases.

249: Even for the same total mass of planetesimals, 

250: the damping effect is stronger with a larger number of smaller planetesimals 

251: \citep{Obr06, Ray06}. 

252: However, the total mass 

253: and mass distribution of the remaining planetesimals are unknown, 

254: unless mass evolution in the runaway stage is followed. 

255: 

256: The hybrid code of \citet{KB06, BK06},

257: is able to follow planetary accretion through both the runaway 

258: and giant impact stages.

259: In their code,  the mass and velocity distributions of planetesimals 

260: contained in multi annuli are solved by 

261: a statistical approach

262: whereas orbits of protoplanets are solved using

263: direct $N$-body calculations, that include the effect of 

264: interaction between planetesimals and protoplanets.

265: Statistical approaches based on the local approximation

266: produce consistent results with those obtained from 

267: direct $N$-body simulations in the runaway stage

268: \citep[e.g.][]{Ina01}.

269: However, it is questionable if statistical approaches 

270: can accurately follow the late 

271: stage of planetary accretion because orbital eccentricities of remnant

272: planetesimals are usually very large.

273:  

274: Here we report results of direct $N$-body simulations beginning with 

275: a planetesimal disk until the end of planetary accretion in the 

276: terrestrial region. We consider compact planetesimal disks 

277: (initial disk widths of $\le 0.5$AU), 

278: whose total masses and total angular momenta 

279: are the same as those of the present terrestrial planets.   

280: These initial conditions are used since 

281: accretion simulations beginning from compact disks are 

282: usually computationally less expensive than those from 

283: wider disks.  This is the case even with the same initial number of particles, 

284: because of the rapid decrease in the number of particles through accretion.   

285: Another reason to adopt compact disks is that total angular momenta of 

286: final planetary systems obtained from most of previous simulations are 

287: much larger than for the terrestrial planets, as 

288: these simulations usually have a super-massive Mars 

289: \citep[e.g.][]{Cha01, Ray06}. 

290: This excess angular momentum is likely due to 

291: initially extended disks. Though Jupiter removes angular momentum, 

292: mostly from the asteroid region, its effect does not seem to be sufficiently strong

293: in the terrestrial region. 

294: As one possibility for this issue, we consider initially compact disks,

295: supposing that they result from, for example, 

296: dust migration due to the gas drag prior to formation

297: of planetesimals \citep[e.g.][]{YS02}.

298: 

299: In \S~2, we explain the numerical methods used in this study.  

300: We show results of simulations in \S~3. 

301: We compare our results with previous simulations in \S~4. 

302: In \S~5, we give some physical interpretations for our simulation results 

303: using analytic estimations.

304: We summarize our results in \S~6.

305: 

306: \section{Methods}

307: 

308: The runaway growth stage is shorter than the giant impact stage, but orbits of

309: a large number of bodies need to be followed. On the other hand, although

310: fewer bodies are necessary 

311: for the giant impact stage (unless the effect of fragmentation is considered), 

312: more care must be taken to accurately follow the orbital evolution over many more

313: dynamical times.

314: Taking these physically different types of the accretion stages into account, 

315: we apply different $N$-body codes to these two stages of the evolution.  

316: 

317: The runaway growth stage is simulated with the parallel 

318: tree-code {\it PKDGRAV} \citep{Ric00, Sta01} 

319: for $10^5$ yr with adopting artificially enhanced radii. 

320: The code uses a fourth-order 

321: multi-pole expansion for the force calculations, and 

322: a second-order leap-frog scheme is used for time integration. 

323: We apply a hierarchical time stepping with the largest time step of 

324: 1.8 days (0.005 yr).  

325: The opening angle of 0.5 is used as a criterion for searching down the tree.

326: The energy error in the runaway stage is 

327: $|\Delta E/E| \sim 10^{-4}$--$10^{-3}$, arising entirely from the integrator.

328: The error due to the force

329: calculation using our tree method is negligibly small.    

330: Using the output of the runaway stage as the initial condition, we simulate

331: the giant impact stage with the hybrid symplectic code 

332: {\it Mercury} \citep{Cha99} 

333: for $2 \times 10^8$ yr without any enhancement of radii. 

334: This code uses a mixed variable symplectic (MVS) method 

335: \citep{Kino91,WH91,ST92} for 

336: orbits around the Sun whereas close encounters are integrated by 

337: the Bulirsch-Stoer method. We use a fixed time step of 6 days, which is 

338: the same as or similar to those adopted in the previous works using  

339: {\it Mercury} \citep{Cha01, Ray06}.

340: The energy error in the giant impact stage is $|\Delta E/E| \sim 10^{-5}$.

341: It usually takes less than one computer day  

342: for a simulation of 

343: the runaway stage with {\it PKDGRAV}, 

344: whereas it can take several months to compute

345: the giant impact stage with {\it Mercury}. 

346: 

347: Whereas MVS type integrators can take much 

348: larger time steps than those used with the leap-frog scheme, 

349: the {\it Mercury} code uses direct summation for calculations of the mutual 

350: gravity force.

351: {\it PKDGRAV} is thus faster than {\it Mercury} as long as the number of 

352: particles is larger than several hundred.  

353: Some comparisons between these two codes are also found 

354: in \citet{Ray05}.  

355: 

356: The enhancement of radii in the runaway stage is used in order to 

357: reduce computational time; we use an enhancement factor of radii $g = 4.3$. 

358: This gives an analogous effect of the gas drag, 

359: and the growth time scale of protoplanets is 

360: reduced by a factor of $\sim g^{2}$ 

361: %\citep[(see also eq.~[\ref{eq:tgrow}]]{KI96, KI02}. 

362: (Kokubo \& Ida 1996, 2002, see also eq.~[\ref{eq:tgrow}]).

363: On the other hand, the growth time scale of protoplanets is 

364: actually reduced by a factor of $\sim \beta^{2}$ by the gas drag,

365: where $\beta$ is the factor for reduction of planetesimal eccentricities \citep{KI00}. 

366: Thus, our simulations approximately mimic 

367: a situation in which the gas disappears suddenly at 

368: $(g/\beta)^2 \times 10^5$ yr.

369: This time scale is probably shorter than the typical 

370: life time of circumsteller disks, $\sim$ a few Myr 

371: \citep{Hai01}, although the 

372: exact time scale for formation of planetesimals from dusty gaseous disks is not known.

373: As long as the number of planetesimals is sufficient, the orbits of protoplanets 

374: are stabilized during the runaway stage by dynamical friction such that the growth mode is 

375: not affected by the enhancement of radii \citep{KI96}. 

376: However, in the transition from the runaway growth 

377: stage to the giant impact stage, faster clean up of 

378: remnant planetesimals, due to this approach, usually 

379: causes final planets to have higher eccentricities.

380: Thus the radii of particles should be set to realistic values

381: before planetesimals are too depleted in order to accurately 

382: examine the effect of remnant planetesimals.  

383: Additional simulations and analytic calculations were performed to assess the sensitivity of 

384: the results to varying the time at which $g$ is reduced to unity. 

385: 

386: We use 10 different initial conditions which are summarized in Table~1.

387: The total mass and angular momentum are assumed to be the same as for  

388: the present terrestrial planets 

389: (1.98$M_{\rm E}$ and 1.86$M_{\rm E}$ AU$^{1/2}$ $\sqrt{GM_{\odot}}$, 

390: respectively, where $M_{\rm E}$, $G$, and $M_{\odot}$ stand for 

391: the mass of Earth, the gravitational constant, and the solar mass,

392:  respectively) 

393: with the central star's mass equal to the solar mass.

394: The initial width of a planetesimals disk $\Delta_{\rm disk}$ 

395: is taken to be 0.3 AU or 0.5AU.

396: The planetesimal mass is assumed to be identical and the 

397: number of planetesimals $N$ varies from 1000 to 5000. 

398: The physical density of all the bodies is assumed to be 

399: $\rho = 2$g cm$^{-3}$.

400: The surface number density $n(a)$ as a function of the semimajor axis $a$ is 

401: given by a power law $n(a) \propto a^{\alpha}$ with $\alpha = -1$ or $-2$.

402: We also conduct two additional simulations for $N=1000$, 

403: where we switch the code and $g$ at $5 \times 10^4$ yr in order to check 

404: whether the outcomes are affected by this timing.

405: 

406: It would be very interesting to investigate the accretion of planets using

407: more extended disks ($\Delta_{\rm disk} > 0.5$AU), but computationally too expensive

408: with our current codes. 

409: The rate at which planetesimals merge is slower in a wider disk, particularly at

410: its extremities,

411: so we need to use the tree method for the gravity calculation

412: for longer period of time.

413: On the other hand, it is not appropriate to use the leap-frog integrator

414: for the long-term orbital evolution for the following reasons.  

415: Firstly, since the leap-frog integrator causes a secular error in the longitude of the 

416: perihelion \citep{KM04},  it does not accurately treat long-term secular

417: interactions. Secondly, since the standard (or explicit) block multi- timestep

418: algorithm used for the leap-frog integrator is not time-symmetric, 

419: the error in the energy accumulates with close encounters \citep{HMM95}.

420: This is also the case for the higher order Hermite-integrator \citep{KM04}.

421: We ensure that the transition from {\it PKDGRAV} is chosen conservatively, thus we

422: achieve high energy conservation as mentioned above.

423: 

424: Although the implicit block time-step algorithm can avoid this problem

425: \citep{MHKS06}, MVS integrators have considerable advantages for 

426: simulations of planetary accretion. 

427: We are therefore implementing the SyMBA integrator \citep{DLL98}

428: into the latest version of {\it PKDGRAV}, which enables us to simulate 

429: planetary accretion in wider disks. These simulation results will be reported later.

430: 

431: \section{Results}

432: 

433: \subsection{An example of evolution: Run~6}

434: 

435: As an example, we first explain time evolution of Run~6.

436: For this simulation the width of the initial disk $\Delta_{\rm disk}$ is

437: $0.5$AU and the initial number of 

438: planetesimals $N$ is $3000$.

439: Figure~1--3 show time evolution of this simulation: 

440: snapshots on the plane of the semimajor axis versus the orbital eccentricity 

441: (Fig.~1),

442: the cumulative number of planetesimals plotted against the mass (Fig.~2), 

443: and the epicyclic velocity plotted against the mass (Fig.~3).

444: For detailed analysis, we  

445: divide the accretional evolution into 

446: four different stages (the runaway, oligarchic, giant impact, 

447: and post giant impact stages), rather than two main stages discussed so far.

448: 

449: \subsubsection{Runaway growth stage ($\sim 10^{4}$ yr)}

450: 

451: In the early stage, most of the mass of the system 

452: is contained in smallest planetesimals. 

453: In this case the epicyclic velocity,  

454: $v = \langle a\Omega\sqrt{e^2+i^2}\rangle$ 

455: (where $\Omega$ is the orbital frequency, and 

456: $e$ and $i$ are the orbital eccentricity and inclination of a planetesimal, 

457: respectively),

458: is regulated by the smallest planetesimals and 

459: is typically as large as their escape velocity; 

460: $v_{\rm esc,0} = \sqrt{2Gm_0/(gr_0)}$ (Fig.~3), where $m_0$ and $r_0$ are 

461: the mass and the radius.

462: If $v$ is much smaller than the escape velocity 

463: of the largest body $v_{\rm esc,p} = \sqrt{2Gm_{\rm p}/(gr_{\rm p})}$

464: (where $m_{\rm p}$ and $r_{\rm p}$ are the mass and the 

465: radius respectively), 

466: and is a decreasing function with mass as shown in Figure~3, 

467: then the largest body starts to grow much faster than nearby objects. 

468: This growth mode is called runaway growth 

469: \citep{WS89, KI96, Wei97}. 

470: At $t \sim 10^4$ yr, the power-law index $q$ 

471: ($dN_{\rm c} \propto m^{q} dm$, where $N_{\rm c}$ is the cumulative number

472: and $m$ is the planetesimal mass) 

473: is about $-2.7$ in our simulation (Fig.~2). 

474: This value is close to  $q \simeq -2.5$ obtained in simulations of 

475: \citet{KI96, KI00} and the analytical estimate $q = -8/3$ by 

476: \citet{Mak98}.

477: (\citet{Mak98} assumed complete

478: energy partitioning ($v \propto m^{-1/2}$) 

479: in the strong gravitational limit ($v \ll v_{\rm esc, 0}$).

480: However, the actual velocity distribution is less steep than this

481: \citep[see also our Fig.~3]{Raf03}. 

482: If we assume $v \propto m^{\gamma}$ with $\gamma \sim -1/4$, which is 

483: a rough approximation of Figure~3 at $t \sim 10^4$ yr, 

484: the formulation of \citet{Mak98} gives $q = -13/6+\gamma \sim -2.4$.)

485: 

486: 

487: \subsubsection{Oligarchic growth stage ($\sim 10^{5}$ yr)}

488: 

489: As large bodies grow, their mutual gravitational interactions leave their

490: orbits separated by 5--10 Hill radius 

491: \citep[see also our Fig.~5]{KI95, KI98}. 

492: The Hill radius $r_{\rm H}$ of a planetesimal of mass $m_{\rm p}$ 

493: is given by

494: \begin{equation}

495: r_{\rm H} = ah_{\rm p} = a \left(\frac{2m_{\rm p}}{3M_{\odot}}\right)^{1/3},

496: \end{equation}

497: where $a$ is its semimajor axis and $h_{\rm p}$ is the reduced 

498: Hill radius.

499: The largest bodies gravitationally 

500: influence the velocity evolution of all the neighboring planetesimals 

501: \citep{IM93}, increasing towards the escape velocity of the protoplanet 

502: $v_{\rm esc,p}$ (see the panel of $t = 10^5$ yr in Fig.~3).

503: On the other hand, the velocities of protoplanets $v_{\rm p}$ are 

504: also influenced by the energy partitioning with surrounding 

505: smaller planetesimals. 

506: Indeed, the value of $v_{\rm p}$ for Run~6 is quite close to the 

507: equilibrium value $v_{\rm p,eq}$ $(\sim v_{\rm esc,0})$, 

508: which is theoretically estimated neglecting 

509: the mutual perturbations of protoplanets 

510: (see eq.~[\ref{eq:eeq2}] in \S~5.3).

511: 

512: Since the growth rate of the largest body slows down at the

513: expense of its nearby neighbors, the largest intermediate 

514: mass objects begin

515: catch up with the largest body \citep{IM93}. 

516: This growth mode is called oligarchic growth  \citep{KI98}.

517: At $t = 10^5$ yr in Run~6, about half of the total mass 

518: is contained in the 10 largest oligarchic bodies. 

519: Since the growth of smaller planetesimals has

520: substantially stalled, protoplanets start to separate from the 

521: continuous size distribution.

522: Therefore, the number of planetesimals decreases mostly by

523:  accretion onto protoplanets and not by mutual collisions.

524: Since $v$ for planetesimals is nearly independent of the mass, 

525: so is their collision probability with protoplanets. 

526: Hence, the power-law index $q$ for the mass distribution of 

527: planetesimals does not change from $-2$ after this stage.

528: 

529: \subsubsection{Giant impact stage ($\sim 10^{6}$ yr)}

530: 

531: Without the damping force by remnant planetesimals and/or gas, 

532: a multiple protoplanet system undergoes an orbital instability after 

533: a certain time $T_{\rm inst}$. 

534: This instability time  $T_{\rm inst}$ depends 

535: on the orbital separation,  eccentricities, and  

536: absolute mass of protoplanets 

537: \citep{Cha96, IT99, Yos99, IO06}.

538: We switch the code and 

539: reduce the value of $g$ from 4.3 to 1 at $t = 10^5$ yr.

540: For Run~6, $T_{\rm inst}$ this time is estimated to be 

541: $10^5 $--$10^6$ yr from above studies.

542: With a decreasing total mass of planetesimals that have higher velocities,

543: the damping due to the dynamical friction of planetesimals

544: becomes less effective. At this point the orbital instability and mutual collisions of 

545: protoplanets start to take place.  

546: We find that the orbital instability starts immediately after $t = 10^5$ yr,

547: and the number of protoplanets

548: decreases from 12 at $t = 10^5$ yr to 10 at $t = 2\times 10^5$ yr

549: (here we assume a protoplanet to be a body with mass 

550: $> 2\times 10^{26}$g $\sim 50m_0$).

551: 

552: During the giant impact stage, the mass distribution changes mainly 

553: owing to collisions between protoplanets, 

554: while the population of small planetesimals does not change so much. 

555: This can be seen in Figure~2; 

556: from $t = 10^5$ yr to $t = 10^6$ yr, 

557: when the number of protoplanets reduces from 

558: 12 to 5, whereas the total number of particles reduces 

559: only from $\sim 400$ to $\sim 300$. 

560: Because the velocities of protoplanets are much smaller than those of 

561: planetesimals, mutual collisions of protoplanets

562: occur quickly. This is similar to results of simulations

563: including the damping force due to the tidal interaction 

564: between a gas disk and protoplanets; \citet{KomI02}

565: find that the giant impact stage becomes shorter with 

566: the stronger damping force.

567: 

568: \subsubsection{Post giant impact stage ($> 10^{7}$ yr)}

569: 

570: The number of protoplanets further reduces to 3 at $t = 10^7$ yr,

571: after two final giant impacts that occur shortly before $t = 10^7$ yr.

572: Through the giant impact stage, the mutual separation 

573: between protoplanets normalized by their Hill radii

574: increases to $\sim 30$. 

575: The mass distribution becomes completely bimodal (Fig.~2) with

576: the masses of protoplanets smaller than the isolation mass 

577: by a factor of $\sim 3$.

578: The isolation mass is the total mass contained in a ring of width 30 Hill radius 

579: with the initial surface density (see eq.~[\ref{eq:mp}]). 

580: This deviation likely  

581: comes from decrease of the surface density (by a factor of $\sim 2$)

582: due to expansion of the disk from its initial diameter 

583: via gravitational scattering of protoplanets.

584: For remnant planetesimals, the power-law index

585: $q$ remains to be $\sim -2$ 

586: and the largest mass is $\sim 50m_0$, which is 

587: similar to the protoplanet's mass during the runaway to

588: oligarchic stages.

589: 

590: Since the mutual interactions between protoplanets after the giant impact stage is 

591: rather weak, the eccentricities of protoplanets are expected to be determined by 

592: the energy partitioning with remnant planetesimals.

593: As we discussed for the oligarchic stage,   

594: $v_{\rm p}$ in the giant impact phase is surprisingly close to the equilibrium value

595: ($\simeq  v_{\rm esc,0}$; see eq.~[\ref{eq:eeq2}]), although some amount of remnant

596: planetesimals may be necessary to achieve full equilibrium.

597: For this simulation we find that the fraction of the total mass 

598: contained in planetesimals is 

599: 0.29 and 0.16 at $t = 10^6$ and $10^7$ yr, respectively.

600: Naively, one might predict that equilibrium occurs once the mass in 

601: planetesimals is comparable to the total mass of protoplanets (see \S~5.2).

602: This slight contradiction might mean that the damping 

603: due to giant impacts themselves or 

604: some other unknown mechanism, works effectively.  

605: 

606: \subsection{Evolution of orbital spacing and eccentricities}

607: 

608: Here we analyze the evolution of each simulation quantitatively and 

609: discuss its dependence on the initial conditions.

610: In order to examine characteristics of the evolution of the largest bodies,

611: we define a planet (or a protoplanet) as 

612: a body having mass larger than $2 \times 10^{26}$g.

613: The physical meaning of this choice is that 

614: a body larger than this mass regulates the velocity evolution 

615: of all the neighboring bodies (see \S~5.3). 

616: In this case the following discussion does not strongly 

617: depends on the definition of the minimum mass of planets.

618: We will explore the following four quantities:

619: [1] the number of planets, $N_{\rm p}$,

620: [2] the mass fraction of planets compared to the total mass, $f_{\rm p}$ ,

621: [3] the orbital spacing of planets normalized by the mutual Hill radius, 

622: $\bar{b}_{\rm p}$,

623: [4] the eccentricity of planets, $\bar{e}_{\rm p}$,

624: and

625: [5] the ratio of the effective mass of planetesimals to the mean mass of

626: protoplanets

627: $m_{\rm eff}/\langle m_{\rm p} \rangle$.

628: The effective mass of planetesimals is defined as 

629: $m_{\rm eff} = \langle m^2 \rangle/\langle m \rangle$, 

630: where $ \langle m^2 \rangle$ and $\langle m \rangle$ are the 

631: mean squared and mean masses of planetesimals, whose masses are

632: smaller than $2 \times 10^{26}$g. 

633: The mass ratio $m_{\rm eff}/\langle m_{\rm p} \rangle$ would be 

634: important for the evolution of $\bar{e}_{\rm p}$, 

635: as its equilibrium value due to the dynamical friction is given by 

636: (eq.~[\ref{eq:epeq}] in \S~5.1)

637: \begin{equation}

638: e_{\rm p,eq} \simeq \sqrt{\frac{4m_{\rm eff}}{3\langle m_{\rm p} \rangle}} 

639: \langle e^2 \rangle^{1/2}. \label{eq:eeq}

640: \end{equation}

641: 

642: We apply the following form for the 

643: averaged orbital spacing normalized by the mutual Hill radius

644: \begin{equation}

645: \bar{b}_{\rm p} =  \sqrt{

646: \frac{\sum^{N_{\rm p}-1}_{j=1} ((a_{j+1}-a_j)/r_{{\rm H},j})^2 \mu_j} 

647: {\sum^{N_{\rm p}-1}_{j=1} \mu_j} \label{eq:aveep}

648: },

649: \end{equation}

650: with the reduced mass

651: \begin{equation}

652: \mu_j = \frac{m_{{\rm p},j} m_{{\rm p},j+1}}{m_{{\rm p},j}+m_{{\rm p},j+1}},

653: \end{equation}

654: and the mutual hill radius 

655: \begin{equation}

656: r_{{\rm H},j} = \frac{1}{2}(a_j + a_{j+1})

657: \left(\frac{m_{{\rm p},j} + m_{{\rm p},j+1}}{3M_{\cdot}}\right)^{1/3}.

658: \end{equation}

659: Here  $a_{j}$, $e_{j}$, and  $m_{{\rm p},j}$ are the semimajor axis,

660: orbital eccentricity, and mass of the protoplanet $j$ 

661: in the order of semimajor axis, respectively. 

662: We use the following form for the averaged eccentricity,

663: which characterizes the energy of epicyclic motion of planets:   

664: \begin{equation}

665: \bar{e}_{\rm p} =  \sqrt{

666: \frac{\sum_{j=1}^{N_{\rm p}} m_{{\rm p},j} e_{j}^2}

667: {\sum_{j=1}^{N_{\rm p}} m_{{\rm p},j}}},

668: \end{equation}

669: where  $e_{j}$ is the orbital eccentricity of the 

670: protoplanet $j$. The mass weighted 

671: eccentricity \citep[e.g.][]{BK06, Ray06}

672: gives a slightly smaller 

673: value than the above eccentricity.

674: 

675: Figure~4 shows the evolution of these four quantities for the case of 

676: $\Delta_{\rm disk} = 0.3$AU.

677: In the third panel from the top, we also plot the orbital instability time as a 

678: function of the orbital spacing for the two cases of 

679: $\bar{e}_{\rm p}/\bar{h}_{\rm p} = 2$ and 4, respectively, 

680: from \citet{Yos99} (see eq.~[\ref{eq:tinst}]). Here 

681: $\bar{h}_{\rm p} = (2\langle m_{\rm p}\rangle/(3M_{\odot}))^{1/3}$ 

682: is the averaged reduced hill radius.

683: These lines indicate the 

684: stability of multiple protoplanet systems; if the orbital spacing 

685: is narrower than these lines, orbital instability will occur. 

686: Several protoplanets form at $\sim 10^4$ yr and the normalized 

687: orbital spacing $\bar{b}_{\rm p}$ is about 10, as pointed out by 

688: \citet{KI98}. Since the orbital instability time 

689: is $10^5$--$10^6$ yr in such systems, giant impacts between 

690: protoplanets start around that time as we discussed in the previous 

691: section. The orbital eccentricity prior to the giant impact stage is

692: $\sim 0.03$ while it increases up to 0.1 during 

693: the giant impact stage. The corresponding normalized 

694: eccentricities, $\bar{e}_{\rm p}/\bar{h}_{\rm p}$, are 2--3 and $\sim 10$,

695: respectively.

696: Through the giant impact stage,  $\bar{b}_{\rm p}$ increases 

697: to $20$--30.  As the radial excursion of planets during the 

698: giant impact stage determines the final separation of planets, 

699: the relation between $\bar{b}_{\rm p}$ in the final state 

700: and $\bar{e}_{\rm p}$ during the giant impact stage

701: can be approximately represented by \citep{KomI02} 

702: \begin{equation}

703: \bar{b}_{\rm p}\bar{h}_{\rm p} \sim 2\bar{e}_{\rm p}. \label{eq:dela}

704: \end{equation}

705: This is roughly consistent with our simulation results.

706: 

707: During the post giant impact stage, the orbital eccentricities 

708: are reduced ($\bar{e}_{\rm p}/\bar{h}_{\rm p} \sim 3$) 

709: and these values have little dependence on initial parameters 

710: (we discuss the weak dependence in detail in \S~5.3).

711: Since $\bar{b}_{\rm p}$ is large enough in the final state, 

712: the mutual interaction of planets is likely to be unimportant. 

713: In this case, final eccentricities of planets are expected to 

714: be determined by the energy partitioning with remnant planetesimals.   

715: The mass ratio $m_{\rm eff}/\langle m_{\rm p} \rangle$ 

716: decreases nearly monotonically with time and is about 0.01--0.02 in the

717: final state (the bottom panel of Fig.~4; the decrease of $m_{\rm eff}/\langle m_{\rm p} \rangle$  

718: is due to the increase of $\langle m_{\rm p} \rangle$, while 

719: $m_{\rm eff}$ is nearly constant except at very early times).

720: Since the mean eccentricity of planetesimals is 

721: $\sim 0.3$ in the post giant impact stage (Fig.~3), the equilibrium 

722: eccentricity of planets (eq.~[\ref{eq:eeq}]) is estimated to be 0.03--0.05. 

723: This is almost the same as the values

724: obtained in our simulations.

725: 

726: While the final giant impact occurs before $t \sim 10^7$ yr 

727: in most of the runs, it happens at $t \sim 5 \times 10^7$ yr

728: for Run~1 ($N=1000$).  Because of that impact, 

729: the orbital spacing for Run~1 becomes even wider.

730: This unstable behavior likely suggests that the dynamical 

731: friction for $N=1000$ is less effective as compared with larger

732: $N$. However, except for this event, the dependence of the evolutions of 

733: all the quantities shown in Figure~4 on $N$ is very small.  

734: Switching $g$ in the earlier time does not affect the quantities 

735: in the final state shown in Figure~4, either, 

736: although Run~1b is dynamically more excited during 

737: the giant impact stage. This is because the mass variation of protoplanets

738: in Run~1b is somewhat large and smaller protoplanets are dynamically 

739: enhanced by larger protoplanets. 

740: 

741: Figure~5 is the same as Figure~4, but for the case of 

742: $\Delta_{\rm disk} = 0.5$AU. The evolution of all the quantities are 

743: very similar to those in Figure~4, although the early evolution is slightly 

744: slower due to the lower initial surface density. 

745: The dependence on $N$ is very small here as well as in Figure~4. 

746: In particular, the final $\bar{b}_{\rm p}$ has converged to $\sim 30$

747: for all the runs. However, the final eccentricity 

748: $\bar{e}_{\rm p}$ is much larger

749: for the case with earlier switching of $g$ (Run~5b).

750: This is in an opposite sense to our prediction because dynamical 

751: friction works more effectively with a higher density of planetesimals. 

752: We will discuss this issue in the next section along with the

753: final configurations of the systems.

754: Except for Run~5b, $\bar{e}_{\rm p}$ and

755: $m_{\rm eff}/\langle m_{\rm p} \rangle$ in Figure~5

756: are slightly smaller than those in Figure~4. 

757: The smaller $m_{\rm eff}$ is due to the smaller surface density 

758: (the relation is roughly given by $m_{\rm eff} \propto \Sigma^{3/2}$, where

759: $\Sigma$ is the initial surface density of planetesimals; see \S~5.3).

760: 

761: \subsection{Final systems}

762: 

763: Here we present the orbital parameters of all the final systems obtained 

764: in our simulations. 

765: Figure~6 shows snapshots of all of our runs on the $a$-$e$ plane at 200Myr.

766: Also, the number of planets $N_{\rm p}$, 

767: the averaged eccentricities of planets $\bar{e}_{\rm p}$,

768: and the angular momentum deficit of planetary systems $S_{\rm d}$ 

769:  at 200Myr are summarized in Table~1. 

770: The angular momentum deficit is defined as 

771: \citep{Las97,Cha01}

772: \begin{equation}

773: S_{\rm d} = \frac{ \sum_{j=1}^{N_{\rm p}} m_{{\rm p},j}\sqrt{a_j}

774: \left(1 - \sqrt{(1-e_{j}^2)}\cos{i_j}\right)}

775: {\sum_{j=1}^{N_{\rm p}} m_{{\rm p},j}\sqrt{a_j}}, \label{eq:sd}

776: \end{equation}

777: where $i_{j}$ is  the orbital inclination of the planet $j$.  

778: We take 5Myr averages for $\bar{e}_{\rm p}$ and $S_{\rm d}$. 

779: For the current terrestrial planets,

780: we take the mean values between the minimum and maximum orbital 

781: eccentricities and inclinations from 3Myr orbital integrations 

782: in \citet{Qui91}. 

783: This operation roughly corresponds to subtracting only 

784: the free eccentricities, 

785: provided that the free eccentricities is larger than the 

786: forced eccentricities due to giant planets 

787: \citep[see][Chap.~7.4]{MD99}. 

788: In fact, the minimum eccentricities and inclinations 

789: obtained in \citet{Qui91} are almost zero except for Mercury. 

790: This suggests that the free eccentricity (inclination)

791: and the forced eccentricity  (inclination) are comparable for the current 

792: terrestrial planets.

793: 

794: We always obtain three similar size planets between

795: 0.5AU and 1.3 AU, except Run~1 has only two planets in this region. 

796: The orbital spacing between planets are also quite similar.

797: We find that more than 95$\%$ of the initial mass and angular momentum

798: are contained in planets at 200Myr in all our simulations.

799: Both the averaged eccentricities and the angular momentum deficits 

800: obtained from our simulations are comparable or even smaller than 

801: those for the current solar system, except for Run~5b and Run~8 (Table~1).  

802: Except these two runs, both $\bar{e}_{\rm p}$ and $S_{\rm d}$ 

803: are smaller for $\Delta_{\rm disk} = 0.5$AU than for 

804: $\Delta_{\rm disk} = 0.3$AU.

805: There seems to be a weak tendency that $\bar{e}_{\rm p}$ and $S_{\rm d}$

806: decrease with increasing $N$. These trends are interpreted in terms of 

807: the effective planetesimal mass $m_{\rm eff}$, 

808: if the final $\bar{e}_{\rm p}$ is 

809: determined by the energy partitioning with planetesimals neglecting 

810: the mutual interaction of planets (\S~5.3). 

811:  

812: Differing from other runs, the mutual interaction between the

813: innermost two planets in Run~5b and Run~8 is important even at the 

814: end of simulations, as their orbital separation is narrow. 

815: This seems related to the spatial distribution of planetesimals

816: during the giant impact stage. In these runs, we find that

817: two innermost planets quickly sweep out planetesimals in 

818: the inner region whilst there are still large numbers of planetesimals 

819: in the outer region. The outer planets with these planetesimals 

820: tend to push the middle planet inward. As a result, the 

821: two innermost planets continue interacting without sufficient 

822: dynamical friction due to surrounding planetesimals. 

823: Although our statistics is not sufficient, such a difference in the 

824: inner and outer region would tend to appear 

825: when $g$ is reduced at earlier times or if the initial 

826: planetesimal mass were concentrated in the inner region.

827: 

828: This fact seems related to the existence of the small outermost planet.

829: When we compare simulations Run~1 and Run~1b, the sizes and locations

830: of two largest planets are very similar. 

831: While Run~1b has inner and outer small planets, 

832: Run~1 has only an outer small planet (near 2.1AU).

833: Similarly, while the sizes and locations of the three largest planets in 

834: Run~5 and Run~5b are similar, only Run~5b has an outermost planet.

835: These facts suggest that the

836: tendency to have small planets in the inner and outer edges

837: is stronger in simulations with earlier switching of $g$.  

838: We interpret this as due to a larger amount of planetesimals that 

839: are scattered inward (outward) at the inner (outer) edge of the disk 

840: before they are accreted by planets. Small planets form from

841: these scattered planetesimals.

842: A similar trend is also found in the simulations starting with 

843: a stronger gradient of the surface density in the radial direction

844: (Runs~4 and 8).

845: In these systems, the inner planets form quickly while large planets 

846: have not grown in the outer region. Then inner planets gravitationally

847: scatter planetesimals and small protoplanets outward. 

848: The orbital eccentricities of protoplanets scattered outward 

849: are reduced by the dynamical friction of similarly scattered planetesimals.

850: Eventually, these protoplanets can have stable orbits near the location 

851: of Mars and slightly grow as they collide with planetesimals.  

852: 

853: \section{Comparison with previous works}

854: 

855: Our simulation results are different from those starting with 

856: only protoplanets.  

857: \citet{Kok06} conducted simulations

858: starting with $\sim 15$ mars-size protoplanets at $0.5<a<1.5$AU and  

859: found that most of the final mass is contained in the largest two planets. 

860: Their final orbital eccentricities are usually higher than those for

861: the current terrestrial planets.

862: \citet{CW98} also showed similar results to 

863: \citet{Kok06} for the case without perturbation of 

864: Jovian planets (their Model~A). They also found that

865: mass concentration within fewer planets is strengthened by 

866: the presence of Jovian planets (their Model~B).

867: The difference between their results and ours suggests 

868: that the number of final planets increases with a stronger damping force,

869: which makes the radial excursion of protoplanets and thus the final separation 

870: between neighboring planets narrower.  

871: Indeed, the same trend was also found by \citet{KomI02}, 

872: who examined the effect of damping due to the gas disk on the accretion of 

873: protoplanets.

874: 

875: In recent direct $N$-body simulations stating with planetary embryos 

876: with small planetesimals, 

877: the total mass of planetesimals 

878: is half or less than half, and the orbital separation of embryos

879: is equal to or less than 10 in units of the Hill radii

880: \citep{Cha01, Obr06, Ray06}.  

881: Since the dynamical friction of surrounding planetesimals is not 

882: strong enough to suppress the orbital instability with these initial

883: conditions (see \S~5.2),

884: giant impacts start immediately before planets substantially grow by 

885: accretion.

886: They adopt a nearly identical size distribution of planetesimals, which

887: in principle, does not change in such enhanced systems.

888: Therefore, the effective mass of planetesimals after giant impacts occur

889: is still given by the initial planetesimal mass. 

890: Thus, the dependence of the equilibrium eccentricity of planets on

891: the initial planetesimal mass 

892: ($ e_{\rm p,eq} \sim \sqrt{m_0/\langle m_{\rm p} \rangle}

893: \langle e^2 \rangle^{1/2} $) 

894: is much stronger than we find.

895: In some of their simulations, mutual interaction amongst the final planets 

896: is insignificant and the final eccentricities 

897: seem to be close to the equilibrium value.  

898: For example, \citet{Obr06} adopt the initial 

899: mass of planetesimals to be $m_0 \simeq 1/400M_{\rm E}$. 

900: In the late stages of their EJS (eccentric orbits of Jupiter and Saturn) 

901: simulations, most of the mass supplied to terrestrial planets is 

902: from small planetesimals with very high speed.  

903: If we convert their impact speed $\sim 20$km s$^{-1}$ 

904: in the late stage to the eccentricity around 1AU,  

905: it gives $\langle e^2 \rangle^{1/2} \sim 0.5$.

906: Supposing that the mean mass of planets is as large as the Earth's mass,

907: we obtain  $ e_{\rm p,eq} \sim 0.025$, 

908: which is even smaller than those for the current terrestrial planets, 

909: and consistent with their results.

910: Therefore, we predict that the

911: final eccentricities of planets would be further reduced if they adopted 

912: smaller planetesimal masses in their simulations.

913: However, such a small effective mass of planetesimals in the beginning of 

914: the giant impact stage might be unlikely if we take the growth of planetesimals

915: in the runaway and oligarchic stages into account.

916: 

917: \citet{KB06} and \citet{BK06} 

918: conducted planetary accretion simulations starting 

919: with very small planetesimals ($r_0 = 1$--5km),

920: using their hybrid code. 

921: In their runs, simulations 

922: starting at $0.86$--1.16AU in \citet{KB06}  have 

923: similar initial conditions to ours, although our 

924: disks are slightly more massive. 

925: The evolution of the number of oligarchic bodies 

926: (with masses $> \sim 10^{25}$--$10^{26}$g in their simulations) 

927: and their orbital separation

928: (their Hill parameter almost corresponds to $1/\bar{b}_{\rm p}$) 

929: are very similar to our results.

930: However, the final eccentricities of planets is more 

931: excited in their simulations

932: (one of their simulations obtained three planets with $e_{\rm p} \sim 0.1$). 

933: In fact, 

934: in all of their simulations starting with wider initial disks ($0.4$--2.0AU), 

935: the final planetary orbits are more eccentric than the current terrestrial

936: planets \citep[see Table~I of][]{KB06},

937: and apparently remnant small planetesimals do not contribute to 

938: damping of eccentricities of planets.

939: Since we have not conducted simulations starting with wide disks, 

940: it is not clear for us if their results obtained with  

941: a hybrid code are consistent with those obtained 

942: from direct $N$-body simulations.

943: We are planning to conduct direct $N$-body simulations with initially 

944: wide disks to clarify this problem.

945: 

946: 

947: 

948: %% To help institutions obtain information on the effectiveness of their

949: %% telescopes, the AAS Journals has created a group of keywords for telescope

950: %% facilities. A common set of keywords will make these types of searches

951: %% significantly easier and more accurate. In addition, they will also be

952: %% useful in linking papers together which utilize the same telescopes

953: %% within the framework of the National Virtual Observatory.

954: %% See the AASTeX Web site at http://www.journals.uchicago.edu/AAS/AASTeX

955: %% for information on obtaining the facility keywords.

956: 

957: %% After the acknowledgments section, use the following syntax and the

958: %% \facility{} macro to list the keywords of facilities used in the research

959: %% for the paper.  Each keyword will be checked against the master list during

960: %% copy editing.  Individual instruments or configurations can be provided 

961: %% in parentheses, after the keyword, but they will not be verified.

962: 

963: 

964: %% Appendix material should be preceded with a single \appendix command.

965: %% There should be a \section command for each appendix. Mark appendix

966: %% subsections with the same markup you use in the main body of the paper.

967: 

968: %% Each Appendix (indicated with \section) will be lettered A, B, C, etc.

969: %% The equation counter will reset when it encounters the \appendix

970: %% command and will number appendix equations (A1), (A2), etc.

971: 

972: 

973: \section{Analytic estimates}

974: In this section,

975: we interpret our simulation results using analytic estimates.

976: 

977: 

978: \subsection{Evolution of velocities of planetesimals 

979: and protoplanets}

980: First we provide analytic formulation for the  

981: evolution of velocities of planetesimals and protoplanets 

982: necessary for subsequent discussions.

983: Consider a situation in which protoplanets are spatially separated 

984: but mutually interact due to the distant perturbations and 

985: each protoplanet is surrounded by a swarm of planetesimals.

986: We first consider equal-mass planetesimals, then the formulation 

987: is extended to 

988: the case of continuous size distribution (protoplanets are always assumed to be 

989: equal-mass).

990: The mass, mean square eccentricity, and surface number density of planetesimals

991: are represented as $m$, $\langle e^2 \rangle$, and $n$, respectively. 

992: Corresponding characters for protoplanets are 

993: $m_{\rm p}$, $\langle e_{\rm p}^2 \rangle$, and $n_{\rm p}$, respectively. 

994: The scattering cross section for planetesimal-planetesimal encounters 

995: $\sigma^{m-m'}_{\rm sca}$ and that for protoplanet-planetesimal encounters    

996: $\sigma^{m_{\rm p}-m'}_{\rm sca}$ are given by \citep{IM93}

997: \begin{equation}

998: \sigma^{m-m'}_{\rm sca} = C_{e}\left(\frac{G(m+m')}

999: {(a\Omega)^2 (\langle e^2 \rangle + \langle e'^2 \rangle)}\right)^2, 

1000: \label{eq:mmsca1}

1001: \end{equation}

1002: \begin{equation}

1003: \sigma^{m_{\rm p}-m'}_{\rm sca} = C_{e}\left(\frac{G(m_{\rm p}+m')}

1004: {(a\Omega)^2 (\langle e_{\rm p}^2 \rangle + \langle e'^2 \rangle)}\right)^2,

1005: \label{eq:mpsca1}

1006: \end{equation}

1007: where $C_{e}$ is the numerical factor of $\sim 40$, 

1008: $G$ is the gravitational constant, $a$ is the distance of the system from 

1009: the Sun,

1010: and $\Omega$ is the orbital frequency. In equation~(\ref{eq:mmsca1}), 

1011: we used primed characters, $m'$ and  $\langle e'^2 \rangle$, to distinguish two 

1012: interacting planetesimal groups (the primed values are averaged later). 

1013: Using the scattering cross sections, 

1014: the change rates of $\langle e^2 \rangle$ and $\langle e_{\rm p}^2 \rangle$ are 

1015: given by 

1016: \begin{eqnarray}

1017: \frac{1}{\Omega}\frac{\langle e^2 \rangle}{dt} &=& 

1018: n \left(\frac{m'}{m + m'}\right)^2  \sigma^{m-m'}_{\rm sca}(\langle e^2 \rangle + 

1019: \langle e'^2 \rangle)

1020: + n_{\rm p} \left(\frac{m_{\rm p}}{m_{\rm p} + m}\right)^2 

1021:  \sigma^{m_{\rm p}-m}_{\rm sca}(\langle e_{\rm p}^2 \rangle + \langle e^2 \rangle)

1022:  \nonumber \\

1023: &=& C_{e} \left(\frac{G}{(a\Omega)^2}\right)^2 

1024: \left(\frac{nm'^2}{\langle e^2 \rangle + \langle e'^2 \rangle} +

1025: \frac{n_{\rm p}m_{\rm p}^2}{\langle e_{\rm p}^2 \rangle + \langle e^2 \rangle} 

1026: \right), \label{eq:mmsca2}\\

1027: \frac{1}{\Omega}\frac{\langle e_{\rm p}^2 \rangle}{dt} 

1028: &=&

1029: \left(\frac{1}{m_{\rm p} + m'}\right)^2 

1030: \left( 4nm'^2 \langle e'^2 \rangle - 3nm'm_{\rm p}\langle e_{\rm p}^2 \rangle + nm'^2

1031: \langle e_{\rm p}^2 \rangle \right)

1032: \sigma^{m_{\rm p}-m'}_{\rm sca} + \frac{\langle e_{\rm p}^2 \rangle}

1033: {\Omega T_{\rm Inst}},  \label{eq:mpsca2}

1034: \end{eqnarray}

1035: where $T_{\rm Inst}$ is the time scale for orbital instability of protoplanets 

1036: (see \S~5.2).

1037: In equation~(\ref{eq:mmsca2}), the first term stands for the viscous stirring due to 

1038: planetesimal-planetesimal encounters while the second term for 

1039: the viscous stirring due to protoplanet-planetesimal encounters 

1040: \citep{IM93}. In equation~(\ref{eq:mpsca2}), the first term 

1041: stands for the sum of viscous stirring and the dynamical 

1042: friction both due to encounters with 

1043: planetesimals \citep{IM92} while the second term approximately 

1044: accounts for 

1045: the enhancement due to distant interactions between protoplanets.

1046: The rate of change of the inclinations are given by similar equations, but here 

1047: we omit them. 

1048: 

1049: The surface number density per unit mass is given by $dn/dm$.

1050: As in the simplest case, we assume that $\langle e'^2 \rangle$ is 

1051: independent of mass.

1052: In this case, after integration of 

1053: equations~(\ref{eq:mmsca2}) and 

1054: (\ref{eq:mpsca2}) over the range of $m'$,

1055: $nm'$ and $nm'^2$ in these equations can be replaced by 

1056: \begin{equation} 

1057: \int m' dn, = \Sigma_{\rm s}, \hspace{1em}

1058: \int m'^2 dn = \Sigma_{\rm s} m_{\rm eff}, \label{eq:meff} 

1059: \end{equation}

1060: respectively.

1061: Here $\Sigma_{\rm s}$ and 

1062: $m_{\rm eff} = \langle m'^2 \rangle/\langle m' \rangle$ 

1063: are the surface density and effective 

1064: mass of planetesimals,

1065: respectively. 

1066: Using these averaged quantities, we have simple implications from 

1067: equations~(\ref{eq:mmsca2}) and (\ref{eq:mpsca2}).

1068: Equation~(\ref{eq:mmsca2}) suggests that the velocity evolution of 

1069: planetesimals is regulated 

1070: by protoplanet-planetesimal encounters rather than by 

1071: planetesimal-planetesimal encounters if 

1072: \begin{equation} 

1073: n_{\rm p} m_{\rm p}^2 > f_e \Sigma_{\rm s} m_{\rm eff}, \label{eq:pdom}

1074: \end{equation} 

1075: where $f_e \le 1$ is the numerical factor associated with the velocity distribution.

1076: Equation~(\ref{eq:mpsca2}) suggests that the eccentricity of planets is given by

1077: \begin{equation} 

1078: \langle e_{\rm p,eq}^2 \rangle = \frac{4m_{\rm eff}}{3 m_{\rm p}-m_{\rm eff}}

1079: \langle e^2 \rangle, \label{eq:epeq} 

1080: \end{equation}

1081: in the equilibrium state provided that the mutual interaction 

1082: between protoplanets is negligible.

1083: 

1084: During the giant impact stage, in which orbits of 

1085: protoplanets cross each other,

1086: the second term in the right hand side of equation~(\ref{eq:mpsca2}) 

1087: can be replaced by

1088: the mutual viscous stirring term:

1089: \begin{equation}

1090: \left(\frac{1}{\Omega}\frac{\langle e_{\rm p}^2 \rangle}{dt}\right)_{\rm mut} 

1091: = \frac{1}{2}\sigma^{m_{\rm p}-m_{\rm p}}_{\rm sca} 

1092: \langle e_{\rm p}^2 \rangle,  

1093: \label{eq:mpsca3}

1094: \end{equation}

1095: with the cross section for the mutual scattering

1096: \begin{equation}

1097: \sigma^{m_{\rm p}-m_{\rm p}}_{\rm sca} = C_{e}\left(\frac{Gm_{\rm p}}

1098: {(a\Omega)^2 \langle e_{\rm p}^2 \rangle}\right)^2.

1099: \label{eq:mpsca4}

1100: \end{equation}

1101: The averaged eccentricity of protoplanets during the giant impact stage

1102: is determined by the balance between the mutual viscous stirring and 

1103: the dynamical friction due to surrounding planetesimals.

1104: Assuming $m_{\rm p} \gg m_{\rm eff}$  and 

1105: $m_{\rm p}\langle e_{\rm p}^2 \rangle \gg m_{\rm eff}\langle e^2 \rangle$, we obtain

1106: \begin{equation}

1107: \langle e_{\rm p}^2 \rangle^{1/2}_{\rm GI} = 

1108: \left(\frac{1}{6}\frac{n_{\rm p}m_{\rm p}}{\Sigma_{\rm s}}\right)^{1/4}

1109: \langle e^2 \rangle^{1/2}.

1110: \label{eq:epgi}

1111: \end{equation} 

1112: The same expression is obtained in \citet{Gol04} except for

1113: a factor of order unity. 

1114: Equation~(\ref{eq:epgi}) somewhat overestimates $\langle e_{\rm p}^2 \rangle^{1/2}$ 

1115: during the giant impact stage as compared with those 

1116: obtained from our simulations. This is probably 

1117: because the motion of all the planets are not enhanced simultaneously

1118: in our simulations as some planets are in stable orbits isolated from others. 

1119:  

1120: \subsection{Comparison of time scales and timing of the onset of giant impacts}

1121: 

1122: Here we discuss how the timing of the onset of the giant impact stage 

1123: is affected by the radius enhancement factor $g$. 

1124: Since the time and distance can be rescaled for 

1125: our $N$-body simulations, $g$ is physically associated with 

1126: the ratio of the physical radius $r$ to the Hill radius $r_{\rm H}$ as

1127: $g \propto r/r_{\rm H} \propto a^{-1}\rho^{-1/3}$.

1128: 

1129: For simplicity in this section, we call the most massive body in its

1130: feeding zone of width of $b_{\rm p}r_{\rm H}$, a planet 

1131: (note that the definition of planets used in the 

1132: main text follows the discussion in \S~5.3). 

1133: Defining the mass ratio of the planet with the total mass in the 

1134: feeding zone to be $f_{\rm p}$, the planet mass $m_{\rm p}$ 

1135: is given as 

1136: \begin{eqnarray}

1137: m_{\rm p} &=& 2 \pi f_{\rm p} a b_{\rm p} r_{\rm H} \Sigma

1138: = \left(2 \pi f_{\rm p}b_{\rm p}\Sigma \right)^{3/2} a^3 (2/3M_{\odot})^{1/2}

1139: \nonumber \\

1140: &\simeq& 0.06 \left(\frac{b_{\rm p}}{10} \right)^{3/2}

1141: \left(\frac{\Sigma}{20 {\rm \hspace{0.2em} g \hspace{0.2em} cm}}

1142: \right)^{3/2}

1143: \left(\frac{f_{\rm p}}{0.25}

1144: \right)^{3/2}

1145: \left(\frac{a}{1 {\rm AU}} \right)^{3} M_{\rm E}, \label{eq:mp}

1146: \end{eqnarray}

1147: where $\Sigma$ is the surface density of all the bodies 

1148: (thus assumed to be constant regardless of $f_{\rm p}$).

1149: For $f_{\rm p} = 1$, $m_{\rm p}$ corresponds to the so-called isolation

1150: mass \citep[e.g.][]{KI00}. 

1151: 

1152: Considering a two component system composed of

1153: planets and surrounding planetesimals,

1154: we introduce the following five timescales, which 

1155: characterize the evolution of planets and surrounding planetesimals.

1156: [1] The growth time scale of planet $T_{\rm grow}$ is given by 

1157: \begin{equation}

1158: T_{\rm grow} = \left|\frac{1}{m_{\rm p}} 

1159: \frac{dm_{\rm p}}{dt} \right|^{-1} =

1160: \frac{m_{\rm p}}{(1-f_{\rm p})\Sigma \sigma_{\rm col} \Omega},  

1161: \label{eq:tgrow}

1162: \end{equation}

1163:  with the collisional cross section $\sigma_{\rm col}$

1164: \citep{GL92}:

1165: \begin{equation}

1166: \sigma_{\rm col} = c_{\rm col}(gr_{\rm p})^2

1167: \left(\frac{1}{3}+\frac{1}{\tilde{v}^2} \right),

1168: \end{equation}

1169: where $c_{\rm col}$ is 

1170: a numerical factor of $\sim 8$ 

1171: (we assume $\langle e^2 \rangle^{1/2} = 2\langle i^2 \rangle^{1/2}$)

1172: and $\tilde{v}$ is the averaged velocity of planetesimals 

1173: normalized by the escape velocity of 

1174: the planet.

1175: [2] The depletion time scale of planetesimals $T_{\rm dep}$ is given by 

1176: \begin{equation}

1177: T_{\rm dep} = \left|\frac{1}{(1-f_{\rm p})\Sigma} 

1178: \frac{d(n_{\rm p}m_{\rm p})}{dt} \right|^{-1} =

1179: \frac{m_{\rm p}}{f_{\rm p}\Sigma \sigma_{\rm col} \Omega}, 

1180: \label{eq:tdep}

1181: \end{equation}

1182: where $n_{\rm p} = f_{\rm p}\Sigma/m_{\rm p}$ is the surface number density 

1183: of planets.

1184: [3] The time scale for the evolution of the velocity of planetesimals 

1185: $T_{\rm sca}$ due to gravitational scattering by planets is given by

1186: \begin{equation}

1187: T_{\rm sca} = \left|\frac{1}{\langle e^2 \rangle} 

1188: \frac{d\langle e^2 \rangle}{dt} \right|^{-1} \label{eq:tsca}

1189: = \frac{m_{\rm p}}{f_{\rm p}\Sigma \sigma_{\rm sca} \Omega}. 

1190: \end{equation}

1191: Here the scattering cross section $\sigma_{\rm sca}$ corresponds to 

1192: $\sigma^{m_{\rm p}-m'}_{\rm sca}$ (eq.~[\ref{eq:mpsca1}]) in the 

1193: limit of $m_{\rm p} \gg m'$ and 

1194: $\langle e_{\rm p}^2 \rangle \ll \langle e'^2 \rangle$

1195: \begin{equation}

1196: \sigma_{\rm sca} = 

1197: c_{\rm sca}(gr_{\rm p})^2 \frac{1}{\tilde{v}^4}, \label{eq:sigsca}

1198: \end{equation}

1199: where $c_{\rm sca}$ is a numerical factor of $\sim 16$.

1200: Note that $\sigma_{\rm sca}$ does not directly depend

1201: on $g$. However, as the velocity can be scaled by the escape velocity 

1202: of the planet, which depends on $g$,

1203: $\sigma_{\rm sca}$ indirectly depends on $g$ as represented by 

1204: equation~(\ref{eq:sigsca}).

1205: [4] The time scale for damping the velocity of planets $T_{\rm damp}$ 

1206: due to dynamical friction of surrounding planetesimals is given by 

1207: \begin{equation}

1208: T_{\rm damp} = \left|\frac{1}{\langle e_{\rm p}^2 \rangle} 

1209: \frac{d\langle e_{\rm p}^2 \rangle}{dt} \right|^{-1} \label{eq:tdamp}

1210: = \frac{m_{\rm p}}{(1-f_{\rm p})\Sigma \sigma_{\rm sca} \Omega}. 

1211: \end{equation}

1212: [5] The time scale for the orbital instability $T_{\rm inst}$

1213: of a multiple protoplanetary 

1214: system without any damping force represents 

1215: either the time of the first collision or the first close encounter 

1216: \citep[see][for comparison of 

1217: $T_{\rm inst}$ with different types of definitions]{Yos99}.

1218: The form of  

1219: $T_{\rm inst}$ is empirically given by \citep{Cha96, Yos99, IT99, IO06}

1220: \begin{equation}

1221: \log \left(\frac{T_{\rm inst}}{T_{\rm orb, 1}}\right) = c_1 b_{\rm p} + c_2,

1222: \label{eq:tinst}

1223: \end{equation}

1224: where $T_{\rm orb, 1}$ is the orbital period of 

1225: the innermost protoplanet, and 

1226: $c_1$ and $c_2$ are numerical coefficients. 

1227: These coefficients  

1228: depend strongly on the orbital eccentricity and relatively weakly on  

1229: the absolute averaged mass and the variation of masses.

1230: The dependence of $c_1$ and $c_2$ on orbital eccentricities is 

1231: summarized in Table~III of \citet{Yos99}.

1232: For $b_{\rm p} = 10$ and $\bar{e}_{\rm p} \sim 4\bar{h}_{\rm p}$,

1233: $T_{\rm inst}/T_{\rm orb, 1} \sim 10^5$-- $10^6$. 

1234: 

1235: For simplicity, we normalize all the timescales as follows:

1236: \begin{equation}

1237: \tilde{T} =  \frac{c_{\rm col} \Sigma (gr_{\rm p})^2}{m_{\rm p}} \Omega T.  

1238: \end{equation}

1239: With this normalization, 

1240: all the time scales except for $\tilde{T}_{\rm inst}$ depend 

1241: only on $\tilde{v}$ and $f_{\rm p}$ as 

1242: \begin{eqnarray}

1243: \tilde{T}_{\rm grow}^{-1} &=& \left(1-f_{\rm p}\right)

1244: \left(\frac{1}{3} + \frac{1}{\tilde{v}^2}\right), \nonumber \\

1245: \tilde{T}_{\rm dep}^{-1} &=& f_{\rm p}

1246: \left(\frac{1}{3} + \frac{1}{\tilde{v}^2}\right), \nonumber \\

1247: \tilde{T}_{\rm sca}^{-1} &=& 2f_{\rm p}

1248: \frac{1}{\tilde{v}^4}, \nonumber \\

1249: \tilde{T}_{\rm damp}^{-1} &=& 2\left(1-f_{\rm p}\right)

1250: \frac{1}{\tilde{v}^4}.

1251: \end{eqnarray}

1252: It should be noted that now the dependence on $g$ 

1253: is included only in $\tilde{T}_{\rm inst}$.

1254: 

1255: In order to obtain the time scales as functions of $f_{\rm p}$,

1256: we consider the evolution of $\tilde{v}$. 

1257: In the early stages the smallest planetesimals dominate the mass of 

1258: the system so that the velocity is as large as the escape velocity 

1259: of planetesimals.  The exact value of $\tilde{v}$ 

1260: at the initial state (when $m_{\rm p} = m_0$) is determined by 

1261: the balance between the mutual scattering and the collisional damping.  

1262: Since the ratio of these time scales is given by the ratio of 

1263: $\sigma_{\rm col}$ to $\sigma_{\rm sca}$, we obtain  

1264: $\tilde{v} = 1.17$ for the initial state. 

1265: Since the dimensional velocity of planets is $1.17v_{\rm esc,0}$,

1266: $\tilde{v}$ decreases as planets grow (or with increasing $f_{\rm p}$).

1267: When $f_{\rm p}$ reaches to a certain value, 

1268: planets start to regulate the velocities of surrounding planetesimals. 

1269: In this case, $\tilde{v}$ evolves as planets grow. 

1270: Thus, $\tilde{v}$ is obtained from 

1271: $\tilde{T}_{\rm grow} = \tilde{T}_{\rm sca}$ 

1272: \citep{Dai06}. 

1273: The value of $\tilde{v}$ increases with $f_{\rm p}$ to 

1274: 1.17 at $f_{\rm p} = 0.5$. 

1275: For $f_{\rm p} > 0.5$, 

1276: $\tilde{T}_{\rm dep}$ becomes shorter than $\tilde{T}_{\rm sca}$,

1277: if $\tilde{v} > 1.17$. 

1278: This means that planetesimals collide with planets 

1279: before their velocities are further enhanced. 

1280: Therefore, $\tilde{v}$ takes a constant 

1281: value, 1.17, for $f_{\rm p} > 0.5$.  

1282: 

1283: We plot the evolution of $\tilde{v}$ as a function of $f_{\rm p}$ in Figure~7.

1284: In the same figure, we also plot 

1285: $\tilde{T}_{\rm grow}$, $\tilde{T}_{\rm sca}$, $\tilde{T}_{\rm damp}$, and 

1286: $\tilde{T}_{\rm inst}$ with the parameters used in 

1287: Run~6 and $b_{\rm p} = 10$. 

1288: For $\tilde{T}_{\rm inst}$, we  

1289: take $T_{\rm inst}/T_{\rm orb} = 5 \times 10^5$ with 

1290: $g = 1$ and 4.3 as examples. This time scale would be appropriate 

1291: judging from our simulations.

1292: It should be noted that $\tilde{T}_{\rm dep} = \tilde{T}_{\rm sca}$ 

1293: for $f_{\rm p} > 0.5$ and $\tilde{T}_{\rm dep}$ further increases 

1294: with decreasing $f_{\rm p}$ for $f_{\rm p} \le 0.5$, although we do not plot 

1295: $\tilde{T}_{\rm dep}$ on Figure~7 in order to avoid confusion due to 

1296: too many lines. 

1297: 

1298: \citet{Iwa02} and \citet{IO06}

1299: showed that orbital instability is prevented 

1300: when $\tilde{T}_{\rm damp} < c_3\tilde{T}_{\rm inst}$, 

1301: with a coefficient $c_3$

1302: of the order of unity (we assume $c_3 = 3$  after \citet{IO06}).

1303: Hence, the value of $f_{\rm p}$ at the onset of giant impacts

1304: is estimated from the equation $\tilde{T}_{\rm damp} = 3\tilde{T}_{\rm inst}$,

1305: and is 0.3--0.4 and 0.8--0.9 for $g = 1$ and 4.3, respectively, in Figure~7.

1306: However, Figure~7 also suggests that  

1307: $\tilde{T}_{\rm inst}(g =4.3) > \tilde{T}_{\rm dep}$, 

1308: which means that remnant planetesimals are depleted before 

1309: the giant impact phase starts, as long as we keep $g = 4.3$. 

1310: Complete depletion of planetesimals keeping  

1311: $b_{\rm p} \sim 10$ is found in simulations with $g = 6$ in 

1312: \citet{KI02} and \citet{LR05}, 

1313: and we also found the same results with additional tests.

1314: Therefore, in order to keep some amount of planetesimals 

1315: at the onset of the giant impact stage, it must start before 

1316: $f_{\rm p}$ reaches 0.5 (strictly speaking, this value is slightly higher than 0.5 for $c_3 > 1$). 

1317: This also corresponds to the condition 

1318: for the damping of enhanced eccentricities of planets during 

1319: the giant impact stage. The condition under which the

1320: eccentricities of planets are damped before depletion of 

1321: planetesimals is given by $\tilde{T}_{\rm damp} < \tilde{T}_{\rm dep}$, 

1322: which gives $f_{\rm p} < 0.5$ (Fig.~7). 

1323: For the case of $g = 1$, this condition is satisfied. 

1324: The threshold value of $g$ to satisfy the condition is roughly estimated 

1325: to be 3.

1326: 

1327: For our simulations (except Runs~1b and 5b), 

1328: we reduced $g$ from 4.3 to 1 when $f_{\rm p}$ is about 0.6 (Figs.~4 and 5).

1329: After reducing $g$, $\tilde{T}_{\rm inst}$ becomes smaller than 

1330: any other timescale (Fig.~7). 

1331: Thus, the giant impact phase rapidly begins

1332: (the instantaneous reduction of $g$ causes drop of $\tilde{v}$ to 

1333: $\sim 1.17/\sqrt{4.3}$, but $\tilde{T}_{\rm sca}$ is shorter than 

1334: $\tilde{T}_{\rm grow}$ by a factor of 4.3 with this small $\tilde{v}$

1335: and thus $\tilde{v}$ increases near to 1.17 again before 

1336: $f_{\rm p}$ substantially increases).

1337: On the other hand, for Runs~1b and 5b, 

1338: $f_{\rm p}$ is about $0.4$ when we reduce $g$ at $5 \times 10^4$ yr.

1339: In these cases giant impacts do not start immediately as 

1340:  $\tilde{T}_{\rm damp} \sim 3\tilde{T}_{\rm inst}$ at $f_{\rm p} \sim 0.4$

1341: (this can be seen in Figs.~4 and 5 as $N_{\rm p}$ does not change 

1342: immediately after reducing $g$). 

1343: In fact, after $f_{\rm p}$ increases to $\sim 0.5$, giant impacts start

1344: in these simulations.

1345: Therefore, it is expected that reducing $g$ at an earlier time 

1346: does not affect the results (we have now conducted some simulations 

1347: keeping fixed $g = 1$ and obtained consistent results; 

1348: these results will be reported in another paper).

1349: 

1350: To summarize, for simulations with a constant $g$ throughout the entire

1351: accretion stage, sweeping up all the planetesimals and 

1352: a subsequent giant impact stage are expected for $g > 3$, 

1353: whereas giant impacts during accretion of planetesimals

1354: and a subsequent damping of eccentricities of planets due to 

1355: the dynamical friction are expected for $g < 3$.

1356: For $g < 3$, the evolution and final state of a system are 

1357: expected to weakly depend on $g$, from the comparison between 

1358: Runs~1 and 1b and that between Runs~5 and 5b. 

1359: 

1360: \subsection{Minimum oligarchic mass and equilibrium eccentricity of protoplanets}

1361: 

1362: Once the mass of the largest body is above a certain critical mass, 

1363: the velocity evolution of 

1364: neighboring planetesimals is primarily regulated by the largest body.

1365: We call this critical mass the minimum oligarchic mass.  

1366: Then, the growth of smaller planetesimals near the largest body 

1367: is stalled and as the largest body grows 

1368: it starts to separate from the continuous size distribution of planetesimals.

1369: Therefore, the planetesimal size distribution after this stage is 

1370: expected to be a continuous distribution with the maximum mass being 

1371: the minimum oligarchic mass. This size distribution determines the equilibrium 

1372: eccentricity of planets after the giant impact stage. 

1373: Here we estimate the planetesimal size distribution, when 

1374: the largest body starts to regulate the velocity evolution of 

1375: all of its neighboring planetesimals.

1376: 

1377: We consider the power-law size distribution for planetesimals 

1378: $dn \propto m^{q}dm$ with the upper and lower cutoff masses 

1379: $m_{\rm p}$ and $m_0$. The condition that the largest body regulates 

1380: the velocity evolution is again given by (eq.~[\ref{eq:pdom}])

1381: \begin{equation}

1382: m_{\rm p}^2 > f_e m_{\rm T} m_{\rm eff}, \label{eq:pdom2}

1383: \end{equation}

1384: where $f_e$ is a factor associated with the velocity distribution

1385: and is slightly smaller than unity and $m_{\rm T}$ ($= \Sigma_{\rm s}/n_{\rm p}$) 

1386: is the total mass of planetesimals (excluding the largest body or a planet) 

1387: in the heated region, where velocities are regulated 

1388: by a planet \citep[see][]{IM93}, with $n_{\rm p}$ being the 

1389: surface number density of planets (the sizes of the heated region and 

1390: the feeding zones are similar).

1391: We also define the cumulative number in the heated region

1392: to be  $N_{\rm h} = n/n_{\rm p}$, 

1393: and now the size distribution is given by $dN_{\rm h} = k m^{q} dm$.

1394: Using the condition for the mass of the largest planetesimals 

1395: (or the second largest body in the heated region) $m_{\rm L}$

1396: \begin{equation}

1397: 2 = \int^{\infty}_{m_{\rm L}} dN_{\rm h} = 2\frac{km_{\rm L}^{q+1}}{-q-1},

1398:  \hspace{1em} ({\rm for} \hspace{0.5em} q < -1)

1399: \end{equation}

1400: the total mass $m_{\rm T}$ and the effective mass $m_{\rm eff}$ are,

1401:  respectively, given by 

1402: \begin{eqnarray}

1403: m_{\rm T} &=& \int^{m_{\rm L}}_{m_0} m dN_{\rm h} = 2\frac{q+1}{q+2} 

1404: \left[\left(\frac{m_{\rm L}}{m_0}\right)^{-q-2} - 1\right] 

1405: m_{\rm L}, \hspace{1em} ({\rm for} \hspace{0.5em} q \ne -2)

1406: \label{eq:mt1}\\

1407: m_{\rm T}m_{\rm eff} &=& \int^{m_{\rm L}}_{m_0} m^2 dN_{\rm h} =

1408: -2\frac{q+1}{q+3} \left[ 1-\left(\frac{m_0}{m_{\rm L}}\right)^{q+3}\right]

1409: m_{\rm L}^2. 

1410: \hspace{1em} 

1411: ({\rm for} \hspace{0.5em} q \ne -3) \label{eq:meff1}

1412: \end{eqnarray} 

1413: For $q = -2$, we have  $m_{\rm T} = 2m_{\rm L}\ln(m_{\rm L}/m_0)$. 

1414: 

1415: Substituting  equation~(\ref{eq:meff1}) and the relation  

1416: $m_{\rm L} = 2^{1/(q+1)}m_{\rm p}$ into equation~(\ref{eq:pdom2}),

1417: the condition approximately becomes  

1418: \begin{equation}

1419: 1 < - f_e \frac{q+1}{q+3} 2^{(q+3)/(q+1)} \label{eq:pf},

1420: \end{equation}

1421: which gives $q > -2.2$ for $f_e=1$. 

1422: Since $f_e$ is expected to be slightly smaller than unity, 

1423: $q$ would be slightly smaller than $-2.2$

1424: when the largest body starts to regulate the velocity evolution. 

1425: This might correspond to the value $q\simeq -2.5$ obtained by \cite{KI96}.

1426: 

1427: The fraction of planetary mass $f_{\rm p}$ is given by  

1428: \begin{equation}

1429: f_{\rm p} = \frac{m_{\rm p}}{\int^{m_{\rm p}}_{m_0} m dN_{\rm h}} = 

1430: \frac{q+1}{q+2}\left[(m_{\rm p}/m_{0})^{-q-2}-1\right]^{-1}. \label{eq:fp2}

1431: \end{equation}

1432: Substituting it 

1433: into equation~(\ref{eq:mp}),  

1434: we obtain the minimum oligarchic mass $m_{\rm p,oli} = m_{\rm p}(q \sim -2.2)$ 

1435: as function of $q$. 

1436: This is shown in Figure~8 as well as the corresponding 

1437: $m_{\rm eff}$ for the parameters of Run~6.  

1438: Using $m_{\rm p}(q)$, we also plot the evolution of 

1439: $q$ as a function of $f_{\rm p}$ in Figure~7. 

1440: As we estimated above,  Figure~7  suggests that 

1441: $\tilde{v}$ starts to increase at $q \sim -2.5$ because of 

1442: the gravitational scattering of planets.

1443: Therefore, when $q \sim -2.5$,  

1444: growth of small planetesimals slows down as well as 

1445: the evolution of their size distribution.

1446: 

1447: However, at this stage, the velocities are still not high enough to suppress the growth 

1448: of large planetesimals.   

1449: Therfore, mass transfer from small planetesimals to large planetesimals 

1450: further increases $q$ to $\sim -2$.  

1451: When $q \sim -2$, the velocity of planetesimals in the heated region 

1452: is almost regulated by a single planet, whereas

1453: the contribution from a planet and all the other bodies are the same for

1454: $q \sim -2.5$. 

1455: As a result, 

1456: the velocity of planetesimals is enhanced toward the escape velocity of 

1457: the planet. Then, actual separation of the planet

1458: from the continuous size distribution starts as we showed in \S~3.1 

1459: (that is also why we keep $q = -2$ for large $f_{\rm p}$

1460: in Fig.~7, although there is no physical reason for $q$ to be strictly $-2$). 

1461: Therefore, an appropriate 

1462: minimum oligarchic mass to give the size distribution after the giant 

1463: impact stage seems  

1464: to be $m_{\rm p,oli}$ for $q \simeq -2$ 

1465: ($m_{\rm p, oli} \sim 60m_0$ for $q=-2$ whereas $\sim 20m_0$ for $q=-2.5$). 

1466: In the case of $q=-2$, 

1467: $m_{\rm p,oli}$ and $m_{\rm eff}$ 

1468: depend on the initial planetesimal mass $m_{0}$

1469: very weakly as $m_{\rm p,oli}(q=-2) \propto [\ln{(m_{\rm p,oli}/m_0)}]^{-3/2}\Sigma^{3/2}$

1470: and $m_{\rm eff}(q=-2)= 0.5m_{\rm p,oli}(q=-2)/\ln{(0.5m_{\rm p,oli}/m_0)}$.

1471: 

1472: Using the form of $m_{\rm eff}$ for $q = -2$, we rewire 

1473: the equilibrium eccentricity (eq.~[\ref{eq:eeq}]) as 

1474: \begin{equation}

1475: e_{\rm p,eq} = \left(\frac{2}{3}\frac{m_{\rm p,oli}/m_0}{\ln{(0.5m_{\rm p,oli}/m_0)}} \right)^{1/2}

1476: \left(\frac{m_0}{m_{\rm p}}\right)^{1/6} v_{\rm esc,0}, 

1477: \hspace{1em} ({\rm for} \hspace{0.2em} q = -2). \label{eq:eeq2}

1478: \end{equation}

1479: Substituting $m_{\rm p,oli}(q = -2)$, 

1480: which is obtained from eqs.~[\ref{eq:mp}] and [\ref{eq:fp2}], into eq.~[\ref{eq:eeq2}],

1481: we calculate $e_{\rm p,eq}$ for various cases 

1482: of $N$ (or $m_0$) and $\Delta_{\rm disk}$ (or $\Sigma$), assuming three final planets 

1483: (thus $m_{\rm p} = 0.66M_{\rm E}$) and the velocity of planetesimals to be the escape velocity 

1484: of a planet ($v = v_{\rm esc, p}$). 

1485: 

1486: The calculated values are plotted in Fig.~9 and compared with our $N$-body simulation results. 

1487: We find good agreements between analytic estimates and simulations, 

1488: as long as the planet-planet interactions are not important in the final state.

1489: Since the dependence of $e_{\rm p,eq}$ on $N$ is very weak for $N > 1000$,

1490: we need more runs for statistics.

1491: We also conducted additional simulations for the case of $N = 100$ and results 

1492: are plotted in the same figure. 

1493: In most of simulations with $N = 100$ (we conducted four 

1494: simulations for each $\Delta_{\rm disk}$), 

1495: giant impacts occur after nearly complete 

1496: sweep up of remnant planetesimals. 

1497: Therefore, it would not be appropriate to apply 

1498: our analytic estimate to the case of $N = 100$. 

1499: Nevertheless, we find coincidental good agreements 

1500: between the averaged $e_{\rm p,eq}$'s for simulations with

1501: $N = 100$ and those from the analytic estimates.

1502:  

1503: Finally, let us discuss what will happen if our simulations started with a very small size 

1504: distribution of planetesimals. In Figure~8, we also plot $m_{\rm p,oli}$ and $m_{\rm eff}$

1505: as functions of $q$ for the parameters used in 

1506: \citet{Wei97} ($m_0 = 4.8 \times 10^{18}$g), who conducted planetary accretion 

1507: simulations in the runaway and oligarchic stages with their multi-zone code.  

1508: For $(m_{\rm p,oli}/m_{0})^{-q-2} \gg 1$ (this is not the case for our simulations),

1509: the minimum oligarchic mass is given by 

1510: \begin{equation}

1511: m_{\rm p,oli} = \left(\frac{q+1}{q+2} 2 \pi  m_0^{-q-2} \Sigma a^2 

1512: b_{\rm p} \left(\frac{2}{3M_{\odot}}\right)^{1/3} \right)^{-1/(q+4/3)}.

1513: \end{equation}

1514: This equation indicates that $m_{\rm p,oli}$ decreases rather strongly

1515: with $m_{0}$ for small $q$, as $m_{\rm p,oli} \propto m_{0}^{(q+2)/(q+4/3)}$.

1516: Therefore, the minimum oligarchic mass with $q = -2.5$ for $m_0 = 4.8 \times 10^{18}$g 

1517: is much smaller than ours ($m_0 = 3.94 \times 10^{24}$g), 

1518: and oligarchic bodies started to regulate the velocity evolution even 

1519: when $f_{\rm p}$ is very small (estimated to be $\sim 10^{-3}$). 

1520: On the other hand, $m_{\rm p,oli}$ and  $m_{\rm eff}$ for $q = -2$

1521: are only one order of magnitude smaller than those for Run~6. 

1522: Indeed, in simulations of \citet{Wei97}, 

1523: the maximum mass of remnant planetesimals is $\sim 10^{25}$g

1524: (whereas oligarchic bodies have masses $\sim 10^{27}$g), 

1525: suggesting $q \sim -2$.

1526: Therefore, from eq.~[\ref{eq:eeq2}] the final equilibrium eccentricities is expected to be 

1527: smaller than those in our simulations only by a factor of 3--4.

1528: In the discussion here, we ignored the effect of damping due to 

1529: mutual collisions,

1530: which would reduce the eccentricities of planetesimals.

1531: If the fast clean up of remnant planetesimals happens due to the collisional 

1532: damping, the final planetary system might be unstable, as in the case of a large $g$.  

1533: 

1534: \section{Conclusions} 

1535: 

1536: We have conducted direct $N$-body simulations of the

1537: formation of terrestrial planets beginning with a compact 

1538: planetesimal disk, with 

1539: the total mass and angular momentum being those  

1540: for the current terrestrial planets. 

1541: In most of the cases, a planetesimal disk results in 

1542: three planets of similar size, and sometimes a fourth

1543: small planet forms around the location of Mars. 

1544: Since a sufficient number of planetesimals remain 

1545: even after giant impacts of protoplanets, 

1546: orbital eccentricities of the final planets 

1547: are as small as those of the current Earth and Venus.

1548: This is a very nice success of our simulations which test

1549: the standard model for planet formation.

1550: 

1551: The final eccentricities of planets are nearly

1552: in the equilibrium state for the energy partitioning with 

1553: remnant planetesimals, meaning that the mutual interactions 

1554: of planets in the final state is unimportant in most cases.

1555: The final eccentricities of planets depends on the initial mass 

1556: of planetesimals only very weakly, and on the surface density 

1557: relatively strongly. These dependences are interpreted in terms of 

1558:  the effective mass of remnant planetesimals 

1559: $(m_{\rm eff} = \langle m^2 \rangle/\langle m \rangle)$, which 

1560: determines the strength of the gravitational scattering effects of planetesimals.

1561: The mass distribution of remnant planetesimals is approximately

1562: represented by a power law distribution, $dn \propto m^{q}dm$, with $q \sim -2$,

1563: with the upper cut off mass (we call it the minimum oligarchic mass)

1564: which increases very weakly with the initial mass of planetesimals.

1565: Therefore, the dependence of the effective mass on the  

1566: initial mass of planetesimals is very weak as well. 

1567: 

1568: In a few of our simulations, planet-planet interactions are important even at the end of 

1569: simulations and the orbits of final planets can be

1570: much more eccentric than for our terrestrial planets. This situation seems to appear when the gradient 

1571: of population of planetesimals in the radial direction is large 

1572: during the giant impact stage. However, the number of our simulations is  

1573: still too small to statistically discuss the conditions that are responsible for

1574: final planetary orbits.

1575:  

1576: 

1577: \acknowledgments

1578: We appreciate an anonymous reviewer for useful comments.

1579: We are grateful to Derek Richardson for providing us with his version of 

1580: {\it PKDGRAV}.

1581: We thank Shigeru Ida, Makiko Nagasawa, and Eiichiro Kokubo for fruitful 

1582: discussions.

1583: Our simulations have been conducted with the zBox1 and zBox2 supercomputers 

1584: at the University of Zurich. 

1585: We thank Doug Potter for his management of the computers.

1586: 

1587: 

1588: 

1589: %% The reference list follows the main body and any appendices.

1590: %% Use LaTeX's thebibliography environment to mark up your reference list.

1591: %% Note \begin{thebibliography} is followed by an empty set of

1592: %% curly braces.  If you forget this, LaTeX will generate the error

1593: %% "Perhaps a missing \item?".

1594: %%

1595: %% thebibliography produces citations in the text using \bibitem-\cite

1596: %% cross-referencing. Each reference is preceded by a

1597: %% \bibitem command that defines in curly braces the KEY that corresponds

1598: %% to the KEY in the \cite commands (see the first section above).

1599: %% Make sure that you provide a unique KEY for every \bibitem or else the

1600: %% paper will not LaTeX. The square brackets should contain

1601: %% the citation text that LaTeX will insert in

1602: %% place of the \cite commands.

1603: 

1604: %% We have used macros to produce journal name abbreviations.

1605: %% AASTeX provides a number of these for the more frequently-cited journals.

1606: %% See the Author Guide for a list of them.

1607: 

1608: %% Note that the style of the \bibitem labels (in []) is slightly

1609: %% different from previous examples.  The natbib system solves a host

1610: %% of citation expression problems, but it is necessary to clearly

1611: %% delimit the year from the author name used in the citation.

1612: %% See the natbib documentation for more details and options.

1613: 

1614: 

1615: 

1616: 

1617: %% Use the figure environment and \plotone or \plottwo to include

1618: %% figures and captions in your electronic submission.

1619: %% To embed the sample graphics in

1620: %% the file, uncomment the \plotone, \plottwo, and

1621: %% \includegraphics commands

1622: %%

1623: %% If you need a layout that cannot be achieved with \plotone or

1624: %% \plottwo, you can invoke the graphicx package directly with the

1625: %% \includegraphics command or use \plotfiddle. For more information,

1626: %% please see the tutorial on "Using Electronic Art with AASTeX" in the

1627: %% documentation section at the AASTeX Web site,

1628: %% http://www.journals.uchicago.edu/AAS/AASTeX.

1629: %%

1630: %% The examples below also include sample markup for submission of

1631: %% supplemental electronic materials. As always, be sure to check

1632: %% the instructions to authors for the journal you are submitting to

1633: %% for specific submissions guidelines as they vary from

1634: %% journal to journal.

1635: 

1636: %% This example uses \plotone to include an EPS file scaled to

1637: %% 80% of its natural size with \epsscale. Its caption

1638: %% has been written to indicate that additional figure parts will be

1639: %% available in the electronic journal.

1640: 

1641: \clearpage

1642: \begin{thebibliography}{}

1643: \bibitem[Agnor \& Ward (2002)]{AW02} 

1644: Agnor, C. B., \& Ward, W. R.

1645: 2002, \apj, 567, 579

1646: 

1647: \bibitem[Agnor et al.(1999)]{Agn99} 

1648: Agnor, C. B., Canup, R. M., \& Levison, H. F.   

1649: 1999, \icarus, 142, 219

1650: 

1651: \bibitem[Bromley \& Kenyon(2006)]{BK06} 

1652: Bromley, B. C.,  \& Kenyon, S. J.  

1653: 2006, \aj, 131, 2737

1654: 

1655: \bibitem[Chambers(1999)]{Cha99} 

1656: Chambers, J. E.  

1657: 1999, \mnras, 304, 793

1658: 

1659: \bibitem[Chambers(2001)]{Cha01} 

1660: Chambers, J. E.  

1661: 2001, \icarus, 152, 205

1662: 

1663: \bibitem[Chambers(2004)]{Cha04} 

1664: Chambers, J. E.  

1665: 2004, E\&PSL, 223, 241

1666: 

1667: \bibitem[Chambers \& Wetherill(1998)]{CW98} 

1668: Chambers, J. E., \& Wetherill, G. W.  

1669: 1998, \icarus, 136, 304

1670: 

1671: \bibitem[Chambers et al.(1996)]{Cha96} 

1672: Chambers, J. E., Wetherill, G. W., \& Boss, A. P.

1673: 1996, \icarus, 119, 261

1674: 

1675: \bibitem[Daisaka et al.(2006)]{Dai06} 

1676: Daisaka, J. K., Tanaka, H., \& Ida, S.

1677: 2006, \icarus, 185, 492

1678: 

1679: \bibitem[Duncan et al.(1998)]{DLL98} 

1680: Duncan, M. J., Levision, H. F., \& Lee, M. H.

1681: 1998, \aj, 116, 2067

1682: 

1683: \bibitem[Greenberg et al.(1978)]{Gre78} 

1684: Greenberg, R., Hartman, W. K., Wacker, J., \& Chapman, C. R.

1685: 1978, \icarus, 35, 1  

1686: 

1687: \bibitem[Greenzewig \& Lissauer(1992)]{GL92} 

1688: Greenzewig, Y., \& Lissauer, J. J.

1689: 1992, \icarus, 100, 440

1690: 

1691: \bibitem[Haisch et al.(2001)]{Hai01} 

1692: Haisch, K. E., Lada, E. A., \& Lada, C. J.

1693: 2001, \apj, 553, L153

1694: 

1695: \bibitem[Hut et al.(1995)]{HMM95}

1696: Hut, P., Makino, J., \& McMillan, S.

1697: 1995, \apj, 443, 93

1698: 

1699: \bibitem[Ida \& Makino(1992)]{IM92} 

1700: Ida, S., \& Makino, J.

1701: 1992, \icarus, 98, 28

1702: 

1703: \bibitem[Ida \& Makino(1993)]{IM93} 

1704: Ida, S., \& Makino, J.

1705: 1993, \icarus, 106, 210

1706: 

1707: \bibitem[Inaba et al.(2001)]{Ina01} 

1708: Inaba, S., Tanaka, H., Nakazawa, K., Wetherill, G. W.

1709: \& Kokubo, E.

1710: 2001, \icarus, 149, 235

1711: 

1712: \bibitem[Ito \& Tanikawa(1999)]{IT99} 

1713: Ito, T., \& Tanikawa, K.

1714: 1999, \icarus, 139, 336

1715: 

1716: \bibitem[Iwasaki \& Ohtsuki(2006)]{IO06} 

1717: Iwasaki, K., \& Ohtsuki, K.

1718: 2006, \aj, 131, 3093

1719: 

1720: \bibitem[Iwasaki et al.(2002)]{Iwa02} 

1721: Iwasaki, K., Emori, H., Nakazawa, K., \& Tanaka, H.

1722: 2002, \pasj, 54, 471

1723: 

1724: \bibitem[Goldreich et al.(2004)]{Gol04} 

1725: Goldreich, P., Lithwick, Y., \& Sari, R.

1726: 2004, \araa, 42, 549

1727: 

1728: \bibitem[Kenyon \& Bromley(2006)]{KB06} 

1729: Kenyon, S. J., \& Bromley, B. C.  

1730: 2006, \aj, 131, 1837

1731: 

1732: \bibitem[Kinoshita et al.(1991)]{Kino91}

1733: Kinoshita, H., Yoshida, H., \& Nakai, H.

1734: 1991, CeMDA, 50, 59

1735: 

1736: \bibitem[Kokubo \& Ida(1995)]{KI95} 

1737: Kokubo, E., \& Ida, S.  

1738: 1995, \icarus, 114, 247

1739: 

1740: \bibitem[Kokubo \& Ida(1996)]{KI96} 

1741: Kokubo, E., \& Ida, S.  

1742: 1996, \icarus, 123, 180

1743: 

1744: \bibitem[Kokubo \& Ida(1998)]{KI98} 

1745: Kokubo, E., \& Ida, S.  

1746: 1998, \icarus, 131, 171

1747: 

1748: \bibitem[Kokubo \& Ida(2000)]{KI00} 

1749: Kokubo, E., \& Ida, S.  

1750: 2000, \icarus, 143, 15

1751: 

1752: \bibitem[Kokubo \& Ida(2002)]{KI02} 

1753: Kokubo, E., \& Ida, S.  

1754: 2002, \apj, 581, 666

1755: 

1756: \bibitem[Kokubo \& Makino(2004)]{KM04} 

1757: Kokubo, E., \& Makino, J.  

1758: 2004, \pasj, 56, 861

1759: 

1760: \bibitem[Kokubo et al.(2006)]{Kok06} 

1761: Kokubo, E., Kominami, J., \& Ida, S.  

1762: 2006, \apj, 642, 1131

1763: 

1764: \bibitem[Kominami \& Ida(2002)]{KomI02} 

1765: Kominami, J., \& Ida, S.  

1766: 2002, \icarus, 157, 43

1767: 

1768: \bibitem[Kominami \& Ida(2004)]{KomI04} 

1769: Kominami, J., \& Ida, S.  

1770: 2004, \icarus, 167, 231

1771: 

1772: \bibitem[Laskar(1997)]{Las97} 

1773: Laskar, J.  

1774: 1997, \aap, 317, L75

1775: 

1776: \bibitem[Leinhardt \& Richardson(2005)]{LR05} 

1777: Leinhardt, Z. M., \& Richardson, D. C.

1778: 2005, \apj, 625, 427

1779: 

1780: \bibitem[Makino et al.(1998)]{Mak98} 

1781: Makino, J., Fukushige, T., Funato, Y., \& Kokubo, E.

1782: 1998, \na, 3, 411 

1783: 

1784: \bibitem[Makino et al.(2006))]{MHKS06}

1785: Makino, J., Hut, P., Kaplan, M., \& Saygin, H.

1786: 

1787: \bibitem[Murray \& Dermott(1999)]{MD99} 

1788: Murray, C. D., \& Dermott, S. F.

1789: 1999, Solar system dynamics (Cambridge: Cambridge Univ. Press)

1790: 

1791: \bibitem[Nagasawa et al.(2005)]{Nag05} 

1792: Nagasawa, M., Lin, D. N. C., \&  Thommes, E. W.

1793: 2005, \apj, 635, 578

1794: 

1795: \bibitem[Nagasawa et al.(2007)]{Nag07} 

1796: Nagasawa, M., Thommes, E. W., Kenyon, S. J., Bromley, B. C.,

1797: \& Lin, D. N. C.  

1798: 2007, in Protostars and Planets V, 

1799: ed. Reipurth, B., Jewitt, D., \& Keil, K.

1800: (Tucson: Univ. Arizona Press), 639 

1801: 

1802: \bibitem[O'brien et al.(2006)]{Obr06} 

1803: O'brien, D. P., Morbidelli, A., \& Levison, H. F.

1804: 2006, \icarus, 184, 39 

1805: 

1806: \bibitem[Ogihara et al.(2007)]{Ogi07} 

1807: Ogihara, M., Ida, S., \& Morbidelli, A.

1808: 2007, \icarus, 188, 522

1809: 

1810: \bibitem[Quinn et al.(1991)]{Qui91}

1811: Quinn, T.,  Tremaine, S., \& Duncan, M.

1812: 1991, \aj, 101, 2287

1813: 

1814: \bibitem[Rafikov(2003)]{Raf03} 

1815: Rafikov, R. R.

1816: 2003, \aj, 126, 2529

1817: 

1818: %\bibitem[Raymond et al.(2004)]{Ray04} 

1819: %Raymond, S. N., Quinn, T., \& Lunine, J. I. 

1820: %2004, \icarus, 168, 1 

1821: 

1822: \bibitem[Raymond(2005)]{Ray05} 

1823: Raymond, S. N.

1824: 2005, PhD dissertation (Seatle: Univ. of Washinton)

1825: 

1826: \bibitem[Raymond et al.(2006)]{Ray06} 

1827: Raymond, S. N., Quinn, T., \& Lunine, J. I. 

1828: 2006, \icarus, 183, 265 

1829: 

1830: \bibitem[Richardson et al.(2000)]{Ric00}

1831: Richardson, D. C., Quinn, T., Stadel, J., \& Lake, G. 

1832: 2000, \icarus, 143, 45 

1833: 

1834: \bibitem[Saha \& Tremaine(1992)]{ST92} 

1835: Saha, P., \& Tremaine, S.

1836: 1992, \aj, 104, 1633

1837: 

1838: \bibitem[Stadel(2001)]{Sta01} 

1839: Stadel, J.

1840: 2001, PhD dissertation (Seatle: Univ. of Washinton)

1841: 

1842: \bibitem[Yoshinaga et al.(1999)]{Yos99} 

1843: Yoshinaga, K., Kokubo, E., \& Makino, J.

1844: 1999, \icarus, 139, 328

1845: 

1846: \bibitem[Youdin \& Shu(2002)]{YS02} 

1847: Youdin, A. N., \& Shu, F. H.

1848: 2002, \apj, 580, 494 

1849: 

1850: \bibitem[Weidenschilling et al.(1997)]{Wei97}

1851: Weidenschilling, S. J., Spaute, D., Davis, D. R., 

1852: Marzari, F., \& Ohtsuki, K.

1853: 1997, \icarus, 128, 429

1854: 

1855: \bibitem[Wetherill \& Stewart(1989)]{WS89} 

1856: Wetherill, G. W., \& Stewart, G. R. 

1857: 1989, \icarus, 77, 330

1858: 

1859: \bibitem[Wetherill \& Stewart(1993)]{WS93} 

1860: Wetherill, G. W., \& Stewart, G. R. 

1861: 1993, \icarus, 106, 190

1862: 

1863: \bibitem[Wisdom \& Holman(1991)]{WH91} 

1864: Wisdom, J., \& Holman, M. 

1865: 1991, \aj, 102, 1528

1866: 

1867: 

1868: 

1869: 

1870: \end{thebibliography}

1871: 

1872: 

1873:   

1874: \clearpage

1875: \begin{figure}

1876: \begin{center}

1877: \includegraphics[width=.85\textwidth]{f1.eps}

1878: \end{center}

1879: \caption{Snapshots of Run~6 on the $a$--$e$ plane. 

1880: Vertical dashed lines are the inner and outer edges of 

1881: the initial planetesimal disk. The circles represent planetesimals and planets,

1882: and plotted radius sizes are proportional to 

1883: the actual radii without artifical enhancement.}

1884: 

1885: 

1886: \end{figure}

1887: 

1888: \clearpage

1889: 

1890: \begin{figure}

1891: \begin{center}

1892: \includegraphics[width=.85\textwidth]{f2.eps}

1893: \end{center}

1894: \caption{Evolution of the cumulative number vs. mass in units 

1895: of the initial mass for Run~6.

1896: The slope of the approximated power-law distribution, 

1897: $dN_{\rm c} \propto m^{q}dm$, is shown in each panel.}

1898: \end{figure}

1899: 

1900: \clearpage

1901: 

1902: \begin{figure}

1903: \begin{center}

1904: \includegraphics[width=.85\textwidth]{f3.eps}

1905: \end{center}

1906: \caption{Evolution of the epicyclic velocity vs. mass 

1907: for Run~6. The velocity is normalized by the Keplerian velocity 

1908: $v_{\rm kep,mid}$ at $a=0.89$AU.

1909: The upper and lower horizontal dashed lines represent 

1910: the escape velocities of the largest body and the smallest body, 

1911: respectively. Note that the escape velocity of the smallest body 

1912: increases by a factor of $\sqrt{g}$ after $10^5$ years as we reduce the

1913: radius enhancement factor $g$ from 4.3 to 1.}

1914: 

1915: \end{figure}

1916: 

1917: \clearpage

1918: \begin{figure}

1919: \begin{center}

1920: \includegraphics[width=.85\textwidth]{f4.eps}

1921: \end{center}

1922: \caption{Time evolution of the averaged quantities for 

1923: $\Delta_{\rm disk} = 0.3$AU (Runs~1, 1b, 2, and 3).

1924: From top to bottom, the panels show

1925: the number, mass fraction, orbital spacing, and orbital eccentricity of planets 

1926: ($m > 2 \times 10^{26}$ g), and the ratio 

1927: of the effective mass of planetesimals to the mean mass of planets.    

1928: In the third panel, the relations between the orbital spacing 

1929: and orbital instability time for $\bar{e}_{\rm p}/\bar{h}_{\rm p} = 2$ and 4 

1930: \citep{Yos99} are shown by lower and upper dashed lines, respectively.}

1931: 

1932: \end{figure}

1933: 

1934: \clearpage

1935: \begin{figure}

1936: \begin{center}

1937: \includegraphics[width=.85\textwidth]{f5.eps}

1938: \end{center}

1939: \caption{Same as Figure~4 but for the case of 

1940: $\Delta_{\rm disk} = 0.5$AU (Runs~5, 5b, 6, and 7).}

1941: \end{figure}

1942: 

1943: \clearpage

1944: \begin{figure}

1945: \begin{center}

1946: \includegraphics[width=.85\textwidth]{f6.eps}

1947: \end{center}

1948: \caption{Snapshot of all of runs on the $a$--$e$ plane at 200Myr.

1949: The vertical lines represents inner and outer edges of 

1950: the initial planetesimal disk.}

1951: \end{figure}

1952: 

1953: \clearpage

1954: \begin{figure}

1955: \begin{center}

1956: \includegraphics[width=.7\textwidth]{f7.eps}

1957: \end{center}

1958: \caption{Evolution of the normalized time scales $\tilde{T}$ 

1959: (upper panel),  the velocity of planetesimals $\tilde{v}$ 

1960: normalized by the escape velocity of 

1961: planets, and the power-law index, $q$, 

1962: for mass distribution of planetesimals 

1963: (lower panel) as functions of mass fraction of planets $f_{\rm p}$ 

1964: to the total mass.  $\tilde{T}_{\rm grow}$,  $\tilde{T}_{\rm scat}$,

1965:  $\tilde{T}_{\rm damp}$, and $\tilde{T}_{\rm inst}(g)$ represent 

1966: time scales for the growth of planets, evolution of 

1967: the planetesimal velocity due to scattering by planets, 

1968: damping of eccentricities of planets due to the dynamical friction of 

1969: planetesimals, and orbital instability for multiple planet systems,

1970: respectively. $\tilde{T}_{\rm grow}$ is slightly shifted downward (by 0.05) 

1971: to avoid overlapped displays of the time scales.} 

1972: 

1973: 

1974: \end{figure}

1975: 

1976: 

1977: \clearpage

1978: \begin{figure}

1979: \begin{center}

1980: \includegraphics[width=.65\textwidth]{f8.eps}

1981: \end{center}

1982: \caption{The minimum oligarchic mass $m_{\rm p, oli}$ (solid lines)

1983: and the effective mass of planetesimals $m_{\rm eff}$ (dashed lines)

1984: verses the power-law index $q$.

1985: The case for our Run~6

1986: ($m_0 = 3.94 \times 10^{24}$g, $\Sigma = 19.1$ g cm$^{-2}$, and 

1987: $a = 0.89$AU), and the case for parameters used in  

1988: simulations of \citet{Wei97}

1989: ($m_0 = 4.8 \times 10^{18}$ g, $\Sigma = 16.7$ g cm$^{-2}$, and 

1990: $a = 1.0$AU) are shown. The latter parameters are also used in 

1991: \citet{WS93} and \citet{Ina01}.}

1992: 

1993: \end{figure}

1994: 

1995: \clearpage

1996: \begin{figure}

1997: \begin{center}

1998: \includegraphics[width=.65\textwidth]{f9.eps}

1999: \end{center}

2000: \caption{Comparison of the final eccentricities of planets between 

2001: $N$-body simulations (open and filled circles; from Table~1 except for $N = 100$) 

2002: and analytic estimates

2003: (solid and dashed lines; eq.~[\ref{eq:eeq2}]). In analytic estimates, we assume 

2004: the mass of planets $m_{\rm p}$ to be $0.66 M_{\rm E}$ and 

2005: $q = -2$ for remnant planetesimals.

2006: MVEM and VEM stand for the values for the 

2007: current terrestrial planets

2008: with and without Mercury, respectively.

2009: }

2010: 

2011: \end{figure}

2012: 

2013: \clearpage

2014: 

2015: %%\begin{table}

2016: %%\begin{center}

2017: %%\caption{Initial conditions and final states of simulations}

2018: %%\begin{tabular}{|c|c|c|c|c|c|c|}  

2019: %%\tableline

2020: %% Run & $\Delta_{\rm disk}$ & $N$ & 

2021: %% $\alpha$ & $N_{\rm p}$ & $\bar{e}_{\rm p}$ ($10^{-2}$)  & $S_{\rm d}$ ($10^{-3}$)\\ 

2022: %%\tableline

2023: 

2024: 

2025: \begin{deluxetable}{ccccccc}

2026: \tabletypesize{\footnotesize}

2027: \tablewidth{0pt}

2028: \tablecaption{Initial conditions and final states of simulations}

2029: \tablehead{

2030: \colhead{Run} & \colhead{$\Delta_{\rm disk}$(AU)} & \colhead{$N$} & 

2031: \colhead{$\alpha$} & \colhead{$N_{\rm p}$} & 

2032: \colhead{$\bar{e}_{\rm p}$ ($10^{-2}$)}  & 

2033: \colhead{$S_{\rm d}$ ($10^{-3}$)}}

2034: \startdata

2035:  1 &  0.3 &  1000  & -1 & 3 & 3.96  & 1.55   \\ 

2036:  1b&  0.3 &  1000  & -1 & 4 & 3.87  & 1.51   \\ 

2037:  2 &  0.3 &  3000  & -1 & 3 & 3.87  & 1.79   \\ 

2038:  3 &  0.3 &  5000  & -1 & 3 & 3.63  & 0.99   \\ 

2039:  4 &  0.3 &  3000  & -2 & 4 & 3.09  & 0.99   \\ 

2040:  5 &  0.5 &  1000  & -1 & 3 & 2.73  & 0.95   \\ 

2041:  5b&  0.5 &  1000  & -1 & 4 & 7.74  & 3.49   \\ 

2042:  6 &  0.5 &  3000  & -1 & 3 & 2.98  & 0.60   \\ 

2043:  7 &  0.5 &  5000  & -1 & 3 & 2.37  & 0.49   \\ 

2044:  8 &  0.5 &  3000  & -2 & 4 & 6.02  & 2.97   \\  \tableline

2045:  MVEM    &        &        &   & 4 & 4.62 & 1.90    \\   

2046:  VEM     &        &        &   & 3 & 3.41 & 1.50    \\   

2047: \enddata

2048: \tablecomments{Parameters $\Delta_{\rm disk}$, $N$,  and $\alpha$

2049: stand for the width, number of planetesimals, and 

2050: power-law index for the surface density

2051: of initial planetesimal disks, respectively,  

2052: and $N_{\rm p}$, $\bar{e}_{\rm p}$, and $S_{\rm d}$ 

2053: stand for the number, averaged orbital eccentricity (eq.~[\ref{eq:aveep}]),

2054: and angular momentum 

2055: deficit of planets (eq.~[\ref{eq:sd}]) at the end of simulations.

2056: In Runs~1b and 5b we switch integrators and 

2057: reduce $g$ to unity at $5 \times 10^{4}$ yr, and

2058: at $10^{5}$ yr for other runs. 

2059: MVEM and VEM stand for the current terrestrial planets

2060: with and without Mercury, respectively.} 

2061: \end{deluxetable}

2062:  

2063: 

2064: 

2065: 

2066: %% Here we use \plottwo to present two versions of the same figure,

2067: %% one in black and white for print the other in RGB color

2068: %% for online presentation. Note that the caption indicates

2069: %% that a color version of the figure will be available online.

2070: %%

2071: 

2072: 

2073: %% If you are not including electonic art with your submission, you may

2074: %% mark up your captions using the \figcaption command. See the

2075: %% User Guide for details.

2076: %%

2077: %% No more than seven \figcaption commands are allowed per page,

2078: %% so if you have more than seven captions, insert a \clearpage

2079: %% after every seventh one.

2080: 

2081: %% Tables should be submitted one per page, so put a \clearpage before

2082: %% each one.

2083: 

2084: %% Two options are available to the author for producing tables:  the

2085: %% deluxetable environment provided by the AASTeX package or the LaTeX

2086: %% table environment.  Use of deluxetable is preferred.

2087: %%

2088: 

2089: %% Three table samples follow, two marked up in the deluxetable environment,

2090: %% one marked up as a LaTeX table.

2091: 

2092: %% In this first example, note that the \tabletypesize{}

2093: %% command has been used to reduce the font size of the table.

2094: %% We also use the \rotate command to rotate the table to

2095: %% landscape orientation since it is very wide even at the

2096: %% reduced font size.

2097: %%

2098: %% Note also that the \label command needs to be placed

2099: %% inside the \tablecaption.

2100: 

2101: %% This table also includes a table comment indicating that the full

2102: %% version will be available in machine-readable format in the electronic

2103: %% edition.

2104: 

2105: %% Tables may also be prepared as separate files. See the accompanying

2106: %% sample file table.tex for an example of an external table file.

2107: %% To include an external file in your main document, use the \input

2108: %% command. Uncomment the line below to include table.tex in this

2109: %% sample file. (Note that you will need to comment out the \documentclass,

2110: %% \begin{document}, and \end{document} commands from table.tex if you want

2111: %% to include it in this document.)

2112: 

2113: %% \input{table}

2114: 

2115: %% The following command ends your manuscript. LaTeX will ignore any text

2116: %% that appears after it.

2117: 

2118: \end{document}

2119: 

2120: %%

2121: %% End of file `sample.tex'.

2122: