1: %% 2007/2/18
2: %%\documentclass[12pt,preprint]{aastex}
3:
4: %% manuscript produces a one-column, double-spaced document:
5:
6: %%\documentclass[manuscript]{aastex}
7:
8: %% preprint2 produces a double-column, single-spaced document:
9:
10: \documentclass[preprint2]{aastex}
11:
12: %% Sometimes a paper's abstract is too long to fit on the
13: %% title page in preprint2 mode. When that is the case,
14: %% use the longabstract style option.
15:
16: %% \documentclass[preprint2,longabstract]{aastex}
17:
18:
19: \newcommand{\bm}[1]{\mbox{\boldmath $#1$}}
20: \newcommand{\kms}{km s$^{-1}$}
21:
22:
23: %% If you wish, you may supply running head information, although
24: %% this information may be modified by the editorial offices.
25: %% The left head contains a list of authors,
26: %% usually a maximum of three (otherwise use et al.). The right
27: %% head is a modified title of up to roughly 44 characters.
28: %% Running heads will not print in the manuscript style.
29:
30: \shorttitle{Evolution of Star Clusters near the Galactic Center}
31: \shortauthors{Fujii et al.}
32:
33: %% This is the end of the preamble. Indicate the beginning of the
34: %% paper itself with \begin{document}.
35:
36: \begin{document}
37:
38:
39:
40: \title{Evolution of Star Clusters near the Galactic Center: Fully
41: Self-consistent $N$-body Simulations}
42:
43: %% Use \author, \affil, and the \and command to format
44: %% author and affiliation information.
45: %% Note that \email has replaced the old \authoremail command
46: %% from AASTeX v4.0. You can use \email to mark an email address
47: %% anywhere in the paper, not just in the front matter.
48: %% As in the title, use \\ to force line breaks.
49:
50: \author{M. Fujii\altaffilmark{1,3}, M. Iwasawa\altaffilmark{2,3},
51: Y. Funato\altaffilmark{2}, and J. Makino\altaffilmark{3}}
52:
53:
54: \altaffiltext{1}{Department of Astronomy, Graduate School of Science, The
55: University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-0033;
56: fujii@cfca.jp}
57:
58: \altaffiltext{2}{Department of General System Studies, College of Arts and
59: Sciences, The University of Tokyo, 3-8-1 Komaba, Meguro, Tokyo 153-8902;
60: iwasawa@margaux.astron.s.u-tokyo.ac.jp, funato@artcompsci.org}
61:
62: \altaffiltext{3}{Division of Theoretical Astronomy, National
63: Astronomical Observatory of Japan, 2-21-1 Osawa, Mitaka, Tokyo,
64: 181-8588; makino@cfca.jp}
65:
66: \begin{abstract}
67: We have performed fully self-consistent $N$-body simulations of star
68: clusters near the Galactic center (GC). Such simulations have not
69: been performed because it is difficult to perform fast and
70: accurate simulations of such systems using conventional methods.
71: We used the Bridge code, which integrates the parent galaxy using the
72: tree algorithm and the star cluster using the fourth-order Hermite
73: scheme with individual timestep. The interaction between the parent
74: galaxy and the star cluster is calculate with the tree algorithm.
75: Therefore, the Bridge code can handle both the orbital and internal
76: evolutions of star clusters correctly at the same time. We
77: investigated the evolution of star clusters using the Bridge code and
78: compared the results with previous studies. We found that 1) the
79: inspiral timescale of the star clusters is shorter than that obtained
80: with "traditional" simulations, in which the orbital evolution of star
81: clusters is calculated analytically using the dynamical friction
82: formula and 2) the core collapse of the star cluster increases the
83: core density and help the cluster survive. The initial conditions of
84: star clusters is not so severe as previously suggested.
85: \end{abstract}
86:
87: \keywords{galaxy: star clusters --- Galaxy: center, kinematics and
88: dynamics --- methods: numerical --- stellar dynamics}
89:
90:
91: \section{Introduction}
92:
93: A few dozens of very young and massive stars have been found in the
94: central parsec of the Galaxy \citep{K95,Pa01,Pa06}. These stars are a
95: few million years old \citep{Pa01,Ghez03} and lie on a disk
96: \citep{Lu06} or two disks \citep{Pa06}. The disks rotate around the
97: central black hole (BH) and are at large angles with each other. One
98: disk rotates clockwise in projection, the other counterclockwise
99: \citep{Pa06}. These disks are coeval to within 1 Myr.
100: The orbit of the stars on the clockwise rotating disk are circular
101: \citep{Pa06} or eccentric have lower-limit eccentricities of
102: $0.0-0.8$ \citep{Lu06}, while those on the counterclockwise disk have
103: high eccentricities at around 0.8 \citep{Pa06}.
104: In the central
105: parsec, in situ formation of these stars seems difficult because of
106: the strong tidal field of the central BH. To overcome this
107: difficulty, two possibilities have been suggested: (1) in situ star
108: formation in a massive accretion disk, or (2) inspiraling young star
109: clusters.
110:
111: The accretion disk scenario was proposed by \citet{LB03}. A dense
112: gaseous disk is formed from a molecular cloud which somehow fell to
113: the neighborhood of the central BH. If the disk is sufficiently
114: massive it can become gravitationally
115: unstable, resulting in fragmentation and formation of stars.
116: However, this scenario is problematic. Observations have
117: shown that two disks are at large angles with respect to each other
118: and these stars on the disks formed almost at the same time. In this
119: scenario, two disks must have existed simultaneously within 1
120: Myr. Moreover, it is difficult to make stars with eccentric orbit from
121: accretion disks \citep{Nay07}.
122:
123: The star cluster inspiral scenario was proposed by \citet{Gerhard01}. A
124: star cluster was formed at a distance of tens of parsecs from the GC and
125: spiraled into the GC due to the dynamical friction.
126: This scenario is supported by the observational fact that two young
127: dense star clusters, the Arches and Quintuplet clusters are observed at
128: the distances of $\sim$ 30 pc from the GC\citep{Fi04}. Although this
129: scenario can explain two stellar disks without difficulty, numerical
130: simulations have shown that it would take too long time for the star
131: cluster to inspiral to the central parsec unless it was very massive or its
132: initial position is very near from the GC (Portegies
133: Zwart et al. 2003, hereafter PZ03; G\"{u}rkan \& Rasio 2005).
134:
135:
136: In these works (PZ03; G\"{u}rkan \& Rasio 2005), the orbit of the
137: cluster within its parent galaxy was calculated using the dynamical
138: friction formula \citep{Ch43}. With this approach, the inspiral
139: timescale might have been overestimated. In \citet{Fj06}, we
140: performed fully self-consistent $N$-body simulations of a satellite
141: galaxy within its parent galaxy and found that the orbital decay of
142: the satellite is much faster than those calculated analytically from
143: the dynamical friction formula. This difference was caused by
144: particles escaped from the satellite. One mechanism is that
145: the direct gravitational forces from escaped particles worked as
146: effective drag force to the satellite. The second mechanism is that
147: escaped particles remain close to the body of the satellite and
148: enhance the dynamical friction. These effect should also work in the
149: case of star clusters. Therefore, a fully self-consistent $N$-body
150: simulation is necessary to obtain correct results for the orbital
151: evolution of star clusters.
152:
153:
154: Kim \& Morris (2003; hereafter KM03) performed
155: self-consistent $N$-body simulations of star clusters near the GC.
156: Their results showed that if the initial central density of a star
157: cluster was initially very high ($\sim 10^8 M_{\sun}$pc$^{-3}$), the
158: cluster can deliver stars to the central parsec of the Galaxy.
159: In these simulations, however,
160: the internal evolution of the star clusters was neglected. The
161: stars in their model of star clusters have an equal mass and a large
162: softening length of 0.025pc. Such star clusters experience neither mass
163: segregation nor core collapse. However, if the core collapse occurs,
164: the core density of a star cluster increases. The initial high density
165: of the core will not be necessary. Thus, it is also important
166: to solve the internal evolution correctly.
167:
168:
169: Such a fully self-consistent $N$-body simulation has been impossible
170: with conventional numerical methods. While star clusters need a very
171: accurate scheme such as the combination of fourth-order Hermite scheme
172: and direct force calculation, galaxies contain too many particles to
173: use the direct force calculation. To solve this problem, we have
174: developed a new tree-direct hybrid scheme, the ``Bridge'' scheme
175: \citep{Fj07}. The Bridge scheme enables us to perform fully
176: self-consistent $N$-body simulations of star clusters within their
177: parent galaxies in a realistic time (less than 2 days with a single
178: GRAPE-6 board).
179:
180: We performed fully self-consistent $N$-body simulations of evolution of
181: star clusters within their parent galaxies using the Bridge scheme. We
182: also performed the ``traditional'' $N$-body simulations, in which the
183: orbital decay of the star cluster is calculated using the dynamical
184: friction formula for comparison. We found that the inspiral timescale of
185: the star cluster is shorter than that obtained in previous studies. In
186: addition, if the initial orbit of the star cluster is eccentric, the
187: timescale of inspiral is much shorter than that for a cluster in the
188: circular orbit of the same apocenter distance.
189:
190: The eccentricities of the stars
191: escaped from the star cluster distribute around the eccentricity of the
192: star cluster. Thus, if a star cluster is initially in an eccentric
193: orbit, it naturally explains the rather high eccentricities of the
194: stars in the observed "disks". Also, because of the mass segregation
195: effect, very massive stars (more than $10M_{\odot}$) remained in the
196: cluster and were brought very close to GC. Thus, we conclude that the
197: timescale problem with the cluster inspiral scenario was partly because of
198: the wrong treatment of the dynamical friction and partly because the
199: limited assumption of the circular orbit, and it is not difficult to
200: make the central stellar disks from inspiraled star clusters.
201:
202:
203: We describe the simulation method and initial conditions in section
204: 2. In section 3 we show the results of simulations.
205: Section 4 is for summary and discussions.
206:
207:
208:
209:
210: \section{Numerical Simulation}
211:
212: \subsection{Models}
213:
214: We adopted a King model with $W_0=3$ for the model of a star cluster.
215: Its core radius, $r_{\rm c}$, half-mass radius, $r_{\rm h}$, and tidal
216: radius, $r_{\rm t}$, are 0.087 pc, 0.13 pc, and 0.47 pc, respectively.
217: It consists of 65536 stars and we assigned each star a mass randomly
218: drawn from a \citet{Sal55} initial mass function between 0.3 and 100
219: $M_{\sun}$, irrespective of position. The total mass of the star
220: cluster, $M_{\rm SC}$, is $7.9\times 10^{4}{M_{\sun}}$. This model
221: imitates the Arches cluster \citep{Nagata95}, whose mass and velocity
222: dispersion are $7\times 10^4{M_{\sun}}$ within 0.23 pc and 22 \kms,
223: respectively \citep{Fi02}. The Arches cluster is located at $\sim 30$
224: pc from the GC. We, however, placed our cluster much closer to the
225: GC, to compare our result with those of PZ03. In table
226: \ref{tb:models}, we summarize the model parameters.
227:
228: We used two galaxy models as a model of the central region of
229: the Galaxy; one includes the central SMBH
230: (galaxy 1) and the other does not (galaxy 2).
231: For the galaxy model 1, we adopted a King model with non-dimensional
232: central potential of $W_0 = 10$. We scaled the
233: density and velocity dispersion of our model at 5 pc from the GC to
234: the those of the Galaxy. The density at 5 pc is $6.8\times 10^3
235: M_{\sun}{\rm pc}^{-3}$ and the one-dimensional velocity dispersion is
236: 64 km s$^{-1}$ in our scales, while the observed density at 5 pc is
237: $\rho (5{\rm pc}) = 6.9 \times 10^3 M_{\sun}{\rm pc}^{-3}$
238: \citep{Genzel03} and the observed velocity dispersion is 54 \kms at
239: $\sim$ 4 pc \citep{Genzel00}. Figure \ref{fig:model}
240: shows the enclosed masses of our model galaxy and the result of
241: \citet{Genzel03}. In this model, the total mass of the galaxy, $M_{\rm
242: G}$, is $8\times 10^7 {\rm M_{\sun}}$, and the core radius, $r_{\rm c,
243: G}$, and the half-mass radius, $r_{\rm h, G}$, are 0.66 pc and 21 pc,
244: respectively. We used $2 \times 10^6$ particles to model the
245: Galaxy. The mass of a particle of the galaxy is 40 $M_{\sun}$. Since our
246: galaxy model has the finite core size of 0.66pc, the orbital evolution
247: of the star cluster should be reasonably accurate as far as its distance
248: from the GC is more than 1pc.
249:
250: The galaxy model 2 is a more realistic model of than model 1.
251: It includes a central super-massive black hole (SMBH). This model is
252: based on King model. We put a SMBH at the center of the galaxy with
253: King model $W_0 = 10$ and integrated it for around two crossing times.
254: The central region of the galaxy evolved and its density profile became
255: cuspy. This model roughly represents the Galactic center between
256: 0.1 and 5 pc. The enclosed mass is shown in figure \ref{fig:model}.
257: The total mass and the particle mass of the galaxy are
258: $2.9\times 10^7 {\rm M_{\sun}}$ and 14 $M_{\sun}$, respectively.
259:
260: The initial position and velocity of the star cluster are shown in table
261: \ref{tb:initial}. We calculated two orbits; one is circular (model C) and
262: the other is eccentric (model E) for the galaxy model 1, and two
263: eccentric orbits (model B1, B2) for the galaxy model 2.
264: In Model B1, the star cluster has almost the same eccentricity as the
265: previous simulation without SMBH. In Model B2, the eccentricity of the
266: star cluster is lower than Model B1.
267: The orbital elements for the eccentric case were chosen so that the
268: star cluster would survive at least for several orbits. At first, we
269: tried the eccentric orbit with the same orbital energy as that of the
270: circular case. However, in this case, the disruption of the cluster
271: was too fast, unless the orbit is close to circular. Therefore we made
272: the eccentric orbit significantly wider.
273:
274: \begin{figure}
275: \epsscale{0.8}
276: \plotone{f1.eps}
277: \caption{Enclosed mass of our model galaxy and the GC \citep{Genzel03}.}
278: \label{fig:model}
279: \end{figure}
280:
281: \onecolumn
282:
283: \begin{table}[htbp]
284: \begin{center}
285: \caption{Models for the galaxy and the star cluster\label{tb:models}}
286: \begin{tabular}{ccccccccccc}
287: \tableline
288: \tableline
289: & King $W_0$& $N$ & $M({\rm M_{\odot}})$ & $M_{\rm BH}({\rm M_{\odot}})$ & $r_{\rm c}$ (pc) &
290: $r_{\rm h}$ (pc) & $r_{\rm t}$ (pc)\\
291: \tableline
292: Star cluster & 3 & 65536 & $7.9 \times 10^4$ & - & 0.087 & 0.13 & 0.47 \\
293: Galaxy 1& 10 & $2 \times 10^6 $ & $8.0 \times 10^7$ & - & 0.66 & 21 & 120 \\
294: Galaxy 2& 10 & $2 \times 10^6 $ & $2.9 \times 10^7$ & $3.6 \times 10^6$ & - & 9.6 & 72 \\
295: \tableline
296: \end{tabular}
297: \end{center}
298: \end{table}
299:
300: \begin{table}[htbp]
301: \begin{center}
302: \caption{Initial Conditions for the cluster orbit\label{tb:initial}}
303: \begin{tabular}{ccccc}
304: \tableline
305: \tableline
306: Simulation & Galaxy model & Orbit & Initial position (pc) & Initial velocity (km/s)\\ \tableline
307: Model C & 1 & Circular & 2 & 130\\
308: Model E & 1 & Eccentric & 5 & 72 \\
309: Model B1 & 2 & Eccentric & 5 & 57\\
310: Model B2 & 2 & Eccentric & 5 & 67 \\
311: \tableline
312: \end{tabular}
313: \end{center}
314: \end{table}
315:
316: \twocolumn
317:
318:
319:
320: \subsection{Fully Self-consistent $N$-body Simulation}
321:
322: We performed fully self-consistent $N$-body simulations using the
323: Bridge code \citep{Fj07}. It is a direct-tree hybrid
324: code. Only the internal motion of the star cluster is calculated by
325: the direct scheme with high accuracy, and all other interactions are
326: calculated by the tree algorithm. The splitting between the direct
327: part and tree part is through the splitting of the Hamiltonian in the
328: way similar to MVS \citep{WH91,KYN91} or RESPA \citep{Tu90}.
329: Thus, the low-accuracy calculation of galaxy particles and its
330: interaction with cluster particles are integrated with time-symmetric
331: leapfrog algorithm, resulting in small long-term error. Hence, we can
332: treat a large-$N$ system with embedded small-scale systems fully
333: self-consistently and accurately.
334:
335: The numerical parameters used for the time integration are summarized
336: in table \ref{tb:param}. For the tree part, the Bridge code needs the
337: same parameters with the Barnes-Hut treecode modified for the use with
338: GRAPE hardware \citep{BH86,M04}. We used the opening angle $\theta =
339: 0.75$ with the center-of-mass (dipole-accurate) approximation. The
340: maximum group size for a GRAPE calculation \citep{M91} is 8192. The
341: stepsize of leapfrog integrator is $\Delta t = 1/512$ (in Heggie
342: unit) $= 2.9 \times 10^{-4}$ (Myr) for galaxy 1 and $\Delta t = 1/1024$
343: (in Heggie unit) $= 1.2 \times 10^{-4}$ (Myr) for galaxy 2. The potential is
344: softened using Plummer softening. The softening length between galaxy
345: particles that between cluster particles and galaxy particles are the
346: same. Both are $\epsilon _{\rm G}=3.9\times 10^{-2}$ pc.
347: For galaxy 2, we adopted the softening length between between the SMBH
348: and galaxy particles, $\epsilon _{\rm G-BH}$, is 0.12 pc and between the
349: SMBH and star cluster particles, $\epsilon _{\rm SC-BH}$, is 0.012 pc.
350:
351:
352: For the direct part, we used the fourth-order Hermite integrator with
353: block timestep, and the timestep criterion is of the standard Aarseth
354: type \citep{MA92} with $\eta = 0.01$. We also used the Plummer softening
355: for the gravitational force between cluster particles, and we did not
356: model physical collisions or binary formation in the calculation
357: reported in this paper. The softening length between star cluster
358: particles, $\epsilon _{\rm SC}$, is $1.0\times 10^{-5}$ pc.
359:
360: We stopped the simulations at $T=0.75-0.8$ (Myr). Since
361: we used the softening and did not model the physical collision and
362: merging of the stars, the structure of the star cluster after the core
363: collapse might not expressed correctly. We ignored the stellar
364: evolution because we treat only very short time ($<$ 1 Myr) in our
365: simulations.
366:
367:
368: We used GRAPE-6 \citep{M03} for force calculation.
369: The total energy was conserved better than $5\times 10^{-5}$ for
370: the circular orbit $8\times 10^{-5}$ for the eccentric orbit throughout
371: the simulations.
372:
373: \begin{table}[htbp]
374: \begin{center}
375: \caption{Parameters for $N$-body Simulation\label{tb:param}}
376: \begin{tabular}{lcc}
377: \tableline \tableline
378: Parameters & Value \\ \tableline
379: $\epsilon _{\rm G}$ & $3.9 \times 10^{-2}$ (pc)\\
380: $\epsilon _{\rm SC}$ & $1.0 \times 10^{-5}$ (pc)\\
381: $\epsilon _{\rm G-BH}$ & $1.2 \times 10^{-1}$ (pc)\\
382: $\epsilon _{\rm SC-BH}$ & $1.2 \times 10^{-2}$ (pc)\\
383: \tableline
384: $\Delta t$ (for galaxy 1)& $2.9 \times 10^{-4}$ (Myr)\\
385: $\Delta t$ (for galaxy 2)& $1.2 \times 10^{-4}$ (Myr)\\ \tableline
386: $\theta$ & 0.75\\
387: $n_{\rm crit}$ & 8192\\ \tableline
388: \end{tabular}
389: \end{center}
390: \end{table}
391:
392:
393: \subsection{$N$-body Simulation with Artificial Dynamical Friction}
394:
395: To compare our result with those of previous works, for model C and E,
396: we also performed $N$-body
397: simulations in which the Galaxy is modeled as a fixed potential and
398: the dynamical friction due to the Galaxy is calculated analytically.
399: This treatment is the same as in PZ03, where the acceleration due to
400: the dynamical friction is calculated using Chandrasekhar's dynamical
401: friction formula \citep{Ch43,BT87,MP03}
402: \begin{eqnarray}
403: \bm{a}_{\rm df} = -4\pi \ln \Lambda \chi G^2 \rho _{\rm G} M_{\rm SC}
404: \frac{\bm{v}_{\rm SC}}{v_{\rm SC}^3}\label{eq:df}.
405: \end{eqnarray}
406: Here $M_{\rm SC}$ and $\bm{v}_{\rm SC}$ are the mass and the
407: center-of-mass velocity of the star cluster, $\rho _{\rm G}$ is the
408: local density of the Galaxy, $\ln \Lambda$ is the Coulomb logarithm,
409: and
410: \begin{eqnarray}
411: \chi \equiv {\rm erf} (X) - \frac{2X}{\sqrt{\mathstrut \pi}} \exp(-X^2),
412: \end{eqnarray}
413: where $X={v_{\rm SC}}/{\sqrt{2} \sigma _{\rm G}}$ and $\sigma _{\rm
414: G}$ is the local one-dimensional velocity dispersion of the Galaxy,
415: assumed to be isotropic and locally Maxwellian. For $M_{\rm SC}$, we
416: adopted the bound mass, which is the total mass of the particles bound
417: to the star cluster. We gave all bound stars the acceleration due
418: to the dynamical friction from the above formula. Unbound stars were
419: not affected by the dynamical friction.
420:
421: In equation (\ref{eq:df}), the Coulomb logarithm, $\ln \Lambda$, is given by
422: \begin{eqnarray}
423: \ln \Lambda = \ln \left(\frac{b_{\rm max}}{b_{\rm min}}\right),
424: \end{eqnarray}
425: where $b_{\rm max}$ and $b_{\rm min}$ are the maximum and minimum impact
426: parameters.
427: It is often set as a constant given by
428: \begin{eqnarray}
429: \ln \Lambda \sim \ln \left(\frac{R_{\rm SC}}{\langle r_{\rm SC}\rangle}\right),
430: \end{eqnarray}
431: where $R_{\rm SC}$ is the distance of the star cluster from the GC,
432: $\langle r_{\rm SC}\rangle$ is the characteristic radius of the star
433: cluster (roughly the half-mass radius).
434: PZ03 adopted a constant value, $\chi \ln \Lambda = 1$.
435: However, \citet{H03} found a constant $\Lambda$
436: overestimates dynamical friction at pericenter and proposed a variable
437: $\Lambda$:
438: \begin{eqnarray}
439: \ln \Lambda = \ln \left(\frac{R_{\rm SC}}{1.4 \epsilon_{\rm SC}}\right),
440: \end{eqnarray}
441: where $\epsilon _{\rm SC}$ is the size of the star cluster.
442: We performed $N$-body simulations with artificial dynamical
443: friction in two ways. One is the
444: same as that used PZ03 (constant $\Lambda$), and the other is the way
445: proposed by \citet{H03} (variable $\Lambda$).
446: We adopted the virial radius of the star cluster for $\epsilon _{\rm SC}$.
447:
448:
449: \section{Simulation Results}
450:
451: \subsection{Circular Orbits}
452:
453: The top panel of figure \ref{fig:snap} shows the snapshots of the star
454: clusters projected onto the x-y plane. The top and bottom panels
455: show model C (circular orbit) and E (eccentric orbit), respectively.
456: In model C, the star cluster is initially located at 2 pc from the GC.
457: Due to the tidal field of the Galaxy, the star cluster
458: becomes elongated. Particles stripped from the star cluster form tidal
459: arms and make ring-like structures. Finally they form a disk-like structure.
460:
461: We investigated the orbital and internal evolutions of the star
462: cluster. Figure \ref{fig:radius_cir} shows the distance of the star
463: cluster from the GC obtained by our $N$-body simulations. The solid
464: curve shows the result of the full $N$-body simulation. The dashed and
465: dotted curves show the result of the ``traditional'' simulations in
466: which the dynamical friction is calculated analytically from the
467: formula with constant $\Lambda$ and variable $\Lambda$, respectively.
468: The orbital decay in the full $N$-body simulation is faster by 30-40\%
469: than that in other simulations. On the other hand, the evolution of
470: the bound mass of the star cluster shown in figure \ref{fig:mass_cir}
471: is almost the same among the three. This result suggests that previous
472: studies underestimated the inspiral timescale of star clusters. This
473: effect is not very large, but not negligible.
474:
475: In figure \ref{fig:mass_cir}, it seems that the mass loss in the very
476: late stage (after more than 80\% of the initial total mass is lost)
477: seems to be significantly slower for the case of full $N$-body
478: simulation than those for other two simulations, even though the cluster
479: is much closer to GC. This difference is probably due to the finite core
480: size of our galaxy model ($r_c = 0.66{\rm pc}$) and not the reality. In
481: order to study the evolution after this stage, we need to use a more
482: realistic model of the mass distribution of the central parsec of the
483: galaxy, which includes the central massive black hole. In addition,
484: stellar collision and merger within the star cluster must be modeled.
485: We are currently working on such an extension of the Bridge code.
486:
487: Figures \ref{fig:core_cir} and \ref{fig:density_cir} show the core
488: radius and the core density of the star cluster. The core collapse
489: occurred at around $T\simeq 0.53$ Myr. The core collapse times are
490: also the same among the three simulations.
491:
492: \onecolumn
493: \begin{figure}
494: \epsscale{0.9}
495: \plotone{f2.eps}
496: \caption{Snapshots of the star clusters projected onto $x-y$
497: plane. The upper six panels are for the run with the circular initial
498: orbit (model C), and the lower six panels are for the run with the eccentric
499: orbit (model E). Times are 0.122, 0.239, 0.356, 0.473, 0.591, and 0.708 Myrs.}
500: \label{fig:snap}
501: \end{figure}
502: \twocolumn
503:
504: \begin{figure}
505: \epsscale{.90}
506: \plotone{f3.eps}
507: \caption{The distance of the star cluster from the GC plotted as a
508: function of time for model C.
509: Solid curve shows the result of the full $N$-body
510: simulation. Dashed and dotted curves show the results of the
511: ``traditional'' simulations with variable $\Lambda$ and constant $\Lambda$,
512: respectively.
513: \label{fig:radius_cir}}
514: \end{figure}
515:
516:
517: \begin{figure}
518: \epsscale{.90}
519: \plotone{f4.eps}
520: \caption{The bound mass of the star cluster plotted as a function of
521: time for model C. Curves have the same meanings as in
522: figure \ref{fig:radius_cir}.\label{fig:mass_cir}}
523: \end{figure}
524:
525:
526: \begin{figure}
527: \epsscale{.90}
528: \plotone{f5.eps}
529: \caption{The core radius of the star cluster, $r_{\rm c}$, plotted as a
530: function of time for model C. Curves have the same
531: meanings as in figure \ref{fig:radius_cir}.\label{fig:core_cir}}
532: \end{figure}
533:
534:
535: \begin{figure}
536: \epsscale{.90}
537: \plotone{f6.eps}
538: \caption{The core density of the star cluster, $\rho _{\rm c}$,
539: plotted as a function of time for model C. Curves have the
540: same meanings as in figure \ref{fig:radius_cir}.\label{fig:density_cir}}
541: \end{figure}
542:
543:
544: \subsection{Eccentric Orbits}
545:
546: The bottom panel of figure \ref{fig:snap} shows the snapshots of
547: model E projected onto the x-y plane. The initial distance of
548: the star cluster from the GC is 5 pc. The star cluster is elongated
549: and particles are stripped due to the tidal force of the Galaxy. The
550: stripped particles form complex tidal tails.
551:
552: Figure \ref{fig:radius_ecc} shows the orbital evolution of the star
553: cluster. The orbital decay of
554: the full $N$-body simulation is faster than the traditional
555: simulations, as was the case in the runs from the circular orbit. In
556: this case also, the evolution of the bound mass is the same among the
557: three simulations (see figure \ref{fig:mass_ecc}).
558: Moreover, this
559: result shows that variable $\Lambda$ works better than constant
560: $\Lambda$.
561:
562: Figures \ref{fig:core_ecc} and \ref{fig:density_ecc} show the core radius
563: and the core density of the star cluster.
564: These show that the core collapse occurred at around $T\simeq 0.55$
565: Myr. The core collapse time is almost the same as that for the case of
566: the circular orbit, even though the bound mass at the collapse time is
567: different by almost a factor of two. The half-mass relaxation time at
568: the time of the collapse was 0.11 Myr and 0.34 Myr, in
569: model C and E, respectively. Thus, if clusters are rapidly
570: losing its mass due to the tidal field, the apparent age measured by
571: the present relaxation time can show large variations, even if they
572: started from the same initial condition and collapsed at the same time.
573: Furthermore, the core collapse times are the same as that in the
574: case without massloss due to the tidal field. The core collapse
575: time, $t_{\rm cc}$, is estimated as $t_{\rm cc} \simeq 0.20 t_{\rm rh}$,
576: where $t_{\rm rh}$ is half-mass relaxation time \citep{PM02}. From this
577: equation, we obtain the core collapse time of our model star cluster as
578: 0.51 Myr. Our results agreed with it.
579:
580: \begin{figure}
581: \epsscale{.90}
582: \plotone{f7.eps}
583: \caption{The distance of the star cluster from the GC plotted as a
584: function of time for model E.
585: Solid curve shows the result of the full $N$-body
586: simulation. Curves have the same meanings as in
587: figure \ref{fig:radius_cir}.\label{fig:radius_ecc}}
588: \end{figure}
589:
590:
591: \begin{figure}
592: \epsscale{.90}
593: \plotone{f8.eps}
594: \caption{The bound mass of the star cluster plotted as a function of
595: time for model E. Curves have the same meanings as in
596: figure \ref{fig:radius_ecc}.\label{fig:mass_ecc}}
597: \end{figure}
598:
599:
600: \begin{figure}
601: \epsscale{.90}
602: \plotone{f9.eps}
603: \caption{The core radius of the star cluster, $r_{\rm c}$, plotted as a
604: function of time for model E. Curves have the same
605: meanings as in figure \ref{fig:radius_ecc}.\label{fig:core_ecc}}
606: \end{figure}
607:
608:
609: \begin{figure}
610: \epsscale{.90}
611: \plotone{f10.eps}
612: \caption{The core density of the star cluster, $\rho _{\rm c}$,
613: plotted as a function of time for model E. Curves have the same
614: meanings as in figure
615: \ref{fig:radius_ecc}.\label{fig:density_ecc}}
616: \end{figure}
617:
618:
619: Figure \ref{fig:radius_BH} - \ref{fig:density_BH} shows the results of
620: model B1 (dashed curves) and B2 (dotted curves), where the Galaxy model
621: has a central BH. Solid curves show the result of model E (no BH; the same
622: as the solid curve in figure \ref{fig:radius_ecc} - \ref{fig:density_ecc}).
623: The orbital evolution of the star cluster in model B1 and B2 is essentially
624: similar to that in model E. However, the orbital decay in model
625: B1 is somewhat slower than that in model E. As is clear
626: in figure \ref{fig:mass_BH}, in model B1, the mass loss at the
627: pericenter passage is much larger than in the case of model E.
628: Therefore, the core collapse did not occur and the core density of the
629: cluster did not increase.
630: This is because the star cluster suffers the strong tidal
631: force from the SMBH at the pericenter. The small mass of the star
632: cluster slows the orbital evolution of the star cluster. Furthermore,
633: the strong tidal field disrupts the star cluster much faster than the
634: case without SMBH.
635:
636: In model B2, the pericenter of the star cluster is farther than that in
637: model B1. The evolution of the bound mass is the same as the case
638: without SMBH. However, the orbital evolution is much slower than that
639: of model B1, because the pericenter of the star cluster is farther.
640:
641:
642: The cluster model and its orbital evolutions in model E and B1 are
643: almost the same as simulation 8 of KM03.
644: Their model of the Galaxy has a power-law density profile with the
645: central SMBH, while our galaxy model 1 for model E has no SMBH. However,
646: the enclosed mass of the galaxy in KM03 is very similar to ours between
647: around 1 and 10pc. In their simulation, the star cluster was totally
648: disrupted before 1 Myr, but in our simulation, the star cluster has
649: 30\% of its initial mass at the end of our simulation (0.75 Myr). The
650: difference is caused by the internal evolution of the star cluster.
651: In our simulation, mass segregation and core collapse of the cluster
652: made the central density much higher, which prevent the complete
653: disruption of the cluster. On the other hand, in KM03, such an
654: evolution was prevented by the numerical method they used.
655:
656: On the other hand, the cluster in model B1 was disrupted before the
657: core collapse occurs by the strong tidal field of the central SMBH.
658: At around 1 pc, the enclosed mass of galaxy model 2 which includes a SMBH
659: is twice as large as that of K04's model. Such a difference of the
660: enclosed masses is caused by the mass of the SMBH. We adopted
661: $3.6\times10^6 M_{\sun}$ \citep{E05}, while KM03 did $2.5\times10^6 M_{\sun}$.
662:
663: Thus, the presence of a SMBH has considerable effect on the evolution of
664: star clusters. The main effect is that the tidal field near the GC
665: becomes much stronger, resulting in faster mass loss from star
666: clusters and preventing its core collapse.
667: In the models we tested, the core collapse time is about the same as the
668: time of the complete disruption in model B1. If the core collapse time is
669: somewhat faster, it is possible that the cluster survives and approaches
670: to the GC. In particular, if an IMBH is formed within the star cluster, it
671: would significantly help the survival of the cluster. We will study this
672: aspect in more details in the forthcoming papers.
673:
674:
675: \begin{figure}
676: \epsscale{.90}
677: \plotone{f11.eps}
678: \caption{The distance of the star cluster from the GC plotted as a
679: function of time. Solid curve shows the result of the galaxy model
680: without the central SMBH (Model E; the same as the solid curve in figure
681: \ref{fig:radius_ecc}). Dashed and dotted curves show the result of the
682: galaxy model with the central SMBH (Model B1 and B2).\label{fig:radius_BH}}
683: \end{figure}
684:
685:
686: \begin{figure}
687: \epsscale{.90}
688: \plotone{f12.eps}
689: \caption{The bound mass of the star cluster plotted as a function of
690: time for the galaxy model with the SMBH. Curves have the same meanings as in
691: figure \ref{fig:radius_BH}.\label{fig:mass_BH}}
692: \end{figure}
693:
694:
695: \begin{figure}
696: \epsscale{.90}
697: \plotone{f13.eps}
698: \caption{The core radius of the star cluster, $r_{\rm c}$, plotted as a
699: function of time for the galaxy model with the SMBH. Curves have the same
700: meanings as in figure \ref{fig:radius_BH}.\label{fig:core_BH}}
701: \end{figure}
702:
703:
704: \begin{figure}
705: \epsscale{.90}
706: \plotone{f14.eps}
707: \caption{The core density of the star cluster, $\rho _{\rm c}$, plotted
708: as a function of time for the galaxy model with the SMBH. Curves have the
709: same meanings as in figure \ref{fig:radius_BH}.\label{fig:density_BH}}
710: \end{figure}
711:
712:
713:
714: \subsection{Eccentricities and Inclinations of the Escaped Stars}
715:
716: For some of the stars in the central parsec, the projected positions
717: and the proper motions have been measured, and their orbital elements
718: have been estimated \citep{Pa06,Lu06}. Young and bright stars
719: apparently belong to one of the two ``disks'' (clockwise and
720: counter-clockwise rotating disks), though the existence of the
721: counter-clockwise disk is controversial \citep{Lu06}. The
722: eccentricities of the stars on the counter-clockwise rotating disk are
723: high, $\sim 0.8$ \citep{Pa06}. For the stars on the clockwise rotating
724: disk, \citet{Pa06} concluded their orbits are circular, while
725: \citet{Lu06} concluded the lower limits of their eccentricities
726: distribute between 0.0 and 0.8. These high eccentricities are
727: difficult to explain with the in-situ formation scenario, and also have
728: been thought to be difficult to explain with cluster inspiral
729: scenario, since in both cases the stars would have close-to-circular
730: orbits.
731:
732: We investigated the eccentricities, $e$, and inclinations, $i$, of
733: stars escaped from the star cluster (i.e. unbound stars) for model C and
734: E. Figures \ref{fig:ei_cir} and \ref{fig:ei_ecc} show the
735: eccentricities and inclinations of escaped stars and their evolutions
736: for model C and E, respectively. In these figures, the $x$-axis is the
737: semi-major axis, $a$, of the star. The eccentricity is defined as
738: $e = (r_{\rm a}-r_{\rm p})/(r_{\rm a}+r_{\rm p})$, where $r_{\rm a}$ and
739: $r_{\rm P}$ are the apocenter and pericenter distances, respectively. We
740: obtained the peri- and apocenter distances by integrating the orbits of
741: the stars
742: in the model potential. The inclination is defined as
743: $i=\cos ^{-1} (h_{z}/h)$, where $h$ and $h_{z}$ are the angular momentum
744: and its $z$ component, respectively. The top panels show the
745: eccentricities and inclinations of the stars escaped before $T=0.15$
746: Myr. The central gap corresponds to the semi-major axis of the star
747: cluster, and the left and right wings are the stars on
748: the leading and trailing arms, respectively. The distributions of the
749: escaped stars on the $a-e$ and $a-i$ plains expand because of the
750: time-varying gravitational force from
751: the star cluster (see the middle panels). The bottom panels show the
752: distribution of all escaped stars at the last snap shots, $T=0.75$
753: Myr. Even in the case of the circular orbit (model C), the
754: eccentricities of the escapers in innermost orbits can reach rather
755: large values. However, in the case of the eccentric orbit (model E), the
756: eccentricities of the escapers are even higher. They distribute between
757: 0.4 and 0.8, being consistent with the observed values of stars in the
758: two ``disks''. Thus, if the cluster is initially in a highly eccentric
759: orbit, the high eccentricities of observed stars are naturally explained.
760:
761: \onecolumn
762: \begin{figure}
763: \epsscale{1.0}
764: \plotone{f15.eps}
765: \caption{The eccentricity (left) and inclination (right) of the stars
766: escaped from the star cluster as the function of their the
767: semi-major axis $a$ for model C (circular orbit).
768: The top panels show the unbound stars at $T=0.15$ Myr and the
769: middle panels show the position of the same stars shown in top
770: panels, but at $T=0.75$ Myr. The bottom panels show the all unbound
771: stars at $T=0.75$ Myr.
772: \label{fig:ei_cir}}
773: \end{figure}
774:
775: \begin{figure}
776: \epsscale{1.0}
777: \plotone{f16.eps}
778: \caption{Same as figure \ref{fig:ei_cir}, but for model E (eccentric
779: orbit). \label{fig:ei_ecc}}
780: \end{figure}
781: \twocolumn
782:
783: \subsection{The Evolution of the Mass Function}
784:
785: Figure \ref{fig:MF1} shows the evolution of the mass function (MF) of
786: the stars bound to the star cluster for model C.
787: We can see that the slope of the mass function for $m> 10M_{\odot}$
788: becomes flatter as the system evolves, while that for $m< 10M_{\odot}$
789: shows rather little change. To see this effect more clearly, in figure
790: \ref{fig:MF} we plot the fraction of the mass retained in the cluster
791: as the function of the stellar mass for model C.
792: Stars with mass $m> 30M_{\odot}$
793: are almost perfectly retained in the cluster, and the retention rate
794: quickly drops in the range of $10M_{\odot}<m< 30M_{\odot}$. For
795: $m< 10M_{\odot}$, retention rate becomes smaller for smaller mass,
796: but the dependence becomes much weaker.
797:
798: Figure \ref{fig:rm} shows the evolution of the enclosed mass in $r$,
799: the distance from the center of the star cluster for model C.
800: The enclosed masses are calculated for five mass ranges.
801: Initially, all mass ranges have same profile. At $T=0.15$ Myr, the most
802: massive stars have sank to the center and the second massive stars also
803: shows some central condensation. At $T=0.45$ Myr, a large fraction of
804: stars more massive than 10 $M_{\sun}$ have sank to the center (within
805: radius 0.02 pc). However, the distribution of stars with mass less than
806: $10 M_{\sun}$ have not changed significantly. Thus, the stars heavier
807: than 10 $M_{\sun}$ remained in the star cluster.
808:
809: We can estimate the critical mass of the star at which this change in
810: the behavior occurs, by calculating the mass
811: of the star at which the dynamical friction just balances the two-body
812: heating. First, we consider the energy change of the stars in the
813: system consisting of two components with the masses $m_1$ and $m_2$.
814: Assuming that the velocity dispersions of both components are
815: Maxwellian, the mean energy change of a star of mass $m_1$ is
816: expressed as
817: \begin{eqnarray}
818: \left< \frac{d}{dt}(m_1E_1) \right> = \frac{4\sqrt{3\pi}G^2m_1m_2n_2\log
819: \Lambda}{(\left< E_1\right> +\left< E_2\right>)^{2/3}}
820: (m_2\left< E_2\right> - m_1\left< E_1 \right>),\label{eq:energy1}
821: \end{eqnarray}
822: where $\left< E_1\right>$ and $\left< E_2\right>$ are mean specific
823: kinetic energy of each components, $n_2$ is the number density of the
824: stars of mass $m_2$, and $\ln \Lambda$ is the Coulomb logarithm
825: \citep{HH03}. At least in the initial model, the velocity dispersion
826: of the stars is independent of the mass. So we can set
827: $\left< E_1\right> = \left< E_2\right> = 1/2 \sigma ^2$.
828: In this case, equation (\ref{eq:energy1}) is rewritten as
829: \begin{eqnarray}
830: \left< \frac{d}{dt}(m_1E_1) \right> = An_2m_1m_2(m_2-m_1),\label{eq:energy2}
831: \end{eqnarray}
832: where $A$ is a constant. Now we consider the case of continuous mass
833: distribution, in which the mass distribution of $m_2$ is given by
834: \begin{eqnarray}
835: \frac{dn}{dm} \propto m^{-\alpha},
836: \end{eqnarray}
837: where $C$ is a constant.
838: The number density of mass $m_2$, $n_2$, is expressed as
839: \begin{eqnarray}
840: n_2 = C m_2^{-\alpha}.\label{eq:MF}
841: \end{eqnarray}
842: By substituting equation (\ref{eq:MF}) to (\ref{eq:energy2}) and
843: integrating it over mass $m_2$, we can obtain the energy change of
844: stars with mass $m_1$ as
845: \begin{eqnarray}
846: \left< \frac{d}{dt}(m_1E_1) \right> &=& A'm_1\int^{m_{\rm max}}_{m_{\rm min}}m_2^{-\alpha+1}(m_2-m_1)dm_2\\
847: &\propto&m_1 \left[\frac{m_2^{-\alpha +3}}{-\alpha +3} -\frac{m_2^{-\alpha+2}}{-\alpha+2}m_1\right]_{m_{\rm min}}^{m_{\rm max}},\label{eq:energy3}
848: \end{eqnarray}
849: where $m_{\rm max}$ and $m_{\rm min}$ are the maximum and minimum mass of
850: the MF and $A'$ is a constant.
851: If the right side of equation (\ref{eq:energy3}) is negative, the star
852: with mass $m_1$ loses energy and sink to the center of the star cluster.
853: The minimum mass with the negative energy change, $m_{\rm sink}$, is
854: expressed as
855: \begin{eqnarray}
856: m_{\rm sink}&=&\frac{-\alpha +2}{-\alpha +3} \hspace{3pt}
857: \frac{m_{\rm max}^{-\alpha+3}-m_{\rm min}^{-\alpha+3}}{m_{\rm
858: max}^{-\alpha+2}-m_{\rm min}^{-\alpha+2}}\\
859: &=&f(\alpha) m_{\rm max}\frac{1-x^{-\alpha +3}}{1-x^{-\alpha +2}},
860: \end{eqnarray}
861: where we defined $x\equiv m_{\rm min}/m_{\rm max}$ and
862: $f(\alpha) \equiv (-\alpha +2)/(-\alpha +3)$.
863: Using the values used for our model, $m_{\rm max}=100M_{\sun}$,
864: $m_{\rm min}=0.3M_{\sun}$ and $\alpha = 2.35$, we obtain
865: $m_{\rm sink}=7.9M_{\sun}$.
866: This value agrees well with the mass where the power-low index of the MF
867: breaks in figure \ref{fig:MF}.
868: If $x \ll 1$, we obtain
869: \begin{eqnarray}
870: \frac{1-x^{-\alpha +3}}{1-x^{-\alpha +2}}=
871: \left\{ \begin{array}{ll}
872: 1 & (\alpha > 3),\\
873: -x^{\alpha-2} & (2<\alpha < 3),\\
874: x & (\alpha < 2).
875: \end{array} \right.
876: \end{eqnarray}
877: Therefore,
878: \begin{eqnarray}
879: m_{\rm sink} &\simeq&
880: \left\{ \begin{array}{ll}
881: f(\alpha) m_{\rm min} & (\alpha > 3),\\
882: -f(\alpha) m_{\rm min}^{\alpha -2} m_{\rm max}^{3-\alpha} & (2<\alpha < 3),\\
883: f(\alpha) m_{\rm max} & (\alpha < 2).
884: \end{array} \right.
885: \end{eqnarray}
886: When $2<\alpha <3$, the value of $m_{\rm sink}$ varies from
887: $m_{\rm min}$ to $m_{\rm max}$.
888:
889:
890: The observed MF of the stars in the central parsec is much flatter
891: than Salpeter \citep{Pa06}. The cluster inspiral model rather
892: naturally explain this flat MF, since only the most massive
893: stars remain bound to the cluster and are carried to the central region of
894: the galaxy.
895:
896: \begin{figure}
897: \epsscale{.90}
898: \plotone{f17.eps}
899: \caption{The mass function of the bound stars for model C. Times are
900: $T=$ 0.15, 0.3, 0.45, 0.6, and 0.75 Myrs.\label{fig:MF1}}
901: \end{figure}
902:
903: \begin{figure}
904: \epsscale{.90}
905: \plotone{f18.eps}
906: \caption{The fraction of the mass function of the bound stars to the
907: initial mass function for model C. Times are the same as figure
908: \ref{fig:MF1}.\label{fig:MF}}
909: \end{figure}
910:
911:
912:
913: \begin{figure}
914: \epsscale{.90}
915: \plotone{f19.eps}
916: \caption{The evolution of the enclosed mass of the bound stars for model
917: C.\label{fig:rm}}
918: \end{figure}
919:
920: \section{Summary and Discussion}
921:
922: \subsection{Summary}
923:
924: We performed fully self-consistent $N$-body simulations of a star
925: cluster within its parent galaxy and compared the orbital and internal
926: evolutions of the star cluster with those obtained by ``traditional''
927: simulations, in which the orbital evolution of the star cluster is
928: calculated from the dynamical friction formula.
929: We confirmed that the inspiral timescale of the star cluster is shorter
930: than that obtained from the ``traditional'' simulations. Furthermore,
931: our results showed that the core collapse make the core density of the
932: cluster increase and helps the cluster survive.
933:
934: We performed simulations of circular and eccentric orbits of the star
935: cluster. In previous studies, most of the simulations were from
936: circular orbits (PZ03; G\"{u}rkan \& Rasio 2005). We found that,
937: however, eccentric orbits are more favorable to explain the distribution
938: of stars around the GC for following reasons.
939:
940: First, eccentric orbits are natural, if the formation of star clusters
941: was triggered by collisions between gas clouds. Second, star clusters
942: with eccentric orbits can approach to the GC much faster than those with
943: circular orbits (KM03). Third, \citet{Pa06} showed that many stars in the
944: counterclockwise rotating disk have high eccentricities
945: ($e\simeq 0.8$), while the distribution of the eccentricities
946: of the stars in the clockwise disk is very broad. Since the
947: eccentricities of the escaped stars distribute around the eccentricity
948: of the star cluster, the star cluster model with eccentric orbits can
949: naturally explain the existence of high-eccentricity stars.
950:
951: The power-low index of the MF of the bound stars to the star cluster
952: breaks at around 7.9 $M_{\sun}$. Stars heavier than this mass sink to
953: the center of the star cluster due to the mass segregation. Since the
954: tidal stripping removes the stars outside of the star cluster, the
955: massive stars selectively remain in the star cluster. As a result, the
956: star cluster carries only massive stars to the GC. The star cluster
957: scenario can reproduce the flat MF in the central parsec, without the
958: need for nonstandard initial mass function.
959:
960:
961: \subsection{Realistic Model of the Galaxy}
962: First, we showed the orbits of star clusters decays faster in
963: full $N$-body simulations than in ``traditional'' simulations, which
964: treat the dynamical friction analytically using a King model $W_{0}=10$
965: as a model of the central region of the Galaxy.
966: The model is sufficient for such comparisons, but not for more realistic
967: comparison between our model and the stars in the GC because within
968: $\sim$1 pc the mass density of our model is much lower than that of
969: the actual Galaxy.
970: Next, we showed the case of more realistic galaxy model with a central
971: BH. The orbital evolution was similar that in
972: the model without a BH. For the comparison between our simulations and
973: the actual stars, however, we need more simulations in various initial
974: conditions. We will report more detail result of runs with the central
975: SMBH in forthcoming papers.
976:
977:
978: \subsection{Formation of an Intermediate-mass Black Hole (IMBH)}
979:
980: IRS 13E consists of seven stars within a projected diameter of $\sim$
981: 0.02 pc and is located at $\sim$ 0.14 pc in projection from the GC
982: and these stars have very similar proper motions \citep{Maillard04,Pa06}.
983: \citet{Maillard04} suggested that IRS 13E is the remnant core of a star
984: cluster that have fallen to the GC and dissolved there and that the members
985: of IRS 13E are bound by a central IMBH.
986: From the analysis of proper motions, the minimum mass of the IMBH was
987: estimated as 1300 $M_{\sun}$ \citep{Maillard04} - $10^4 M_{\sun}$
988: \citep{Schodel05}.
989:
990: In this paper we simulated the evolution of star clusters only before
991: the core collapse because of the limitation of our present code. Our
992: code currently cannot treat the post-collapse evolution since we use the
993: softened potential. Collisions and mergers between stars would have
994: occurred and an IMBH would have be formed, if our code can handle these
995: events. The star cluster inspiral scenario might reveal the origin of
996: the IRS 13E. We are currently working to implement collisions and
997: mergers. The result will be reported in the future papers.
998:
999:
1000:
1001: \acknowledgments
1002:
1003: The authors thanks Piet Hut, Keigo Nitadori, and Ataru Tanikawa for
1004: useful comments and discussions.
1005: M. F. is financially supported by Research Fellowships of the Japan
1006: Society for the Promotion of Science (JSPS) for Young Scientists.
1007: This research is partially supported by
1008: the Special Coordination Fund for Promoting Science and Technology
1009: (GRAPE-DR project), Ministry of Education, Culture, Sports, Science and
1010: Technology, Japan.
1011: Part of calculations were done using the
1012: GRAPE system at the Center for Computational Astrophysics (CfCA) of
1013: the National Astronomical Observatory of Japan.
1014:
1015:
1016:
1017: \begin{thebibliography}{}
1018:
1019: \bibitem[Barnes \& Hut(1986)]{BH86} Barnes, J., \& Hut, P. 1986,
1020: \nat, 324, 446
1021:
1022: \bibitem[Binney and Tremaine(1987)]{BT87}
1023: Binney, J., \& Tremaine S. 1987, Galactic Dynamics (Princeton:
1024: Princeton Univ. Press), 425
1025:
1026: \bibitem[Chandrasekhar(1943)]{Ch43}
1027: Chandrasekhar, S. 1943, \apj, 97, 255
1028:
1029: \bibitem[Eisenhauer et al.(2005)]{E05}
1030: Eisenhauer, F., et al.\ 2005, \apj, 628, 246
1031:
1032:
1033: \bibitem[Figer et al.(2002)]{Fi02}
1034: Figer, D. F. et al. 2002, \apj, 581, 258
1035:
1036: \bibitem[Figer(2004)]{Fi04}
1037: Figer, D. F. 2004, in ASP Conf. Ser. 322, in The formation and
1038: evolution of massive young star clusters,
1039: ed. H. J. G. L. M. Larmers, L. J. Smith, \& A. Nota, 49,
1040: astro-ph/0403088
1041:
1042: \bibitem[Fujii et al.(2006)]{Fj06}
1043: Fujii, M., Funato, Y., \& Makino, J. 2006. \pasj, 58, 743
1044:
1045: \bibitem[Fujii et al.(2007)]{Fj07}
1046: Fujii, M., Iwasawa, M., Funato, Y., \& Makino, J. 2007, PASJ, 59, 1095
1047:
1048: \bibitem[Fukushige et al.(2005)]{Fk05}
1049: Fukushige, T., Makino, J., \& Kawai, A. 2005, \pasj, 57, 1009
1050:
1051: \bibitem[Genzel et al.(2000)]{Genzel00}
1052: Genzel, R., Pichon, C., Eckart, A., Gerhard, O. E., \& Ott, T. 2000,
1053: \mnras, 317, 348
1054:
1055: \bibitem[Genzel et al.(2003)]{Genzel03}
1056: Genzel, R. et al. 2003, \apj, 594, 812
1057:
1058: \bibitem[Gerhard(2001)]{Gerhard01}
1059: Gerhard, O. 2001, \apj, 546, L39
1060:
1061: \bibitem[Ghez(2003)]{Ghez03}
1062: Ghez, A. M., et al. 2003, \apj, 586, L127
1063:
1064: \bibitem[G\"{u}rkan \& Rasio(2005)]{GR05}
1065: G\"{u}rkan, M. A., \& Rasio, F. A. 2005, \apj, 628, 236
1066:
1067: \bibitem[Hashimoto et al. (2003)]{H03}
1068: Hashimoto, Y., Funato, Y., \& Makino, J. 2003, \apj, 582, 196
1069:
1070: \bibitem[Heggie \& Mathieu(1986)]{HM86}
1071: Heggie, D. C., \& Mathieu, R. D. 1986, in The Use of
1072: Supercomputers in Stellar Dynamics, ed. P. Hut, and S.
1073: McMillan (Lecture Notes in Physics 267; Berlin: Springer), 233
1074:
1075: \bibitem[Heggie \& Hut(2003)]{HH03}
1076: Heggie, D. C. \& Hut, P. 2003, The gravitational million-body
1077: problem (Cambridge: Cambridge University Press), 159
1078:
1079: \bibitem[Hernquist \& Katz(1989)]{HK89}
1080: Hernquist, L. \& Katz, N. 1989, \apjs, 70, 419
1081:
1082: \bibitem[Kim \& Morris(2003)]{KM03}
1083: Kim, S. S. \& Morris, M. 2003, \apj, 597, 312
1084:
1085: \bibitem[Kim et al.(2004)]{Kim04}
1086: Kim, S. S., Figer D. F.,\& Morris, M. 2004, \apj, 607, L123
1087:
1088: \bibitem[Kinoshita, Yoshida, \& Nakai(1991)]{KYN91}
1089: Kinoshita, H., Yoshida, H., \& Nakai, H. 1991, Cel. Mech. and
1090: Dyn. Astr., 50, 59
1091:
1092: \bibitem[Krabbe et al.(1995)]{K95}
1093: Krabbe, A. et al. 1995, \apj, 447, L95
1094:
1095: \bibitem[Lacy et al.(1991)]{Lacy91}
1096: Lacy, J. H., Achtermann, J. M., \& Serabyn, E. 1991, 380, L71
1097:
1098: \bibitem[Levin \& Beloborodov(2003)]{LB03}
1099: Levin, Y. \& Beloborodov, M. 2003, \apj, 590, L33
1100:
1101: \bibitem[Lo \& Claussen(1983)]{LC83}
1102: Lo, K. Y., \& Claussen, M. J. 1983, \nat, 306, 647
1103:
1104: \bibitem[Lu et al.(2006)]{Lu06}
1105: Lu, J. R., Ghez, A. M., Hornstein, S. D., Morris, M., Thompson,
1106: D. J., \& Becklin, E. E. 2006, JPhCS, 54, 279
1107:
1108: \bibitem[Maillard et al.(2004)]{Maillard04}
1109: Maillard, J. P., Paumard, T., Stolovy, S. R., \& Rigaut, F. 2004,
1110: \aap, 423, 155
1111:
1112: \bibitem[Makino(1991)]{M91}
1113: Makino, J. 1991, \pasj, 43, 621
1114:
1115: \bibitem[Makino \& Aarseth(1992)]{MA92}
1116: Makino, J. \& Aarseth, S. J. 1992, \pasj, 44, 141
1117:
1118: \bibitem[Makino et al.(2003)]{M03}
1119: Makino, J., Fukushige, T., Koga, M. \& Namura, K. 2003, \pasj, 55,
1120: 1163
1121:
1122: \bibitem[Makino (2004)]{M04}
1123: Makino, J. 2004, \pasj, 56, 521
1124:
1125: \bibitem[McMillan \& Aarseth(1993)]{MA93}
1126: McMillan, S. L. W. \& Aarseth, S. J. 1993, \apj, 414, 200
1127:
1128: \bibitem[McMillan \& Portegies Zwart(2003)]{MP03}
1129: McMillan, S. L. W. \& Portegies Zwart, S. F. 2003, \apj, 596, 314
1130:
1131: \bibitem[Nagata et al.(1995)]{Nagata95}
1132: Nagata, T., Woodward, C. E., Shure, M. \& Kobayashi, N. 1995, \aj,
1133: 109, 1676
1134:
1135: \bibitem[Nayakshin \& Sunyaev(2005)]{NS05}
1136: Nayakshin, S. \& Sunyaev, R. 2005, \mnras, 364, L23
1137:
1138: \bibitem[Nayakshin et al.(2007)]{Nay07}
1139: Nayakshin, S., Cuadra, J., \& Springel, V. 2007, astro-ph/0701141
1140:
1141: \bibitem[Paumard et al.(2001)]{Pa01}
1142: Paumard, T., Maillard, J. P., Morris, M., \& Rigaut, F. 2001 \aap,
1143: 366, 466
1144:
1145: \bibitem[Paumard et al.(2004)]{Pa04}
1146: Paumard, T., Maillard, J. P., \& Morris, M. 2004, \aap, 426, 81
1147:
1148: \bibitem[Paumard et al.(2006)]{Pa06}
1149: Paumard, T. et al. 2006 \apj, 643, 1011
1150:
1151: \bibitem[Portegies Zwart \& McMillan(2002)]{PM02}
1152: Portegies Zwart, S. F. \& McMillan, S. L. W. 2002, \apj, 576, 899
1153:
1154: \bibitem[Portegies Zwart et al.(2003)]{PZ03}
1155: Portegies Zwart, S. F., McMillan, S. L. W., \& Gerhard, O. 2003,
1156: \apj, 593, 352
1157:
1158: \bibitem[Salpeter(1955)]{Sal55}
1159: Salpeter, E. E. 1955, \apj, 121, 161
1160:
1161: \bibitem[Sch\"{o}del et al.(2005)]{Schodel05}
1162: Sch\"{o}del, R., Eckart, A., Iserlohe, C., Genzel, R., \& Ott,
1163: T. 2005, \apj, 625, L111
1164:
1165: \bibitem[Spitzer \& Hart(1971)]{SH71}
1166: Spitzer, L. J. \& Hart, M. H. 1971, \apj, 164, 399
1167:
1168: \bibitem[Tuckerman et al.(1990)]{Tu90}
1169: Tuckerman, M. E., Martyna, G. J., \& Berne, B. J. 1992,
1170: J. Chem. Phys. 97, 1990
1171:
1172: \bibitem[Wisdom \& Holman(1991)]{WH91}
1173: Wisdom, J. \& Holman, M. 1991, AJ, 102, 152
1174:
1175: \end{thebibliography}
1176:
1177:
1178:
1179: \end{document}
1180:
1181:
1182: