1: %\documentclass[10pt,preprint]{aastex}

2: \documentclass[onecolumn]{emulateapj}

3: 

4: \usepackage{multirow,graphicx}

5: 

6: \renewcommand{\vec}[1]{\mathbf{#1}}

7: \newcommand{\el}{$2003~{\rm EL_{61}}$}

8: 

9: \begin{document}

10: 

11: \title{L\'{e}vy Flights of Binary Orbits due to Impulsive Encounters}

12: \author{Benjamin F. Collins\altaffilmark{1} and Re'em Sari\altaffilmark{1,2}}

13: \altaffiltext{1}{California Institute of Technology, MC 130-33,

14:         Pasadena, CA 91125}

15: \altaffiltext{2}{Racah Institute of Physics, Hebrew University, Jerusalem 91904,

16: Israel}

17: \email{bfc@tapir.caltech.edu}

18: 

19: 

20: \begin{abstract}

21: We examine the evolution of an almost circular Keplerian orbit

22: interacting with unbound perturbers.  We calculate the change in

23: eccentricity and angular momentum that results from a single

24: encounter, assuming the timescale for the interaction is shorter than

25: the orbital period.  The orbital perturbations are incorporated into a

26: Boltzmann equation that allows for eccentricity dissipation.  We

27: present an analytic solution to the Boltzmann equation that describes

28: the distribution of orbital eccentricity and relative inclination as a

29: function of time.  The eccentricity and inclination of the binary do

30: not evolve according to a normal random walk but perform a L\'{e}vy

31: flight.  The slope of the mass spectrum of perturbers dictates whether

32: close gravitational scatterings are more important than distant tidal

33: ones.  When close scatterings are important, the mass spectrum sets

34: the slope of the eccentricity and inclination distribution functions.

35: We use this general framework to understand the eccentricities of

36: several Kuiper belt systems: Pluto, \el, and Eris.  We use the model

37: of \citet{TBGE07} to separate the non-Keplerian components of the

38: orbits of Pluto's outer moons Nix and Hydra from the motion excited by

39: interactions with other Kuiper belt objects.  Our distribution is

40: consistent with the observations of Nix, Hydra, and the satellites of

41: \el and Eris.  We address applications of this work to objects outside

42: of the solar system, such as extrasolar planets around their stars and

43: millisecond pulsars.

44: \end{abstract}

45: 

46: \keywords{Kuiper Belt --- planets and satellites --- minor planets, asteroids}

47: 

48: \section{Introduction}

49: 

50: Several binary Kuiper belt objects (KBOs) have well-measured small

51: orbital eccentricities \citep{NGC07}.  \citet{SBL03} investigate

52: numerically the forcing of the eccentricity of the Pluto-Charon orbit

53: by interloping KBOs.  They find that the system almost never possesses

54: an eccentricity as high as the observed value of 0.003 \citep{TBGE07};

55: depending on the model of tidal damping used, they find median values

56: of $10^{-5} - 10^{-4}$.  Our goal is to develop an analytic theory that

57: describes the effects of a population of unbound perturbers on a

58: binary orbit and can be applied simply to any binary, in the Kuiper

59: belt or elsewhere.

60: 

61: The interaction of a binary system with its environment has been

62: studied extensively in the literature \citep{H75,SS95,Y02,MME07,SHM07}.  

63: One interesting context is

64: white dwarf-pulsar binaries, which are expected to be circular.  For

65: these objects pulse timing produces very accurate measurements of

66: their orbital motion; such measurements reveal that their

67: eccentricities are typically very small but finite, around

68: $10^{-4}-10^{-5}$ \citep{S04}.  \citet{P92} investigated the effects

69: of passing stars on the orbit of such a binary and found that for

70: Galactic pulsars, the perturbations are sub-dominant compared to the

71: effects of atmospheric fluctuations in the companion star.  The higher

72: density environment of a globular cluster however can induce an order

73: of magnitude higher eccentricity.  \citet{RH95} and \citet{HR96}

74: present a detailed account of the changes in orbital parameters for

75: binaries in a stellar cluster.  The work of these authors focuses on

76: the regime where a perturbing body interacts with the binary on

77: timescales longer than the orbital period of the binary.  In the

78: Kuiper belt, a single interaction between a binary and an unbound

79: object occurs over a shorter timescale than the orbital period of the

80: binary.  We focus on this regime, where the perturbations 

81: to the orbital dynamics can be approximated as discrete impulses.

82: 

83: The main result of this work is that we have identified the

84: perturbative evolution of the eccentricity and relative inclination of

85: a nearly circular binary orbit as a L\'{e}vy flight, a specific type of 

86: random walk through phase space \citep{SZF95}.  The entire

87: distribution function of the eccentricity and inclination is then

88: determined by calculating the frequency of perturbations as a function

89: of their magnitude.    We find a simple analytic expression for this

90: distribution function.

91: 

92: We take the following steps to arrive at 

93: our conclusion.  In section \ref{secSingleEnc} we

94: calculate the effect of one perturber on a two-body orbit, examining

95: separately the tidal effects of distant scatterings, close encounters

96: with a single binary member, and direct collisions.  We describe the

97: effects of many such encounters in section \ref{secBE}, and write a

98: Boltzmann equation that describes the distribution function of the

99: orbital eccentricity and the inclination of the binary relative to its

100: initial plane.  The quantitative description of the binary's evolution 

101: given by this distribution function reveals its nature as a L\'{e}vy flight.

102: In section \ref{secMassSpec}, we allow for a

103: distribution of perturbing masses and discuss the different L\'{e}vy

104: distributions that result.

105: 

106: We then use the analytic theory to examine the orbits of binary KBOs

107: being perturbed by the other members of the Kuiper belt.  Section

108: \ref{secKBB} applies our analysis to several specific Kuiper belt

109: binaries.  We briefly discuss the relevance of this theory to other

110: astrophysical systems in section \ref{secApp}, and summarize our

111: conclusions in section \ref{secConclusions}.

112: 

113: \section{A Single Encounter}

114: \label{secSingleEnc}

115: 

116: We use the following terminology to describe the geometry of the

117: encounter between a single perturber and a two-body orbit.  We refer

118: to the two bound bodies as ``the binary.''  The members of the

119: binary have masses $m_1$ and $m_2$, with a total mass labeled

120: $m_b=m_1+m_2$ and $m_1 \geq m_2$.  The position of body 2 relative to body

121: 1 is given by $\vec r_b$, and the relative velocity by $\vec v_b$.  We

122: distinguish between the magnitude and direction of a vector with the

123: notation $\vec r_b = r_b \hat r_b$.  We assume

124: $v_b \approx \Omega r_b$, where $\Omega$ is the orbital frequency of

125: the binary.  We write the orbital period as $T_{\rm orb}=2 \pi/\Omega$.

126: 

127: We label the mass of the perturber $m_p$.  The position of the

128: perturber as a function of time, $\vec r_p(t)$, is described by two

129: vectors: $\vec r_p(t) = \vec b + \vec v_p t$.  The vector $\vec b$

130: specifies the closest point of the perturber's trajectory to body 1,

131: and $\vec v_p$ is velocity of the perturber relative to body 1.  Each

132: encounter geometry is uniquely specified by $\vec b$ and $\vec v_p$

133: under the constraint $\vec b \cdot \vec v_p =0$.  Figure

134: \ref{figCartoon} depicts the arrangement of the vectors $\vec r_b,

135: \vec v_b, \vec r_p(t), \vec b,$ and $\vec v_p$.  We assume $T_{\rm

136:   orb} \gg b/v_p$ so that we may ignore the motion of the binary

137: during the interaction.  We further assume that the effects of the

138: gravity of the binary on the perturber are small; the perturber then

139: travels along a straight path with a constant $\vec v_p$.  This

140: assumption requires the criterion of $v_p^2 \gg G (m_b+m_p)/b$.  If

141: $b$ is small, the perturber may collide with a member of the binary.

142: In this case the assumption that the path of the perturber is

143: unaffected by the gravity of the binary is true under the condition

144: that $v_p$ is much greater than the escape velocity of that member of

145: the binary.  The escape velocity from body 1 is defined $v^2_{{\rm

146:     esc,1}}=2 G m_1/ R_1$, where $R_1$ is the radius of body 1.

147: 

148: \begin{figure}[t!]

149: \center

150: \includegraphics[angle=-90]{b-1-3.ps}

151: \caption{An illustration of the notation we use to denote the geometry of each

152: perturbation.  The dotted line is the almost circular orbit of the binary viewed 

153: at an angle.  The dashed line is the path of the perturber, given by 

154: $\vec r_p(t)= \vec b + \vec v_p t$.}

155: \label{figCartoon}

156: \end{figure}

157: 

158: We are assuming that the timescale of the interaction is much shorter than

159: the orbital timescale, such that the perturbation instantaneously changes the

160: velocities of the binary objects.

161: The impulse provided to a specific member of the binary is found by

162: integrating the acceleration caused by the perturber over its path:

163: 

164: \begin{equation}

165: \label{eqDeltaV1}

166: \Delta \vec v_j = \int_{-\infty}^{\infty} 

167: \frac{G m_p(\vec b_j + \vec v_p t)}{|\vec b_j+ \vec v_p t|^3}

168:  dt = 2 \frac{G m_p}{v_p}

169: \frac{\hat b_j}{b_j},

170: \end{equation}

171: 

172: \noindent

173: where the index $j$ specifies whether the impulse $\Delta \vec v_j$

174: and impact parameter $\vec b_j$ are 

175: with respect to either the primary ($j=1$) or the secondary

176: ($j=2$).  For the primary, $\vec b_1 = \vec b$ as we have defined it

177: above.  For encounters with the secondary, $\vec b_2$ is related to

178: $\vec b$ by enforcing that it is also perpendicular to $\vec v_p$.

179: Thus we find $\vec b_2 = \vec b - \vec r_b + \hat v_p ( \vec r_b \cdot

180: \hat v_p)$.

181: 

182: 

183: We

184: consider the effects of such impulses on the full Laplace-Runge-Lenz

185: vector, $\vec e = (\vec v_b \times \vec H)/G m_b - \hat r$, where

186: $\vec H=\vec r_b \times \vec v_b$, the angular momentum per unit mass

187: of the binary.  The vector $\vec e$ has a magnitude equal to the

188: eccentricity of the orbit, and points from body 1 towards

189: the periapse.  It responds to a small impulse $\Delta \vec v$ according to

190: the formula

191: 

192: \begin{equation}

193: \label{eqEdot}

194: \Delta \vec e = \frac{1}{G m_b} \left[ 2 \vec r_b (\Delta \vec v \cdot \vec v_b)

195: - \vec v_b (\Delta \vec v \cdot \vec r_b) 

196: - \Delta \vec v (\vec r_b \cdot \vec v_b)

197: \right],

198: \end{equation}

199: 

200: \noindent

201: keeping terms up to linear order in $\Delta \vec v$.

202: Since we have assumed the binary has very small eccentricity, the

203: third term in equation \ref{eqEdot} is negligible compared to the

204: other two.  

205: 

206: The orbital plane of the binary is defined by the angular momentum

207: vector $\vec H$, and evolves according to $\Delta \vec H = \vec

208: r_b \times \Delta \vec v$.  The impulses affect the

209: direction of the angular momentum vector, and therefore alter the

210: orientation of the orbital plane of the binary.  We use the

211: two-dimensional vector $\vec i$ to denote the components of $\hat H$

212: in the plane defined by the initial angular momentum.  This vector,

213: $\vec i$, has a magnitude equal to $\sin i$, the sine of the

214: inclination of the binary with respect to the initial orbital plane,

215: and points from body 1 towards the longitude of the ascending node.

216: 

217: The change in relative velocity given by a general gravitational scattering 

218: is given by $\Delta \vec v = \Delta \vec v_2 - \Delta \vec v_1$.  The 

219: resulting change in the eccentricity vector is

220: 

221: \begin{equation}

222: \label{eqDeltaegeneral}

223: \Delta \vec e = 2 \frac{m_p}{m_b} \frac{v_b}{v_p} 

224: \left[ 2 \hat r_b \left( \frac{\hat b_2 \cdot \hat v_b}{b_2/r_b} 

225: - \frac{\hat b \cdot \hat v_b}{b/r_b} \right)

226: - \hat v_b \left( \frac{\hat b_2 \cdot \hat r_b}{b_2/r_b} -

227: \frac{\hat b \cdot \hat r_b}{b/r_b} \right) \right].

228: \end{equation}

229: 

230: \noindent 

231: %We simplify this expression with the abbreviation 

232: %$|\Delta \vec e| = A_e (m_p/m_b)(v_b/v_p)$.  

233: %Here $A_e$ contains all of the 

234: %dependence of the perturbation on the relative orientation of the binary 

235: %and perturber as well as the dependence on impact parameter $b$.  

236: The change in $\vec i$ is

237: 

238: \begin{equation}

239: \label{eqDeltaigeneral}

240: \Delta \vec i = - 2 \frac{m_p}{m_b} \frac{v_b}{v_p}

241: \left[ \hat v_b \left(\frac{\hat b_2 \cdot \hat n}{b_2/r_b} 

242: - \frac{\hat b \cdot \hat n}{b/r_b} \right) \right],

243: \end{equation}

244: 

245: \noindent

246: where $\hat n$ is the unit normal vector to the binary's orbital 

247: plane.

248: %We abbreviate this expression as $|\Delta \vec i|=A_i (m_p/m_b)(v_b/v_p)$.

249: For both the farthest perturbers and the closest, the dependence of 

250: equations \ref{eqDeltaegeneral} and \ref{eqDeltaigeneral} on the impact parameter

251: can be simplified.  We discuss these limits in the following sections.

252: 

253: \subsection{Close Encounters}

254: \label{secVClose}

255: 

256: Interactions with impact parameters greater than the radius of the

257: primary or secondary but much less than the semi-major axis of the

258: binary belong to what we call the ``close-encounter regime.''  By

259: definition the encounters in this regime of impact parameter are much

260: closer to one member of the binary than the other.  As a result the

261: relative impulse experienced is dominated by the single impulse

262: delivered to that body, $|\Delta \vec v| \approx |\Delta \vec v_j|$.

263: The changes in $\vec e$ and $\vec i$ are then given not by the

264: difference of the impulses on each body, as in equations

265: \ref{eqDeltaegeneral} and \ref{eqDeltaigeneral}, but by the effects of

266: only the largest impulse.  For the change in eccentricity we find,

267: 

268: \begin{equation}

269: \label{eqEClose}

270: \Delta \vec e = 2 \frac{m_p}{m_b} \frac{v_b}{v_p} \frac{r_b}{b}

271: \left[ 2 \hat r_b (\hat b_j \cdot \hat v_b) - \hat v_b (\hat b_j \cdot \hat r_b)

272: \right],

273: \end{equation}

274: 

275: \noindent

276: and for the inclination,

277: 

278: \begin{equation}

279: \label{eqIClose}

280: \Delta \vec i = - 2 \frac{m_p}{m_b} \frac{v_b}{v_p} \frac{r_b}{b}

281: \left[ \hat v_b (\hat b_j \cdot \hat n) \right].

282: \end{equation}

283: %\noindent

284: %We abbreviate the angular information of close encounters with the 

285: %quantity $B_e$ such that $|\Delta \vec e|=B_e (m_p/m_b)(v_b/v_p)(r_b/b)$, 

286: %leaving the scaling with $b$ explicit.  

287: %We use angled brackets

288: %to denote the average over the possible angular configurations of the

289: %encounter:

290: %$\langle B_e \rangle = \int B_e d \Omega/ 8 \pi^2 = E(-3) \approx 1.2$,

291: %where $E(-3)$ is the complete Elliptic integral and $d \Omega$ denotes

292: %the differential of the angular phase space.  Inclination?

293: 

294: 

295: \subsection{Distant Encounters}

296: 

297: For interactions where $b \gg r_b$, the impulse delivered to each

298: member of the binary is almost the same.  In this limit only the tidal

299: difference in impulse affects the eccentricity of the binary.  The

300: perturbation delivered to the lowest order in $r_b/b$ is

301: 

302: \begin{equation}

303: \label{eqDeltae}

304: \Delta \vec e = 2 \frac{m_p}{m_b} \frac{v_b}{v_p} \left(\frac{r_b}{b}\right)^2

305: \left[ \hat r_b \left( 4 (\hat r_b \cdot \hat b)( \hat v_b \cdot \hat b)

306: + 2 (\hat r_b \cdot \hat v_p)( \hat v_b \cdot \hat v_p) \right)

307:   + \hat v_b \left(1 - 2 (\hat r_b \cdot \hat b)^2 - (\hat r_b \cdot \hat v_p)^2 

308: \right) \right].

309: \end{equation}

310: 

311: \noindent

312: %To shorten this expression, we write $|\Delta \vec e|=C_e

313: %(m_p/m_b)(v_b/v_p)(r_b/b)^2$.  

314: \citet{P92} derives the special case of

315: this formula for interactions that take place entirely in the plane of

316: the binary.  This formula is also equivalent to equation A24 of

317: \citet{HR96}.

318: 

319: The change in $\vec i$ due to distant encounters is given by:

320: 

321: \begin{equation}

322: \label{eqDeltah}

323: \Delta \vec i = \frac{m_p}{m_b} \frac{v_b}{v_p} \left(\frac{r_b}{b}\right)^2

324: \hat v_b \left[ 4(\hat r_b \cdot \hat b)(\hat b \cdot \hat n)

325: + 2 (\hat r_b \cdot \hat v_p)(\hat v_p \cdot \hat n) \right].

326: \end{equation}

327: 

328: 

329: \subsection{Collisions}

330: \label{secCollisions}

331: 

332: Physical collisions between perturbers and body 1 or 2 cause the orbit

333: to evolve impulsively.  We define collisions to be any encounters

334: where the impact parameter is smaller than the radius of the primary

335: or secondary: $b<r_1$ or $b_2<r_2$.  In this case the impulse is given

336: by the conservation of linear momentum of the encounter: $\Delta \vec

337: v = \chi (m_p/m_j) \vec v_p$, where $m_j$ is the mass of the binary

338: member involved in the collision ($j=1$ or $2$).  The coefficient

339: $\chi$ accounts for the final momentum of the perturber.  For an

340: inelastic collision with $m_p \ll m_j$, $\chi=1$.  If the perturber is

341: perfectly reflected, $\chi=2$.  The momentum loss from an impact

342: crater can enhance this factor above 2 depending on the properties of

343: the colliding bodies \citep{MNZ94}.  For simplicity we

344: assume that the mass of each binary

345: member remains unchanged after each collision.

346: 

347: The collisional impulse changes the eccentricity according to equation

348: \ref{eqEdot} and the orbital plane according to the change in angular

349: momentum $\vec r_b \times \Delta \vec v$. 

350: 

351: \section{Boltzmann Equation}

352: \label{secBE}

353: 

354: The evolution of the eccentricity and inclination (relative to the

355: initial orbital plane) is given by the sum of the perturbations the

356: binary receives as it travels through a swarm of perturbers.  From the

357: average properties of the perturbing population, we can calculate a

358: distribution function that describes the evolution of the orbit in a

359: statistical sense.

360: 

361: \subsection{Eccentricity}

362: \label{secBEecc}

363: 

364: 

365: Since the perturbation in eccentricity is a two-dimensional vector,

366: each component is added to the components of the existing eccentricity

367: vector separately.  As the binary experiences many perturbations, its

368: eccentricity vector travels throughout this two-dimensional space.  We

369: write a distribution function $f(\vec e,t)$ that describes the

370: probability that the binary will have an eccentricity 

371: in a small region $d^2 \vec e$.  Assuming isotropic perturbations, there

372: is no preferred longitude of periapse for the binary.  It follows that

373: $f(\vec e,t)=f(e,t)$ and the likelihood of finding the

374: eccentricity in a small range $de$ around $e$ is $2 \pi e f(e, t) de$.

375: 

376: We define ${\cal R}(e')$ to be the frequency at which the binary

377: experiences perturbations of magnitudes between $e'$ and $e'+de'$.

378: The frequency of perturbations with magnitudes on the order of 

379: $|\Delta \vec e|=e'$ is given

380: schematically by $e' {\cal R}(e') \sim n v b^2$, where

381: $n$ is the number density of the perturbers, $v$ is the speed at

382: which the binary encounters those perturbers, and $b$ is the

383: distance at which the binary encounters perturbers that cause a

384: perturbation of strength $e'$.  We make this calculation precise with 

385: the following integral:

386: 

387: \begin{equation}

388: \label{eqRofe}

389: {\cal R}(e')= \int \delta(| \Delta \vec e(\vec v_p,\vec b,m_p)| - e')

390: {\cal F}(\vec v_p,m_p) v_p \delta(\vec b \cdot \hat v_p) d^3 \vec b d^3 \vec v_p dm_p,

391: \end{equation}

392: 

393: \noindent 

394: where ${\cal F}(\vec v_p,m_p)$ is the phase space density per unit

395: mass of the perturbers.  The integral of ${\cal F}(\vec v_p,m_p)$ over

396: $d^3 \vec v_p dm_p$ is the number density of the perturbers.  We

397: assume this density is uniform in the spatial dimensions and isotropic

398: in velocity.  It is normalized such that the total mass density of

399: perturbers is given by $\rho= \int m_p {\cal F}(\vec v_p,m_p) d^3 \vec

400: v_p dm_p$.  The factor of $v_p$ in the integrand of equation

401: \ref{eqRofe} represents the velocity at which the binary encounters

402: perturbers.  The second delta function in equation \ref{eqRofe}

403: converts the volume element $d^3 \vec b$ to an element of

404: cross-sectional area.  The first delta function, $\delta(| \Delta \vec

405: e(\vec v_p,\vec b,m_p)| - e')$, restricts the integral to include only

406: the combinations of $\vec b$, $\vec v_p$, and $m_p$ that cause a

407: $|\Delta \vec e|=e'$.

408: 

409: The evolution of the distribution function as a result of these perturbations 

410: is given by a Boltzmann equation that links the rate of change of $f(e,t)$

411: to the interaction frequency.  We write this equation as:

412: 

413: \begin{equation}

414: \label{eqFofe}

415: \frac{\partial f(e,t)}{\partial t} 

416: = \int p(\vec e') \left[ f(|\vec e'+\vec e|) - f(e)\right] d^2\vec e'

417: \end{equation}

418: 

419: \noindent

420: The function $p(\vec e')$ describes the frequency per unit of 

421: eccentricity space ($d^2 \vec e'$) at which a binary with

422: eccentricity $\vec e$ is perturbed to the value $\vec e+\vec e'$.

423: Since there is no preferred direction for the encounters, this

424: function is axisymmetric, $p(\vec e') = p(e')$.  It is related to

425: ${\cal R}(e')$ by integrating over the angular 

426: direction of the phase space, ${\cal R}(e') = \int p(e') e' d \omega = 2 \pi e' p(e')$.

427: 

428: We first derive $p(e')$ for a simple scenario: a population of

429: perturbers each with mass $m_p$ and velocity $v_p$.  To clarify this

430: derivation, we present a qualitative treatment.  The eccentricity

431: excited by such a perturber with an impact parameter of order $b

432: \gg r_b$ is about $e' \sim (m_p/m_b)(v_b/v_p)(r_b/b)^2$ (Section

433: \ref{secSingleEnc}).  Since the frequency of encounters with impact

434: parameters $b$ is proportional to $b^2$, and the size of the

435: perturbation $e' \propto b^{-2}$, the frequency at which the binary is

436: perturbed by an amount of order $e'$ is therefore a power law: 

437: $e'^2 p(e') \propto e'^{-1}$.   

438: This power law is valid from very low

439: $e'$, caused by the farthest possible impulsive encounter, to $e' \sim

440: (m_p/m_b)(v_b/v_p)$, the rare encounters with $b \sim r_b$.  We take

441: into account the very rare occurrence of a physical collision, which

442: excite eccentricities of order $e' \sim (m_p/m_j)(v_p/v_b)$, in section

443: \ref{secCollisions}.

444: 

445: Evaluating equation \ref{eqRofe} using $\Delta \vec e (\vec v_p,\vec b,m_p)$

446: given by equation \ref{eqDeltae} provides the exact form of 

447: $p(e')$ for this scenario.

448: We find:

449: 

450: \begin{equation}

451: \label{eqPofE}

452: p(e') = %\frac{148.9}{32 \pi^3} G n_{\rm 3d} m_p T_{\rm orb} \frac{1}{e^3},

453: \frac{\langle C_e \rangle}{4 \pi} G \rho T_{\rm orb} \frac{1}{e'^3},

454: \end{equation}

455: 

456: \noindent 

457: where $T_{\rm orb}$ is the orbital period of the binary, and $\langle

458: C_e \rangle = 1.89$ is the average value of the angular terms of

459: equation \ref{eqDeltae} (see Appendix).   

460: We note that the frequency of perturbations depends not on

461: $m_p$, but only on the total mass density of perturbers.  It is also

462: independent of $v_p$, as the lowered effectiveness of the faster

463: perturbations is directly canceled by their higher frequency.

464: These properties are typical of distant encounters with 

465: binaries, as evident in earlier work on binary dynamics \citep{BHT85}.

466: 

467: We can generalize equation \ref{eqFofe} by including a term to

468: account for dissipation of the binary's eccentricity: $\partial f(\vec

469: e,t)/\partial t = {\rm -div}(f(\vec e,t) {\dot \vec e})$.  We restrict

470: our attention to mechanisms that reduce $\vec e$ at a timescale that

471: is independent of $\vec e$, ${\dot \vec e} = - \vec e / \tau_d$.  The

472: tidal dissipation of eccentricity obeys this form and is our main

473: motivation for including such terms.

474: 

475: Since $p(e')$ is a power-law, we can look for self-similar solutions

476: to the time-dependent integro-differential Boltzmann equation,

477: equation \ref{eqFofe}.  The frequency of perturbations $p(e')$ does

478: not depend on any special eccentricity, so the distribution function

479: should depend only on the time $t$.  We separate the

480: distribution function into three parts: the time-dependent

481: normalization, $F(t)$, the time-independent shape of the function,

482: $g(x)$, and the time-dependent eccentricity scale, $e_c(t)$.  These

483: quantities obey the relation $f(e,t)=F(t)g(e/e_c(t))$.  We choose the

484: normalization of $g(x)$ such that $\int g(x) d^2 x=1$.  We further

485: choose that $f(e,t)$ be normalized to 1 for all times; this constrains

486: the normalization function to be $F(t)=1/e_c(t)^2$.

487: 

488: Substituting $f(e,t)=e_c(t)^{-2}g(e/e_c(t))$ into equation \ref{eqFofe},

489: we find two equations.  The first specifies the time-independent

490: shape of the distribution as a function of the dimensionless 

491: parameter $x \equiv e/e_c(t)$:

492: 

493: \begin{equation}

494: \label{eqFofg}

495: 2 g(x)

496: +x \frac{d g(x)}{d x}

497: + \frac{1}{2 \pi} \int\int 

498: \frac{g(x_n)-g(x)}{|\vec x_n - \vec x|^3} d^2\vec {x}_n = 0,

499: \end{equation} 

500: 

501: \noindent 

502: The solution to this equation has been presented in several earlier

503: works, where we investigate the eccentricity distribution of the

504: oligarchs in a protoplanetary disk \citep{CS06,CSS07}:

505: 

506: \begin{equation}

507: \label{eqGofx}

508: g(x)=\frac{1}{2\pi} (1+ x^2)^{-3/2}.

509: \end{equation}

510: 

511: \noindent

512: This function is the two-dimensional Cauchy distribution.  

513: The median and mode of this distribution are

514: $x_{\rm med}=\sqrt{3}$ and $x_{\rm mode}=1/\sqrt{2}$.  The mean of

515: this distribution is formally divergent; assuming there is a maximum 

516: value of $x$, $x_u \gg 1$, then $x_{\rm mean}\approx 2.3 \log_{10} (0.74 x_u)$.

517: 

518: The eccentricity scale $e_c(t)$ is set by an ordinary differential 

519: equation,

520: 

521: \begin{equation}

522: \label{eqDiffeoft}

523: \dot e_c(t) = - e_c(t)/\tau_d + \langle C_e \rangle G \rho T_{\rm orb}/2 

524: %+ \sum_{i=1,2} \chi D_e G \frac{\rho}{\Lambda} T_{\rm orb}

525: %\left(\frac{ A v_p}{v_{\rm esc}}\right)^2 \frac{r_i}{r_b}.

526: \end{equation}

527: %+  G \rho T_{\rm orb} \left( \frac{C_e}{2} 

528: %+ \sum_{i=1,2} \frac{D_e}{\Lambda} \chi 

529: %\left(\frac{ A v_p}{v_{\rm esc}}\right)^2 \frac{r_i}{r_b}

530: 

531: \noindent

532: We note that $\tau_d$ and the terms on the right

533: hand side of equation \ref{eqDiffeoft} do not need to be constant in time;

534: evolution of the binary ($T_{\rm orb}(t)$), the perturbing swarm

535: ($\rho(t)$), or the damping mechanism ($\tau_d(t)$)

536: can be treated by including the time-dependence of these quantities.

537: 

538: We offer a reminder that $e_c(t)$ is the characteristic value of the

539: entire distribution of eccentricity that the binary may attain.  The

540: probability is highest that the binary will have an eccentricity near

541: the mode of the distribution, which is smaller than $e_c(t)$ by a

542: factor of 0.7.  The distribution is somewhat wide, and the confidence

543: levels around the median value are large.  The 66 percent confidence

544: interval of $x$ is $0.67-5.8$, and the 95 percent interval is

545: $0.23-40.0$.

546: 

547: Equations \ref{eqGofx} and \ref{eqDiffeoft} present a new picture of

548: the stochastic evolution of the binary's eccentricity.  Often the

549: evolution of a random variable is characterized by Brownian motion,

550: in which the distribution of the random variable is

551: set by the long term accumulation of many small perturbations.  The

552: typical value of such a variable grows as the square-root of time

553: (written $\sqrt{\langle x^2 \rangle} \propto t^{1/2}$), and the

554: probability of finding the system very far away from the typical value

555: is exponentially low.  The eccentricity of the binary evolves

556: differently.  The probability of finding the binary with an

557: eccentricity larger than $e_c(t)$ only diminishes as a power law

558: (equation \ref{eqGofx}).  Physically, this reflects the probability

559: that the binary received a single large perturbation to that state.

560: The characteristic eccentricity, $\sim e_c(t)$ corresponds to the size

561: of the perturbation that occurs with a frequency of about $1/t$.  The

562: linear growth of $e_c(t)$ demonstrated by equation

563: \ref{eqDiffeoft} reveals that the eccentricity of the binary does not

564: reflect the accumulation of many small perturbations, but the single

565: largest perturbation occurring in its history.  This kind of random

566: walk is called a ``L\'{e}vy flight'' \citep{SZF95}.

567: 

568: \subsection{Inclination}

569: \label{secBEinc}

570: 

571: The same analysis applies to the changes in angular momentum of the

572: binary.  Since $|\Delta \vec i| \sim |\Delta \vec e|$, it follows that

573: $p(i') \sim p(e')$.  The evolution of inclination differs only in the

574: coefficients that depend on the geometrical configuration of the

575: encounter.  The calculation of the coefficients is described in the

576: Appendix.  The self-similar distribution shape is a function of the

577: dimensionless variable $i/i_c(t)$, where $i_c(t)$ is the

578: time-dependent characteristic inclination.  The following equation

579: describes the evolution of $i_c(t)$:

580: 

581: \begin{equation}

582: \label{eqDiffioft}

583: \dot i_c(t) = - i_c(t)/\tau_{d,i}+ \langle C_i \rangle G \rho T_{\rm orb}/2

584: %+ \sum_{i=1,2} \chi D_i G \frac{\rho}{\Lambda} T_{\rm orb}

585: %\left(\frac{ A v_p}{v_{\rm esc}}\right)^2 \frac{r_i}{r_b},

586: \end{equation}

587: 

588: \noindent

589: where we have used $\tau_{d,i}$ to distinguish the timescale at which

590: the inclination of the binary is damped, and $ \langle C_i \rangle =

591: 0.75$, the average of the angular terms in equation \ref{eqDeltah}.

592: The inclination is always measured relative to the orbital plane at

593: $t=0$. The distribution given by equation \ref{eqGofx} then describes

594: the probability of the binary being inclined by $i=x~ i_c(t)$ relative

595: to its original orbital plane.

596: 

597: \section{A Spectrum of Colliding Perturbers}

598: \label{secMassSpec}

599: 

600: For many physical applications we must consider a range of perturbing

601: masses and velocities and the effects of collisions onto the binary.

602: In the single mass case discussed in section

603: \ref{secBEecc}, the interaction frequency $p(e')$ is set by the

604: likelihood that the binary encounters a perturber at the impact parameter that

605: causes such a change of $e'$.  For perturbers that have different

606: masses, the chance of experiencing a perturbation of magnitude $e'$

607: depends on the combined likelihood that the perturber has the required

608: impact parameter and the required mass to excite such a change.

609: 

610: To extend our analysis we set up several pieces of notation.

611: Assuming that the mass and velocity distributions are independent, we 

612: consider ${\cal F}(m_p,v_p) = {\cal F}_v(v_p) {\cal F}_m(m_p)$.  We

613: restrict our analysis to velocity distributions with a characteristic

614: value, $v_0$, such as a Gaussian distribution.  We consider systems

615: with differential mass spectra characterized by a power law: ${\cal

616:   F}_m(m_p) \propto m_p^{-\gamma}$, valid from a minimum mass $m_{\rm

617:   min}$ to a maximum $m_{\rm max}$.  These functions are consistent

618: with conditions in the Kuiper belt, where a power law mass spectrum

619: and roughly Gaussian velocity spectrum are observed \citep{LJ02}.

620: We define the differential mass spectrum by 

621: 

622: \begin{equation}

623: \label{eqMassSpec}

624: {\cal F}_m(m_p)=(n_0 (\gamma-1)/m_0) (m_0/m_p)^{\gamma}, 

625: \end{equation}

626: 

627: \noindent

628: where $n_0$ is the number density of bodies larger than mass $m_0$.

629: In the literature the differential size spectrum of Kuiper belt

630: objects is characterized as a power law in

631: radius with index $q$; this is related to our index by $\gamma =

632: (q+2)/3$.  In this section we discuss the $p(e')$ and $p(i')$ that

633: result from several values of $\gamma$.

634: 

635: \subsection{$\gamma<2$}

636: \label{secShallow}

637: 

638: The total mass density of perturbers for $\gamma < 2$ is dominated by

639: the perturbers with the largest mass, $m_{\rm max}$.  While

640: perturbations of size $e'$ are excited by all of the perturbers, the

641: most likely perturber to cause a perturbation of this strength is the

642: largest mass perturber.  The dynamics of the binary are then the same as described

643: in section \ref{secBEecc} with $m_p=m_{\rm max}$.  The power law of

644: $p(e') \propto e'^{-3}$, based on distant encounters, is valid up to

645: the eccentricity excited by a perturber of mass $m_{\rm max}$

646: interacting at a $b \sim r_b$, or for $e' \ll (m_{\rm

647:   max}/m_b)(v_b/v_p)$ (equation \ref{eqDeltae}).  It is necessary only

648: to know the total mass density $\rho$ of the perturbing swarm in order

649: to calculate the excitation frequency in this scenario, given by

650: equation \ref{eqPofE}.

651: 

652: \subsection{$\gamma=2$}

653: \label{secEq2}

654: 

655: The power law $\gamma=2$ describes a special mass distribution where

656: the frequency of encountering the few large perturbers at large impact

657: parameters is the same as encountering the more abundant smaller perturbers at

658: smaller impact parameters.  Thus each logarithmic interval in impact

659: parameter contributes the same amount to the frequency of perturbations by 

660: $e'$, $p(e')$.

661: The upper limit of impact parameters that can contribute to

662: excitations of a given $e'$, however, is given by the maximum mass

663: perturber.  The total range of contributing impact parameters then

664: diminishes as $e'$ approaches the eccentricity caused by the largest

665: perturber interacting with $b \sim r_b$, $e'_{\rm max}\equiv (m_{\rm

666:   max}/m_b)(v_b/v_0)$.  Mathematically this behavior is determined by

667: the integral of equation \ref{eqRofe}, which yields an excitation

668: frequency of:

669: 

670: \begin{equation}

671: \label{eqRdistanteq2}

672: p(e') = \frac{G n_0 m_0 T_{\rm orb}}{e'^3} 

673: \frac{\log \left( 2.1 (e'_{\rm max}/e') \right)\langle C_e \rangle}

674: {4 \pi},

675: \end{equation}

676: 

677: \noindent

678: for $e' \ll e'_{\rm max}$.  The

679: equivalent formula for the inclination excitations is:

680: 

681: \begin{equation}

682: \label{eqRIdistanteq2}

683: p(i') = 

684: \frac{G n_0 m_0 

685: T_{\rm orb}}{i'^3} \frac{\log \left((e'_{\rm max}/i') \right)\langle C_i \rangle}{4 \pi}.

686: \end{equation}

687: 

688: \noindent

689: For the smallest $e'$ and $i'$, the entire range of perturbing masses

690: contributes to the interaction frequency.  This occurs for excitations of the order

691: $(m_{\rm min}/m_b)(v_b/v_0)$, below which the perturbation frequency is given by 

692: equation \ref{eqPofE}. 

693: 

694: 

695: \subsection{$2<\gamma<3$}

696: \label{secInt}

697: 

698: The mass density of the perturbers when $2<\gamma<3$ is

699: dominated by perturbers of the smallest mass, $m_{\rm min}$.  Distant

700: encounters by perturbers with this mass produce very small

701: perturbations; for very low $e'$ then, $p(e') \propto e'^{-3}$, given

702: by the simple model of section \ref{secBEecc}.  The upper limit of

703: $e'$ caused by these perturbers interacting with impact parameters $b

704: \sim r_b$ is $e' \sim (m_{\rm min}/m_b)(v_b/v_p)$.  

705: 

706: Perturbers with

707: $m_{\rm min}$ cause eccentricity changes larger than this via close

708: encounters, but these encounters are less frequent than

709: interactions with perturbers of a higher mass and an impact parameter

710: of order $r_b$.  Perturbations with a strength $(m_{\rm

711:   min}/m_b)(v_b/v_p) \gg e' \gg (m_{\rm max}/m_b)(v_b/v_p)$ are most

712: often excited by perturbers with impact parameters of $\sim r_b$ and

713: masses $m \sim e' (v_p/v_b) m_b$.  In other words the frequency of

714: perturbations is directly proportional to the slope and normalization

715: of the mass spectrum.

716: 

717: In this case, the functions $p(e')$ and $p(i')$ cannot be determined

718: using the simplifications to equation \ref{eqDeltaegeneral} afforded

719: by very small or very large impact parameters.  In general, the

720: perturbation frequency for a mass spectrum of $2<\gamma<3$ follows the

721: power law $p(e') \propto e'^{-(\gamma+1)}$.  As an example we present

722: the perturbation frequency for $\gamma=25/12$.  This corresponds to

723: $q=4.25$, the best fit to observations of the Kuiper belt size

724: distribution presented by \citet{FKH08}.  We calculate from equation

725: \ref{eqRofe},

726: 

727: \begin{equation}

728: \label{eqPintermediate}

729: p(e')=2.6 \frac{G n_0 m_0 T_{\rm orb}}{e'^{37/12}} 

730: \left( \frac{m_0}{m_b} \frac{v_b}{v_0} \right)^{1/12}.

731: \end{equation}

732: 

733: \noindent

734: It is simple to understand the relationship between equations

735: \ref{eqPofE} and \ref{eqPintermediate} with the following argument.

736: A perturbation of size $e'$ that occurs via an interaction at a distance

737: $r_b$ requires a perturber of mass about $m' \sim e' (v_0/v_b) m_b$.

738: If we interpret the total density in equation \ref{eqPofE} as only

739: the density in bodies around $m'$, then $\rho' \sim m' {\cal F}_m(m') \sim

740: n_o (m_0/m')^{\gamma-1}$, and we recover the scaling of equation

741: \ref{eqPintermediate}.

742: 

743: The integral over

744: $b$ and the angular variables of equation \ref{eqDeltaigeneral} yield

745: a different coefficient for the perturbations to inclination:

746: 

747: \begin{equation}

748: \label{eqPofIintermediate}

749: p(i')= \frac{G n_0 m_0 T_{\rm orb}}{i'^{37/12}} 

750:  \left( \frac{m_0}{m_b} \frac{v_b}{v_0} \right)^{1/12}.

751: \end{equation}

752: 

753: \noindent

754: We relegate to the appendix the details of the integrals that 

755: produce the coefficients of equations \ref{eqPintermediate} and 

756: \ref{eqPofIintermediate}.

757: 

758: %\subsection{$\gamma>3$}

759: 

760: %For all $\gamma >3$, the distribution is given by a Gaussian.  This is 

761: %because the accumulation of the smaller excitations overwhelms the 

762: %rare largest excitations.  The eccentricity and relative inclination 

763: %then random walk diffusively, causing $e_c(t) \propto t^{1/2}$.

764: 

765: \subsection{Collisional Perturbations}

766: \label{secColRate}

767: 

768: The integral of equation \ref{eqRofe} over impact parameters from 0 to

769: $r_j$ produces the frequency of perturbations to the binary by

770: collisions on member $j$.  Since the size of the impulse from a

771: collision does not depend on the impact parameter, it is the mass of the

772: perturber that dictates the size of the eccentricity perturbation.

773: Accordingly, the frequency of perturbations as a function of $e'$

774: reflects the frequency of collisions as a function of $m_p$.  The frequency

775: of collisional perturbations does not depend on $m_{\rm max}$ or

776: $m_{\rm min}$ regardless of the slope.  However, the limits of the mass 

777: distribution specify the lowest and highest perturbations achievable 

778: via collisions: $\chi (m_{\rm min}/m_j)(v_0/v_b) \leq e' \leq \chi (m_{\rm max}/m_j)(v_0/v_b)$.

779: In this range of $e'$, for any value of $\gamma$, the

780: perturbation frequency due to collisions is

781: 

782: \begin{equation}

783: \label{eqPcol}

784: p(e') = \frac{G n_{0} m_b T_{\rm orb} }{e'^{\gamma+1}}

785: \left (\chi \frac{m_0}{m_j} \right)^{\gamma-1}

786: \left(\frac{v_0}{v_b}\right)^{\gamma} \left(\frac{r_j}{r_b}\right)^2

787: V_{\gamma} \frac{ (\gamma-1) \langle D_e^{\gamma-1}\rangle}{2 \pi},

788: \end{equation}

789: 

790: \noindent

791: where $\langle D_e^{\gamma-1} \rangle$ is the average of the angular

792: dependence of $\Delta \vec e$ from collisions to the power of

793: $\gamma-1$, and $V_\gamma \equiv v_0^{-\gamma} \int v_p^{\gamma+2}

794: {\cal F}_v(v_p) dv_p$.  If ${\cal F}_v(v_p)$ is proportional to a

795: delta function, $\delta(v_p-v_0)$, then $V_{\gamma}=1$ for all

796: $\gamma$.  If the velocity spectrum were Gaussian, such that ${\cal

797:   F}_v(v_p) \propto \exp (-(v_p/v_0)^2)$, then $V_\gamma=2

798: \Gamma((3+\gamma)/2)/\sqrt{\pi}$.  

799: The frequency of perturbations to the relative inclination by

800: collisions is the same as equation \ref{eqPcol}, replacing the

801: integrated coefficient $\langle D_e^{\gamma-1} \rangle$ with the

802: appropriate calculation made from the coefficients of $|\Delta \vec

803: i|$.

804: 

805: Although we use $r_j$ to represent either member of the binary, it is

806: clear from equation \ref{eqPcol} that the collisions onto the smallest

807: body have the largest effect on the orbit.  The ratio of the

808: perturbation frequency through collisions, $p(e')_{\rm collisions}$

809: (equation \ref{eqPcol}) to the frequency of gravitational scatterings,

810: $p(e')_{\rm gravity}$ (equation \ref{eqPintermediate}), is, for mass

811: distributions of $2 < \gamma <3$,

812: 

813: \begin{equation}

814: \label{eqRatio}

815: \frac{p(e')_{\rm collisions}}{p(e')_{\rm gravity}}

816: = 0.03 \left( \frac{r_j}{r_b} \right)^2

817: \left[\chi \frac{m_b}{m_j} \left( \frac{v_0}{v_b} \right)^2 \right]^{\gamma-1},

818: \end{equation}

819: 

820: \noindent

821: where we have evaluated the coefficients for $\gamma=25/12$.  The choice 

822: of $\gamma$ does not change these coefficients dramatically.

823: 

824: %The total frequency of perturbations to the binary is the 

825: %sum of the gravitational scattering and the collisional perturbation frequencies.

826: %The distribution set by both follows equation

827: %\ref{eqFofe} with the sum of both $p(e)$ in the scattering integrand.

828: %For $2<\gamma<3$ the dependence on eccentricity is cancelled as seen above

829: %in equation \ref{eqRatio}.  For shallower mass distributions, the 

830: %gravitational scattering rate, which goes as $e^{-3}$, is

831: %overwhelmed by the effects of collisions, whose eccentricity 

832: %dependence follows the mass distribution, $e^{-(\gamma+1)}$. 

833: 

834: %The contribution of collisions with this mass spectrum to the

835: %excitation rate, $p_{\rm col}(e)$, relative to that of gravitational

836: %interactions, $p_{\rm grav}(e)$, is given by the ratio

837: 

838: %\begin{equation}

839: %\label{eqGravVsCol}

840: %\frac{p_{\rm col}(e)}{p_{\rm grav}(e)}=\frac{2}{\Lambda} 

841: %\frac{\langle D_e \rangle}{\langle C_e \rangle} \chi

842: %\left( \frac{ A v_p}{v_{\rm esc}} \right)^2 \frac{r_i}{r_b}

843: %\end{equation}

844: 

845: %For the Pluto

846: %Charon binary collisions contribute about 10 percent of the

847: %excitation rate of distant encounters (see section \ref{secPluto}).  

848: %For the outer satellites that are at least 10 times smaller but have

849: %orbits larger by a factor of several, the collisions increase the 

850: %perturbation rate by about 25 percent.

851: 

852: \subsection{Eccentricity Distributions}

853: 

854: %The distribution given by equation \ref{eqGofx} and the equation to

855: %calculate $e_c(t)$ were derived in the context of $p(e') \propto

856: %e'^{-3}$.  The fact that $p(e')$ is a power law allows the distribution 

857: %function $f(e,t)$ to be self-similar; its shape then follows the same 

858: %power-law at high $e$, and its characteristic value $e_c(t)$ evolves 

859: %linearly with time.  It is these two properties that identify the 

860: %evolution of the eccentricity as a L\'{e}vy flight: a power-law 

861: %distribution function, and an average traversal of phase space

862: %that is faster than $t^{1/2}$ \citep{SZF95}.

863: 

864: The distribution given by equations \ref{eqGofx} and \ref{eqDiffeoft}

865: were derived in the context of $p(e') \propto e'^{-3}$.  As long as

866: $p(e')$ follows a power law with $e'$, we can write a self-similar

867: distribution function $f(e,t)$.  We write a generic function,

868: $p(e')=P_0 e'^{-(1+\eta)}$, to account for the different slopes caused

869: by different mass distributions (for $3 > \gamma > 2$, $\eta =

870: \gamma$; for $\gamma < 2$, $\eta=2$).  The derivation of the

871: distribution function proceeds analogously as in section

872: \ref{secBEecc}.  Equation \ref{eqFofe} becomes two equations: a

873: dimensionless integro-differential equation that specifies the shape,

874: and an ordinary differential equation to specify the evolution of the

875: eccentricity scale $e_c(t)$.  The general version of equation

876: \ref{eqDiffeoft} is:

877: 

878: \begin{equation}

879: \label{eqDiffeoftgeneral}

880: \dot e_c(t) = - e_c(t)/\tau_d + 2 \pi P_0 / e_c(t)^{\eta-2}. 

881: \end{equation}

882: 

883: \noindent 

884: In the limit of no eccentricity dissipation ($\tau_d \rightarrow

885: \infty$), equation \ref{eqDiffeoftgeneral} shows that $e_c(t) \propto

886: t^{1/(\eta-1)}$. For all of the $p(e')$ discussed in section \ref{secMassSpec},

887: the growth of $e_c(t)$ is always faster than $t^{1/2}$.

888: 

889: The shape of the distribution function is determined through a Fourier

890: transform of the general version of equation \ref{eqFofg}.  For slopes

891: of $1<\eta<3$, $g(x)=\int \cos (\vec k \cdot \vec x) \exp(-|\vec

892: k|^{\eta-1})d^2 \vec k $ \citep{S99,CSS07}.  While there is only a

893: closed form solution for $\eta=2$, given by equation \ref{eqGofx}, all

894: of these functions are flat at low $x$ and fall off like

895: $x^{-(\eta+1)}$.  In fact, it is easy to show from equation

896: \ref{eqFofe} that the high $e$ tail is given by

897: 

898: \begin{equation}

899: \label{eqHighTail}

900: f(e \gg e_c(t)) = p(e) t / (\gamma-1),

901: \end{equation}  

902: 

903: \noindent

904: when $t \ll \tau_d$.  For equilibrium distributions where $\dot

905: e_c(t)=0$, $t$ is replaced with

906: $\tau_d$, the timescale for the dissipation.

907: 

908: When $p(e') \propto e'^{-4}$ or steeper, the accumulation of the

909: smallest perturbations over time is more effective at raising the

910: eccentricity of the binary than single large perturbations.  In this

911: case, the evolution of the eccentricity follows standard Brownian

912: motion, where the distribution function is a Gaussian, and $e_c(t)

913: \propto t^{1/2}$.  

914: 

915: \section{Kuiper Belt Binaries}

916: \label{secKBB}

917: 

918: In this section we compute $e_c(t)$ and $i_c(t)$ for several Kuiper

919: belt binaries.  The ``binary'' of section \ref{secSingleEnc} now refers to 

920: a bound pair of Kuiper belt objects, and the ``perturbers'' are all of the 

921: other members of the Kuiper belt.

922: 

923: For the highest mass KBOs, the size spectrum is well determined to be

924: a power law with an index slightly greater than $q=4$.  The lowest

925: mass bodies, of about 30 km in radius, are less frequent than

926: predicted by a single power law, however the parameters of a more

927: general model are still under investigation

928: \citep{TB01,LJ02,PS05,FKH08,FH08}.  For this section we use the best

929: fit of a single power law model to the high mass part of the spectrum

930: provided by \citet{FKH08}, who find $q=4.25$ and a number density of 1

931: body per square degree brighter than magnitude 23.4.  We assume an

932: average distance of 40 AU to the Kuiper belt and a depth of 20 AU to

933: find a volumetric number density $n_0=3 \times 10^{-41} ~{\rm

934:   cm^{-3}}$.  To convert the magnitudes of the objects to physical

935: sizes, we assume a constant geometric albedo of 0.04, a constant

936: physical density of 1 g ${\rm cm^{-3}}$, and take the R-band apparent

937: magnitude of the Sun to be -27.6.  We find that the magnitude 23.4

938: corresponds to a mass $m_0=1.75\times 10^{21}~{\rm g}$, equivalent to

939: a radius of 75 km.  Most of the objects found between 30-50 AU are

940: inclined by about 5-15 degrees relative to the plane of the solar

941: system, and have heliocentric eccentricities of 0.1-0.2.

942: 

943: \subsection{Perturbations by a Disk}

944: \label{secDisk}

945: 

946: Our analysis so far has treated the perturbing bodies as unbound

947: objects moving relative to the binary with a constant velocity.  When

948: the perturbers are part of a disk orbiting the central star, the

949: orbital elements of the disk set the parameters of the perturbation

950: frequencies we calculate in section \ref{secBE}.

951: 

952: The relative velocity between KBOs, when they interact, is set by the

953: size of their eccentricities and inclinations, $v_p \sim e_{H} a

954: \Omega_H$, where the subscript ``H'' denotes a heliocentric orbital

955: quantity.  We assume a constant perturbing velocity with $v_p = 1~{\rm

956:   km/s}$, which corresponds to the typical heliocentric eccentricities

957: and inclinations of KBOs.  We assume that these encounters occur

958: isotropically in the frame of a binary, however this is not accurate.

959: A more detailed calculation of the angular distribution of relative

960: velocities will only affect the coefficients of the perturbations.

961: The disk does not specify a special direction for the perturbation vector

962: $\Delta \vec e$, so the perturbing frequency and the distribution

963: function retain their axisymmetry.  The influence of the central 

964: star on the binary and the perturbers adds

965: another constraint to our assumption of impulsive encounters: the

966: timescale for an interaction must be shorter than the orbital period

967: around the star: $b/v_p \ll 1/\Omega_H$, or equivalently, $b \ll e_H

968: a$.  This guarantees that the relative velocity is constant 

969: during the interaction.

970: 

971: If the orbit of the binary is much different than the typical KBO

972: orbit, there are several modifications to perturbation frequencies

973: experienced by the binary.  One modification is due to the finite

974: height of the disk of perturbers.  This height is set by their

975: inclinations around the central star; for the Kuiper belt we refer to

976: the average inclination as $\langle i \rangle_{KB}$.  A binary with

977: heliocentric inclination $i_{\rm CoM} \ll \langle i \rangle_{KB}$

978: never travels above or below the perturbing disk height and therefore

979: experiences the maximal frequency of perturbations.  If $i_{\rm CoM} \gg \langle

980: i \rangle_{KB}$, the binary spends most of its orbit outside of the

981: perturbing swarm.  The frequency of perturbations to such a binary is

982: reduced by the fraction of the time the binary leaves the disk,

983: proportional to $\langle i \rangle_{KB}/i_{\rm CoM}$.  The

984: eccentricity of the binary in the disk reduces the effective density

985: of perturbers in a similar manner if the epicycle of the binary

986: carries it outside of the region populated by perturbers.

987: 

988: If the heliocentric eccentricity or inclination of the binary is much

989: greater than the typical values for the Kuiper belt, the relative

990: velocity between the binary and a perturber is primarily due to 

991: the non-circular heliocentric motion of the binary.

992: Gravitational interactions depend weakly on $v_0$ so their frequency

993: does not change much in this case.  Perturbations by collisions,

994: however, become more important if $v_0$ is increased due to this

995: effect (equation \ref{eqRatio}).

996: 

997: \subsection{Pluto et. al.}

998: \label{secPluto}

999: 

1000: Pluto is the second largest known Kuiper belt object, with a radius of

1001: about 1100 km.  It has a semi-major axis of 39.5 AU and its orbit is

1002: inclined relative to the ecliptic by $17^{\circ}$.  Its largest

1003: satellite, Charon, contains about one tenth of the total mass of the

1004: system.  Recent observations have revealed two smaller satellites, Nix

1005: and Hydra \citep{W06}.  These satellites have small eccentricities and

1006: are roughly co-planar with Charon.  Numerical simulations of

1007: collisions between similarly sized objects by \citet{C05} produce

1008: binaries with orbits similar to Pluto and Charon.  The circularity and

1009: co-planarity of Nix and Hydra lend additional weight to a collisional

1010: origin of the system.

1011: 

1012: The triple system of Pluto and its moons is a valuable test case for

1013: the dynamics we have presented.  For an isolated binary it is

1014: impossible to know the initial orbital plane.  The relative

1015: inclinations of the moons of Pluto can be measured directly assuming

1016: their formation was co-planar.  Furthermore, the perturbing swarm for

1017: all three Pluto-moon pairs is the same.  A major issue in comparing

1018: our analytic calculations to the observations is that the large mass

1019: ratio of Charon to Pluto causes significant non-Keplerian effects in

1020: the orbits of the outer satellites.  We first re-examine the published

1021: observational model of their orbits to separate the relevant motion

1022: of the outer satellites from the forced motion due to Charon.  We then

1023: compare the resulting eccentricity with our predicted values.

1024: 

1025: \subsubsection{Orbital Model of Tholen et al}

1026: 

1027: A model of the observations of the Pluto system

1028: has been presented by \citet{TBGE07}, who fit the parameters of a

1029: four-body numerical integration such that the simulation agrees with

1030: the observations.  Such work is necessary, as it has been 

1031: shown that the observations cannot be consistently modeled by 

1032: three non-interacting two-body orbits \citep{W06}.

1033: 

1034: The model of \citet{TBGE07} presents a full set of osculating elements

1035: describing the orbits of Charon, Nix, and Hydra.  The orbit of Charon

1036: is virtually unaffected by Nix and Hydra; \citet{TBGE07} measure the

1037: eccentricity of Charon to be $3.48 \pm 0.04 \times 10^{-3}$, and the 

1038: period of its orbit is $6.387$ days. Since

1039: the combined potential of Charon and Pluto is significantly

1040: non-Keplerian, the elements of Nix and Hydra vary significantly during

1041: their orbits.  \citet{TBGE07} average the osculating semi-major axis

1042: to find an orbital period for these satellites of 25.49 days and 38.73

1043: days for Nix and Hydra respectively.  The osculating eccentricities of

1044: Nix and Hydra both oscillate between zero and about 0.2; for each

1045: satellite oscillations at the frequencies of its own orbit and

1046: that of Charon are visible (their figure 4).  The orbital planes of the 

1047: satellites relative to Charon's are tilted by 0.15 degrees for Nix and 

1048: 0.18 degrees for Hydra.  Each plane precesses relative to the plane of Charon, 

1049: however the angle of the offset remains constant.

1050: 

1051: \subsubsection{A Different Interpretation}

1052: \label{secInterpret}

1053: 

1054: \begin{figure}[t!]

1055: \center

1056: \includegraphics[angle=-90,width=0.8\columnwidth]{binaries-fig3.ps}

1057: \caption{The distance of Nix (lower panel) and Hydra (upper panel) from the Pluto-Charon barycenter, in units of Pluto radii, as a function of time, in 

1058: an integration of the parameters found by \citet{TBGE07}.  Nix and Hydra are treated

1059: as massless test particles.  The origin of the time coordinate is arbitrary.}

1060: \label{figRNixHydra}

1061: \end{figure}

1062: 

1063: 

1064: %The osculating Keplerian elements of Nix and Hydra are misleading.

1065: For two body motion, the Keplerian elements are constant and indicate

1066: the shape of the orbit in space.  Osculating elements that describe

1067: motion in significantly non-Keplerian potentials, such as the combined

1068: potential of Pluto and Charon, may vary on timescales shorter than the

1069: orbital period of the satellite.  When this is true, relating the

1070: osculating elements to the shape of the orbit can be misleading. 

1071: The average value of the osculating eccentricity of Nix is

1072: 0.015 in the model of \citet{TBGE07}, however the motion of Nix

1073: relative to Pluto never resembles an ellipse with such an

1074: eccentricity.

1075: 

1076: We re-examine the model provided by \citet{TBGE07} by reproducing the

1077: numerical integration based on the Pluto-centric positions and

1078: velocities of Charon, Nix, and Hydra published in their table 1.  We

1079: set the masses of Nix and Hydra to zero to eliminate their secular

1080: interactions with each other.  Instead of examining the

1081: osculating elements, we adopt the approach of \citet{LP06}

1082: and characterize the orbits of Nix and Hydra based

1083: on their position as a function of time from the Pluto-Charon

1084: barycenter, plotted in figure \ref{figRNixHydra}.  The units of distance

1085: are Pluto radii, defined as $R_P = 1147~{\rm km}$.  

1086: 

1087: Although short oscillations on the timescale of Charon are visible in

1088: the top panel of figure \ref{figRNixHydra}, they are very small

1089: compared to the oscillations that occur on the timescale of Hydra's

1090: orbital period.  To parametrize Hydra's orbit we fit the function

1091: $r_0(1+e \cos (\kappa_1 t + \omega_1))$ to the first 200 days of the

1092: numerical model.  Because for a non-Keplerian potential the radial

1093: epicyclic frequency differs from the orbital frequency, we calculate

1094: the average angular frequency by fitting a straight line to the

1095: angular position of Hydra as a function of time, $f(t) = \Omega_1

1096: t+\lambda_0$.  The results are written in table \ref{tabFitResults}.

1097: We interpret $e_1$ as the orbital degree of freedom in the combined

1098: potential of Pluto and Charon that is analogous to the eccentricity of

1099: a two-body orbit.

1100: 

1101: \begin{deluxetable}{lcccccccc}

1102: \tablecaption{Best fit values to the epicyclic models of the radial motion

1103: of Nix and Hydra.}

1104: \tablehead{

1105: \colhead{} & \colhead{$r_0/R_P$} & \colhead{$e_1$} 

1106: & \colhead{$2 \pi/\kappa_1$}& \colhead{$e_2$} 

1107: & \colhead{$2 \pi/\kappa_2$}&  \colhead{$e_3$} 

1108: & \colhead{$2 \pi/\kappa_3$}& \colhead{$2 \pi/\Omega_1$} \\

1109: \colhead{} & \colhead{} & \colhead{$\times 10^{-3}$} & \colhead{(days)} 

1110: & \colhead{$\times 10^{-3}$} & \colhead{(days)} 

1111: & \colhead{$\times 10^{-3}$} & \colhead{(days)} & \colhead{(days)}}

1112: \startdata

1113: Nix & 46.805(5) & 2.96(3) &  25.22(2) &

1114:  1.25(3) &  8.599(8) &  1.38(3) & 4.298(1) & 24.8505(5) \\

1115: Hydra &  62.237(1) &  5.595(2) & 38.535(15) & & & & & 38.20(1) \\

1116: \enddata

1117: \tablecomments{The motion of Nix is fit with three epicyclic 

1118: terms, while the motion of Hydra is only fit with one.  The parenthesis 

1119: indicate the 95 \% confidence level of the fit around the last digits.}

1120: \label{tabFitResults}

1121: \end{deluxetable}

1122: 

1123: The motion of Nix (bottom panel of figure \ref{figRNixHydra}) appears

1124: more irregular than that of Hydra. We find the position of Nix to be

1125: well-described by a model of three epicycles with different

1126: frequencies: $r(t)=r_0(1+\sum_{k=1,2,3} e_k \cos (\kappa_k t +

1127: \omega_k))$.  The best fit values are printed in table 1.  We

1128: distinguish the cause of each epicycle by its period.  The combined

1129: potential of Pluto and Charon oscillates with frequency of

1130: $\Omega_{\rm Charon}-\Omega_{\rm Nix}$; motion being forced by this

1131: potential should occur on integer multiples of this frequency.  

1132: Using the numbers in table \ref{tabFitResults}, we see that 

1133: $2 \pi/(\Omega_{\rm Nix}+\kappa_2) = 2\pi/(\Omega_{\rm Nix}+\kappa_3/2)=

1134: 6.39$ days.  The second and third epicycles in our fit correspond 

1135: to motion at the first and second harmonic of Nix's relative 

1136: orbital frequency.  We therefore interpret the first term, with a size of 

1137: $e_1=3 \times 10^{-3}$ and a period close to Nix's orbital period,

1138: as analogous to the two-body eccentricity.

1139: 

1140: We perform another integration of the best fit initial conditions from

1141: \citet{TBGE07} to investigate the secular effects between Nix and

1142: Hydra.  We use the best fit masses from \citet{TBGE07} for the two

1143: outer satellites.  Since the motion of Hydra is dominated by a single

1144: epicyclic frequency, the variation in the size of its epicycle is

1145: apparent on the timescale of several years.  To determine the effect

1146: of secular variations on Nix, we fit the same three-component

1147: epicyclic model to five orbits at $t \sim 5$ years.  In the best-fit model to these

1148: later orbits, the only difference compared to the model of table

1149: \ref{tabFitResults} is in $e_1$, the epicycle with a frequency close

1150: to Hydra's orbital frequency.  This is further confirmation that the

1151: degree of freedom represented by $e_1$ is analogous to the two-body

1152: eccentricity.  

1153: %The difference between the epicyclic motion we find for

1154: %Nix and Hydra, both of order $10^{-3}$, and the osculating

1155: %eccentricities that vary between 0 and 0.2 offer a strong warning

1156: %against the physical interpretation of osculating orbital elements.

1157: 

1158: \subsubsection{Theoretical Distribution}

1159: 

1160: To compute the distribution of eccentricities and inclinations

1161: expected of Pluto's moons, we solve equation \ref{eqDiffeoftgeneral}

1162: for each of the moons, given the interaction frequencies specified by

1163: equations \ref{eqPintermediate} and \ref{eqPofIintermediate}.  The

1164: only remaining parameters to evaluate are the damping timescales for

1165: the eccentricity and inclinations of each satellite.  We use the standard 

1166: formula for the damping of eccentricity due to the tidal 

1167: force of the primary acting on a secondary that is in synchronous 

1168: rotation \citep{YP81,MD99}:

1169: 

1170: \begin{equation}

1171: \label{eqTidalDamping}

1172: \tau_{d,2} = \frac{4}{63} Q_2 (1+{\tilde \mu_{2}}) \frac{m_2}{m_1}

1173: \left( \frac{r_b}{r_2} \right)^5 \frac{1}{\Omega},

1174: \end{equation}

1175: 

1176: \noindent

1177: where $Q_2$ is the dissipation function of the secondary, and 

1178: ${\tilde \mu_2} = 19 \mu r_2 / (2 \rho G m_2)$ 

1179: is its effective rigidity, a ratio between the 

1180: material strength of the secondary and its self-gravity. 

1181: The damping rate of

1182: eccentricity due to tides of the primary acting on the secondary,

1183: $\tau_{d,1}$, if the primary is also rotating synchronously with the

1184: orbit of the satellite, is given by equation \ref{eqTidalDamping} with

1185: the quantities specific to the primary switched with those of the

1186: secondary and vice versa.

1187: 

1188: Pluto and Charon are known to be in a double-synchronous state of

1189: rotation, where the spin period of each body is equal to the 6.4 day

1190: orbital period.  In many binaries, only the spin of the secondary is

1191: synchronous with the orbital period.  Tides on the primary then raise

1192: the eccentricity.  Double-synchronous systems, however, experience

1193: damping due to both the tides on the secondary and those on the

1194: primary.  Assuming a water-ice composition for Pluto 

1195: ($\mu=4~\times 10^{10}~{\rm dynes~cm^{-2}}$), we calculate the

1196: eccentricity damping timescale due to tides raised by Charon,

1197: $\tau_{d,1}$ from equation \ref{eqTidalDamping} to be 5.1 Myrs.  The

1198: shortest damping timescale due to tides from Pluto acting on Charon,

1199: $\tau_{d,2}$ is found by assuming Charon is also made of water-ice; we

1200: find in this case a timescale of 8.2 Myr.  The longest timescale

1201: assumes a rocky composition ($\mu=6.5~\times 10^{11}~{\rm dynes~cm^{-2}}$); 

1202: we find this corresponds to 133 Myr.  The

1203: overall damping of the system is given by the sum of the damping

1204: rates.  The short damping timescale of tides on Pluto prevents Charon

1205: from contributing significantly to the combined effect of both tides,

1206: reducing the importance of its composition.  The longest eccentricity

1207: damping timescale that results from both tides is 4.9 Myr.  The

1208: inclinations of the outer satellites relative to the Pluto-Charon

1209: plane are also damped by tidal dissipation. For a circular synchronous

1210: orbit the timescale for inclination damping is longer than the

1211: timescale for eccentricity damping by a factor of $\sim i^{-2}$.  We

1212: ignore the damping of inclinations in equation \ref{eqDiffioft} for

1213: all three satellites.

1214: 

1215: As discussed in \citet{TBGE07} and section \ref{secInterpret}, secular

1216: interactions between the satellites are visible in the long term

1217: calculations of their orbits.  For the best-fit values of the masses

1218: of Nix and Hydra, their eccentricities are modulated on the order of

1219: $10 \%$ over timescales of years; we neglect these fluctuations for

1220: this work.  It is more important in this model to determine whether

1221: secular evolution can cause the eccentricity of Nix or Hydra dissipate

1222: via Charon's orbit.

1223: 

1224: We use linear secular theory to describe the coupled evolution of the

1225: eccentricity and longitude of periapse of each satellite \citep{MD99}.

1226: We find that the undamped secular evolution agrees qualitatively with

1227: the numerical orbit determinations.  We add a term to the differential

1228: equations describing Charon's eccentricity that reduces it at a

1229: constant timescale ($\dot e_{\rm Charon} = - e_{\rm Charon}/\tau_d$).

1230: The frequencies of the oscillations of the eigenmodes of the solution

1231: are practically unchanged by this term, however each eigenmode gains a

1232: dissipative factor.  Quantitatively, only one eigenmode is damped on

1233: timescales shorter than than 4.5 Gyr.  By integrating the damped

1234: secular equations with different initial periapses, we determined that

1235: the secular interactions do not cause substantial damping of Nix and

1236: Hydra.

1237: 

1238: Equation \ref{eqRatio} gives the

1239: frequency of perturbations due to collisions of perturbers onto each

1240: moon relative to the frequency of perturbations caused by

1241: gravitational scattering, equation \ref{eqPintermediate}.  For Charon,

1242: the collisional perturbations increase $p(e)$ by only 2 percent.

1243: Since Nix and Hydra are smaller, perturbations by collisions have a

1244: greater relative effect; however it is only a 20 percent contribution

1245: to the total perturbation frequency for Nix and 15 percent for Hydra.

1246: We solve equation \ref{eqDiffeoftgeneral} to find $e_c(t)$ and

1247: $i_c(t)$ for each of Pluto's moons. 

1248: 

1249: For Charon we find $e_c=2.6 \times 10^{-6}$, and $i_c=0.029^{\circ}$.

1250: This value of $e_c$ corresponds to the most likely perturbation during

1251: a damping timescale of 4.9 Myr, and is much smaller than the

1252: observed value of $3.5 \times 10^{-3}$ \citep{TBGE07}.  Using equation

1253: \ref{eqHighTail}, we calculate that given this value of $e_c$, the

1254: probability of Charon's eccentricity being as high as its observed

1255: value is 0.2 percent.

1256: 

1257: For Nix we calculate $e_c(4.5 ~{\rm Gyr})=4.8\times 10^{-3}$ and

1258: $i_c(4.5 ~{\rm Gyr})=0.1^{\circ}$, and for Hydra, $7.1 \times 10^{-3}$

1259: and $0.15^{\circ}$ respectively.  The distributions specified by these

1260: values are quite consistent with the free eccentricity we determine in

1261: table \ref{tabFitResults}.

1262: 

1263: \subsection{Other Interesting KBOs}

1264: \label{secOthers}

1265: 

1266: Two other Kuiper belt objects have satellites on low eccentricity

1267: orbits: \el, and Eris.  Along with Pluto these are three of the four

1268: most massive KBOs known, all with radii of about 1000 km.

1269: \el has two known satellites.  The largest has a 50 day orbit and a

1270: measured orbital eccentricity of $0.050 \pm 0.003$ \citep{BBR05}.  An

1271: additional smaller satellite orbits \el~ with a period of about 35

1272: days \citep{BVB06}.  The orbital parameters of the inner satellite are

1273: unconstrained, however the relative inclination between the two is

1274: about $40^{\circ}$.  The masses of the satellites are negligible compared

1275: to the mass of \el.  The heliocentric inclination of the system is

1276: $28^{\circ}$.

1277: 

1278: \citet{BBR05} argue that if the tidal response of \el~ and its large

1279: satellite are fluid-like, tidal interactions should damp their

1280: eccentricity on a timescale of about 300 Myr.  With these parameters

1281: we use equation \ref{eqDiffeoftgeneral} 

1282: to calculate an equilibrium $e_c=4.3 \times 10^{-4}$.  The

1283: distribution with this eccentricity scale predicts an observed

1284: eccentricity of 0.05 at a probability of three percent.  However, for

1285: smaller bodies, internal elastic forces dominate the tidal

1286: deformation of their shape; it is more reasonable to assume that the

1287: tidal response of the satellite is characterized by its material

1288: strength.  Then, the tides raised on the primary have the greatest effect

1289: and the eccentricity of the system grows on the same timescale as the

1290: growth of the semi-major axis.  Forced eccentricity growth and an

1291: evolving orbital period can be incorporated into equation

1292: \ref{eqDiffeoftgeneral}.  However, these corrections are only an order

1293: unity correction since the growth timescale, by definition, is

1294: comparable to the age of the system.  Assuming $T_{\rm orb}$ is fixed

1295: and ignoring the eccentricity growth, we calculate $e_c(4.5~{\rm

1296:   Gyr})=0.0052$.  The 95 percent confidence interval around this $e_c$ is 

1297: 0.001-0.2; the observed eccentricity of \el is within this range.

1298: 

1299: The dwarf planet Eris is orbited by the satellite Dysnomia.

1300: Observations have shown an upper limit to their eccentricity of 0.013

1301: \citep{BVB06}.  The system has a 15 day orbital period, and orbits the

1302: sun at a semi-major axis of 67.7 AU with an eccentricity of 0.44 and a

1303: heliocentric inclination of $44^{\circ}$.  In addition to the

1304: reduction in effective perturbing density caused by the inclination,

1305: the high eccentricity reduces the effective perturber density by an

1306: additional factor of 0.09.  The semi-major axis of the binary is

1307: consistent with 4.5 Gyr of tidal evolution away from an initially very

1308: close orbit; if the tidal response of the secondary that of a

1309: strength-less fluid, then its eccentricity is damped on a timescale of

1310: 50 Myr.  These parameters yield an $e_c=2.2\times 10^{-6}$.  However,

1311: if the material strength of the secondary is stronger than its own

1312: self-gravity, then the tides raised on the primary cause the

1313: eccentricity of the satellite to grow.  In this case the relevant

1314: timescale is the age of the system, and we find that $e_c(4.5~{\rm

1315:   Gyr})=1.0\times 10^{-4}$.  Both values are below the observed upper

1316: limit.

1317: 

1318: In addition to the high mass ratio and low eccentricity Kuiper belt

1319: binaries, there are other known binaries of almost equal mass on

1320: moderately eccentric orbits.  The binary $1998~{\rm WW}_{31}$ is an

1321: example of such an object: both members have a radius of about 50 km,

1322: an orbital period of 574 days, and a mutual eccentricity is 0.817

1323: \citep{VPG02}.  Even though our analysis is derived in the

1324: low eccentricity limit, we can use equation \ref{eqDiffeoftgeneral}

1325: to estimate approximately the eccentricity expected from impulsive

1326: encounters; we find $e_c(4.5~{\rm Gyr}) = 0.31$.  This moderate

1327: characteristic eccentricity is consistent with the high observed

1328: value.  Other binaries with orbital periods on the order of a year

1329: will have acquired large eccentricities through their interactions

1330: with the other Kuiper belt objects.

1331: 

1332: \section{Other Binary Systems}

1333: \label{secApp}

1334: 

1335: Our analysis holds for any two-body orbit perturbed isotropically in

1336: the impulsive limit.  As binary orbits are prevalent in astrophysics,

1337: we briefly discuss several other examples.

1338: 

1339: The asteroid belt harbors many binaries with well determined

1340: eccentricities.  The mass spectrum of the asteroid belt, however, is

1341: much shallower than that of the Kuiper belt: the largest asteroid,

1342: Ceres, contains a third of the total mass of all asteroids.  A binary

1343: asteroid is then perturbed mostly by the largest objects that it

1344: encounters.  To calculate $p(e')$ accurately, it is necessary to model

1345: the neighborhood of that binary.  The asteroid belt is also

1346: collisionally active so its binaries may not be coeval with the whole

1347: solar system. We postpone a detailed analysis of the binary asteroid

1348: population for a future work.

1349: 

1350: A well-measured class of binaries outside the solar system are

1351: millisecond pulsars with white dwarf companions.  The tidal damping

1352: between the pulsar and its companion in the phase before the companion

1353: becomes a white dwarf is very short, indicating that during this phase

1354: the eccentricity of the binary should be smaller than the observed

1355: values of around $10^{-4}-10^{-5}$ \citep{S04}.  To explain the

1356: observations, \citet{P92} presents the following model.  As the

1357: companion star becomes a white dwarf, random fluctuations in the

1358: atmosphere of the star cause irregular motion in the orbit of the

1359: binary.  These motions are reflected by a small eccentricity that

1360: remains since the tidal interactions between the white dwarf and the

1361: neutron star cannot damp the system.  The model of \citet{P92} produces

1362: eccentricities for these systems that match the observations well.

1363: 

1364: These binaries are perturbed by encounters with other stars in the

1365: galaxy; we can calculate the contribution to their eccentricities by

1366: the distant stellar interactions.  The perturbation of these systems

1367: by other stars falls into the simple regime of only distant

1368: interactions described in section \ref{secBEecc}.  A typical

1369: volumetric mass density for field stars is $0.1 {\rm M_{\odot}~

1370:   pc^{-3}}$ \citep{HF00}.  Given this density, we calculate the

1371: characteristic eccentricity of these systems to be

1372: 

1373: \begin{equation}

1374: \label{eqEcpulsars}

1375: e_c(t)= 1.2 \times 10^{-9}  \left( \frac{T_{\rm orb}}{1~{\rm day}}\right)

1376: \left( \frac{t}{1~{\rm Gyr}} \right)

1377: \left( \frac{\rho}{0.1 M_{\odot}~{\rm pc}^{-3}}\right).

1378: \end{equation}

1379: 

1380: \noindent

1381: Typical orbital periods are between 1 and 10 days, and the ages of

1382: these systems are on the order of Gyrs.  We find then that $e_c(t)$ is

1383: several orders of magnitude lower than the observed eccentricities.

1384: \citet{P92} also concludes that the perturbations from other stars

1385: cannot be responsible for the eccentricities of the binary pulsars.

1386: Since we have calculated the distribution, however, we can estimate

1387: more accurately the likelihood of achieving these eccentricities

1388: by only distant stellar perturbations: less than 0.1 percent.

1389: 

1390: Globular clusters can have densities many orders of magnitudes higher than

1391: the average galactic density, such that distant perturbations to the 

1392: binaries may be important.  However, in a cluster the 

1393: interactions between a binary and a star are not

1394: typically in the impulsive interaction regime.  Instead the orbits of

1395: the perturbers are affected by the gravity of the binary, and the

1396: interactions occur over several orbital periods.  Analytic work on the

1397: eccentricity perturbations in this regime has been performed by

1398: \citet{RH95} and \citet{HR96}.

1399: 

1400: The characteristic eccentricity caused by distant stellar passages on

1401: the orbits of extra-solar planets is also given by equation

1402: \ref{eqEcpulsars}.  These eccentricities are too low to be

1403: reflected in the current sample of known extra-solar planets.  As with

1404: the pulsar binaries, the distant interactions may play a role in

1405: setting the eccentricity distribution of long period planets found in a dense

1406: stellar cluster.  For most extra-solar planets however, planet-disk 

1407: interactions \citep{GS03} or planet-planet scatterings \citep{RF96}

1408: are probably the source of their eccentricity.

1409: 

1410: \section{Conclusions}

1411: \label{secConclusions}

1412: 

1413: We have calculated the effects of impulsive perturbations and

1414: collisions on a nearly circular Keplerian orbit.  If the swarm of

1415: perturbers encounter the binary isotropically in space, we can write a

1416: distribution function that describes the probability density for the

1417: binary to have a given eccentricity or inclination relative to its

1418: initial plane. The growth rate of the binary's likeliest eccentricity 

1419: and inclination depends on the mass spectrum of the perturbers.  For 

1420: shallow mass distributions ($q<4$) it is the distant encounters that 

1421: set the binary's eccentricity and only the total mass density of perturbers

1422: is important to the evolution of the binary.  For steeper mass distributions of 

1423: $q=4-7$, it is the interactions at about the semi-major axis of the binary 

1424: that dominate the frequency of perturbations.  Only the normalization and 

1425: slope of the 

1426: mass spectrum set the distribution of eccentricities in this regime.

1427: 

1428: The assumptions of this model are valid in the Kuiper belt.  Our

1429: calculations match the observations of Nix and Hydra very well.  For

1430: Eris and \el, the observations lie within the 95 percent confidence

1431: intervals of the distributions we calculate, assuming the tidal

1432: response of the secondaries is dominated by material strength.  For

1433: Charon our theory is consistent with the numerical simulations of

1434: \citet{SBL03}, predicting an eccentricity about 3 order of magnitudes

1435: smaller than observed.  However, our analysis alleviates their need

1436: for numerical simulations as well as predicts the entire distribution

1437: of the eccentricity.  The distributions measured by \citet{SBL03} are

1438: not all correct as their model includes only impact parameters out to

1439: twice the semi-major axis.  In their simulations where $q=3.5$ and 4.0

1440: this excludes the impacts that are most relevant over an eccentricity

1441: damping timescale.  Our results show that for $q=3.5$ the interactions

1442: that dominate Charon's eccentricity are Pluto-sized perturbers

1443: interacting at about 200 times the semi-major axis!

1444: 

1445: %For Eris, we predict an eccentricity of the order $10^{-4}$, several

1446: %orders of magnitudes below current observational limits.  The main

1447: %satellite of the binary \el~ has an eccentricity much higher than

1448: %predicted by this model, indicating that another process has

1449: %interfered with its orbit.  The presence of an additional satellite

1450: %that is highly inclined may be evidence that its violent past

1451: %continues to dictate the dynamics of its satellites.

1452: 

1453: Even without eccentricity dissipation through tides, perturbations

1454: from other Kuiper belt objects are too weak to excite eccentricities

1455: of order 1 or inclination changes of order a radian for binaries that

1456: have orbital periods of a few days or weeks.  It is not likely that

1457: the orbital planes of the close binaries have been affected

1458: significantly by other Kuiper belt objects given our current

1459: understanding of the history of the Kuiper belt.  It falls on theories

1460: of binary formation to explain the distribution of orbital

1461: inclinations relative to the ecliptic for close binaries.  Since

1462: $e_c(t)$ grows faster for binaries with large orbital periods, it is

1463: plausible that the smaller wide binaries ($1998 {\rm WW}_{33}$ for

1464: example) have been brought to large eccentricities and inclinations by

1465: interacting with the rest of the Kuiper belt.

1466: 

1467: When many binaries share the same perturbing swarm, such as in the

1468: Kuiper belt, we can use the eccentricities of all the binaries to

1469: probe the properties of the entire system.  For example, if the mass

1470: spectrum is steeper than $q=4$, the distribution of 

1471: eccentricity is directly related to the slope and

1472: normalization of the mass spectrum. 

1473: Conversely, the observed eccentricity can be used to place limits on

1474: the damping timescale of a binary and therefore the rigidity of those

1475: bodies.  The small sample of Kuiper belt binaries with well measured

1476: eccentricities limits the current effectiveness of such a calculation.

1477: However, the Pan-STARRS project plans to detect around 20000 more

1478: members of the Kuiper belt \citep{KAB02}; from these the number of

1479: orbit-determined Kuiper belt binaries will surely increase.

1480: 

1481: The distribution we describe with equation \ref{eqGofx} is a special

1482: case of a L\'{e}vy distribution \citep{S99}.  This class

1483: of functions arise in the generalization of the central limit theorem

1484: to variables distributed with an infinite second moment.

1485: Alternatively, these functions can be characterized by the properties

1486: of the L\'{e}vy flight they describe.  For the eccentricity of the

1487: binaries discussed in this work, the frequency of a step is inversely

1488: proportional to a power of its size that depends on the

1489: mass spectrum of perturbers.  It follows that the largest single step

1490: dominates the growth from accumulated smaller steps, causing, in the

1491: absence of damping, the typical eccentricity to grow faster than in a

1492: normal diffusive random walk.  The slope of the distribution of

1493: excitations dictates the shape of the distribution.  This explains the

1494: coincidence of the distribution we derive in this work being exactly

1495: that of the distribution of eccentricity of protoplanets in a

1496: shear-dominated planetesimal disk, where the probability of changing the

1497: eccentricity of a protoplanet is inversely proportional to the size of

1498: that change \citep{CS06,CSS07}.

1499: 

1500: The authors thank Dmitri Uzdensky and Scott Tremaine for valuable

1501: discussions.  R.S. is a Packard Fellow and an Alfred P. Sloan Fellow.

1502: This research was partially supported by the ERC.

1503: 

1504: \appendix

1505: 

1506: To calculate the excitation rates presented in sections \ref{secBE}

1507: and \ref{secMassSpec}, it is necessary to integrate over all possible

1508: configurations of angles $\vec b$ and $\vec v_p$ relative to $\vec

1509: r_b$ and $\vec v_b$.  In this appendix we clarify the relation between

1510: the coefficients and equations \ref{eqDeltaegeneral} through

1511: \ref{eqDeltah}.

1512: 

1513: We choose spherical polar

1514: coordinates for $\vec b$ and $\vec v_p$ to integrate 

1515: equation \ref{eqRofe}.  This requires a

1516: polar and azimuthal angle for $\vec b$, $\theta_b$ and $\phi_b$, and a

1517: polar and azimuthal angle for $\vec v_p$, $\theta_v$ and $\phi_v$.

1518: By defining $\theta_v$ relative to $\vec b$, the requirement that

1519: $\vec b$ and $\vec v_p$ be perpendicular fixes $\theta_v = \pi/2$.

1520: 

1521: The magnitude of the perturbation only depends on the vectors

1522: $\vec b$ and $\vec v_p$ relative to $\hat r_b$ and $\hat v_b$, so we use

1523: these vectors and their cross product, $\hat n$ to describe the

1524: components of $\hat b$: $\hat b = b_r \hat r_b + b_v \hat v_b + b_n

1525: \hat n $.  The components are related to $\theta_b$ and $\phi_b$ in

1526: the typical way: $b_r=\cos \phi_b \sin \theta_b$, $b_v=\sin \phi_b

1527: \sin \theta_b$, and $b_n= \cos \theta_b$.  

1528: We define the components

1529: of $\vec v_p$ relative to the same unit vectors.

1530: The angle $\phi_v$ describes the direction of $\vec v_p$ in

1531: the plane given by $\hat b$; the components of $\vec v_p$ follow

1532: from a rotation of this plane to align with $\hat n$.

1533: We find the relations:

1534: %\begin{eqnarray}

1535: %\label{eqVconversion}

1536: %v_r & =& \cos \theta_b \cos \phi_v - ( \sin^2 \theta_b \sin \phi_b ) ( \cos \phi_b \sin \phi_v - \sin \phi_b \cos \phi_v) / (1+\cos \theta_b), \nonumber \\

1537: %v_v & =& \cos \theta_b \sin \phi_v + (\sin^2 \theta_b \cos \phi_b)  (\cos \phi_b \sin \phi_v - \sin \phi_b \cos \phi_v) / (1+\cos \theta_b),\\

1538: %v_n& =& -\sin \theta_b (\cos \phi_b \cos \phi_v + \sin \phi_b \sin \phi_v). \nonumber

1539: %\end{eqnarray}

1540: 

1541: \begin{eqnarray}

1542: \label{eqVconversion}

1543: v_r & =& b_n\cos \phi_v 

1544: - b_v ( b_r \sin \phi_v - b_v \cos \phi_v) / (1+b_n), \nonumber \\

1545: v_v & =& b_n \sin \phi_v +  b_r (b_r \sin \phi_v - b_v \cos \phi_v) 

1546: / (1+ b_n),\\

1547: v_n& =& - b_r \cos \phi_v - b_v \sin \phi_v. \nonumber

1548: \end{eqnarray}

1549: 

1550: The coefficient from equations \ref{eqPofE} and \ref{eqRdistanteq2},

1551: $\langle C_e \rangle$, is defined to be the integral of   

1552: $|\Delta \vec e|/(8 \pi^2 (m_p/m_b)(v_b/v_p)(r_b/b)^2)$ as given by

1553: equation \ref{eqDeltae}:

1554: 

1555: \begin{equation}

1556: \label{eqAvgCe}

1557: \langle C_e \rangle = 

1558: \frac{1}{4 \pi^2} \int_0^{2 \pi} \int_0^{2 \pi} \int_{0}^{\pi} 

1559: \left[ (4 b_r b_v +2 v_r v_v)^2+(1-v_r^2-2 b_r)^2

1560: \right]^{1/2} \sin \theta_b d\theta_b d\phi_b d\phi_v =1.89

1561: \end{equation}

1562: 

1563: We similarly define $\langle C_i \rangle$ from equation \ref{eqDeltah}:

1564: 

1565: \begin{equation}

1566: \label{eqAvgCi}

1567: \langle C_i \rangle = 

1568: \frac{1}{4 \pi^2} \int_0^{2 \pi} \int_0^{2 \pi} \int_{0}^{\pi} 

1569: | 2 b_r b_n+ v_r v_n | \sin \theta_b 

1570:  d\theta_b d\phi_b d\phi_v = 0.75.

1571: \end{equation}

1572: 

1573: To calculate the coefficients used in the collisional excitation rate,

1574: equation \ref{eqPcol}, we use the $|\Delta \vec e|$ discussed in 

1575: section \ref{secCollisions}.

1576: 

1577: \begin{equation}

1578: \label{eqAvgDe}

1579: \langle D_e^{\gamma-1} \rangle = 

1580: \frac{1}{4 \pi^2}  \int_0^{2 \pi} \int_0^{2 \pi} \int_{0}^{\pi}

1581: (4 v_v^2+v_r^2)^{(\gamma-1)/2}

1582: \sin \theta_b  d\theta_b d\phi_b d\phi_v

1583: \end{equation}

1584: 

1585: \noindent

1586: For $\gamma=2$, the integral has a closed form solution of 

1587: $\langle D_e \rangle = E(-3)$, the complete Elliptic integral.

1588: For the inclination,

1589: 

1590: \begin{equation}

1591: \label{eqAvgDi}

1592: \langle D_i^{\gamma-1} \rangle

1593: = \frac{1}{4 \pi^2} \int_0^{2 \pi} \int_0^{2 \pi} \int_{0}^{\pi} 

1594: |(v_z)^{\gamma-1}| 

1595: \sin \theta_b  d\theta_b d\phi_b d\phi_v = \frac{1}{\gamma}

1596: \end{equation}

1597: 

1598: The coefficients for the excitation rates in the regime of 

1599: $2<\gamma<3$ are more complicated as the dependence on 

1600: $b/r_b$ cannot be factored out of the coefficient.  In addition to 

1601: integrating over all angles, we must integrate over impact parameter.

1602: For any $\gamma$, equation \ref{eqPintermediate} is written:

1603: 

1604: \begin{equation}

1605: \label{eqPIntGeneral}

1606: p(e) = \frac{G n_0 m_0 T_{\rm orb}}{e^{\gamma+1}}

1607: \left( \frac{m_0}{m_b} \frac{v_b}{v_0} \right)^{\gamma-2}

1608: \frac{\gamma-1}{2 \pi} V_{2-\gamma} \langle A_e^{\gamma-1} \rangle,

1609: \end{equation}

1610: 

1611: \noindent

1612: where $V_{\gamma-2}$ is discussed in section \ref{secColRate}; for a

1613: Gaussian distribution of perturber velocities, $V_{\gamma-2}=2

1614: \Gamma((1+\gamma)/2)$.  The term $\langle A_e^{\gamma-1} \rangle$

1615: again contains the angular information.  Excitations for $2 < \gamma <

1616: 3$ are most important at $b \sim r_b$ so we can not assume that $\vec

1617: b_2 \approx \vec b$. We introduce explicit notation for the the

1618: components of the unit vector $\hat b_2 = b_{2,r} \hat r_b + b_{2,v}

1619: \hat v_b + b_{2,n} \hat n$.  Then the angular average coefficient is:

1620: 

1621: \begin{equation}

1622: \label{eqAvgAe}

1623: \langle A_e^{\gamma-1} \rangle =

1624: \frac{1}{8 \pi^2} 

1625: \int_0^{2 \pi} \int_0^{2 \pi} \int_{0}^{\pi} \int_{0}^{\infty}

1626: \left[ 

1627: 16 \left(\frac{b_{2,v}}{x_2}-\frac{b_v}{x_1} \right)^2 

1628: + 4 \left(\frac{b_{2,r}}{x_2}-\frac{b_{r}}{x_1}\right)^2

1629: \right]^{(\gamma-1)/2} x_1 \sin \theta_b d x_1 d \theta_b d \phi_b d \phi_v,

1630: \end{equation}

1631: 

1632: \noindent

1633: with $x_1=b/r_b$ and $x_2=b_2/r_b$.  The magnitude and components 

1634: of $\vec b_2$ are related to $\vec b$ and $\vec v_p$ as described 

1635: in section \ref{secSingleEnc}: 

1636: $\vec b_2 = \vec b - \vec r_b + \hat v_p (\vec r_b \cdot \hat v_p)$.

1637: For $\gamma=25/12$ as discussed in \ref{secInt}, 

1638: $\langle A_e^{13/12} \rangle \approx 15$.  For other $\gamma$ between 

1639: 2 and 3, this factor is of the same order, 10-15.

1640: 

1641: \bibliographystyle{apj}

1642: \bibliography{ms}

1643: 

1644: \end{document}

1645: