1: \def\etal{{\it et al.\thinspace}}
2: \def\eg{{\it e.g.\ }}
3: \def\etc{{\it etc.\ }}
4: \def\ie{{\it i.e.\ }}
5: \def\cf{{\it c.f.\ }}
6: \def\gsim{~\rlap{$>$}{\lower 1.0ex\hbox{$\sim$}}}
7: \def\lsim{~\rlap{$<$}{\lower 1.0ex\hbox{$\sim$}}}
8: \def\v#1{{\bf #1}}
9: \def\u#1{{\mbox{[#1]}}}
10: \def\bdelta{{\pmb \delta}}
11: \def\bphi{{\pmb \phi}}
12: \def\deg{{\rm o}}
13: \def\idm#1{{\mbox{\scriptsize #1}}}
14: \def\D#1{\frac{\mbox{d}\,#1}{\mbox{d\,t}}}
15: \def\Dtau#1{\frac{\mbox{d}^2\,#1}{\mbox{d}\,t^2}}
16: \newcommand\Ym{\langle Y\rangle}
17: \newcommand\Chi{{(\chi^2_\nu)^{1/2}}}
18: \def\astrobj#1{#1\ }
19: \def\bibcode#1{(\texttt{#1})}
20: %\def\url#1{\texttt{#1}}
21: \newcommand\pstar{{\astrobj{HD~74156}}}
22:
23: \documentclass[12pt,preprint]{aastex}
24: %\usepackage{natbib}
25: %\usepackage{setspace}
26: %\doublespacing
27:
28: \begin{document}
29:
30: \title{The Successful Prediction of the Extrasolar Planet HD 74156 d}
31:
32: \author{Rory Barnes\altaffilmark{1}, Krzysztof Go\'zdziewski\altaffilmark{2}, Sean N. Raymond\altaffilmark{3,4}}
33:
34: \altaffiltext{1}{Lunar and Planetary Laboratory, University of Arizona,
35: Tucson, AZ 85721, rory@lpl.arizona.edu}
36: \altaffiltext{2}{Toru\'n Centre for Astronomy, N. Copernicus University, Toru\'n, Poland}
37: \altaffiltext{3}{NASA Postdoctoral Program Fellow}
38: \altaffiltext{4}{Center for Astrophysics and Space Astronomy, University of Colorado, Boulder, CO 80309}
39:
40: \keywords{methods: N-body simulations, stars: planetary systems, stars individual: HD 74156}
41:
42: \begin{abstract}
43: Most of the first-discovered extrasolar multi-planet systems were found to lie
44: close to dynamically unstable configurations. However a few observed
45: multi-planet systems (\eg HD 74156) did not show this trait. Those systems could share this property if they contain an additional planet in between those that are
46: known. Previous investigations identified the properties of hypothetical planets that would place
47: these systems near instability. The hypothetical planet in HD 74156 was
48: expected to have a mass about equal to that of Saturn, a semi-major axis
49: between 0.9 and 1.4 AU, and an eccentricity less than 0.2. HD 74156 d, a
50: planet with a mass of 1.3 Saturn masses at 1.04 AU with an eccentricity of
51: 0.25, was recently announced.
52: We have reanalyzed all published data on this system in order to
53: place tighter constraints on the properties of the new planet. We find two
54: possible orbits for this planet, one close to that already identified and
55: another (with a slightly better fit to the data) at $\sim 0.89$ AU. We
56: also review the current status of other planet predictions, discuss the
57: observed single planet systems, and suggest other systems which may
58: contain planets in between those that are already known. The confirmation
59: of the existence of HD 74156 d suggests that planet formation is an
60: efficient process, and planetary systems should typically contain many
61: planets.
62: \end{abstract}
63:
64: \section{Introduction}
65:
66: The detection of extrasolar planets has provided an unprecedented
67: opportunity to test models of planet formation. One such model is the
68: ``Packed Planetary Systems'' (PPS) hypothesis (Barnes \& Raymond 2004 (hereafter RB04); Raymond \& Barnes 2005 (hereafter RB05); Raymond \etal
69: 2006; Barnes \& Greenberg 2007; see also Barnes \& Quinn 2001, 2004)
70: which posits that, between the innermost and outermost giant planets,
71: planetary systems formed such that they were filled to capacity; an
72: additional planet would create an unstable system. Other
73: investigations gave consistent results (\eg Rivera \& Lissauer 2000;
74: Bois \etal 2003; Go\'zdziewski \etal 2006). These analyses were
75: grounded in considerable theoretical work to examine the (in)stability
76: of planetary systems (\eg Chambers \etal 1996; Weidenschilling \&
77: Marzari 1996; Ida \& Lin 1997).
78:
79: The PPS hypothesis was suggested because five of the first six known
80: extrasolar multi-planet systems showed evidence for packing:
81: $\upsilon$ And (Butler \etal 1999), 47 UMa (Fischer \etal 2002), GJ
82: 876 (Marcy \etal 2001b), HD 82943 (Mayor \etal 2004), and HD 168443
83: (Marcy \etal 2001a) were packed (Barnes \& Quinn 2001, 2004; BR04),
84: but HD 74156 (Naef \etal 2004) was not. This high frequency of packing
85: suggests that it must be a common feature of planetary systems. The
86: PPS hypothesis takes this suggestion a step further and proposes that
87: all planetary systems have tended to form dynamically
88: full.
89:
90: For the PPS model to be correct, then seemingly non-packed systems
91: require at least one additional planet to fill them up. This
92: conjecture led BR04 and RB05 to search for stable regions in between
93: known planets in several systems, including HD 74156. These
94: investigations presumed that, by identifying a stable region in
95: between the known planets, it would be possible to predict the mass
96: and orbit of a previously unknown planet. BR04 explained the reasons to
97: expect additional planets and outlined how to predict planets
98: through numerical simulations of test particles. RB05 then placed
99: massive companions in the gaps identified in BR04 in order to predict
100: the likely orbital elements of planets that could be detected. Note
101: that BR04 used the initially reported orbital elements and found HD
102: 74156 was packed. RB05 used the revised orbits, published in Naef
103: \etal (2004), and found the system was not packed.
104:
105: An additional planet would require a mass and orbit that guaranteed
106: dynamical stability for the age of the system. The greatest difficulty
107: with such an analysis is that the size of the stable zone and the mass
108: of the hypothetical planet are degenerate; lower mass planets have a
109: wider stable zone, larger planets have a narrower zone. In order to
110: estimate the mass, RB05 considered the mass of the two known
111: planets. As both these planets are larger than Jupiter (2 and 8
112: Jupiter masses), then a planet in between the two would also need to
113: be large in order for the protoplanetary disk to have a plausible
114: surface density profile (\eg a power law). At the same time, the mass
115: would have to be small enough to have avoided detection by the initial
116: observations. In order to help quantify the predicted mass of the
117: hypothetical planets, RB05 considered Saturn-mass, Jupiter-mass and 10
118: Jupiter-mass objects. They found the HD 74156 system had an 80\%
119: probability of being able to support a Saturn-mass planet, a 40\%
120: probability for a Jupiter-mass planet, and a 10\% probability of 10
121: Jupiter-mass planet. The larger masses would likely have been detected
122: previously, so RB05 settled on a Saturn mass as the likely mass of the
123: putative companion.
124:
125: Bean \etal (2008) discovered HD 74156 d, an extra-solar planet with a mass
126: 40\% greater than Saturn, at 1.04 AU with an eccentricity of
127: 0.25. This planet is consistent with the PPS model: RB05 predicted a
128: Saturn-mass planet with semi-major axis, $a$, in the range $0.9 \lsim
129: a \lsim 1.4$ AU and eccentricity, $e$, in the range $0 \le e \lsim
130: 0.2$. Here we reanalyze all available data for HD 74156
131: ($\S$ 2), discuss the current status of the PPS model in
132: $\S$ 3, and draw general conclusions in $\S$ 4.
133:
134: \section{The Orbit of HD 74156 d}
135: The best-fit orbits published by Bean \etal (2008) for the planets
136: orbiting HD 74156 are actually unstable on $\sim 10^5$ year
137: timescales. The instability arises due to strong gravitational
138: interactions resulting from close approaches between the inner two
139: planets, b and d. In this section we consider all published
140: observations and identify four plausible stable fits to the data.
141:
142: Currently, all published radial velocity (RV) data for \pstar{} appear
143: in papers by Naef \etal (2004) (51 observations gathered with the
144: ELODIE spectrometer with a mean dispersion of 12.7~ms$^{-1}$, and 44
145: measurements of CORALIE with a mean dispersion of 8.5~ms$^{-1}$), and
146: in Bean \etal (2008), who published 82 precision measurements from the
147: Hobby-Eberly Telescope (HET), with mean dispersion of
148: 2.7~ms$^{-1}$. The combined data set covers $\sim 9.33$~yr. We binned
149: a few measurements in the CORALIE and ELODIE data sets which were done
150: during one night. We also shifted ELODIE and CORALIE observations with
151: respect to the mean RV in both set. We rescaled data
152: errors by adding jitter of $4$~ms$^{-1}$ in quadrature. In all
153: calculations, we adopted the mass of the parent star to be
154: $1.24$~M$_{\sun}$, following Naef \etal (2004). This procedure is
155: slightly different than (but consistent with) that in Bean \etal (2008).
156:
157: First, we searched for the best Keplerian fits to the combined data
158: set with the hybrid optimization code (Go\'zdziewski \& Migaszewski
159: 2006) relying on the genetic algorithm (Charbonneau 1995).
160: Because the number of planets in the system is unknown, we considered
161: models with two, three and four planets. The so-called velocity
162: offsets of the telescopes are the free parameters in the model. Hence,
163: the $N$-planet model depends on $5N+3$ primary parameters, \ie tuples
164: ($K,P,e,\omega,\tau$) comprising of velocity amplitude $K$ in
165: ms$^{-1}$, orbital period $P$ in days, eccentricity, longitude of
166: pericenter $\omega$ in degrees, and time of periastron passage $\tau$
167: in JD-245000, for each planet, respectively, as well as the
168: offsets. The fit quality is measured in terms of reduced $\Chi$ and an
169: rms.
170:
171: The best fit to the two-planet model yields $\Chi = 1.63$ and an rms
172: $= 11.5$~ms$^{-1}$. The four-planet model could not converge on any
173: solution in which the RV contribution from all planets exceeded the
174: mean uncertainty of the measurements (assuming detectable masses), and
175: therefore could not produce any reasonable fit to the data. However,
176: as we will see below, the three-planet model gives statistically
177: better fits to the data, so we will focus on it.
178:
179: We found the best three-planet fit yields $\Chi = 1.28$ and drops
180: the rms to $= 9.4$~ms$^{-1}$, which represent a significant
181: improvement over those of the two-planet fit. The three-planet fit,
182: given in terms of parameter tuples is the following: (113.296, 51.641,
183: 0.640, 175.318, 1671.231) for planet~b, (13.226, 277.378, 0.351,
184: 294.266, 4221.105), for the new planet~d, and (108.5127, 2433.177,
185: 0.331, 275.455, 3497.965) for planet~c. The velocity offsets are
186: $(-2.94, 3.66, -6.01)$~ms$^{-1}$ for ELODIE, CORALIE and HET,
187: respectively. Curiously, the period of the new planet~d is $\sim
188: 277$~days, and is significantly different from the best-fit solution
189: quoted in Bean et al. (2008), who report $P_d = 347$~days. This
190: discrepancy most likely results from our different handling of the
191: errors, as well as the inclusion of 10 additional data
192: points. However, we also find a similar local minimum at $P_d =
193: 349$~days (close to the fit in Bean \etal 2008), as well as $P_d =
194: 900$~days, but with large $e_d$ (0.6).
195:
196: Because $e_{\idm{d}}$ can reach large values in the Keplerian fit we
197: cannot assume that the best-fit configurations are dynamically
198: stable. To account for the mutual interactions and test for stability,
199: the ensemble of Keplerian solutions has been used as initial
200: conditions in an $N$-body model (Laughlin \& Chambers 2001). The
201: Keplerian fits are not a good representation of the system because
202: they lead to significant degradation of $\Chi$. Hence, we refined
203: these Keplerian fits to account for the mutual interactions. In
204: addition, we tested formal stability (as understood through chaotic or
205: quasi-periodic character of orbital configurations) of these refined
206: solutions with MEGNO (Cincotta \& Sim\'o 2000; Cincotta, Giordano \&
207: Sim\'o 2003; Go\'zdziewski \etal 2001). MEGNO has been computed over
208: $3\cdot 10^4$ orbital periods of the outermost planet which is long
209: enough to account for the destabilizing influence of short-term mean
210: motion resonances.
211:
212: The dynamical analysis reveals two narrow strips of dominant solutions
213: around $a_d = 0.89$ AU (``Fit 1'', $\Chi = 1.30$ with rms 9.55
214: ms$^{-1}$) and 1.02~AU (``Fit 2'', $\Chi = 1.32$ with rms 9.58
215: ms$^{-1}$) (see Fig.\ \ref{fig:fit}), with some hint of a solution
216: around 2~AU. These are the Newtonian fits to the data. Table 1
217: presents these two fits, where $m$ is mass, $M$ is mean anomaly and
218: $T_0$ is the epoch of the first observation, JD~2,445,823.5570. To be
219: certain that the MEGNO signature is well time-calibrated (in the sense
220: that it will identify instabilities generated by strong mean motion
221: resonances), we integrated the best 16 solutions with the
222: Bulirsh-Stoer integrator from the MERCURY package (Chambers 1999) over
223: 100~Myr each. All survived without any noticeable change of the
224: regular character. Fig.\
225: \ref{fig:stable} shows stable and chaotic regions near the two best
226: stable fits.
227:
228: \begin{center}Table 1 - Best-fits Orbits and Masses for the Planets of HD 74156
229: \end{center}
230: \begin{center}
231: \begin{tabular}{ccccccc}
232: \hline\hline
233: Fit & Planet & $m$ (M$_{Jup}$) & $a$ (AU) & $e$ & $\omega$ ($^\circ$) & $M(T_0)$ ($^\circ$)\\
234: \hline
235: 1 & b & 1.847 & 0.292 & 0.635 & 175.32 & 210.90 \\
236: & d & 0.396 & 0.892 & 0.240 & 226.50 & 334.00 \\
237: & c & 7.774 & 3.822 & 0.361 & 272.66 & 328.06 \\
238: \hline
239: 2 & b & 1.847 & 0.292 & 0.629 & 176.45 & 211.64\\
240: & d & 0.412 & 1.023 & 0.227 & 191.81 & 67.51\\
241: & c & 7.995 & 3.848 & 0.426 & 262.17 & 340.20\\
242: \hline
243: 3$^a$ & b & 1.882 & 0.292 & 0.640 & 175.37 & 210.88\\
244: & d & 0.407 & 0.893 & 0.281 & 228.76 & 334.81\\
245: & c & 7.830 & 3.818 & 0.363 & 272.364 & 327.39\\
246: \hline
247: 4$^a$ & b & 1.882 & 0.292 & 0.635 & 176.20 & 211.55\\
248: & d & 0.409 & 1.022 & 0.262 & 177.66 & 78.73\\
249: & c & 8.176 & 3.853 & 0.427 & 261.25 & 341.43\\
250: \end{tabular}
251: \end{center}
252: $^a$ All planets in this fit have an inclination of $80^\circ$.\\
253:
254: \begin{figure}
255: \includegraphics{fig1.eps}
256: \caption{Fit parameters of the coplanar Newtonian model to the data published in
257: Naef \etal (2004) and Bean \etal (2008). Osculating elements at the epoch of the first observation are projected onto the $(a_{\idm{d}},e_{\idm{d}})$-plane. Gray curves are the collision lines of the orbits computed with the elements of the
258: innermost and the outermost companions fixed at their best fit values. The best fit solution (Fit 1) is marked by the intersection of the horizontal and vertical lines. For reference, the best fit (unstable) configuration is marked with a white crossed circle. Light-gray circles are for fits with$\Chi<1.35$ (an rms limit of $\sim 10$~ms$^{-1}$). Darker gray circles are for
259: solutions with $\Chi<1.33$ and $\Chi<1.31$, respectively. Stable fits are
260: marked with red circles. }
261: \label{fig:fit}
262: \end{figure}
263:
264: \begin{figure}
265: \includegraphics{fig2.eps}
266: \caption{The dynamical map for configurations near Fit 2 in terms of the Spectral Number
267: $\log (SN)$ (Michtchenko \& Ferrraz-Mello 2001). The character of the solutions is color-coded: yellow means chaotic systems
268: and black means regular systems. Fit 2 is marked with a crossed
269: circle. Most significant mean motion resonances
270: between the two innermost planets are labeled.}
271: \label{fig:stable}
272: \end{figure}
273:
274: Finally, we tried to estimate a limit for the inclination of the
275: three-planet system. We selected a sample of 200 Newtonian solutions
276: from the previous search. For each fit we increased the system's
277: inclination in steps of $1^{\circ}$ and then refined this fit with the
278: Levenberg-Marquart scheme. Simultaneously, we also tested stability
279: of these fits with MEGNO. Obviously, the inclination of the system is
280: unconstrained. Overall, the $N$-body model does not improve the formal
281: $\Chi$ of Keplerian -- but it is important for the dynamical analysis
282: and for finding stable configurations. Moreover, the stable fits can
283: be found down to $i \sim 40^{\circ}$ with simultaneous rough scaling
284: with the masses by the Doppler factor $f = 1/\sin i$. We also note
285: that the border of stable motions for $e_{\idm{d}}$ is at $\sim
286: 0.3$. Still, fits with $a_{\idm{d}}\sim 0.89$~AU (``Fit 3'', $\Chi =
287: 1.30$ with rms 9.6 ms$^{-1}$) yield a slightly smaller $\Chi$ than
288: solutions with $a_{\idm{d}}\sim 1.02$~AU (``Fit 4'', $\chi^2 = 1.31$
289: with rms 9.66 ms$^{-1}$).
290:
291: RB05 predicted a Saturn-mass planet was most likely to be discovered
292: in the range $0.9 \lsim a \lsim 1.4$ AU, and $0 \le e \lsim 0.2$. The
293: four most likely fits identified by this analysis show that the
294: prediction of the PPS model has been borne out. HD 74156 d is the
295: first extrasolar planet to have its mass and orbit predicted. Fits 1
296: and 3 have the same $\Chi$ value, but Fit 1 has a slightly smaller
297: rms, so it should be considered the best overall fit to the
298: observations.
299:
300: \section{Discussion}
301:
302: In addition to HD 74156, similar predictions of new planets had been
303: made for the systems 55 Cnc, HD 37124 and HD 38529 by BR04 and
304: RB05. Subsequently, the orbital architectures of three of these four systems have been updated in ways that are consistent with the PPS picture.
305:
306: A fifth planet was recently discovered in orbit about 55 Cnc in the
307: stable zone identified in BR04 and RB05 (Fischer \etal 2008). This new
308: planet, 55 Cnc f, is at the inner edge of the stable zone, and is
309: consequently packed in with the inner three. 55 Cnc f is therefore
310: also consistent with the PPS model. However, there is still room for
311: additional planets in this system (Raymond \& Barnes 2008). The HD
312: 37124 planetary system was revised from two to three planets, on
313: orbits markedly different from the initially-announced configuration
314: (Vogt \etal 2005; see also Go\'zdziewski \etal 2008). The current
315: best-fit orbits of the planets in HD 37124 (Go\'zdziewski \etal 2008)
316: suggest the system is packed and is therefore also consistent with the
317: PPS model. The final system considered in BR04 and RB05, HD 38529, has
318: yet to produce another planet, but the observations presented in
319: Moro-Mart\'in \etal (2007) suggest there is no observational evidence
320: for or against an additional planet in HD 38529.
321:
322: The nature of the single planet systems appears at odds with the PPS
323: model. About 85\% of known exoplanets are currently observed to be
324: the sole companion of their host
325: star\footnote{http://exoplanets.org}. If the PPS hypothesis is to be
326: believed, then we would naturally expect planetary systems should all
327: be multiple. There are three possible explanations for the current
328: observations of single planet systems in the context of the PPS
329: hypothesis: 1) The orbits of additional companions have not been
330: robustly detected yet, 2) the additional planets are too small to be
331: detected, and/or 3) the additional planets have significant
332: inclinations. Wright \etal (2007) find that at least 30\% of single
333: planet systems show evidence of additional companions. Therefore it
334: may be that, as more data are obtained, the fraction of single planet
335: systems will drop precipitously. Another possibility stems from the
336: observed distribution of planet masses, which rises steeply at low
337: mass (Marcy \etal 2005). This result implies many planets may lie in
338: these gaps, but surveys lack the precision to detect them. A third
339: possibility is that mutual inclinations could be significant,
340: resulting in only one detectable planet. For example, Marzari \&
341: Weidenschilling (2002) showed that scattering can pump up inclinations
342: to large values while simultaneously increasing
343: eccentricities. Therefore it may be that some systems appear to have
344: only one planet because the other planets are too inclined to the line
345: of sight to be detected by radial velocity surveys.
346:
347: In addition to packed exoplanet systems, numerical experiments testing
348: the stability of the giant planets in our Solar System suggest that
349: they, too, are packed (\eg Varadi \etal 1999, Michtchenko \& Ferraz-Mello 2001; Barnes \& Quinn
350: 2004). The apparently high frequency of packed giant planet systems
351: naturally motivates research to identify more planets in stable
352: locations. Stability analyses show that several systems stand
353: out as non-packed. A simple method to determine the proximity to
354: dynamical instability, $\beta$, is described in Barnes \& Greenberg
355: (2007) based on the concept of ``Hill stability'' (Marchal \& Bozis
356: 1982; Gladman 1993). In Barnes \& Greenberg's formulation, the
357: stability boundary lies at $\beta = 1$, and stable systems require
358: $\beta > 1$, except in the case of mean motion resonances (note that
359: numerical analyses of these systems do suggest that they, too, are packed
360: [Go\'zdziewski \& Maciejewski 2001; Barnes \& Quinn 2004; Barnes \&
361: Greenberg 2007]). Barnes \& Greenberg 2007 estimated that when $\beta
362: > 2$, then the system may contain an additional planet, but more work
363: is needed to verify this possibility.
364:
365: Several two-planet systems are known with $\beta$ values greater than
366: 2 (note that HD 38529 has a $\beta$ value of
367: 2.07). HD 217107 (Vogt \etal 2005) has a $\beta$ value of 7.1, HD 68988
368: (Wright \etal 2007) has $\beta = 12.5$ and HD 187123 (Wright \etal
369: 2007) has $\beta = 15.3$\footnote{see http://www.lpl.arizona.edu/$\sim$rory/research/xsp/dynamics/ for an up-to-date table of $\beta$ values.}. These systems have gaps that are probably large enough to
370: contain multiple additional companions. We also note, however, that
371: the outer companions of the latter two systems have orbits which are
372: poorly constrained. Additionally, the inner planets have been tidally
373: circularized (Rasio \etal 1996) and consequently may have formed with
374: significantly larger values of $e$ and $a$ (Jackson\etal 2008). It may
375: be many years before the known planets' orbits are measured with
376: significant precision, let alone unknown companions are detected.
377:
378: \section{Conclusions}
379: The correct prediction of the mass and orbit of HD 74156 d is a major
380: step forward in our understanding of the nature of exoplanets. It also
381: represents the first successful prediction of the mass and orbit of a
382: planet since Neptune was predicted independently by LeVerrier and
383: Adams in the 1840s. Although their approach was markedly different
384: than that of BR04 and RB05, both predictions relied on the dynamical
385: properties of other planets in the system.
386:
387: As mentioned in $\S$ 1, of the first 6 multiple extrasolar planet
388: systems detected, only HD 74156 appeared to not be dynamically
389: packed. The detection of planet d by Bean \etal (2008), and its
390: confirmation presented in $\S$ 2, now means that the first 6 multi-planet
391: systems to be detected were packed. Although there may be some bias
392: toward detecting packed systems because radial velocity surveys are
393: more likely to find planets close to their host star, HD 74156 d was a
394: difficult planet to detect because of its low mass. Therefore its
395: detection is a strong indicator that most, if not all, multiple planet
396: systems are dynamically packed. But, of course, future observations
397: will ultimately reject or confirm the PPS theory. We
398: encourage observers to focus on HD 38529, HD 217107, HD 68988 and HD
399: 187123 in order to determine if additional companions lay between the
400: two that are known. Furthermore, additional companions need to be
401: detected in one-planet systems.
402:
403: We also encourage future work to continue to explore mechanisms which
404: may lead to packed planetary systems. It could be that packing is a
405: natural consequence of planet formation (Laskar 2000; Scardigli
406: 2007). One such model is the dynamical relaxation of planets in a
407: damped medium (Adams \& Laughlin 2003), but more work is needed to
408: verify this possibility.
409:
410: The detection of HD 74156 d suggests that planet formation is an
411: efficient process and that planets may be common. We have suggested
412: several systems which are important litmus tests for the PPS
413: theory. We also encourage programs aimed at detecting astrometric
414: signals from planets that have been detected by radial velocity
415: surveys.
416:
417: %\medskip
418: R.B. acknowledges support from NASA's PG\&G grant NNG05GH65G and TPFFS
419: grant 811073.02.07.01.15. K.G. is supported by the Polish Ministry of
420: Sciences and Education, Grant No. 1P03D-021-29. S.N.R. was supported
421: by an appointment to the NASA Postdoctoral Program at the University of
422: Colorado Astrobiology Center, administered by Oak Ridge Associated
423: Universities through a contract with NASA. We thank Barbara McArthur,
424: Jacob Bean, Fritz Benedict, Steven Soter, Thomas Quinn, Chris Laws and
425: Richard Greenberg for helpful discussions.
426:
427:
428: \references
429: Adams, F.C. \& Laughlin, G. 2003, Icarus, 163, 290\\
430: Barnes, R. \& Greenberg, R. 2007, ApJ, 665, L67\\
431: Barnes, R. \& Quinn, T.R. 2001, ApJ, 554, 884\\
432: Barnes, R. \& Quinn, T.R. 2004, ApJ, 611, 494\\
433: Barnes, R. \& Raymond, S.N. 2004 ApJ, 617, 569 (BR04)\\
434: Bean, J.L., McArthur, B.E., Benedict, G.F. \& Armstrong, A. 2008, ApJ, 672, 1202\\
435: Bois, E. Kiseleva-Eggleton, L., Rambaux, N. \& Pilat-Lohinger, E. 2003, ApJ, 598, 1312\\
436: Butler, R.P. \etal 1999, ApJ, 526, 916\\
437: Chambers, J., 1999, MNRAS, 304, 793\\
438: Charbonneau, P. 1995, ApJS, 101, 309\\
439: Cincotta, P.M., Giordano, C.M. \& Simo, C. 2003, Physica D Nonlinear Phenomenon, 182, 151\\
440: Cincotta, P.M. \& Sim\'o, C. 2000, A\&AS, 147, 205\\
441: Fischer, D. \etal 2002, ApJ, 564, 1028\\
442: Gladman, B. 1993, Icarus, 106, 247\\
443: Go\'zdziewski, K., Breiter, S. \& Borczyk, W. 2008, MNRAS, 383, 989\\
444: Go\'zdziewski, K., Maciejewski, A.~J.2002, ApJ, 563, L81.\\
445: Go\'zdziewski, K., Maciejewski, A.~J. \& Migaszewski, C. 2007, ApJ, 657, 546\\
446: Go\'zdziewski, K. \& Migaszewski, C. 2006, A\&A, 449, 1219\\
447: Grosser, M. 1962, \textit{The Discovery of Neptune}, Harvard UP\\
448: Jackson, B., Greenberg, R. \& Barnes, R. 2008, ApJ, 678, 1396\\
449: Laskar, J. 2000, PhRvL, 84, 3240\\
450: Laughlin, G. \& Chambers, J.E. 2001, ApJ, 551, L109\\
451: Marchal, C. \& Bozis, G. 1982, CeMDA, 26, 311\\
452: Marcy, G.W. \etal 2001a, ApJ, 555, 418\\
453: Marcy, G.W. \etal 2001b, ApJ, 556, 296\\
454: Marcy, G.W. \etal 2005, PThPS, 158, 24\\
455: Marzari, F. \& Weidenschilling, S. 2002, Icarus, 156, 570\\
456: Michtchenko, T.A. \& Ferraz-Mello, S. 2001, AJ, 122, 474\\
457: -----------. 2001, Icarus, 149, 357.\\
458: Naef, D. \etal 2004, A\&A, 414, 351\\
459: Rasio, F.A., Tout, C. A., Lubow, S. H., \& Livio, M. 1996, ApJ, 470, 1187\\
460: Raymond, S.N. \& Barnes, R. 2005, ApJ, 619, 549 (RB05)\\
461: Raymond, S.N. \& Barnes, R. 2008, ApJ, submitted\\
462: Raymond, S.N., Barnes, R. \& Kaib, N.A. 2006, ApJ, 644, 1223\\
463: Raymond, S.N., Quinn, T.R., \& Lunine, J.I. 2007, Astrobiology, 7, 66\\
464: Rivera, E.J. \& Lissauer, J.J. 2000, ApJ, 530, 454\\
465: Scardigli, F. 2007, FoPh, 37, 1278\\
466: Varadi, F. Ghil, M., \& Kaula, W. M. 1999, Icarus, 139, 286.\\
467: Vogt, S.S. \etal 2005, ApJ, 632,638\\
468: Wright, J.T. \etal 2007, ApJ, 657, 533
469: \end{document}
470:
471:
472:
473:
474:
475:
476:
477:
478:
479:
480:
481:
482: