0803.2935/ms.tex
1: %% This is a sample manuscript marked up using the
2: %% AASTeX v5.x LaTeX 2e macros.
3: 
4: %emulateapj
5: %\documentclass{emulateapj}
6: 
7: %\documentclass[onecolumn]{emulateapj}
8: \documentclass[12pt,preprint]{aastex}
9: 
10: %% manuscript produces a one-column, double-spaced document:
11: %\documentclass[manuscript]{aastex}
12: 
13: %% preprint2 produces a double-column, single-spaced document:
14: %\documentclass[preprint2]{aastex}
15: 
16: %% Sometimes a paper's abstract is too long to fit on the
17: %% title page in preprint2 mode. When that is the case,
18: %% use the longabstract style option.
19: %% \documentclass[preprint2,longabstract]{aastex}
20: 
21: %\newcommand{\vdag}{(v)^\dagger}
22: %\newcommand{\myemail}{skywalker@galaxy.far.far.away}
23: 
24: %% You can insert a short comment on the title page using the command below.
25: 
26: %\slugcomment{Not to appear in Nonlearned J., 45.}
27: 
28: 
29: \shorttitle{Dynamics of HD~17156c}
30: \shortauthors{Short et al.}
31: 
32: %% This is the end of the preamble.  Indicate the beginning of the
33: %% paper itself with \begin{document}.
34: 
35: \def\POFP{{\scshape{pofp}}}
36: 
37: \begin{document}
38: 
39: 
40: \title{Orbital Dynamics Of A Second Planet In HD~17156}
41: 
42: %% Use \author, \affil, and the \and command to format
43: %% author and affiliation information.
44: %% Note that \email has replaced the old \authoremail command
45: %% from AASTeX v4.0. You can use \email to mark an email address
46: %% anywhere in the paper, not just in the front matter.
47: %% As in the title, use \\ to force line breaks.
48: 
49: \author{Donald Short} %\altaffilmark{1}
50: \affil{Department of Mathematics, San Diego State University,
51: 5500 Campanile Drive, San Diego, CA 92182}
52: \email{dshort@rohan.sdsu.edu}
53: 
54: %\and
55: 
56: \author{William F. Welsh, Jerome A. Orosz, and Gur Windmiller} 
57: \affil{Department of Astronomy, San Diego State University,
58: 5500 Campanile Drive, San Diego, CA 92182-1221}
59: \email{wfw@sciences.sdsu.edu}
60: %\email{orosz@sciences.sdsu.edu}
61: %\email{windmill@rohan.sdsu.edu}
62: 
63: \begin{abstract}
64: 
65: In this letter we report the possible existence of a second planet in the
66: transiting extrasolar planet system HD~17156 and its interactive dynamics
67: with the previously known planet. The analysis is achieved through the
68: \POFP\ optimization software which is based on a full integration of
69: the system's multiple-body Newtonian equations of motion.
70: The two-planet solution yields a significantly improved fit to the
71: previously published radial velocities.
72: The two planets are strongly interacting and exchange angular momentum
73: in a 5:1 mean motion resonance,
74: yet remain stable as they mutually excite orbital eccentricities and
75: periastron advances.
76: 
77: \end{abstract}
78: 
79: 
80: \keywords{planetary  systems -- celestial mechanics -- gravitation --
81: instabilities -- methods: N-body simulations -- methods: numerical}
82: 
83: 
84: \section{Introduction}
85: 
86: By virtue of its unusual characteristics, HD~17156b is one of the most
87: valuable extrasolar planets for understanding planet formation and
88: orbital dynamics. Discovered via the Doppler technique by the N2K
89: consortium (Fischer et al.~2007), the planet was found to transit its
90: host star by the {\scshape TransitSearch.org} collaboration (Barbieri
91: et al.~2007). Additional transit observations and refined system
92: parameters are given by Gillon et al.~(2008), Narita et al.~(2008), and
93: Irwin et al.~(2008). The planet radius and mass are approximately 1
94: R$_{\rm Jup}$ and 3 M$_{\rm Jup}$, giving a density of $\sim$ 3.5--4.0
95: $\rm \rm{g} \ \rm{cm}^{-3}$, nearly three times that of Jupiter. More
96: interesting however are the planet's orbital characteristics:
97: (i) The 21.2~d orbital period is roughly a factor 7 times longer than
98: most\footnote{see The Extrasolar Planets Encyclopedia by Jean
99: Schneider at http://exoplanet.eu}
100: other transiting planets; (ii) While most hot Jupiter planets have low
101: eccentricities (the majority consistent with zero), HD~17156b has a very
102: high eccentricity of e=0.67. Although the high eccentricity coupled
103: with the small semimajor axis (0.16~$\rm AU$) does not necessarily
104: require the presence of a third body perturbing the planet's orbit
105: (Gillon et al.~2008), it suggests that such a body may be present.
106: In this work we examine the published radial velocities of Fischer et
107: al.~(2007) and the transit times as given in Irwin et al.~(2008)
108: using a Newtonian (not Keplerian) 3-body integrator and conclude that a 
109: second planet is not only possible, but probable. 
110: Notably, while the two planets exhibit strong dynamical interaction 
111: leading to large eccentricities and periastron advances, they remain 
112: stable via an elegant exchange of orbital angular momentum.
113: 
114: 
115: \section{ Initial Analysis of Radial Velocities }
116: Our initial investigation of the HD~17156 system began with a 1-planet
117: fit to the Keck and Subaru radial velocity data from Fischer 
118: et al.~(2007). We omitted the radial velocity data of Narita et 
119: al.~(2008) since these data are influenced by the Rossiter-McLaughlin 
120: effect, but their inclusion has essentially no effect on the results.
121: We reproduced the parameters of the planet HD~17156b as reported in 
122: Fischer et al.~(2007), though we found that an additional 0.14 $\rm 
123: ms^{-1}$ offset added to the Subaru data set gave a slightly better fit.
124: We computed a power spectrum of the residuals of the radial velocities 
125: after removing the one-planet fit.
126: A Monte Carlo technique was used to allow us to assess the significance 
127: of any peaks. The power spectrum indicated a possible peak with a period 
128: of approximately 115 days. This prompted us to undertake a much more 
129: thorough and realistic 3-body investigation described below.
130: In addition to the power spectra, we computed Keplerian periodograms: 
131: for a given period the parameters of a Keplerian fit to the radial 
132: velocities were optimized (K, e, $\omega$, $T_{0}$, $\gamma$), i.e.~the  
133: Lehmann--Filh\'{e}s equation (see Hilditch 2001 for 
134: example). While a traditional Fourier power spectrum has no free 
135: parameters, it uses sines and cosines as basis functions. However, an 
136: eccentric orbit is far from being sinusoidal and as a consequence a 
137: Fourier power spectrum will require power at many periods to match the 
138: radial velocities. In contrast, by using the Lehmann--Filh\'{e}s equation 
139: as the basis function we obtain an optimal periodogram. Furthermore,  
140: unlike a power spectrum, fitting a Keplerian orbit at each trial period 
141: allows one to use the error bars on each observation as weights. Thus no 
142: Monte Carlo or bootstrap sampling is necessary to assess the quality of 
143: the fit: the periodogram provides a $\chi^{2}$ versus period directly. 
144: Such periodograms clearly showed the 21.2~d signal from planet b, and the 
145: residuals again indicated the presence of a possible second planet 
146: with a period of $\sim$106-116~d. 
147: 
148: 
149: \section{ HD~17156c }
150: To investigate characteristics of a possible second planet we used the 
151: Planetary Orbit Fitting Process (\POFP), an optimization software 
152: written and operated in {\scshape{matlab}}  --- see  Windmiller, Short 
153: $\&$ Orosz (2007) and Short, Windmiller $\&$ Orosz (2008). The \POFP\ 
154: provides multi-body solutions based on a full integration of the 
155: Newtonian equations of motion. 
156: 
157: In addition to the radial velocity data, we used \POFP\ to fit the 
158: 4 transit times as listed in Irwin et al.~(2008). The inclusion of the 
159: transit times was crucial because it not only greatly improved the 
160: period of the known planet b, but the precise timings strongly constrain 
161: the orbital configuration and interaction between the planets.
162: Since the system transits, we assumed the second planet's orbit is 
163: edge-on and co-planar with the first.
164: A careful reading of Fischer et al.~(2007) led us to use caution with   
165: the Subaru radial velocities, thus we considered two data set 
166: combinations: The first consisted of the 24 high-quality radial 
167: velocities from Keck plus the 4 observed transit times, and the second 
168: consisted of the 33 combined Keck and Subaru radial velocities 
169: plus the 4 transit times. 
170: Following Fischer et al.~(2007), we set the stellar mass to 1.2 
171: $M_{\odot}$ and the stellar jitter to 3 $\rm ms^{-1}$. 
172: Using \POFP\, we fit the two data sets with a single planet model
173: giving two solutions denoted here by \POFP1 K+TT (Keck plus transit 
174: times) and \POFP1 K+S+TT (Keck, Subaru, plus transit times). This 
175: provided a baseline set of models to compare against when considering 
176: a two-planet solution. The orbital parameters are given in Table~1, 
177: along with the Fischer et al.~(2007) and Irwin et al.~(2008) 
178: solutions.
179: 
180: We then fit using the Keck and Subaru radial velocities plus transit 
181: times with a two-planet model, giving the solution \POFP2. Because the 
182: \POFP2 solution is fully Newtonian and the planets interact, the 
183: tabulated parameters are valid only at a specific time, which is taken to 
184: be the time of the first Keck radial velocity measurement. The fitness 
185: for the 1-planet and 2-planet solutions is given in Table 2. Note that in 
186: this table only the following are optimized: the 1-planet model using 
187: Keck radial velocities, the 1-planet model using Keck+Subaru velocities, 
188: and the 2-planet model using Keck+Subaru velocities; in all cases the 
189: transit times were used. These optimized values are shown in bold-face; 
190: the other entries result when these models are evaluated and matched 
191: against the listed data set and are given for comparison.
192: Some reduced $\chi^{2}$ values are less than 1.0, suggesting that the 
193: 3~$\rm ms^{-1}$ jitter estimate of Fischer et al.~(2007) is slightly too 
194: large. The key point of Table~2 is the following: 
195: including two planets significantly improves the fit in all cases 
196: (Keck only, Keck+TT, Keck+Subaru, Keck+Subaru+TT).
197: 
198: In our 2-planet solution we found that the inner planet maintains an 
199: almost constant semi-major axis of 0.1596 $\rm AU$, a slowly oscillating 
200: eccentricity having a period of 141.1 yr with $e$ between 0.665 and 
201: 0.670, and a full rotation of its line of apsides every 33,050~yr. 
202: Thus, there are two long term cycles of approximately 141.1~yr 
203: and 33,050~yr. Meanwhile, the outer planet HD~17156c exhibits an 
204: oscillation in eccentricity with a pronounced amplitude, varying between 
205: $e=0.10$ and $e=0.50$, in a near-mirror image of the eccentricity 
206: oscillation of planet~b (see Fig.~1). These complementary oscillations of 
207: eccentricities provide a very interesting example of planetary angular 
208: momentum interchange. The semi-major axis of planet c varies between 
209: $\sim$ 0.46 and 0.49 $\rm{AU}$ and the position of its periastron 
210: completes a 141.1 yr cycle in step with the change in the eccentricities. 
211: Furthermore, for each 141.1~yr cycle, there will be an orbit of maximum 
212: eccentricity for planet c. Over 33,050~yr the periastron of this maximum 
213: eccentricity orbit will advance slowly and synchronously with the 
214: precession of planet b's periastron. As a result of this resonance 
215: coupling between the orbits, the stability of this system is 
216: maintained by keeping an approximately 0.14 $\rm AU$ buffer between the 
217: two planets' orbital paths. This is illustrated in Fig.~2, showing the 
218: change in the orbital configuration caused by planetary interaction. 
219: Over a span of half the 141.1 yr cycle we see planet c's eccentricity 
220: change dramatically. The lower left panel shows a much larger advance 
221: in time and is a snapshot of a maximum eccentric orbit of planet c; 
222: notice the simultaneous precession of both planets' periastra at the 
223: longer 33,050~yr cycle. 
224: 
225: In addition to any goodness of fit analysis, the two-planet model
226: must show long-term stability. We have used the symplectic 
227: integration package\footnote{http://janus.astro.umd.edu/HNBody/} 
228: {\scshape{HNBody}} of Rauch and Hamilton~(2002) to do a long term 
229: integration of this system. Over a time interval of 100 million days
230: ($\sim$274,000 yr), we have verified stability and that the planets 
231: exhibit a 5:1 mean motion resonance.
232: Following the lead of Barnes~(2006), we plotted  
233: $e_{b}e_{c}\sin(\Delta\omega)$ vs. $e_{b}e_{c}\cos(\Delta\omega)$ in 
234: Fig.~3, where $\Delta\omega$ is the difference in the longitude of the 
235: planets' periastra. In this figure, if the angle between the periastra 
236: was constant, all the points would be co-linear, extending radially 
237: outward from the origin.
238: The closed loop shape arises because the angle between periastra grows 
239: and spans a full $2\pi$ every 141.1~yrs; there is no libration.
240: From inspection of the figure, the maximum of $e_{b} e_{c}$ occurs when 
241: $\Delta\omega=0$, and since $e_{b} \approx 0.67$, the orbits of planet c 
242: reach their maximum eccentricity when the periastra are aligned.
243: Hence the specific cases illustrated in Fig.~2 are in fact typical.
244: 
245: 
246: The 1--$\sigma$ confidence limits for the parameters of the 2--planet 
247: solution were calculated following Press et al.~(1986) and are given in 
248: Table~3. Note that we quote the parameter values to more significant 
249: digits than is warranted by the uncertainties so that others can exactly 
250: reproduce our 2-planet \POFP2 numerical solution. Also note that while 
251: the short-term orbital characteristics of planet c can be determined, the 
252: timescales of the much longer precession cycles are poorly constrained by 
253: the data (e.g.~an earlier \POFP2 solution yielded long term cycles of 
254: 212.7~yr and 37,700~yr).
255: 
256: 
257: The question remains, does HD~17156c exist? Noting the small
258: number of observations, the short time span of these observations,
259: and the small mass of the inferred second planet, caution must be taken. 
260: Using the $F$ statistic on the fitness values from Table~2 to test if
261: the extra parameters of the two-planet model are warranted, 
262: we find that the probability for including those additional parameters
263: is greater than 99.9$\%$.
264: While the mass of HD~17156c is poorly determined, it is not consistent
265: with zero, further supporting the F statistic result. 
266: The two-planet model also is physically realizable as a stable system.
267: These facts led to our characterization of HD~17156c as not only possible, 
268: but probable. Obviously, to resolve the remaining uncertainty, 
269: additional observations are required. In \S 4, we point
270: out observations that would allow distinguishing between the one and 
271: two-planet models.
272: 
273: 
274: \section{ Discussion }
275: 
276: Additional radial velocity and transit time data would greatly help 
277: confirm and constrain specific parameters of the second planet. However, 
278: the general dynamics of the 3-body interactions are already well 
279: determined. Using the current observations and our two-planet \POFP\ model 
280: we can make several observable predictions. The divergence between the 
281: velocities derived from the single-planet reproduction and the two-planet 
282: solutions is shown in Fig.~4. In about 2 years, the differences in radial 
283: velocity predictions between the two models become larger than 10 $\rm 
284: ms^{-1}$ at the peaks of the graphs, occurring at the time of periastron 
285: of planet b. In particular, for the epoch of planet b's periastron at HJD 
286: 2455266.04 (UT 2010 Mar 10), a difference of 18 $\rm ms^{-1}$ is 
287: predicted. 
288: % The radial velocities predicted at that time are -86.0 $\rm 
289: % ms^{-1}$ for the one-planet solution and -67.7 $\rm ms^{-1}$ for the 
290: % two-planet solution, referenced to the Keck radial velocity baseline.
291: 
292: Another observable is planet b's time of transit. While this is slowly 
293: advancing on the 33,050 yr cycle, there is a much larger and rapid 
294: modulation in planet b's {\em instantaneous period,} i.e.~the time 
295: between sequential transits.  The near-future predicted instantaneous 
296: period is shown in Fig.~5. 
297: Planet-planet interactions, especially when near apastron of planet b, 
298: result in sharp changes in the instantaneous period.
299: The average time interval between successive conjunctions is 
300: $\sim$26.2~d which is $\sim$24\% longer than planet b's period.
301: Thus conjunctions close to the apastron of planet b occur roughly every 
302: $\sim$5 orbits. The exchange of angular momentum at these times causes 
303: the jumps in period of planet b shown in Fig.~5. 
304: A longer $\sim$1000~d cycle results from the longitude of planet c 
305: shifting by -6.75 degrees during each successive apastral conjunctions.
306: A predicted O-C diagram using the mean period at the current epoch is 
307: shown in the lower panel of Fig.~5. Unlike the instantaneous period 
308: changes, an O-C diagram includes the sum of the effects of the period 
309: changes since the fiducial epoch, and thus is akin to the integral of the 
310: instantaneous period. The deviations from a linear ephemeris are quite 
311: large and thus monitoring planet b's transit times should reveal the 
312: presence of planet c and be very helpful in further refinement of the 
313: system's parameters.
314: %
315: Finally, planet c may in principle also be detected via the transit 
316: method. Unfortunately, however, planet c is not expected to transit 
317: the host star if its orbit is coplanar with that of planet b 
318: (e.g.~$i=86.5^{\circ}$, Irwin et al.~2008). 
319: 
320: Whereas the parameters of the probable planet c are poorly constrained, 
321: the qualitative dynamical aspects of such a system are of great interest. 
322: This case provides an example of a strong dynamical interaction between 
323: a high eccentricity ``hot Jupiter'' planet and a second planet found 
324: further away from the star. The predicted near-term observables should 
325: provide guidance to future observations of this system and may provide a 
326: possible explanation to any variations seen in observed transit times.
327: 
328: 
329: 
330: \acknowledgments
331: 
332: We thank the anonymous referee for suggestions that helped improve this 
333: work. We thank Rory Barnes of the Lunar and Planetary Laboratory
334: of the University of Arizona for his timely and thoughtful comments on
335: the dynamics of the two-planet model.
336: 
337: %------------
338: 
339: 
340: \begin{thebibliography}{}
341: 
342: \bibitem[Barbieri et al. (2007)]{Bar07} Barbieri, M. et al. 2007, \aap,
343: 476 L13
344: 
345: \bibitem[Barnes et al. (2006)]{Bas06} Barnes, M., Greenberg, R. 2006, ApJ,
346: 652 L53
347: 
348: \bibitem[Fischer et al. (2007)]{Fis07} Fischer, D. A. et al., 2007,
349: \apj, 669, 1336
350: 
351: \bibitem[Gillon et al. (2008)]{Gil08} Gillon, M. et al. 2008, \aap,
352: (submitted) [arXiv:0712.2073]
353: 
354: \bibitem[Hilditch (2001)]{Hilditch01} Hilditch, R. W., 2001
355: An Introduction to Close Binary Stars (Cambridge: Cambridge University 
356: Press)
357: 
358: \bibitem[Irwin et al. (2008)]{Irw08} Irwin, J. et al. 2008, \apj,
359: (in press) [arXiv:0801.1496]
360: 
361: \bibitem[Narita et al. (2008)]{Nar08} Narita, N. et al. 2008, PASJ
362: Letters, 60, (in press) [arXiv: 0712.2569]
363: 
364: \bibitem[Press et al. (1986)]{Pre86}  Press, W. et al.  1986,
365: Numerical Recipes, Vol.~1 (2nd ed.; Cambridge: Cambridge University Press)
366: 
367: \bibitem[Rauch et al. (2002)]{Rau02} Rauch, K. P., Hamilton, D. P.,
368: 2002, DDA 33.0802R
369: 
370: \bibitem[Short et al. (2008)]{Sho08} Short, D., Windmiller, G., \&
371: Orosz, J. A., 2008, \mnras, tmp L37S
372: 
373: \bibitem[Windmiller et al. (2007)]{Win07} Windmiller, G. Short, D., \&
374: Orosz, J. A., 2007, [arXiv:0708.0898]
375: 
376: \end{thebibliography}
377: 
378: %\clearpage
379: 
380: %----------------------------------------------------------------------
381: % TABLE 1
382: 
383: \begin{deluxetable}{lclclllcc}
384: \tabletypesize{\scriptsize}
385: \tablecolumns{11}
386: \tablewidth{0pt}
387: \tablecaption{HD~17156 Planet Parameters\label{tbl-1}}
388: \tablehead{
389: \colhead{ Solution} & 
390: \colhead{ Epoch\tablenotemark{a}} & 
391: \colhead{ P}  &
392: \colhead{ $T_{0}$\tablenotemark{b} } & 
393: \colhead{ e} & 
394: \colhead{ $\omega$} &
395: \colhead{ K} &
396: \colhead{ a} & 
397: \colhead{ Mass\tablenotemark{d}} \\
398: \colhead{ } & 
399: \colhead{ (\rm HJD-2450000)} & 
400: \colhead{ (\rm days)}  &
401: \colhead{ (\rm HJD-2450000)} & 
402: \colhead{ } & 
403: \colhead{ (\rm deg)} &
404: \colhead{ ($\rm{ms^{-1}}$) } &
405: \colhead{ ($\rm AU$) } & 
406: \colhead{ (M$\rm_{Jup}$) } 
407: }
408: \startdata
409: Fischer et al.~2007    &  ...        &  21.2      & 3738.529  & 
410: 0.67   &  121     & 275        &  0.15    &  3.12    \\
411: %
412: Irwin et al.~2007 (K+S+TT) &  ...    &  21.2169   & 3738.605  & 
413: 0.670  &  121.3   & 273.8      &  0.160    &  3.13    \\
414: %
415: \POFP1 K+TT             &  ...        &  21.2167    & 3738.614  & 
416: 0.670  &  121.4   & 273.8        &  0.160    &  3.12   \\
417: %
418: \POFP1 K+S+TT           &  ...        &  21.2168    & 3738.614  & 
419: 0.670  &  121.6   & 273.3        &  0.160    &  3.13   \\
420: %
421: \POFP2 Planet b         & 3746.7582   &  21.2144   & 3738.593  & 
422: 0.670  &  120.9   & 273.0      &  0.160    &  3.13   \\
423: ~~~and Planet c      & 3746.7582   & 111.394     & 3736.880  & 
424: 0.136  &  273.6   & 2.6        &  0.481    &  0.07   \\
425: \enddata
426: \tablenotetext{a}{Time at which tabulated parameters are valid for the 
427: \POFP2 solutions}
428: \tablenotetext{b}{Time of periastron passage}
429: \tablenotetext{c}{Assuming a stellar mass of 1.2 $M_{\odot}$
430: from Fischer et al.~2007}
431: \end{deluxetable}
432: 
433: % ==============================================
434: % TABLE 2
435: 
436: \begin{deluxetable}{lcccccc}
437: \tabletypesize{\scriptsize}
438: \tablecolumns{11}
439: \tablewidth{0pt}
440: \tablecaption{Fitness\tablenotemark{a} \ of  
441: Solutions\tablenotemark{b}\label{tbl-2}}
442: \tablehead{
443: \colhead{ }                    & 
444: \colhead{Number of}            &
445: \colhead{ rms}                 & 
446: \colhead{ Keck RV only}        &
447: \colhead{ Keck+TT}             &
448: \colhead{ Keck+Subaru RV only} & 
449: \colhead{ Keck+Subaru+TT}      \\
450: \colhead{ Solution}       &
451: \colhead{ parameters}     &
452: \colhead{ ($m s^{-1}$)}   &
453: \colhead{ N=24}           &
454: \colhead{ N=28}           &
455: \colhead{ N=33}           &
456: \colhead{ N=37}
457: }
458: \startdata
459: Fischer et.~al 2007 & ... & 3.97 & ... & ... & (1.37; 1.08)\tablenotemark{c} 
460: & ...\\
461: \tableline
462: \POFP1 K+TT     &  5    & {\bf{3.65}} & 26.47 (1.39) & {\bf{27.45 (1.19)}}
463:                & 30.41 (1.13) & 33.12 (1.07)      \\
464: %
465: \POFP1 K+S+TT   &  6    & {\bf{3.75}} &  26.49 (1.39)  & 29.99 (1.30)
466:                & 30.48 (1.13) & {{\bf31.51  (1.02) }} \\
467: %
468: \POFP2 K+S+TT   & 11    & {\bf{3.14}} & 16.56 (1.18) &   17.90 (0.99) 
469:                & 19.96 (0.91) & {\bf{21.38 (0.82) }} \\
470: \enddata
471: \tablenotetext{a}{Reduced $\chi^{2}$ values are in parentheses.}
472: \tablenotetext{b}{Optimized solutions are shown in bold; other values 
473: show the solution evaluated against the specific data set.}
474: \tablenotetext{c}{Note Fischer et al.~(2007) give two values for the 
475: reduced $\chi^{2}$.}
476: \end{deluxetable}
477: 
478: 
479: % -------------
480: % TABLE 3
481: 
482: \begin{deluxetable}{lrrccccr}
483: \tabletypesize{\scriptsize}
484: \tablecolumns{11}
485: \tablewidth{0pt}
486: \tablecaption{HD~17156 \POFP2 Error 
487:  Estimates\tablenotemark{a}\label{tbl-3}}
488: \tablehead{
489: \colhead{ Planet} & 
490: \colhead{ P}  &
491: \colhead{ $T_{0}$} & 
492: \colhead{ e} & 
493: \colhead{ $\omega$} &
494: \colhead{ K} &
495: \colhead{ a} & 
496: \colhead{ Mass} \\
497: %second line
498: \colhead{ } & 
499: \colhead{ (\rm days)}  &
500: \colhead{ (\rm HJD)} & 
501: \colhead{ } & 
502: \colhead{ (\rm deg)} &
503: \colhead{ ($\rm{ms^{-1}}$) } &
504: \colhead{ ($\rm AU$) } & 
505: \colhead{ (M$\rm_{Jup}$) }
506: }
507: \startdata
508: ~~~~~ $+1 \sigma$  & $+0.0031$  & $+0.05$    & $+0.003$  & $+0.14$ & 
509:       $+1.77$ & $+0.0001$  & $+0.02$  \\
510: Planet b           & 21.2144    & 3738.593   & 0.669614  & 120.89  & 
511:       272.987 & 0.159505   & 3.125    \\
512: ~~~~~ $-1 \sigma$  & $-0.0025$  & $-0.05$    & $-0.0005$ & $-0.03$ & 
513:       $-1.75$  & $-0.0001$  & $-0.02$  \\
514:   &         &      &            &            &     &       &      \\
515: ~~~~~ $+1 \sigma$  &  $+0.43$   &  $+0.58$   & $+0.0029$ & $+0.17$ &
516:       $+0.79$  &  $+0.001$  &  $+0.021$\\
517: Planet c           & 111.3937   & 3736.880  & 0.136492  & 273.57 & 
518:       2.580   & 0.481478   & 0.068    \\
519: ~~~~~ $-1 \sigma$  & $-0.09$    & $-1.61$    & $-0.0012$ & $-0.17$ & 
520:       $-0.71$ & $-0.0003$  & $-0.019$\\
521: \enddata
522: \tablenotetext{a}{Full precision is given so the initial conditions of 
523: the solution are exactly specified.
524: }
525: \end{deluxetable}
526: 
527: 
528: %---------------------------------------------------------
529: % FIG 1
530: 
531: \begin{figure}
532: \epsscale{0.8}
533: %\epsscale{0.5} 
534: \plotone{f1.eps}
535: \caption{The cyclical nature of the planets' orbital properties is seen 
536: in this \POFP2 integration spanning $10^{6}$ d.
537: The panels show, from top to bottom, the eccentricity of planet b
538: ($e_b$), the eccentricity of planet c ($e_c$), 
539: and the longitude of periastron of planet c ($\omega_c$). 
540: The 141.1 yr (51,550~d) periodicity is manifest in the small
541: modulation of $e_b$ (upper panel) and the much
542: larger changes in $e_c$ and $\omega_c$.
543: The anti-phasing of the eccentricities is a result of the exchange of
544: angular momentum between the planets.
545: \label{fig1}}
546: \end{figure}
547: 
548: %---------------------------------------------------------
549: % FIG 2
550: 
551: \begin{figure}
552: \begin{center}
553: \epsscale{1.0}
554: %\epsscale{0.5}
555: \plotone{f2.eps}
556: \caption{
557: The orbital configuration of the \POFP2 two-planet solution is shown 
558: at four separate epochs. Clockwise from the upper left, the panels show 
559: the advance of planet~c's cycle (141.1~yr) at approximately 1/4 cycle
560: intervals. The lower left panel illustrates the precession of planet b's
561: orbit, on its much longer cycle of 33,050 yr.
562: The cross marks the periastron of planet c's orbit, and the
563: arrow in the lower right corner marks the direction to the observer.
564: \label{fig2}}
565: \end{center}
566: \end{figure}
567: 
568: %---------------------------------------------------------
569: % FIG 3
570: 
571: \begin{figure}
572: \begin{center}
573: \epsscale{1.0}
574: %\epsscale{0.5}
575: \plotone{f3.eps}
576: \caption{
577: The resonant coupling of the eccentricities of the two planets is shown, 
578: following the suggestion of Barnes~(2006) of plotting the product of the 
579: instantaneous eccentricities times the sine and cosine of the angle 
580: between the two planets' periastra ($\Delta \omega$).
581: The orbits were sampled every 50,000 days over the $10^{8}$~d interval 
582: of the {\scshape{HNBody}} integration. 
583: Since the eccentricity of planet b is almost constant, the figure shows 
584: that the orbit of maximum eccentricity of planet c occurs when the  
585: periastra of the two planets are approximately aligned, i.e., 
586: along the abscissa where $\Delta \omega \sim 0$.
587: \label{fig3}}
588: \end{center}
589: \end{figure}
590: 
591: 
592: %---------------------------------------------------------
593: % FIG 4
594: 
595: \begin{figure}
596: \epsscale{0.70}
597: %\epsscale{0.50}
598: \plotone{f4.eps}
599: \caption{The difference between the radial velocity predictions of 
600: the \POFP\ one-planet and two-planet solutions is shown. The bottom
601: panel is an enlargement of a 50-day interval from the upper panel. 
602: The largest differences (peaks in the lower panel) occur around the time 
603: of planet b's periastron, and become readily measurable around 
604: HJD 2455200 ($\sim$ 2010 Jan).
605: \label{fig4}}
606: \end{figure}
607: 
608: %---------------------------------------------------------
609: % FIG 5
610: 
611: \begin{figure}
612: \epsscale{0.85}
613: %\epsscale{0.5}
614: \plotone{f5.eps}
615: \caption{
616: {\it Upper panel:}
617: The {\em instantaneous} period of planet b in the \POFP2 two-planet 
618: solution plotted against time. 
619: Angular momentum exchange between the planets results in jumps 
620: in the orbital period.
621: Changes of up to $\sim$3~min in the transit-to-transit period are 
622: predicted. {\it Lower panel:} A predicted O-C diagram for the times of 
623: transit of planet b, with the transits times from Irwin et al.~(2007) 
624: superposed, using the ephemeris 
625: T(E) = 2453738.32783  +  21.216159~E (HJD).
626: \label{fig5}}
627: \end{figure}
628: 
629: %---------------------------------------------------------
630: 
631: 
632: \end{document}
633: