1: \documentclass[12pt,preprint]{aastex}
2: %\documentclass[manuscript,letter,epsfig,amsmath]{aastex}
3: %\documentclass{emulateapj}
4:
5:
6: %\newcommand{\etal}{{et al.\ }}
7: \newcommand{\be}{\begin{equation}}
8: \newcommand{\ee}{\end{equation}}
9: \newcommand{\bea}{\begin{eqnarray}}
10: \newcommand{\eea}{\end{eqnarray}}
11:
12: \shorttitle{Transit Timing \& Extrasolar Trojans}
13: \shortauthors{Ford \& Holman}
14: \slugcomment{submitted to ApJL}
15:
16: \begin{document}
17:
18: \title{Using Transit Timing Observations to Search for Trojans of Transiting Extrasolar Planets}
19: \author{Eric B.\ Ford\altaffilmark{1} and Matthew J. Holman}
20: \affil{Harvard-Smithsonian Center for Astrophysics, Mail Stop 51, 60 Garden Street, Cambridge, MA 02138}
21: \email{eford,mholman@cfa.harvard.edu}
22:
23: \altaffiltext{1}{Hubble Fellow}
24:
25: \begin{abstract}
26: Theoretical studies predict that Trojans are likely a frequent
27: byproduct of planet formation and evolution. We examine the
28: sensitivity of transit timing observations for detecting Trojan
29: companions to transiting extrasolar planets. We demonstrate that this
30: method offers the potential to detect terrestrial-mass Trojans using
31: existing ground-based observatories. We compare the transit timing
32: variation (TTV) method with other techniques for detecting extrasolar
33: Trojans and outline the future prospects for this method.
34: %
35: \end{abstract}
36: %
37: \keywords{techniques: photometric --- planetary systems: formation ---
38: celestial mechanics }
39: %
40: \section{Introduction}
41:
42: %\subsection{Transit Timing}
43:
44: For centuries, theories of planet formation had been
45: designed to explain our own Solar System, but the first
46: extrasolar planetary systems discovered were very different than our
47: own (e.g., Mayor \& Queloz 1995). These discoveries led to the
48: realization that planet formation theory must be generalized to
49: explain a much greater diversity of planetary systems.
50: %
51: Stable Trojan companions to extrasolar planets may be common. In our
52: solar system, Mars, Jupiter, and Neptune each share their orbit with
53: asteroids orbiting near the stable (L4/L5) Lagrange points that
54: lead/trail the planet by $\simeq~60^{\circ}$.
55: While the mass
56: ratios of the Trojan systems in our solar system are rather extreme
57: ($\le7\times10^{-9}$), it is possible that extrasolar planets may have
58: much more massive Trojans. Indeed, theorists have already outlined
59: several possible mechanisms to form Trojans with mass ratios
60: potentially including unity (Laughlin \& Chambers 2002; Chiang \& Lithwick 2005; Morbidelli et al.\ 2005; Thommes 2005; Cresswell \& Nelson 2006; Ford \& Gaudi 2006).
61:
62: Trojans of both Jupiter and Neptune have provided clues about our own
63: solar system's history (Michtchenko, Beauge \& Roig 2001; Kortenkamp,
64: Malhotra \& Michtchenko 2003; Chiang \& Lithwick 2005; Morbidelli et
65: al.\ 2005). Similarly, the detection of extrasolar Trojans would be
66: useful for constraining theories of planet formation and migration.
67: While all the above mechanisms predict that Trojans would survive the
68: migration process, there are alternative models of planet migration
69: that predict Trojans would not survive (Rasio \& Ford 1996; Wu \& Murray 2003;
70: Gaudi 2003; Ford \& Rasio 2006; Ford \& Gaudi 2006).
71: %
72: The detection of a Trojan companion to a short-period planet would
73: present a serious challenge to these mechanisms for forming ``hot
74: Jupiters'' and would imply that the planet in such a system was formed
75: via migration through a dissipative disk rather than tidal
76: circularization after approaching the star on a highly eccentric
77: orbit. Thus, searching for extrasolar Trojans can test models of
78: planet formation.
79: %
80:
81: %\subsection{Existing methods and previous constraints}
82:
83: Previously, three methods have been proposed to identify extrasolar
84: Trojans. If a Trojan is sufficiently massive and has a sufficiently
85: large libration amplitude, then its presence could be inferred using
86: the deviations from a Keplerian perturbation to the stellar radial
87: velocity or astrometric signal caused by a single planet. Laughlin \&
88: Chambers (2002) have shown that two comparable mass giant planets
89: occupying a 1:1 mean motion resonance would typically have strong
90: planet-planet gravitational interactions on a secular timescale.
91: However, these signatures may not be unique: a reanalysis of the RV
92: observations of HD 128311 and HD 82943 have shown that both of the
93: current data sets are also consistent with a pair of planets in a 1:1
94: mean motion resonance (Gozdziewski \& Konacki 2006), as well as the
95: originally published orbital solutions.
96:
97:
98: If a Trojan transits its parent star, then
99: photometric or spectroscopic monitoring of stars with transiting
100: planets (particularly at times offset from the planet transit by
101: $\sim~P$/6) may reveal the Trojan transit via the decrease in stellar
102: flux (Rowe et al.\ 2000) or anomalous RV excursions due to the
103: Rossiter-McLaughlin effect (Gaudi \& Winn 2006). Both these methods
104: are more sensitive to large Trojans. While ground-based observations are
105: not sensitive to Earth-sized planets, space observations could detect
106: such Trojans. However, it is not guaranteed that a Trojan will transit its
107: parent star, as it may have a significant inclination (e.g.,
108: Morbidelli et al.~ 2005). Further, since the libration period can be quite
109: large, long-term monitoring would be required to ensure detection.
110:
111: Ford \& Gaudi (2006) proposed a method for detecting a Trojan
112: companion based on combining radial velocity observations and
113: photometric observations of a transiting planet. Even if the Trojan
114: itself were not transiting, it could reveal it's presence via a time
115: lag between the radial velocity null and the time of central transit.
116: Existing observations already place significant (99.9\%) upper limits
117: on the mass of Trojan companions to HD 209458b and HD 149026b of
118: $\simeq~13M_\oplus$ and $\simeq~25M_\oplus$.
119:
120: Here, we present another method for detecting Trojan companions to
121: extrasolar planets using only photometric observations of transiting
122: extrasolar planets.
123: %
124: Once a transiting planet has been identified, higher precision
125: follow-up observations and modeling can precisely measure the mid-time
126: of subsequent transits (currently with a precision $\sim$10s; e.g.,
127: Agol \& Steffen 2006; Holman et al.\ 2006; Winn et al.\ 2006). If the
128: star and the transiting giant planet are the only bodies in the
129: system, then the transits will be strictly periodic, i.e., $t_i = t_0
130: + i\times P_s + \delta t_i$, where $t_i$ is the time of the $i$th
131: transit, $P_s$ is the transiting planet's sidereal orbital period, and
132: any transit timing variations ($\delta t_i$) are due to measurement
133: error. However, if an additional planet orbits the star, then the
134: times of the giant planet's transits will be affected (Miralda-Escude
135: 2002; Holman \& Murray 2005; Agol et al.\ 2006; Heyl \& Gladman 2006).
136: %
137: % Similarly, the gravitational perturbation of a Trojan librating about the L4/L5 fixed point would introduce a periodic signal into the transit timing.
138: %
139: By analyzing the deviations of the observed TTVs from
140: a strictly periodic model ($\delta t_i$), astronomers can search for
141: additional planets orbiting the star. Here, we show that a sub-Earth
142: mass Trojan planet could induce a transit timing signal that is easily
143: measurable using existing ground-based observatories.
144:
145:
146: \section{Observational Constraints on Trojans}
147: %
148: We consider a three body system and denote the stellar mass
149: ($m_{\star}$), the planet mass ($m_p$), and the Trojan mass ($m_T$).
150: We refer to all bodies librating about the L4 or L5 fixed point of a
151: planet as ``Trojans''.
152: %
153: If there are no
154: other massive bodies in the system, then the L4/L5 fixed points are
155: stable for circular orbits if the ratio,
156: $\mu=(m_p+m_T)/(m_\star+m_p+m_T)$, is less than a critical threshold
157: $\mu_c$, where $0.03812\le\mu_{c}\le0.03852$ and $\mu_c$ depends on
158: the ratio, $\epsilon\equiv m_T/(m_p+m_T)$ (Laughlin \& Chambers 2002).
159: If the planet and Trojan have equal eccentricities and the Trojan
160: resides exactly at the L4/L5 fixed point, then the transit timing
161: signature for the primary planet would be indistinguishable from a
162: similar system without a Trojan.
163: %
164: More generally, for a Trojan on an orbit that is librating about the
165: L4/L5 fixed point, the times of the primary planet's transits will
166: deviate from being strictly periodic. Here, we focus our attention on
167: Trojans that undergo small librations about the L4/L5 fixed points and
168: are significantly less massive than the currently known planet.
169:
170: The libration can be approximated as a linear superposition of two
171: epicyclic motions. The star-Trojan separation can oscillate
172: about the semi-major axis of the planet ($a_p$) with the amplitude,
173: $\delta a \ll \mu^{1/2} a_p$ on a timescale $\tau_{\rm fast}\simeq
174: P_s$, and the guiding center of the Trojan can oscillate on a longer
175: timescale, $\tau_{\rm slow}\simeq P_s\sqrt{27/(4\mu)}$ (Murray \&
176: Dermott 2000). The timescale of the libration of the guiding center
177: makes this motion most readily detectible by transit timing
178: observations.
179:
180: For a transiting planet, both $P_s$ and each $t_i$ can be measured
181: precisely using photometry alone. Considering a series of continuous
182: photometric observations with uncorrelated Gaussian uncertainties of
183: magnitude $\sigma_{ph}$, taken at a rate $\Gamma$ around a single
184: transit, the mid-transit time can be measured with an accuracy of
185: $\sigma_{t_i}\simeq~\sqrt{t_e/2\Gamma}\sigma_{ph}\rho^{-2}$, where
186: $t_e$ is the duration of ingress/egress and $\rho$ is the ratio of the
187: planet radius to stellar radius (Ford \& Gaudi 2006). For typical
188: parameters (e.g., $\sigma_{ph}\sim~10^{-3}$), $t_i$ can be measured to
189: $\simeq10$s (e.g., Brown et al.\ 2001; Holman et al.\ 2006). The
190: period can be measured much more accurately than $t_i$, from
191: observations of multiple transits separated by many orbits.
192: %
193: For small amplitude libration about L4/L5 and circular orbits, the
194: transit timing perturbation is given by $\delta t_i \simeq \epsilon
195: P_s \Delta M(t_i) / (2\pi)$, where $\Delta M(t_i)$ is the angular
196: displacement of the Trojan from L4/L5 at the time of the $i$th
197: transit. The TTVs can be modeled by a sinusoid,
198: %
199: $\delta t_i = K_{\rm tt} \sin\left(2\pi\left(t-t_0\right)/P_{\rm TTV}+\phi\right)$,
200: where
201: $K_{\rm tt}$ is the amplitude of the transit timing variations
202: and $P_{\rm TTV}\sim\tau_{\rm slow}$.
203: If the dominant periodicity of the transit timing variations ($P_{\rm
204: TTV}$) is well determined, then the remaining parameters can be
205: determined via linear least squares fitting to the observed transit
206: times.
207: %
208: The transit timing variations will have an amplitude
209: %
210: \be
211: %
212: K_{\rm tt} \simeq 60s \left(\frac{P}{4d}\right) \left(\frac{m_T}{m_\oplus}\right) \left(\frac{0.5 M_{\rm Jup}}{m_p+m_T}\right) \left(\frac{K_{\Delta M}}{10^\circ}\right), \\
213: %
214: \ee
215: where $K_{\Delta M}$ is the amplitude of the Trojan's angular displacement from the Lagrange point. For small amplitude libration, $K_{\Delta M}\simeq\mathrm{max}\left|\Delta M\right|$ and $\mathrm{rms}(\delta t_i)\simeq K_{\rm tt} / \sqrt{2}$ (see Fig. 1)
216: %
217: Libration amplitudes of $K_{\Delta M}\sim~5-25^\circ$ are common for
218: Trojans orbiting near the Sun-Jupiter Lagrange points (Murray \&
219: Dermott 2000).
220:
221: The Lomb-Scargle periodogram can be easily adapted to efficiently scan
222: a range of putative periods and identify any significant periodicities
223: (Cumming 2004). If we assume that there are many ($N_{\rm tt}$)
224: transit timing observations with uncorrelated Gaussian uncertainties
225: $\sigma_{t_i}=\sigma_{tt}$, that the transit timing observations are
226: evenly distributed, and the duration of observations ($T_{\rm obs}$)
227: is greater than than $P_{\rm TTV}$, then a periodogram-style analysis
228: results in a 50\% chance of detecting a Trojan if
229: %
230: $K_{tt}\ge K_{1/2} \simeq \sigma_{tt}\left(\frac{4}{N_{\rm tt}}
231: \log\left[T_{\rm obs} / \left(2 F P_s\right) \right] \right)^{1/2}$
232: (Cumming 2004), where $F$ is the false alarm probability, which we set
233: to $10^{-3}$. For $N_{\rm tt} = T_{\rm obs}/P_s = 40$, $K_{1/2}
234: \simeq \sigma_{tt}$, so sub-Earth-mass Trojans could be readily
235: detected. We note that all published transit timing data sets have
236: $N_{\rm tt}<20$, which results in a significantly reduced sensitvity,
237: if $P_{\rm TTV}$ is unknown {\em a priori}. In this small-$N_{\rm
238: tt}$ regime, a simple $\chi^2$ test of the null hypothesis ($\delta
239: t_i=0$) is more sensitive for detecting transit timing variations.
240: However, if only a single periodicity (e.g., $\tau_{\rm slow}$) is to
241: be tested, then even a modest number of observations can be quite
242: sensitive (e.g., $K_{1/2}\simeq 2.5\sigma_{\rm tt}$ even for $N_{\rm
243: tt}=13$).
244: %
245:
246: Once a Trojan has been detected, a Fisher information analysis (e.g.,
247: Gaudi \& Winn 2006) reveals that the uncertainty in $K_{\rm tt}$ will
248: approach
249: $\sigma_{K_{\rm tt}} = \sqrt{4/N_{tt}} \sigma_{tt}$.
250: If a Trojan were present, then the uncertainty in
251: $K_{\rm tt}$ would set the uncertainty in the measurement of the
252: mass of the Trojan to be
253: %
254: \be
255: \label{EqnMass}
256: \frac{\sigma_{m_T}}{m_\oplus} \simeq \frac{0.5}{\sqrt{N_{\rm tt}}} \left(\frac{\sigma_{\rm tt}}{10s}\right) \left(\frac{4d}{P}\right) \left(\frac{m_p}{0.5 M_{\rm Jup}}\right) \left(\frac{10^\circ}{K_{\Delta M}}\right) \\
257: \ee
258: %
259: Thus, transit timing observations can be very sensitive to
260: sub-Earth-mass Trojan companions. However, due to the degeneracy
261: between $m_T$ and $K_{\Delta M}$, the amplitude of transit timing
262: detections due to a Trojan would not provide a strong upper limit on
263: the Trojan mass. For Trojans with significant libration amplitudes,
264: this degeneracy could be resolved by combining the amplitude with the
265: measured $P_{\mathrm{TTV}}$ (see Fig.\ 1 center).
266: For Trojans with small libration amplitudes, the measured $P_{\mathrm{TTV}}$
267: will provide an upper limit for the libration amplitude and
268: hence a lower limit to the Trojan-planet mass ratio. Regardless, the
269: transit timing observations could be used to predict the time of
270: transit of a Trojan and targeted photometric or spectroscopic
271: follow-up observations could place an upper limit on the radius of the
272: Trojan (e.g., Rowe et al.\ 2006; Gaudi \& Winn 2006). Combining such
273: observations with planetary structure models (e.g., Valencia et al.\
274: 2007) could provide an upper limit on the Trojan mass and hence a
275: lower limit for the libration amplitude.
276: %If we were to demand a measurement of $\Delta t>3.291\sigma_{\Delta
277: %t}$ to claim the detection of a Trojan, then a total of $\simeq~$XXX
278: %high precision transit timing measurements would be required to detect
279: %a $\simeq~XM_{\oplus}$ Trojan, assuming a host star with and
280: %measurement uncertainties added in quadrature. While challenging, it
281: %is remarkable that current ground based instruments have the necessary
282: %precision to detect such a low mass Trojan with a plausible amount of
283: %observing time.
284:
285: \section{Discussion}
286:
287: % \subsection{Complications}
288:
289: One long-term goal of immense scientific and public interest is to
290: discover and study rocky planets, and eventually terrestrial planets
291: that have masses, sizes, orbits, atmospheres, and perhaps even surface
292: conditions similar to those of the Earth.
293: %
294: Previous studies have demonstrated that the magnitude and timescale of
295: transit timing variations due to Earth-mass planets are readily
296: detectable if they orbit near an interior or exterior low-order mean
297: motion resonance (Holman \& Murray 2005; Agol et al.\ 2005). The TTV
298: method is quite sensitive to planets near resonances (e.g., the
299: planetary systems GJ 876, HD 128311, HD 73526, 55 Cnc, and HD 82943;
300: Butler et al.\ 2006) that are particularly valuable for studying
301: orbital dynamics and planet formation (Lee \& Peale 2002) and
302: challenging for radial velocity and astrometric searches. The TTV
303: method would also be able to confirm some planet candidates (likely to
304: be identified by future transit searches) by detecting the orbital
305: interactions of the planets, similar to the methods used for
306: confirming the planets around PSR1257+12 (Rasio 1992; Malhotra 1993)
307: and PSR1620-26 (Ford et al.\ 2000). This could prove particularly
308: valuable for planet candidates that have small masses and/or orbit
309: faint stars, so that radial velocity confirmation is impractical
310: (e.g., most of the 16 transiting planet candidates orbiting faint
311: stars recently published by Sahu et al.\ 2006, and the many transiting
312: planets expected to be found by future space missions). \hfill
313:
314:
315: It would be extremely exciting to detect a {\em transiting} Earth-mass
316: planet. Such a detection would enable follow-up observations to study
317: the physical properties of the planet, such as the planet's radius and
318: density (Brown et al.\ 2001, Sato et al.\ 2005, Charbonneau et al.\
319: 2006), the atmospheric composition (Charbonneau et al.\ 2002, Deming
320: et al.\ 2005, Bozorgnia et al.\ 2006), and possibly even ``resolve''
321: surface/atmospheric features (Ford et al.\ 2001; Harrington et al.\
322: 2006; Gaidos et al.\ 2006).
323:
324:
325: We have demonstrated that a sub-Earth-mass Trojan planet could also
326: result in a transit timing signal that can be readily measured with
327: ground based observatories. Since the orbital planes are likely
328: nearly aligned, the fact that a giant planet is already known to
329: transit the star increases the odds that other planets orbiting that
330: star will also transit (Holman \& Murray 2005). Thus, the transit
331: timing method is particularly good at searching for transiting
332: Earth-like Trojan planets that would enable extremely interesting
333: follow-up observations. Our technique could be applied to search for
334: terrestrial-mass Trojans of giant planets orbiting in the habitable
335: zone of their stars (Schwarz et al.\ 2005), particularly for low mass
336: stars where the habitable zone can be $\simeq~0.015$~AU away from the
337: star. Once transitting terrestrial mass planets are discovered, this
338: technique could be extended to search for extrasolar Trojans with
339: asteroid-like masses.
340:
341:
342:
343:
344: While \S2 and previous work have emphasized the sensitivity of transit
345: timing observations, we caution that solving the inverse problem of
346: determing planet properties from transit timing observations is likely
347: to pose a significant challenge and be more difficult than
348: interpreting other types of extrasolar planet observations. For
349: example, in the radial velocity method, the dominant periodicity in
350: the observed time series is readily identified with the orbital period
351: of a massive companion and the amplitude of the variations is
352: proportional to the mass of the companion (Konacki \& Maciejewski
353: 1999). However, in TTV data, the dominant periodicity could be due to
354: any one of several physical effects (see Fig.\ 2), including the
355: reflex motion of the star due to the second planet (with a period
356: equal to the orbital period of the second planet), the long-term
357: mutual gravitational perturbations between the planets (with a period
358: much longer than either orbital period), the short-term gravitational
359: perturbations on the orbit of the transiting giant planet (on an
360: intermediate timescale), or the light travel time due to a distant
361: companion (e.g., Borkovitz et al.\ 2003; Heyl \& Gladman 2006).
362: Therefore, even once a periodicity has been identified, it is not
363: obvious what physical effect is causing the periodicity.
364: %
365: Further, TTV signatures are more complex than the signatures of other
366: dynamical detection techniques. For example, radial velocity
367: observations of a multiple planet system can often be modeled by the
368: linear superposition of multiple Keplerian orbits (Butler et al.\
369: 2006; Ford et al.\ 2006). However, for transit timing observations,
370: the signal is often dominated by the deviations from such a simplified
371: model.
372: %
373: Therefore, it is necessary to perform n-body simulations to accurately
374: calculate the TTV signature of each possible model (Holman \& Murray
375: 2005; Agol et al.\ 2005; Steffen \& Agol 2005). Given the
376: computational requirements of each n-body integration, practical
377: algorithms must explore the high-dimensional ($\simeq7\times~N_{pl}$)
378: parameter space very efficiently and rapidly converge on all physical
379: models consistent with the observations.
380:
381:
382: A Trojan companion with small libration amplitude will induce a TTV
383: signature that can be well approximated by a single sinusoid (Fig.\ 2,
384: top). This contrasts with the TTV perturbations due to a planet near
385: a different resonance (Fig.\ 2, middle and bottom). A moon could also
386: result in a nearly sinusoidal TTV signature, but on a timescale this
387: is typically much shorter than $\tau_{\mathrm{slow}}$ (due to
388: dynamical stability constraints). TTV perturbations with a timescale
389: near $\tau_{\mathrm{slow}}$ and due to non-resonant planets will have
390: small amplitude, unless the outer planet is quite massive and
391: potentially detectable by other methods. Therefore, we suggest that a
392: large amplitude sinusoidal signal with a period near
393: $\tau_{\mathrm{slow}}$ might allow Trojans to be uniquely identified.
394: We suggest future investigations to test this conjecture.
395:
396: We caution that the TTV signature of an extrasolar Trojan could also
397: be non-sinusoidal. For example, Trojans with large libration
398: amplitudes can become significantly non-sinusoidal (reducing the rms
399: TTV by upto $\sim40\%$). A Trojan planet in a horseshoe-shaped orbit
400: would produce much larger TTV perturbations with a very different
401: shape. If the primary and Trojan planets have different
402: eccentricities, then there will be additional longer term
403: periodicities in the TTV signal due to secular perturbations (Fig.\ 1,
404: right). If there are Trojans at both L4 and L5, then the transit
405: timing signature could be approximated as the sum of two such signals
406: (similar frequency, but different amplitudes and phases). Similarly,
407: swarms of Trojan companions librating about L4 and L5 could be modeled
408: as the superposition of many such signals, provided that their mutual
409: interactions are negligible. Additional planets could also perturb
410: the time of central transit (Holman \& Murray 2005; Agol et al.\ 2005)
411: such that the offset will vary from transit to transit. Therefore,
412: many transits should be observed to verify that any observed offsets
413: are not due to perturbations by a more distant giant planet.
414:
415: The interpretation of actual TTV observations will be further
416: complicated by constrained sampling (observations only possible during
417: transit), incomplete sampling (due to available telescope time, and
418: weather; Agol \& Steffen 2006) and measurement errors, all of which
419: increases the uncertainties in the number, masses and orbits of
420: planets. These limitations underscore the need for powerful
421: statistical methods to interpret TTV observations. In cases where multiple
422: orbital models are consistent with TTV data, additional
423: observational constraints (e.g., radial velocities,
424: secondary transit, changing transit duration due
425: to inclination librations) could help identify the correct model. We
426: encourage further research in such methods, so that ongoing TTV
427: observations can be appropriately analyzed.
428:
429:
430:
431: %
432: \acknowledgments
433: %
434: We thank Eric Agol, Dan Fabrycky, Scott Gaudi, Jason Steffen, and Josh Winn, for helpful comments.
435: %
436: Support for EBF was provided by NASA through Hubble Fellowship grant
437: HST-HF-01195.01A awarded by the Space Telescope Science Institute,
438: which is operated by the Association of Universities for Research in
439: Astronomy, Inc., for NASA, under contract NAS 5-26555.
440: %
441: MJH acknowledges support for this work NASA Origins grant
442: NG06GH69G.
443: %
444: \begin{thebibliography}{}
445: %
446: \bibitem[Agol et al.(2005)]{2005MNRAS.359..567A} Agol, E., Steffen, J.,
447: Sari, R., \& Clarkson, W.\ 2005, \mnras, 359, 567
448:
449: \bibitem[Agol \& Steffen(2007)]{9001} Agol, E, Steffen, J.H. 2007, \mnras, 374, 941. % HD 209458b
450:
451: \bibitem[Bozorgnia et al.(2006)]{2006PASP..118.1252B} Bozorgnia, N., Fortney, J.~J., McCarthy, C., Fischer, D.~A., \& Marcy, G.~W.\ 2006, \pasp, 118, 1252
452:
453:
454: \bibitem[Brown et al.(2001)]{2001ApJ...552..699B} Brown, T.~M., Charbonneau, D., Gilliland, R.~L., Noyes, R.~W., \& Burrows, A.\ 2001, \apj, 552, 699
455:
456: \bibitem[Butler et al.(2006)]{2006ApJ...646..505B} Butler, R.~P., et al.\ 2006, \apj, 646, 505
457:
458: \bibitem[Charbonneau et al.(2002)]{2002ApJ...568..377C} Charbonneau, D., Brown, T.~M., Noyes, R.~W., \& Gilliland, R.~L.\ 2002, \apj, 568, 377
459:
460: \bibitem[Charbonneau et al.(2006)]{2006ApJ...636..445C} Charbonneau, D., et al.\ 2006, \apj, 636, 445 % 149026b
461:
462:
463: \bibitem[Chiang \& Lithwick(2005)]{2005ApJ...628..520C} Chiang, E.~I., \& Lithwick, Y.\ 2005, \apj, 628, 520
464:
465:
466: \bibitem[Cresswell \& Nelson(2006)]{2006A&A...450..833C} Cresswell, P., \& Nelson, R.~P.\ 2006, \aap, 450, 833
467:
468: \bibitem[Cumming(2004)]{9002} Cumming, A. 2004, \mnras, 354, 1165
469:
470: \bibitem[Deming et al.(2005)]{2005Natur.434..740D} Deming, D., Seager, S., Richardson, L.~J., \& Harrington, J.\ 2005, \nat, 434, 740
471:
472: \bibitem[Ford et al.(2000)]{2000ApJ...528..336F} Ford, E.~B., Joshi, K.~J.,
473: Rasio, F.~A., \& Zbarsky, B.\ 2000, \apj, 528, 336
474:
475: \bibitem[Ford et al.(2001)]{2001Natur.412..885F} Ford, E.~B., Seager, S., \& Turner, E.~L.\ 2001, \nat, 412, 885
476:
477: \bibitem[Ford \& Gaudi(2006)]{FordGaudi06} Ford, E.~B., \& Gaudi, B.~S. 2006, \apjl, 652, L137
478:
479: \bibitem[Ford \& Rasio(2006)]{2006ApJ...638L..45F} Ford, E.~B., \& Rasio, F.~A.\ 2006, \apjl, 638, L45
480:
481: \bibitem[Ford \& Chiang(2007)]{FordChiang06} Ford, E.~B., \& Chiang, E.~I. 2007, \apj, in press [astro-ph/0701745]
482:
483: \bibitem[Gaidos et al.(2006)]{2006dies.conf..153G} Gaidos, E., Moskovitz, N., \& Williams, D.~M.\ 2006, IAU Colloq.~200: Direct Imaging of Exoplanets: Science \& Techniques, 153
484:
485: \bibitem[Gaudi(2003)]{2003astro.ph..7280G} Gaudi, B.~S.\ 2003, preprint (astro-ph/0307280)
486:
487: \bibitem[Gaudi \& Winn(2006)]{2006astro.ph..8071G} Gaudi, B.~S., \& Winn,
488: J.~N.\ 2006, preprint (astro-ph/0608071)
489:
490: \bibitem[Go{\'z}dziewski \& Konacki(2006)]{2006ApJ...647..573G} Go{\'z}dziewski, K., \& Konacki, M.\ 2006, \apj, 647, 573
491:
492: \bibitem[Harrington et al.(2006)]{2006Sci...314..623H} Harrington, J., Hansen, B.~M., Luszcz, S.~H., Seager, S., Deming, D., Menou, K., Cho, J.~Y.-K., \& Richardson, L.~J.\ 2006, Science, 314, 623
493:
494: \bibitem[Heyl \& Gladman(2006)]{2006astro.ph.10267H} Heyl, J.~S., \& Gladman, B.~J.\ 2006, ArXiv Astrophysics e-prints, arXiv:astro-ph/0610267
495:
496: \bibitem[Holman \& Murray(2005)]{2005Sci...307.1288H} Holman, M.~J., \&
497: Murray, N.~W.\ 2005, Science, 307, 1288
498:
499: \bibitem[Holman et al.(2006)]{2006astro.ph..7571H} Holman, M.~J., et al.\
500: 2006, \apj, 652, 1715 % TLCP I XO-1b
501:
502: \bibitem[Konacki \& Maciejewski(1999)]{1999ApJ...518..442K} Konacki, M., \& Maciejewski, A.~J.\ 1999, \apj, 518, 442
503:
504: \bibitem[Laughlin \& Chambers(2002)]{2002AJ....124..592L} Laughlin, G., \&
505: Chambers, J.~E.\ 2002, \aj, 124, 592
506:
507: \bibitem[Lee \& Peale(2002)]{2002ApJ...567..596L} Lee, M.~H., \& Peale, S.~J.\ 2002, \apj, 567, 596
508:
509: \bibitem[Malhotra(1993)]{1993ApJ...407..266M} Malhotra, R.\ 1993, \apj, 407, 266
510:
511: \bibitem[Michtchenko et al.(2001)]{2001AJ....122.3485M} Michtchenko, T.~A., Beaug{\'e}, C., \& Roig, F.\ 2001, \aj, 122, 3485
512:
513: \bibitem[Miralda-Escud{\'e}(2002)]{2002ApJ...564.1019M} Miralda-Escud{\'e}, J.\ 2002, \apj, 564, 1019
514:
515:
516: \bibitem[Morbidelli et al.(2005)]{2005Natur.435..462M} Morbidelli, A.,
517: Levison, H.~F., Tsiganis, K., \& Gomes, R.\ 2005, \nat, 435, 462
518:
519:
520: \bibitem[Murray \& Dermott(2000)]{2000ssd..book.....M} Murray, C.~D., \&
521: Dermott, S.~F.\ 2000, Solar System Dynamics, Cambridge, UK: Cambridge University Press.
522:
523: \bibitem[Rasio \& Ford(1996)]{1996Sci...274..954R} Rasio, F.~A., \& Ford,
524: E.~B.\ 1996, Science, 274, 954
525:
526: \bibitem[Rasio et al.(1992)]{1992Natur.355..325R} Rasio, F.~A., Nicholson, P.~D., Shapiro, S.~L., \& Teukolsky, S.~A.\ 1992, \nat, 355, 325
527:
528: \bibitem[Rowe et al.(2006)]{2006ApJ...646.1241R} Rowe, J.~F., et al.\ 2006,
529: \apj, 646, 1241
530:
531: \bibitem[Sahu et al.(2006)]{2006Natur.443..534S} Sahu, K.~C., et al.\ 2006, \nat, 443, 534
532:
533: \bibitem[Schwarz et al.(2005)]{2005AsBio...5..579S} Schwarz, R.,
534: Pilat-Lohinger, E., Dvorak, R., {\'E}rdi, B., \& S{\'a}ndor, Z.\ 2005,
535: Astrobiology, 5, 579
536:
537: \bibitem[Steffen \& Agol(2005)]{9003} Steffen, J.H. \& Agol, E. 2005, \mnras, 364, L96. % TrES-1b
538:
539: \bibitem[Thommes(2005)]{2005ApJ...626.1033T} Thommes, E.~W.\ 2005, \apj, 626, 1033
540:
541: \bibitem[Valencia(2007)]{9004} Valencia, D., Sasselov, D.D., O'Connell, R.J. 2007, \apj, 656, 545
542:
543: \bibitem[Winn et al.(2007)]{9005} Winn, J.N., Holman M.J., Roussanova, A. 2007, \apj, in press [astro-ph/0611404]
544:
545: \bibitem[Wu \& Murray(2003)]{2003ApJ...589..605W} Wu, Y., \& Murray, N.\
546: 2003, \apj, 589, 605
547:
548: \end{thebibliography}
549: %
550:
551:
552: \newpage
553:
554: \begin{figure}[htbp]
555: %\epsscale{0.7}
556: \plotone{f1.eps}
557: \caption{
558: %\hbox{\plotone{f1.eps}}
559: %\noindent Fig.\ 1:
560: Transit Timing Signatures:
561: We show the root mean square
562: deviations of the TTVs from a strict periodicity (top) and
563: the period of the transit timing variations (bottom)
564: based on direct n-body integrations of a system with
565: a $0.5M_{\rm Jup}$ planet and a Trojan orbiting a $M_\odot$ star.
566: Dotted lines are analytic expressions from \S2.
567: %
568: (Left) The planet and the Trojan companion with mass $m_T$ are
569: initially placed on circular orbits with a mean orbital separation of
570: 0.05AU and $\Delta M_T=10^{\circ}$.
571: %
572: (Center) As before, but as a function of $K_{\Delta M}$, the amplitude of the angular displacement from the Lagrange point, for a fixed Trojan mass of $1M_{\oplus}$.
573: %
574: (Right) As before, but as a function of initial eccentricity of the Trojan for a fixed Trojan mass of $1M_{\oplus}$ and initial $\Delta M_T=0^{\circ}$. The dotted curve shows the analytic model, $\mathrm{rms}(\delta t_{tr}) = \epsilon P_s e_T /(\pi\sqrt{2})$.
575: %
576: %
577: \label{fig_ttv_amp_mass} }
578: \end{figure}
579:
580:
581:
582:
583: \begin{figure}[htbp]
584: %\epsscale{0.7}
585: \plotone{f2.eps}
586: \caption{
587: %\hbox{\plotone{f2.eps} }
588: %
589: %\noindent Fig.\ 2:
590: Similar Transit Timing Signatures due to Very Different
591: Perturbing Planets: We plot the TTV residuals (disks) versus time for
592: three hypothetical planetary systems. (The dotted lines merely guide
593: the eye.) Each contains a typical transiting giant planet (0.5
594: Jupiter masses, orbital period of $4.09$days) and a second
595: planet. The perturbations are due to: top) a $1M_{\oplus}$
596: Trojan companion, middle) the
597: perturbations are due to a $28M_{\oplus}$ (or 0.3 Saturn-mass) planet
598: with a period of $\simeq~8.7$days (outside the 2:1 mean-motion
599: resonance), and bottom) a
600: $\simeq~4.8$ Earth-mass planet with a period of $\simeq~5.91$days
601: (inside the 3:2 mean-motion resonance). Each planetary systems
602: results in a TTV signature that has a dominant periodicity of
603: $71.40$days and a root-mean-square amplitude of $73.4$s (based 1000
604: transits). Thus, interpreting TTV observations will require combining
605: dynamical analyses with advanced statistical methods.
606: %
607: % P= 4.09166d P_TTV=71.4986d
608: %
609: \label{fig_ttv_ex_comp} }
610: \end{figure}
611:
612:
613:
614: \end{document}
615:
616:
617: