1: \documentclass{emulateapj}
2: \usepackage{amsmath}
3: \usepackage{graphicx}
4:
5: \shorttitle{Eccentricity Affects Detection of Transiting Planets}
6: \shortauthors{Christopher J. Burke}
7: \slugcomment{Submitted for publication in the Astrophysical Journal}
8: \received{2007 November 09}
9: \begin{document}
10: \title{Impact of Orbital Eccentricity on the Detection of Transiting Extrasolar Planets}
11: \author{Christopher J. Burke}
12: \affil{Space Telescope Science Institute, 3700 San Martin Dr., Baltimore, MD, 21218}
13: \email{cjburke@stsci.edu}
14:
15: \begin{abstract}
16: For extrasolar planets with orbital periods, P$>$10 days, radial
17: velocity surveys find non-circular orbital eccentricities are common,
18: $\langle e\rangle\sim 0.3$. Future surveys for extrasolar planets
19: using the transit technique will also have sensitivity to detect these
20: longer period planets. Orbital eccentricity affects the detection of
21: extrasolar planets using the transit technique in two opposing ways:
22: an enhancement in the probability for the planet to transit near
23: pericenter and a reduction in the detectability of the transit due to
24: a shorter transit duration. For an eccentricity distribution matching
25: the currently known extrasolar planets with P$>$10 day, the
26: probability for the planet to transit is $\sim 1.25$ times higher than
27: the equivalent circular orbit and the average transit duration is
28: $\sim 0.88$ times shorter than the equivalent circular orbit. These
29: two opposing effects nearly cancel for an idealized field transit
30: survey with independent photometric measurements that are dominated by
31: Poisson noise. The net effect is a modest $\sim 4\%$ increase in the
32: transiting planet yield compared to assuming all planets have circular
33: orbits. When intrinsic variability of the star or correlated
34: photometric measurements are the dominant source of noise, the transit
35: detectability is independent of the transit duration. In this case
36: the transit yield is $\sim$25\% higher than that predicted under the
37: assumption of circular orbits. Since the Kepler search for
38: Earth-sized planets in the habitable zone of a Solar-type star is
39: limited by intrinsic variability, the Kepler mission is expected to
40: have a $\sim$25\% higher planet yield than that predicted for circular
41: orbits if the Earth-sized planets have an orbital eccentricity
42: distribution similar to the currently known Jupiter-mass planets.
43: \end{abstract}
44:
45: \keywords{eclipses --- planetary systems --- techniques: photometric}
46:
47: \section{Introduction}
48:
49: The known extrasolar planets possess a broad distribution of orbital
50: eccentricity \citep{BUT06} (see Figure~\ref{fig:ecchist}). The short
51: period, Hot Jupiter (P$<$10 day) planets predominately have circular
52: orbits. However, at longer orbital periods circular orbits become a
53: minority and the median eccentricity for extrasolar planets
54: e$\sim$0.3. At the extreme eccentricity end, there are three planets,
55: HD 80606b \citep{NAE01}, HD 20782b \citep{JON06}, and HD 4113b
56: \citep{TAM07}, having e$>$0.9. HD 80606b comes closer to its stellar
57: host (a=0.033 AU) than many of the circular orbit Hot Jupiter planets.
58: \citet{FOR07} recently reviewed the various mechanisms invoked to explain the
59: distribution of eccentricities for the known extrasolar planets.
60: Interactions with a stellar companion, planetary companion, passing
61: star, gaseous disk, planetesimal disk, and stellar jets have all been
62: proposed to modify the orbital eccentricity of extrasolar planets.
63:
64: Limited discussions in the literature have been given to the impact
65: orbital eccentricity has on a transit survey for extrasolar planets.
66: \citet{TIN05} discuss the impact of eccentricity on their $\eta$
67: parameter ($\eta$ is the ratio of the observed transit duration to an
68: estimate of the transit duration). As expected, they find transits
69: occur near pericenter (apocenter) are shorter (longer) in duration
70: than the circular orbit case, and they show that the transit duration
71: of an eccentric is typically shorter than a circular orbit of the same
72: period. However, their discussion was focused on the impact of
73: orbital eccentricity on their $\eta$ parameter. \citet{MOU06} also
74: discuss how the transit duration is affected by orbital eccentricity,
75: but they do not quantify the impact this will have on transit surveys.
76: Recently, \citet{BAR07} derives the probability for a planet on an
77: eccentric orbit to transit and conclude that the photometric precision
78: of current surveys and future surveys, such as Kepler, is insufficient
79: to determine the orbital eccentricity solely from the light curve.
80: \citet{BAR07} concludes that without knowledge of the eccentricity
81: from radial velocity data or independent measurement of the stellar
82: host radius, the habitability of planets detected with Kepler will
83: remain unknown.
84:
85: Neglecting the impact eccentricity has on transit detections is
86: justified for the current sample of transiting planets given the
87: predominance of circular orbits for the Hot Jupiters and the strong
88: bias of transit surveys against finding long period planets on
89: circular orbits \citep{GAU05}. The announced transit for the planet
90: orbiting HD 17156\footnote{First detected by the radial velocity
91: technique \citep{FIS07}} with a 21 day period and eccentric orbit
92: \citep{BARB07} is a precursor for the kinds of planets detectable in
93: transit surveys. As transit surveys continue, longer period
94: transiting planets may be discovered. More importantly, the recently
95: launched COROT mission will surpass current surveys for sensitivity to
96: longer period planets \citep[P$\sim$ several months; ][]{BORD03}.
97: Also, the Kepler mission, scheduled for launch in 2009, has a goal to
98: find Earth-sized objects at 1 AU from their host star \citep{BORU04}.
99: The main purpose of this paper is to show that given the distribution
100: of eccentricities for the currently known extrasolar planets,
101: eccentricity should not be ignored in assessing the detectability of
102: transiting giant planets when the transit survey is sensitive to
103: planets with P$>$ 10 day.
104:
105: In addition to longer period planets, the COROT and Kepler missions
106: also will detect transiting planets with small, Earth-sized radii.
107: The eccentricity distribution for planets less massive than the
108: currently known Jupiter-mass planets is beginning to be explored.
109: \citet{RIB07} and \citet{FOR07} provide tentative evidence for the
110: tendency of lower mass planets to have lower eccentricities in the
111: current sample of radial velocity planets. There is theoretical
112: agreement that a proto-planetary gas disk strongly damps the
113: eccentricity of non-gap opening embedded low-mass planets
114: \citep{GOL80, CRE07}. However, the theoretical models show that once
115: the gas disk dissipates, dynamical interactions amongst the planets
116: results in a random-walk diffusion that leads to an increasing
117: eccentricity that takes on a Rayleigh-distribution similar to what is
118: observed \citep[dotted line in Figure~\ref{fig:ecchist};
119: ][]{JUR07,ZHO07}. Opposing the increases in eccentricity from
120: planet-planet scattering, late stage interactions with planetesimals
121: can preferentially damp the eccentricities of lower mass planets
122: enabling theoretical models to achieve the low eccentricities of the
123: terrestrial planets of the Solar System \citep{RAY06} and possibly
124: explain the current trend of lower eccentricities for lower mass
125: planets \citep{FOR07}. Given the number of potential physical processes that can affect
126: orbital eccentricity, it is premature to assume that the typical
127: Earth-like planet has an eccentricity near zero like the Solar System.
128:
129: In this study, \S~\ref{sec:eccdist} reviews the observed eccentricity
130: distribution of the known radial velocity extrasolar planets. The
131: broad distribution of orbital eccentricity has two main effects on the
132: sensitivity of a transit survey. First, the planet-host separation
133: varies along an eccentric orbit, enhancing the probability to transit
134: when the planet is relatively closer to the stellar host.
135: \S~\ref{sec:tranprob} quantifies the net affect on the transit
136: probability for a population of planets with non-circular orbits.
137: Second, the planet velocity varies along and eccentric orbit,
138: resulting in a reduction or lengthening of the transit duration.
139: \S~\ref{sec:trandur} quantifies the distribution of transit durations
140: resulting from a population of planets on eccentric orbits.
141: \S~\ref{sec:disc} describes how the transit duration affects transit
142: detection for transit surveys in the limit of various noise sources.
143: \S~\ref{sec:conc} concludes by quantifying the net result of the two
144: aforementioned effects, the enhanced probability to transit and the
145: reduced detectability, on the yield from transit surveys.
146:
147: \section{Eccentricity Distribution}\label{sec:eccdist}
148:
149: The solid histograms in Figure~\ref{fig:ecchist} show the normalized
150: distribution of eccentricities for the known extrasolar planets with
151: $P>10$ day \citep{BUT06}. The top panel shows the eccentricity
152: distribution from planets reported before November 2006 and the bottom
153: panel shows the eccentricity distribution including more recent
154: discoveries (September 2007). The more recent radial velocity
155: discoveries over the last year have increased the sample of low
156: eccentricity systems. Results from this study are given for both
157: epochs of the observed eccentricity distribution in order to
158: characterize how the uncertainty in the underlying orbital
159: eccentricity distribution affects the results. No attempt was
160: made to select planets with well determined eccentricities, avoidance
161: of multiple planet or multiple star systems, correction for observational
162: biases against high eccentricity planets
163: \citep{CUM04}, or evolution of orbital elements due to tidal
164: circularization. Planets with $P<10$ day are not included in order to
165: examine the potential issues that future transit surveys sensitive to
166: longer period planets will encounter. In this period regime (P$>$10
167: day) high eccentricity planets are quite common ($\langle
168: e\rangle=0.3$ and $\sigma_{e}=0.2$).
169:
170: \begin{figure}
171: %\plotone{ecchist.ps}
172: %\plotone{f1.ps}
173: \includegraphics[scale=0.5]{f1.ps}
174: \caption{Normalized histogram of the eccentricity distribution for known extrasolar planets with P$>$10 day. The eccentricity distribution is given for the known extrasolar planets at two epochs: November 2006 ({\it Top Panel}) and September 2007 ({\it Bottom Panel}). For this study a two-piece model parameterizes the eccentricity distribution ({\it solid line}). The model has 2 parameters: $e_{\rm crit}$ and $e_{\rm max}$. The model distribution is uniform at low eccentricity up to $e_{\rm crit}$; and toward higher eccentricity, the model linearly decreases to zero-value at $e_{\rm max}$. Also shown in the top panel is the Rayleigh distribution that describes the eccentricity distribution from the planet-planet scattering calculations of \citep{JUR07} ({\it dotted line}). The dotted histogram in the bottom panel shows the eccentricity distribution from the latest epoch (September 2007) that excludes planets with P$<$ 20 day.\label{fig:ecchist}}
175: \end{figure}
176:
177:
178: A two-piece model parameterizes the eccentricity distribution from the
179: earlier epoch (top panel of Figure~\ref{fig:ecchist}). The first
180: piece at low eccentricity is flat up to $e_{\rm crit}$. The second piece matches the first piece at $e_{\rm crit}$, linearly decreases toward
181: higher eccentricities, and is zero at $e_{\rm max}$. The normalized two-piece model is given by the equation,
182:
183: \begin{equation}
184: P(e)de=\left\{ \begin{array}{ll}
185: \frac{2}{e_{\rm crit}+e_{\rm max}} & 0 \le e \le e_{\rm crit} \\
186: \frac{2}{e_{\rm crit}+e_{\rm max}}\frac{(e-e_{\rm max})}{(e_{\rm crit}-e_{\rm max})} & e_{\rm crit} < e \le e_{\rm max} \\
187: 0 & e_{\rm max} < e < 1 \end{array}
188: \right. .\label{eq:eccdist}
189: \end{equation}
190:
191: A $\chi^{2}$ minimization yields the best model parameters $e_{crit}=0.25$ and
192: $e_{max}=0.92$. The $\chi^{2}$ minimization applied to the more recent
193: epoch eccentricity distribution (bottom panel of Figure~\ref{fig:ecchist}),
194: yields $e_{crit}=0.0$ and $e_{max}=0.91$. The relative
195: increase in the number of low eccentricity planets discovered in the
196: last year results in best parameters that are effectively a single-piece model.
197:
198: \citet{JUR07} and \citet{ZHO07} predict that the eccentricity distribution of dynamically active planetary
199: systems approaches a Rayleigh-distribution due to planet-planet
200: scattering, and they show the Rayleigh-distribution is similar to the
201: observed planet eccentricity distribution. The dotted curve in
202: Figure~\ref{fig:ecchist} shows the Rayleigh-distribution with
203: $\sigma_{e}=0.3$ that best describes the outcome of the planet-planet
204: scattering calculations from
205: \citet{JUR07}. The Rayleigh-distribution under represents the
206: low eccentricity systems in the most recent eccentricity data.
207: \citet{JUR07} based their study on the sample of planets known at the time (April 2006). Subsequently, relatively more planets
208: with low eccentricities have been announced. In addition,
209: \citet{JUR07} included systems with P$>$20 day
210: whereas this sample has P$>$10 day. The dotted histogram in
211: the lower panel of Figure~\ref{fig:ecchist} shows the eccentricity
212: distribution at the most recent epoch (September 2007), but with
213: systems with P$<$20~day removed; apparently the difference is slight
214: and will hereafter be ignored. Although this study concentrates on
215: the two-piece model to describe the eccentricity
216: distribution, results are also given for the
217: Rayleigh-distribution with $\sigma_{e}=0.3$. To provide results with
218: the Rayleigh-distribution, the distribution is set to zero for $e>0.95$.
219:
220: \section{Transit Probability}\label{sec:tranprob}
221:
222: \citet{BAR07} derives the impact of orbital eccentricity on the
223: probability for a planet to transit its stellar host \citep[see
224: also][]{SEA03}. The probability to transit depends on the planet-star
225: separation during transit. For an eccentric orbit, a transit occurring
226: during pericenter enhances the transit probability and a transit
227: occurring during apocenter decreases the transit probability. When
228: averaged over observing angles, the net result is an enhancement of
229: the probability to transit over the circular orbit case,
230: \begin{equation}
231: {\rm Prob}_{{\rm T}e}=\frac{{\rm Prob}_{{\rm T}o}}{(1-e^{2})},
232: \end{equation}
233: where ${\rm Prob}_{{\rm T}o}$ is the transit probability of the
234: circular orbit case \citep{BAR07}. A planet with $e$=0.6 is $\sim$1.5
235: times more likely to transit than a planet on a circular orbit.
236:
237: With
238: the observed eccentricity distribution from \S~\ref{sec:eccdist}, it
239: is possible to calculate the average enhancement in the transit
240: probability over the circular orbit case,
241: \begin{equation}
242: \langle {\rm Prob}_{{\rm T}e}\rangle=\int^{1}_{0} {\rm Prob}_{{\rm T}e}P(e)de.
243: \end{equation}
244: The resulting integral using Equation~\ref{eq:eccdist} for the eccentricity distribution is,
245: {\scriptsize
246: \begin{equation}
247: \begin{split}
248: & \langle {\rm Prob}_{{\rm T}e}\rangle=\frac{{\rm Prob}_{{\rm T}o}}{(e_{\rm max}^2-e_{\rm crit}^2)}\left[2(e_{\rm max}-e_{\rm crit}){\rm arctanh}(e_{\rm crit})\right. \\
249: & \left. +\ln\left( \frac{(1-e_{\rm max}^2)}{(1-e_{\rm crit}^2)}\right) +e_{\rm max}\ln\left(\frac{(1-e_{\rm crit})(1+e_{\rm max})}{(1+e_{\rm crit})(1-e_{\rm max})}\right) \right].
250: \end{split}
251: \label{eq:probtrane}
252: \end{equation}}
253: Using the numerical values for $e_{\rm crit}$ and $e_{\rm max}$ from
254: \S~\ref{sec:eccdist}, $\langle {\rm Prob}_{{\rm T}e}\rangle$=1.26 and
255: 1.23 for the earlier and current epochs of observed eccentricities,
256: respectively. The Rayleigh distribution results in $\langle {\rm
257: Prob}_{{\rm T}e}\rangle$=1.31. Figure~\ref{fig:probtran} illustrates
258: $\langle {\rm Prob}_{{\rm T}e}\rangle$ for other choices of the
259: eccentricity distribution model parameters. Figure~\ref{fig:probtran}
260: is a function of $e_{\rm max}$ and the curves are for selected values
261: of $e_{\rm crit}$ as labeled. Thus, it is expected that the yield
262: from a transit survey that is sensitive to P$>$10 day planets will be
263: $\sim$25\% larger than expectations that assume circular orbits for
264: all planets. This section discusses only the probability to transit,
265: and it does not address whether the planet has a transit signal that
266: is detectable. Thus, the results from this section assumes the
267: enhanced probability for the planet to transit does not affect its
268: detectability, which is the topic of the next section.
269:
270: \begin{figure}
271: %\plotone{probtran.ps}
272: %\plotone{f2.ps}
273: \includegraphics[scale=0.9,viewport=0 100 350 350]{f2.ps}
274: \caption{Average transit probability as a function of the observed eccentricity distribution model parameters. The transit probability is scaled to the equivalent transit probability for a circular orbit. The abscissa indicates $e_{\rm max}$, and the curves are for selected values of $e_{\rm crit}$ as
275: labeled. For the model parameters shown in Figure~\ref{fig:ecchist}, $\langle{\rm Prob}_{{\rm T}e}\rangle\sim 25\%$ higher than assuming all planets have circular orbits ({\it dotted lines}).\label{fig:probtran}}
276: \end{figure}
277:
278:
279: \section{Transit Duration}\label{sec:trandur}
280:
281: \subsection{Edge-on Transit}\label{sec:i90}
282: In addition to enhancing the probability for a transit to occur,
283: orbital eccentricity results in the transit duration varying according
284: to the planet's longitude of pericenter during transit. The
285: transit duration is shortest (longest) if the transit occurs when the
286: planet is at pericenter (apocenter) when transiting its stellar host.
287: A short transit duration may reduce the detectability of the transit
288: event. Assuming constant velocity during transit, the transit
289: duration scaled to the edge-on ($i=90\degr$) circular orbit case has
290: extrema of
291: \begin{equation}
292: \tau_{\frac{p}{a}}=\frac{t_{\frac{p}{a}}}{t_{o}}=\sqrt{\frac{(1\mp e)}{(1\pm e)}}\sim 1\mp e+e^2/2,\label{eq:tranextrema}
293: \end{equation}
294: where the sign on top and bottom corresponds to the pericenter and
295: apocenter transit durations, respectively \citep{BAR07}. When
296: $e$=0.6, the transit duration at apocenter is twice the circular orbit
297: case. Given the range of eccentricities observed for extrasolar
298: planets with P$>$10 day, we expect significant variations in the
299: transit duration compared to that of a circular orbit.
300:
301: \citet{TIN05} (their Equation 7) provides a simplified form of the
302: transit duration, including longitude of pericenter, $\varpi$, and
303: orbital inclination, ${\it i}$. In deriving the transit duration,
304: \citep{TIN05} assume constant orbital velocity and planet-star
305: separation during transit and the planet crosses the stellar disk
306: along a straight path ($a\gg R_{\star}$). An exact calculation (of a
307: numerical nature given the need to solve Kepler's Equation) of the
308: transit duration finds Equation 7 of
309: \citet{TIN05} is accurate to better than 5\% for
310: $e$=0.9 and orbital separation $a\geq 0.1$ AU (P$\sim$10 day). The
311: equation is accurate to better than 1\% for $e$=0.9 and $a\geq 1.0$
312: AU.
313:
314: After scaling to the transit duration of the edge-on ($i=90\degr$), circular orbit with the same period and separation, Equation 7 of \citet{TIN05} becomes,
315: \begin{equation}
316: \tau=\frac{\sqrt{(1-e^2)}}{(1+e\cos(\varpi))}\sqrt{1-\rho^{2}\left[ \frac{(1-e^{2})}{(1+e\cos(\varpi))}\right] ^{2}\cos(i)^{2}},
317: \label{eq:durateq}
318: \end{equation}
319: where $\rho=a/(R_{\star}+R_{p})$, $R_{\star}$ is the radius of the star and $R_{p}$ is the radius of the planet. The definition for $\varpi$ is with respect to the line of sight. Thus, $\varpi=0$ means the pericenter is aligned with the observers line of sight, and $\varpi=180\degr$ means the apocenter is aligned with the observers line of sight. This varies from the definition of the argument of pericenter, $\omega$, which is defined with respect to the line of nodes on the plane of the sky ($\varpi=\omega-90\degr$).
320:
321: The $i=90\degr$ case illustrates the first order impact of orbital eccentricity on transit duration. In this case, Equation~\ref{eq:durateq} simplifies to
322: \begin{equation}
323: \tau_{90\degr}=\frac{\sqrt{(1-e^2)}}{(1+e\cos(\varpi))}.
324: \label{eq:trandurati90}
325: \end{equation}
326: The left panel of Figure~\ref{fig:duratboth} shows the transit
327: duration with respect to the edge-on circular orbit case for a variety of
328: orbital eccentricities as a function of $\varpi$. For a uniform
329: distribution in $\varpi$, the shallow slope in transit
330: duration at pericenter and apocenter results in a distribution of
331: transit durations peaked at these extrema (given by Equation~\ref{eq:tranextrema}) with low probability for a
332: transit duration in between these extrema.
333:
334: \begin{figure}
335: %\plotone{duratboth.ps}
336: %\plotone{f3.ps}
337: \includegraphics[scale=0.5,viewport=0 100 350 350]{f3.ps}
338: \caption{{\it Left}: Edge-on transit duration, $\tau$, scaled to the edge-on circular orbit case as a function of the longitude of pericenter, $\varpi$, for several eccentricities as labeled. {\it Right}: For a uniform distribution of $\varpi$, solid curves show the distribution for transit duration scaled to the edge-on circular orbit case assuming all orbits are edge-on (i.e. $i=90\degr$).\label{fig:duratboth}}
339: \end{figure}
340:
341:
342: Assuming a uniform distribution of $\varpi$, $p(\varpi)d\varpi=d\varpi/\pi$, the transformation law of probabilities \citep[Equation 7.2.4 in][]{PRE92} yields the distribution of transit durations,
343: \begin{equation}
344: p(\tau)d\tau=\left| \frac{\partial \varpi}{\partial \tau} \right| p(\varpi)d\tau,
345: \end{equation}
346: resulting in
347: \begin{equation}
348: p(\tau)d\tau = \frac{\sqrt{1-e^{2}}d\tau }{\pi \tau\sqrt{(e\tau)^{2}-(\sqrt{1-e^{2}}-\tau )^{2}}}.
349: \end{equation}
350: The right panel of Figure~\ref{fig:duratboth} shows the probability
351: density for transit duration with $i=90\degr$ for various orbital
352: eccentricities. As expected the probability density is heavily
353: weighted toward the extrema. The singularities in the probability
354: density at the extrema are integrable, making the probability density
355: normalizable.
356:
357: \subsection{Affect of Varying Inclination Angle}\label{incdist}
358:
359: The previous section shows the expected transit duration for fixed
360: orbital eccentricity when the inclination angle, $i=90\degr$. Varying
361: the inclination impacts the transit duration in two important ways.
362: First, as the inclination decreases, the path of the planet across the
363: stellar disc shortens, resulting in a reduction of the transit
364: duration. Second, for an eccentric orbit, the planet is closer to the
365: star at pericenter and farther away at apocenter. The smaller
366: separation at pericenter increases the range of inclination angles
367: that result in a transit, and conversely, a planet at apocenter will
368: have a reduction in the range of inclination angles for a transit to
369: occur. Thus, long transit duration events when the planet is at
370: apocenter become increasingly rare as the eccentricity increases.
371: This section quantifies these effects on the transit duration
372: distribution.
373:
374: Beginning with a simple joint distribution that is uniform in $\cos(i)$
375: and uniform in $\varpi$ enables the more complicated
376: distribution for transit duration to be derived. For this section, $e$ remains
377: fixed. The beginning joint distribution is,
378: \begin{equation}
379: P(\varpi,0 \le \cos(i) \le \cos(i_{min})| e)d(\cos(i))d\varpi=Ad(\cos(i))d\varpi,
380: \label{eq:cosweven}
381: \end{equation}
382: where $\cos(i_{min})$ is cosine of the minimum inclination angle
383: necessary for a transit to occur, and $A$ is the normalization
384: constant. Setting $\tau=0$ in Equation~\ref{eq:durateq} and solving
385: for $\cos(i)$ yields $\cos(i_{min})=(1+e\cos(\varpi))/\rho (1-e^{2})$.
386: Integrating Equation~\ref{eq:cosweven} over the range of variables
387: yields the normalization constant, $A=\rho(1-e^{2})/\pi$.
388:
389: The transformation law of probabilities for multiple dimensions
390: \citep[Equation 7.2.4 in][]{PRE92} provides the joint probability
391: density in terms of new variables $\tau$ and $\varpi$ when
392: starting with the probability density for $\cos(i)$ and
393: $\varpi\prime$.
394: \begin{equation}
395: P(\tau,\varpi | e)d\tau d\varpi=\left\| \begin{array}{cc}
396: \frac{\partial \cos(i)}{\partial \tau} & \frac{\partial \cos(i)}{\partial \varpi} \\
397: \frac{\partial \varpi\prime}{\partial \tau} & \frac{\partial \varpi\prime}{\partial \varpi} \end{array}\right\|
398: P(\cos(i),\varpi\prime)d\tau d\varpi,
399: \label{eq:ptauomega}
400: \end{equation}
401: In Equation~\ref{eq:ptauomega}, $\partial \varpi / \partial \tau=0$ and $\partial \varpi\prime / \partial \varpi =1$. Thus, only $\partial \cos(i) / \partial \tau$ remains, and Equation~\ref{eq:ptauomega} simplifies to
402: \begin{equation}
403: P(\tau ,\varpi |e)d\tau d\varpi=\left| \frac{\partial \cos(i)}{\partial \tau} \right| \rho (1-e^{2})/\pi d\tau d\varpi.
404: \label{ptauomega2}
405: \end{equation}
406:
407: The requisite derivative needed in Equation~\ref{ptauomega2} is
408: obtained by solving Equation~\ref{eq:durateq} for $\cos(i)$ yielding
409: \begin{equation}
410: \cos(i)=\frac{(1+e\cos(\varpi))}{\rho(1-e^2)^{3/2}}\sqrt{(1-e^2)-\tau^{2}(1+e\cos(\varpi))^{2}},
411: \label{cosi}
412: \end{equation}
413: and taking
414: the partial derivative with respect to $\tau$. Performing this operation yields
415: \begin{equation}
416: \begin{split}
417: & P(0\leq \tau \leq \tau_{a}, \varpi_{\rm min}\leq \varpi \leq \pi|e)d\tau d\varpi \\
418: & =\frac{\tau (1+e\cos(\varpi))^{3}}{\pi \sqrt{1-e^{2}} \sqrt{(1-e^{2})-\tau^{2}(1+e\cos(\varpi))^{2}}} d\tau d\varpi ,
419: \end{split}
420: \label{eq:ptauomega3}
421: \end{equation}
422: where the lower limit to $\varpi$ is necessary to avoid an imaginary result and is given by
423: \begin{equation}
424: \varpi_{\rm min}=\cos^{-1}\left[ {\rm MIN}\left( 1.0,\frac{\sqrt{1-e^{2}}-\tau}{\tau e} \right) \right].
425: \end{equation}
426: For $\tau \leq \tau_{peri}$, $\varpi_{\rm min}=0$, but as $\tau>\tau_{peri}$,
427: $\varpi_{\rm min}> 0$ since only transits occurring closer to apocenter
428: result in a transit duration long enough.
429:
430: Finally, the conditional
431: probability density for transit duration alone is obtained by
432: integrating over $\varpi$,
433: \begin{equation}
434: P(\tau | e) d\tau = \int^{\pi}_{\varpi_{\rm min}} P(\tau ,\varpi |e)d\tau d\varpi.
435: \label{eq:ptauconde}
436: \end{equation}
437: A solution to the above integral is possible in terms of a summation
438: of the incomplete elliptic integrals of the first, second, and third
439: kind \citep{BYR54}. In practice, I choose to solve the integral
440: numerically, which is readily solved using the
441: Romberg open ended algorithm which takes into account the singularity
442: at the lower limit of integration at $\varpi=\varpi_{\rm min}$ \citep{PRE92}.
443: Tests of convergence show Equation~\ref{eq:ptauconde} has a singularity at
444: $\tau=\tau_{peri}$, but it is integrable elsewhere over the range $0\leq \tau \leq
445: \tau_{a}$.
446:
447: The left panel in Figure~\ref{fig:duratinc} shows the probability
448: density for transit duration scaled to the edge-on circular orbit case
449: at several values of eccentricity. By comparison to
450: Figure~\ref{fig:duratboth}, the main impact of orbital inclination is to
451: strongly enhance the probability of observing a short duration event
452: at pericenter relative to a long duration event at apocenter. The
453: probability density also allows arbitrarily short events due to the
454: potential for grazing events.
455:
456: \begin{figure}
457: %\plotone{duratincnew.ps}
458: %\plotone{f4.ps}
459: \includegraphics[scale=0.5,viewport=0 100 350 350]{f4.ps}
460: \caption{{\it Left:} Transit duration distribution scaled to the edge-on circular orbit case for a uniform distribution of $\cos(i)$ and fixed eccentricity. The cases $e$=0.2, 0.4, 0.6, and 0.8 are shown. {\it Right:} Average transit duration scaled to the average transit duration of a circular orbit as a function of orbital eccentricity ({\it solid line}).\label{fig:duratinc}}
461: \end{figure}
462:
463: The probability density for transit duration can be summarized by finding the average transit duration at fixed eccentricity,
464: \begin{equation}
465: \langle \tau_{e}\rangle =\int^{\tau_{a}}_{0}\tau P(\tau | e) d\tau.\label{eq:avgtau}
466: \end{equation}
467: The right panel in Figure~\ref{fig:duratinc} shows $\langle \tau_{e}\rangle $ scaled
468: to the average transit duration of the circular orbit case,
469: $\langle \tau_{0}\rangle =\pi /4$, as the solid line. The function,
470: \begin{equation}
471: \langle\tau_{e}\rangle=\frac{\pi}{4}\sqrt{1-e^2},
472: \end{equation}
473: fits the relation to better than $10^{-6}$ (similar to the numerical
474: integration precision), which very strongly suggests this is the
475: analytical solution to the integral in Equation~\ref{eq:avgtau}.
476: \citet{TIN05} find the same result in terms of the average value for
477: their $\eta$ parameter at fixed eccentricity (see their Equation 18).
478:
479: \subsection{Transit Duration Distribution For Observed Eccentricity Distribution}\label{sec:tranprobecc}
480:
481: The previous section derives the transit duration
482: distribution at fixed eccentricity. This section derives the
483: transit duration distribution assuming planets follow the
484: observed eccentricity distribution. As in the previous section,
485: deriving the transit duration distribution begins with the simple
486: distribution that is uniform in $\cos(i)$, uniform in $\varpi$, and
487: the distribution of $e$ follows the observed distribution as given in
488: \S\ref{sec:eccdist}. The transformation law of probabilities enables
489: transforming the simple distribution in $\cos(i)$, $\varpi\prime$, and
490: $e\prime$ into the distribution expressed in $\tau$, $\varpi$, and
491: $e$.
492:
493: The initial distribution is given by,
494: \begin{equation}
495: \begin{split}
496: & P(0\le \cos(i) \le \cos(i_{min}), 0 \le \varpi\prime \le \pi , 0 \le e\prime \le e_{max}) \\
497: & d\cos(i) d\varpi\prime de\prime =AP(e\prime)d(\cos(i))d\varpi\prime de\prime ,
498: \end{split}
499: \end{equation}
500: where the normalization constant $A=\rho\gamma/\pi$, where
501: \begin{equation}
502: \begin{split}
503: & \gamma=(e_{\rm max}^2-e_{\rm crit}^2)\left[2(e_{\rm max}-e_{\rm crit}){\rm arctanh}(e_{\rm crit})\right. \\
504: & \left. +\ln\left( \frac{(1-e_{\rm max}^2)}{(1-e_{\rm crit}^2)}\right) +e_{\rm max}\ln\left(\frac{(1-e_{\rm crit})(1+e_{\rm max})}{(1+e_{\rm crit})(1-e_{\rm max})}\right) \right]^{-1}.
505: \end{split}
506: \end{equation}
507: The Jacobian transformation matrix simplifies as before, such that the transit duration distribution scaled to the edge-on, circular orbit transit duration is given by,
508: \begin{equation}
509: \begin{split}
510: & P(0 \le \tau \le \tau_{a}, \varpi_{\rm min} \le \varpi \le \pi , e_{\rm min} \le e \le e_{\rm max})d\tau d\varpi de \\
511: & =\left| \frac{\partial \cos(i)}{\partial \tau}\right| P(\cos(i),\varpi\prime ,e\prime)d\tau d\varpi de,
512: \end{split}
513: \end{equation}
514: where the lower limit to the eccentricity, $e_{min}$, becomes necessary for $\tau > 1$, when too small of an eccentricity cannot produce a transit duration as long as $\tau$. Solving $\tau_{a}$ for $e$ yields $e_{\rm min}={\rm MAX}[0.0,(\tau^2-1)/(\tau^2+1)]$. Overall, the joint distribution is given by
515: \begin{equation}
516: \begin{split}
517: & P(\tau ,\varpi ,e)d\tau d\varpi de= \\
518: & \frac{\gamma P(e)}{\pi}\frac{\tau (1+e\cos(\varpi))^{3}}{(1-e^{2})^{3/2}\sqrt{1-e^{2}-\tau^{2}(1+e\cos(\varpi))^{2}}}d\tau d\varpi de.
519: \end{split}
520: \end{equation}
521:
522: Integrating over $\varpi$ and $e$ provides the final probability
523: density for $\tau$ for the assumed distribution of orbital
524: eccentricities. Given the additional complication of integration over
525: two variables, an analytical solution was not forthcoming. As in
526: \S~\ref{incdist}, the singularities in the integrand are integrable,
527: and the Romberg open ended algorithm which takes into
528: account the singularity at the lower limits of
529: integration \citep{PRE92} provides the solution.
530:
531: The solid and long-dashed lines in Figure~\ref{fig:duratecc} shows the
532: distribution of transit duration scaled to the edge-on circular orbit
533: case for a population of extrasolar planets that follows the observed
534: orbital eccentricity distributions shown in top and bottom panels of
535: Figure~\ref{fig:ecchist}, respectively. The probability density has a
536: broad flat top with $\tau=0.4$ just as likely to occur as $\tau=1.0$.
537: For comparison, the transit duration distribution that would result
538: from a nearly uniform distribution of orbital eccentricity up to high
539: eccentricities ($e_{\rm crit}=0.9$ and $e_{\rm max}=0.95$) is shown as
540: the short-dashed line in Figure~\ref{fig:duratecc}. Given the bias against
541: detecting $e>0.6$ extrasolar planets in radial velocity studies
542: \citep{CUM04}, a large population of $e\sim 0.9$ planets cannot be
543: ruled out. The transit duration distribution resulting from the Rayleigh
544: distribution of orbital eccentricity (dotted line in
545: Figure~\ref{fig:duratecc}) has relatively fewer planets with $\tau\sim
546: 1$ due to its relatively fewer circular orbits.
547:
548: \begin{figure}
549: %\plotone{durateccnew.ps}
550: %\plotone{f5.ps}
551: \includegraphics[scale=0.5,viewport=0 100 350 350]{f5.ps}
552: \caption{{\it Left}: Transit duration distribution scaled to the edge-on circular orbit case for a uniform distribution of $\cos(i)$ and the observed distribution of $e$ as shown in the top panel ({\it solid line}) and bottom panel ({\it long dashed line}) of Figure~\ref{fig:ecchist}. Given the bias against detecting high eccentricity planets in radial velocity surveys, a uniform distribution of orbital eccentricities up to high eccentricity ($e_{\rm crit}=0.9$ and $e_{\rm max}=0.95$) will be highly skewed towards $\tau\sim 0.3$ ({\it short dashed line}). A Rayleigh distribution of orbital eccentricities results in relatively fewer transit durations $\tau\sim 1$ due to the relatively fewer objects on circular orbits ({\rm dotted line}). {\it Right:} Average transit duration scaled to the average transit duration of a circular orbit as a function of the observed eccentricity distribution model parameters. The abscissa indicates $e_{\rm max}$, and the curves are for selected values of $e_{\rm crit}$ as labeled. For the model parameters shown in Figure~\ref{fig:ecchist}, $\langle\tau\rangle\sim 0.88$ times shorter than assuming all planets have circular orbits ({\it dotted line}).\label{fig:duratecc}}
553: \end{figure}
554:
555: In the future, if a statistically large sample of transiting planets
556: with orbital $P>10$ days is available with accurate stellar
557: parameters, histograms of observed transit duration scaled to the
558: edge-on circular orbit case may help characterize the underlying
559: eccentricity distribution. After accounting for the selection
560: effects, a large number of $\tau\sim 0.3$ detections relative to
561: $\tau\sim 0.8$, as illustrated in the right panel of
562: Figure~\ref{fig:duratecc}, would indicate $e=0.9$ planets are as
563: common as circular orbits. Work toward understanding the sensitivity
564: of Kepler for constraining the underlying eccentricity distribution is
565: underway (E. Ford, private communication).
566:
567: The right panel of Figure~\ref{fig:duratecc} summarizes the transit
568: duration distribution by showing $\langle \tau_{e}\rangle $ in terms of the
569: parameters characterizing the eccentricity distribution, $e_{\rm
570: crit}$ and $e_{\rm max}$. Each line corresponds to a fixed value of
571: $e_{\rm crit}$ as labeled, and the abscissa indicates $e_{\rm
572: max}$ of the eccentricity distribution. For the eccentricity
573: distribution in the top panel of Figure~\ref{fig:ecchist}
574: $\langle \tau_{e}\rangle /\langle \tau_{o}\rangle =0.88$ and the bottom panel
575: $\langle \tau_{e}\rangle /\langle \tau_{o}\rangle =0.89$. The Rayleigh distribution results in $\langle \tau_{e}\rangle /\langle \tau_{o}\rangle =0.86$
576:
577: \section{Discussion: Application to Transit Surveys}\label{sec:disc}
578:
579: The results from \S~\ref{sec:trandur} quantify the impact orbital
580: eccentricity has on the transit duration. Overall, orbital
581: eccentricity results in shorter transit durations than the circular
582: orbit case, and the short transit duration reduces the
583: transit detectability. This section quantifies the reduction in
584: transit detectability for various noise models of transit
585: surveys.
586:
587: In a transit survey with independent photometric measurements, the transit signal to noise ratio is,
588: \begin{equation}
589: {\rm SNR}=\frac{\Delta F}{\sigma}\sqrt{N_{\rm obs}},
590: \end{equation}
591: where $\Delta F$ is the transit depth (the transit is modeled as a
592: box-car shape), $\sigma$ is the error on a photometric measurement,
593: and $N_{\rm obs}$ is the number of measurements during one or more
594: transit(s). A shorter transit duration reduces $N_{\rm
595: obs}\propto\tau$, thus, ${\rm SNR}\propto\sqrt{\tau}$. The observed
596: eccentricity distribution (Figure~\ref{fig:ecchist}) results in
597: $\tilde{\tau}=\langle
598: \tau_{e}\rangle /\langle \tau_{o}\rangle =0.88$, and on average the
599: reduced transit duration results in an effectively smaller ${\rm
600: SNR_{eff}}=\sqrt{\tilde{\tau}}=0.94$ per transit than assuming all
601: planets are on circular orbits.
602:
603: The above impact on the transit signal ${\rm SNR}$ due to a shorter
604: transit duration is for an individual star in a survey. However, the
605: reduced ${\rm SNR}$ will have a larger impact on the overall transit
606: detectability in an ideal field transit survey. As described in
607: \citet{GAU05} and \citet{GAU07}, a specified ${\rm SNR}$ criteria for
608: transit detection, ${\rm SNR_{\rm min}}$, in a field transit survey
609: corresponds to a maximum distance, $\ell_{\rm max}\propto {\rm
610: SNR}_{\rm min}^{-1}$, out to which a planet is detectable. This
611: proportionality assumes white noise and the dominant source of
612: photometric error is Poisson noise. In the studies of \citet{GAU05}
613: and \citet{GAU07}, $\ell_{\rm max}$ is a function of the planet radius
614: and stellar host spectral type (i.e. $\ell_{\rm max}$ is a smaller
615: distance for a smaller radius planet or larger radius star). For this study only the dependence of $\ell_{\rm max}$ on
616: transit duration is of interest.
617:
618: Overall, the yield from a transit survey is
619: proportional to the number of objects in the survey that meets ${\rm
620: SNR_{\rm min}}$, which is $N_{\rm obj}\propto
621: \ell_{\rm max}^{3}$ for stars distributed uniformly in the survey
622: volume as appropriate for nearby stars. The effective ${\rm
623: SNR}_{\rm min,eff}={\rm SNR}_{\rm min}\sqrt{\tilde{\tau}}$ larger than
624: assuming all planets are on circular orbits. Thus, the number of
625: objects in an idealized transit survey where a transit is detectable
626: is $N_{\rm obj,e}=\tilde{\tau}^{3/2}N_{\rm obj,o}=0.82N_{\rm obj,o}$
627: times smaller than the case where the detectability of a transit is
628: based on assuming all planets have circular orbits, $N_{\rm obj,o}$.
629: Despite the reduced detectability of transits, this is offset by the
630: higher probability for the planet to transit in the case of
631: significant eccentricity
632: \S~\ref{sec:tranprob}. The overall yield of the idealized transit
633: survey is discussed in \S~\ref{sec:conc} taking both the reduced
634: detectability and enhanced probability to transit into account.
635:
636: In practice, transit surveys typically are affected by correlated
637: measurements \citep{PON06}. In this regime, the correlation time
638: scale is similar to the transit duration and repeated measurements do
639: not add independent information. When correlated measurements
640: dominate the photometric error, the ${\rm SNR}=(\Delta F/\sigma_{\rm
641: cor})\sqrt{N_{\rm tr}}$, where $N_{\rm tr}$ is the number of transits
642: detected, and the correlated measurement error, $\sigma_{\rm cor}$, no
643: longer depends on the stellar luminosity, but is constant
644: \citep{GAU07}. When dominated by correlated measurements, the ${\rm
645: SNR}$ is independent of $\tau$ and the only requirement is to have a
646: short enough sampling cadence to detect the shortest transit duration
647: expected. Thus, a shorter $\tau$ due to orbital eccentricity has no
648: impact on the transit detectability in a survey which is dominated by
649: correlated measurements (i.e. $N_{\rm obj,e}=N_{\rm obj,o}$).
650: However, the Poisson limited, white noise transit survey will have a
651: higher planet yield than a survey dominated by correlated measurements
652: \citep{PON06}, but in the latter case, orbital eccentricity does not
653: impose any additional reduction in the transit detectability.
654:
655: The Kepler space-based transit survey whose goal is to detect several
656: Earth-sized planets \citep{BORU04} contends with stellar intrinsic
657: variability as the dominant source of noise \citep{JEN02}. Using
658: integrated flux measurements of the Sun to model intrinsic
659: variability of stars, \citet{JEN02} characterizes the detectability of
660: Earth-sized transiting planets with Kepler. The Sun has low noise on
661: 7 hr time scales typical of transiting Earth-sized planets in 1 yr
662: orbital periods. However, the solar noise increases by $\sim$ 4
663: orders of magnitude by 10 day time scales \citep[see Figure 2 of
664: ][]{JEN02}. The rapidly increasing intrinsic noise toward longer time
665: scales cancels any benefit of increased transit signal ${\rm SNR}$ for
666: transit durations $>$4 hr for the Kepler mission \citep[see Figure 9
667: of ][]{JEN02}. Thus, as long as the transit duration remains above 4
668: hr, orbital eccentricity will not reduce the detectability of
669: Earth-sized planets with Kepler.
670:
671: \section{Conclusion}\label{sec:conc}
672:
673: Orbital eccentricity results in an enhanced probability for a planet
674: to transit and potentially a reduction in the transit detectability.
675: The overall yield from a transit survey is given by $N_{\rm
676: det}\propto{\rm Prob}_{{\rm T}}\times N_{\rm obj}$. The results from
677: this study can be used to scale the overall yield from a transit
678: survey based on assuming all planets are on circular orbits for an
679: assumed distribution of orbital eccentricity. The enhanced
680: probability for a planet to transit, ${\rm Prob}_{{\rm T}}$ with a
681: distribution of orbital eccentricity scaled to the circular orbit case
682: is given by Equation~\ref{eq:probtrane}. The reduced number of
683: transiting planets detectable $N_{\rm obj}$ scaled to the circular
684: orbit case for an ideal transit survey where the photometric noise is
685: white and dominated by Poisson error is given in \S~\ref{sec:disc}.
686: Multiplying these two factors provides the overall yield of an ideal
687: transit survey scaled to assuming all planets are on circular orbits.
688: Figure~\ref{fig:tranyield} show that these two opposing effects nearly
689: cancel over the parameters that characterize the eccentricity
690: distribution. Thus, for an idealized transit survey with the
691: currently observed orbital eccentricity distribution, the overall
692: yield will be 4\% greater than assuming all planets are on circular
693: orbits. The result for an idealized transit survey with a Rayleigh
694: distribution of orbital eccentricities gives a similar enhancement
695: (4\%).
696:
697:
698: \begin{figure}
699: %\plotone{overallnew.ps}
700: %\plotone{f6.ps}
701: \includegraphics[scale=0.9,viewport=0 100 350 350]{f6.ps}
702: \caption{The overall yield from an idealized transit survey as function of the observe eccentricity distribution model parameters scaled to the assumption that all planets have circular orbits. The transit survey is assumed to have independent photometric measurements that are dominated by Poisson
703: noise. The abscissa indicates $e_{\rm max}$, and the curves are for selected values of $e_{\rm crit}$ as labeled. For the model parameters shown in Figure~\ref{fig:ecchist}, $N_{\rm det}\sim 4\%$ higher than assuming all planets have circular orbits ({\it dotted line}).\label{fig:tranyield}}
704: \end{figure}
705:
706:
707: In ground-based transit surveys, correlated measurements limit transit
708: detectability \citep{PON06} and intrinsic variability of the star will
709: limit the detectability for Earth-sized planets for the Kepler mission
710: \citep{JEN02}. In both cases the transit detectability is independent
711: of the transit duration, in which case $N_{\rm obj}$ is independent of
712: the orbital eccentricity. For these cases, the transit survey will
713: have higher returns by a factor of $\langle {\rm Prob}_{{\rm
714: T}e}\rangle$ (Equation~\ref{eq:probtrane}) than estimated by assuming
715: all planets are on circular orbits. However, the reduced planet yield
716: in a transit survey due to intrinsic stellar variability or correlated
717: measurements must be properly accounted for. If every dwarf star has
718: an Earth-sized planet orbiting in the habitable zone, then assuming
719: circular orbits, the Kepler mission expects to detect 100 Earth-sized
720: planets in the habitable zone \citep{BORU04}. The work presented here
721: indicates Kepler will have $\langle {\rm Prob}_{{\rm T}e}\rangle\sim
722: 25\%$ higher yield if Earth-sized planets in the habitable zone have a
723: planet eccentricity distribution similar to the currently known sample
724: of giant planets from radial velocity surveys. However, if
725: Earth-sized planets with e$\sim$0.9 are as common as e$\sim$0 then,
726: the yield of Earth-sized planets could be 80\% higher from the Kepler
727: mission assuming the high-e, short transit duration planets are still
728: detectable.
729:
730: The dependence of the transit survey yield on the uncertain underlying
731: orbital eccentricity distribution implies an uncertainty in measuring
732: the frequency of terrestrial planets in the habitable zone \citep[a
733: major goal of the Kepler mission][]{BORU04} . An analysis of the
734: transit yield from a transit survey that assumes all planets are on
735: circular orbits will overestimate the frequency of habitable planets
736: if high eccentricities are common and not taken into account.
737: In practice a variety of
738: noise regimes affect a transit survey and accurate yields necessitate
739: an accurate understanding of the photometric noise, stellar sample,
740: and underlying eccentricity distribution \citep{BUR06,GOU06,FRE07}.
741:
742: \acknowledgements
743: This paper benefited from discussions with Scott Gaudi, Will Clarkson,
744: Peter McCullough, and Eric Ford. This work is funded by NASA Origins
745: grant NNG06GG92G.
746:
747: \begin{thebibliography}
748:
749: \bibitem[Barbieri et al.(2007)]{BARB07} Barbieri, M., et al.\
750: 2007, ArXiv e-prints, 710, arXiv:0710.0898
751:
752: \bibitem[Barnes(2007)]{BAR07} Barnes, J.~W.\ 2007, ArXiv
753: e-prints, 708, arXiv:0708.0243
754:
755: \bibitem[Bord{\'e} et al.(2003)]{BORD03} Bord{\'e}, P., Rouan,
756: D., \& L{\'e}ger, A.\ 2003, \aap, 405, 1137
757:
758: \bibitem[Borucki et al.(2004)]{BORU04} Borucki, W., et al.\
759: 2004, Second Eddington Workshop: Stellar Structure and Habitable Planet Finding, ESA SP-538, Noordwijk: ESA Publications Division, 177
760:
761: \bibitem[Burke et al.(2006)]{BUR06} Burke, C.~J., Gaudi,
762: B.~S., DePoy, D.~L., \& Pogge, R.~W.\ 2006, \aj, 132, 210
763:
764: \bibitem[Butler et al.(2006)]{BUT06} Butler, R.~P., et al.\
765: 2006, \apj, 646, 505
766:
767: \bibitem[Byrd \& Friedman(1954)]{BYR54} Byrd, P.~F., \&
768: Friedman, M.\ 1954, Mitteilungen der Astronomischen Gesellschaft Hamburg,
769: 5, 99
770:
771: \bibitem[Cresswell et al.(2007)]{CRE07} Cresswell, P.,
772: Dirksen, G., Kley, W., \& Nelson, R.~P.\ 2007, ArXiv e-prints, 707,
773: arXiv:0707.2225
774:
775: \bibitem[Cumming(2004)]{CUM04} Cumming, A.\ 2004, \mnras,
776: 354, 1165
777:
778: \bibitem[Fischer et al.(2007)]{FIS07} Fischer, D.~A., et al.\
779: 2007, \apj, 669, 1336
780:
781: \bibitem[Ford \& Rasio(2007)]{FOR07} Ford, E.~B., \& Rasio,
782: F.~A.\ 2007, ArXiv Astrophysics e-prints, arXiv:astro-ph/0703163
783:
784: \bibitem[Fressin et al.(2007)]{FRE07} Fressin, F., Guillot,
785: T., Morello, V., \& Pont, F.\ 2007, ArXiv e-prints, 704, arXiv:0704.1919
786:
787: \bibitem[Gaudi(2007)]{GAU07} Gaudi, B.~S. 2007, in Michelson Summer Workshop 2007, (Moffett Field, CA: NASA Ames Conf. Center) http://msc.caltech.edu/workshop/2007/Gaudi\_stat.pdf
788:
789: \bibitem[Gaudi et al.(2005)]{GAU05} Gaudi, B.~S., Seager, S.,
790: \& Mallen-Ornelas, G.\ 2005, \apj, 623, 472
791:
792: \bibitem[Goldreich \& Tremaine(1980)]{GOL80} Goldreich, P.,
793: \& Tremaine, S.\ 1980, \apj, 241, 425
794:
795: \bibitem[Gould et al.(2006)]{GOU06} Gould, A., Dorsher, S.,
796: Gaudi, B.~S., \& Udalski, A.\ 2006, Acta Astronomica, 56, 1
797:
798: \bibitem[Jenkins(2002)]{JEN02} Jenkins, J.~M.\ 2002, \apj,
799: 575, 493
800:
801: \bibitem[Jones et al.(2006)]{JON06} Jones, H.~R.~A., Butler,
802: R.~P., Tinney, C.~G., Marcy, G.~W., Carter, B.~D., Penny, A.~J., McCarthy,
803: C., \& Bailey, J.\ 2006, \mnras, 369, 249
804:
805: \bibitem[Juric \& Tremaine(2007)]{JUR07} Juric, M., \&
806: Tremaine, S.\ 2007, ArXiv Astrophysics e-prints, arXiv:astro-ph/0703160
807:
808: \bibitem[Moutou et al.(2006)]{MOU06} Moutou, C., Pont, F., \&
809: Halbwachs, J.-L.\ 2006, Formation plan{\'e}taire et exoplan{\`e}tes, Ecole
810: th{\'e}matique du CNRS, Goutelas (Loire), 23 - 27 mai 2005 Edited by
811: J.-L.~Halbwachs, D.~Egret, and J.-M.~Hameury.~Strasbourg: Observatoire
812: astronomique de Strasbourg et Soci{\'e}t{\'e} Fran{\c c}aise d'Astronomie
813: et d'Astrophysique (SF2A), 2006, p.~55-79
814: http://astro.u-strasbg.fr/goutelas/g2005/, 28, 55
815:
816: \bibitem[Naef et al.(2001)]{NAE01} Naef, D., et al.\ 2001,
817: \aap, 375, L27
818:
819: \bibitem[Pont et al.(2006)]{PON06} Pont, F., Zucker, S., \&
820: Queloz, D.\ 2006, \mnras, 373, 231
821:
822: \bibitem[Press et al.(1992)]{PRE92} Press, W.~H., Teukolsky,
823: S.~A., Vetterling, W.~T., \& Flannery, B.~P.\ 1992, Cambridge: University
824: Press, |c1992, 2nd ed.
825:
826: \bibitem[Raymond et al.(2006)]{RAY06} Raymond, S.~N., Quinn,
827: T., \& Lunine, J.~I.\ 2006, Icarus, 183, 265
828:
829: \bibitem[Ribas \& Miralda-Escud{\'e}(2007)]{RIB07} Ribas, I.,
830: \& Miralda-Escud{\'e}, J.\ 2007, \aap, 464, 779
831:
832: \bibitem[Seagroves et al.(2003)]{SEA03} Seagroves, S.,
833: Harker, J., Laughlin, G., Lacy, J., \& Castellano, T.\ 2003, \pasp, 115,
834: 1355
835:
836: \bibitem[Tamuz et al.(2007)]{TAM07} Tamuz, O., et al.\ 2007, \aap L, in press (astro-ph/0710.5028)
837:
838: \bibitem[Tingley \& Sackett(2005)]{TIN05} Tingley, B., \&
839: Sackett, P.~D.\ 2005, \apj, 627, 1011
840:
841: \bibitem[Zhou et al.(2007)]{ZHO07} Zhou, J.-L., Lin,
842: D.~N.~C., \& Sun, Y.-S.\ 2007, \apj, 666, 423
843:
844: \end{thebibliography}
845:
846: \end{document}