1: \documentclass[12pt,preprint]{aastex}
2: %%\documentclass[manuscript]{aastex}
3: %%\documentclass[preprint2]{aastex}
4: %%\documentclass[preprint2,longabstract]{aastex}
5: %%\documentclass{emulateapj}
6: %%\documentclass[onecolumn]{emulateapj}
7:
8: %%\slugcomment{Not to appear in Nonlearned J., 45.}
9:
10: \shorttitle{Signature of the Ice Line and Modest Migration}
11: \shortauthors{Schlaufman et al.}
12:
13: \begin{document}
14:
15: \title{The Signature of the Ice Line and Modest Type I Migration in the
16: Observed Exoplanet Mass-Semimajor Axis Distribution}
17:
18: \author{Kevin C. Schlaufman\altaffilmark{1} and D. N. C. Lin\altaffilmark{2}}
19: \affil{Astronomy and Astrophysics Department, University of California,
20: Santa Cruz, CA 95064; kcs@ucolick.org and lin@ucolick.org}
21:
22: \and
23:
24: \author{S. Ida}
25: \affil{Tokyo Institute of Technology, Ookayama, Meguro-ku, Tokyo 152-8551,
26: Japan; ida@geo.titech.ac.jp}
27:
28: \altaffiltext{1}{NSF Graduate Research Fellow}
29: \altaffiltext{2}{Kavli Institute of Astronomy and Astrophysics, Peking
30: University, Beijing, China}
31:
32: \begin{abstract}
33: Existing exoplanet radial velocity surveys are complete in the planetary
34: mass-semimajor axis ($M_p-a$) plane over the range 0.1 AU $< a <$ 2.0 AU where
35: $M_p \gtrsim 100~M_{\oplus}$. We marginalize over mass in this complete
36: domain of parameter space and demonstrate that the observed $a$ distribution
37: is inconsistent with models of planet formation that use the full Type I
38: migration rate derived from a linear theory and that do not include the effect
39: of the ice line on the disk surface density profile. However, the efficiency of
40: Type I migration can be suppressed by both nonlinear feedback and the barriers
41: introduced by local maxima in the disk pressure distribution, and we confirm
42: that the synthesized $M_p-a$ distribution is compatible with the observed data
43: if we account for both retention of protoplanetary embryos near the ice line
44: and an order-of-magnitude reduction in the efficiency of Type I migration. The
45: validity of these assumption can be checked because they also predict a
46: population of short-period rocky planets with a range of masses
47: comparable to that of the Earth as well as a ``desert" in the $M_p-a$
48: distribution centered around $M_p \sim 30-50~M_{\oplus}$ and $a <1$ AU.
49: We show that the expected ``desert" in the $M_p-a$ plane will be
50: discernible by a radial velocity survey with 1 m s$^{-1}$ precision and
51: $n \sim 700$ radial velocity observations of program stars.
52: \end{abstract}
53:
54: \keywords{planetary systems --- planetary systems: formation --- planetary
55: systems: protoplanetary disks}
56:
57: \section{Introduction}
58:
59: Over 200 planets with reliable mass ($M_p$) and semimajor axis ($a$)
60: measurements have been discovered around nearby FGK stars in the past decade.
61: At the same time, attempts to build a comprehensive deterministic
62: theory of planet formation have lead to the development of population
63: synthesis models based on the sequential accretion scenario.
64: In \citet{ida04}, two of us studied the growth of planetesimals
65: into dynamically isolated embryos as well as their tidal interactions with their
66: parent disks. Using the observed ranges of disk mass, size, and accretion
67: rate we showed that a fraction of embryos evolve into cores with more than a
68: few Earth masses ($M_\oplus$), accrete massive envelopes, open up gaps near
69: their orbits, and attain asymptotic masses comparable to that of Jupiter.
70: In some massive and persistent disks, the newly formed gas giant planets may
71: migrate toward the proximity of their host stars. In the end these
72: simulations produced the distribution of dynamical and structural properties of
73: planets. Presently, the observed sample of extrasolar planetary properties
74: has become large enough to enable direct comparisons between the theoretically
75: predicted and observed $M_p-a$ distributions that not only delineate the
76: dominant physical mechanisms at work in planet formation, but also
77: provide quantitative constraints on the efficiencies of those processes.
78:
79: In the latest update of the planet formation models two of us have
80: incorporated the effect of Type I migration \citep{ida08a}. This
81: process is a direct consequence of a protoplanetary core's tidal
82: interaction with its parent protoplanetary disk. The efficiency of Type I
83: migration was first determined by a linear theory \citep{gol80,war86,tan02}
84: that neglected the embryo's perturbation on the
85: surface density distribution of its parent protoplanetary disk.
86: In the environment of a minimum mass nebula though, this efficiency factor
87: would imply that a protoplanetary embryo with mass a fraction of an Earth mass
88: would migrate from $\sim 1$ AU into its host star
89: before the severe depletion of the disk gas that is known to occur over a time
90: scale of several Myr. Although this critical mass at which Type I migration
91: causes AU-scale migration on the disk depletion time increases with $a$,
92: it is still difficult to retain sufficiently massive cores for the onset of
93: dynamical accretion of gas. This argument implies that gas giants should
94: be very rare \citep{ida08a} and this paradox has led to many in-depth
95: analyses of the Type I migration process. Numerical nonlinear simulations of
96: Type I migration were reviewed by \citet{pap06} and many potential
97: explanations for slow Type I migration are in the literature: intrinsic
98: turbulence in the disk \citep{lau04,nel04}, self-induced unstable flow
99: \citep{kol04,li05}, nonlinear radiative and hydrodynamic feedback
100: \citep{mas06a}, and variation in surface density and temperature gradients
101: \citep{mas06b}. \citet{dob07} has shown that in some
102: situations the nonlinear Type I migration rate can be less than 10\%
103: of the linear prediction.
104:
105: \defcitealias{ida08b}{IL}
106: Another issue studied by two of us \citep[IL hereafter]{ida08b} is the
107: critical embryo mass $M_{crit} >$ at least a few $M_{\oplus}$ required by
108: current models for runaway gas accretion. The embryo is limited by its
109: isolation mass $M_{iso}$, and the solid surface density profile of the minimum
110: mass solar nebula (MMSN) $\Sigma_d \propto a^{-3/2}$ requires that the
111: $M_{iso}$ scales like $a^{3/4}$. On the other hand, the timescale for growth
112: $\tau_{c,acc}$ scales with $a^{27/10}$. As a result, after the
113: characteristic gas depletion time $\tau_{dep}$ the most massive
114: embryos near the ice line have masses $M_c \sim M_{iso} < M_{crit}$.
115: However, since we observe many exoplanets with Jupiter masses at $a \sim 1$
116: AU, there must be some physical process that is neglected in this
117: simple analysis. \citet{kre07} outlined one possible solution to this
118: problem in which solids are trapped near regions of the disk where the
119: local pressure requires the gas to rotate with super-Keplarian
120: velocities. When the combined contribution of both Lindblad and
121: corotation resonances are taken into account \citep{mas06b}, the
122: migration of protoplanetary cores may be suppressed as well
123: \citepalias{ida08b}.
124:
125: In this paper, we utilize the observed data to calibrate the
126: population synthesis models. In \S2 we quantitatively show that the
127: existing distribution of exoplanets cannot be explained by models of
128: planet formation that apply the full Type I migration rate predicted
129: from linear theory and that do not include the effects of the ice
130: line. We also point out that the existing observed and synthesized $M_p-a$
131: distributions are in agreement with each other if we take into account
132: the effect of an ice line barrier and assume a reduction in the magnitude
133: of Type I migration. In addition, we describe the parameters of a radial
134: velocity survey capable of verifying the existence of ``desert" in the $M_p-a$
135: diagram predicted by \citetalias{ida08b}. In \S3 we consider the implications
136: of these models and suggest methods to test our assumptions. In \S4 we
137: summarize our findings.
138:
139: \section{Analysis}
140:
141: We use the ideas presented in \citet{nar05} and the formalism developed in
142: \citet{cum04} to approximately reproduce the result of \cite{cum08} that
143: showed that the current California and Carnegie Planet Search
144: (CCPS) has announced all planets with orbital period $P < 2000$ days, stellar
145: reflex velocities $K > 20$ m s$^{-1}$, and eccentricities $e \lesssim 0.6$. In
146: particular, we determine which of the simulated exoplanet systems from
147: \citetalias{ida08b} would be detectable by a radial velocity survey with
148: precision and cadence similar to present-day radial velocity surveys like
149: the CCPS and the High Accuracy Radial Velocity Planetary Search Project
150: (HARPS). In this approach, we combine Equation (26) and (30) of \cite{cum04}
151: and declare that all the synthesized planetary systems from \citetalias{ida08b}
152: with mass $M_p > M_{50}$ where
153:
154: \begin{eqnarray}\label{eq1}
155: M_{50} & \approx & \frac{70~M_{\oplus}}{\sqrt{N}\sin{i}}
156: \left(\frac{\sigma}{\mbox{m/s}}\right)
157: \left(\frac{P}{\mbox{yr}}\right)^{1/3}
158: \left(\frac{M_{\ast}}{M_{\odot}}\right)^{2/3}
159: \left[\frac{\ln{(M/F)}}{9.2}\right]^{1/2}
160: \end{eqnarray}
161:
162: \noindent
163: are detectable. In Equation~(\ref{eq1}), $N$ is the number
164: of radial velocity observations, $\sigma$ is the RMS of spectrograph
165: precision and stellar jitter, $i$ is the inclination
166: of the exoplanet's orbit, $P$ is the orbital period in years, and $M_{\ast}$
167: is the host stellar mass in solar masses. In the limit of large $N$,
168: $M \approx 100$ is the number of independent frequencies searched and
169: $F \approx 0.01$ is the false alarm probability -- the numerical values are
170: correct to order-of-magnitude and in any case they only very weakly influence
171: our estimate of $M_{50}$. We note that Equation~(\ref{eq1}) is formally
172: correct only for single planetary systems in circular orbits; however,
173: \citet{cum04} shows that in the limit of large $N$ Equation~(\ref{eq1}) applies
174: to multiple planet systems and in the case where $e \lesssim 0.6$. Therefore,
175: we set $M_{50} = \infty$ if $e > 0.6$. We assume a velocity resolution of
176: 1 m s$^{-1}$.
177:
178: We then carry out a Monte Carlo simulation in which we assign each of the
179: simulated planetary systems from \citetalias{ida08b} a random host stellar mass,
180: stellar jitter, eccentricity, inclination, and number of radial velocity
181: observations. In this prescription, we use the empirical distributions for
182: host stellar mass, stellar jitter, eccentricity, and number of radial
183: velocity observations given in the updated \citet{but06}
184: catalog\footnote{Maintained at http://www.exoplanets.org} of all known
185: exoplanets; we use the standard
186: distribution for random inclinations. We compute which planetary systems have
187: $M_p > M_{50}$ and declare that these systems are detectable in this iteration.
188:
189: We repeat this process 1000 times. We consider all planetary systems
190: that are detectable according to the $M_{50}$ criterion in 90\% of the
191: Monte Carlo iterations robustly detectable. Averaged over host
192: stellar mass, stellar jitter, eccentricity, inclination, and number of
193: radial velocity observations we find that all of the simulated planets
194: from \citetalias{ida08b} in the range 0.1 AU $< a <$ 2.0 AU with
195:
196: \begin{eqnarray}\label{eq2}
197: M_p & \gtrsim & \left[174\left(\frac{a}{\mbox{au}}\right)+47\right] M_{\oplus}
198: \end{eqnarray}
199:
200: \noindent
201: are robustly detectable. We include the results of this calculation in
202: Figure~\ref{fig1}. The robustly detectable systems are marked with solid
203: circles and we plot the observed $M_p-a$ distribution of known planetary
204: systems as the solid squares.
205:
206: \citetalias{ida08b} generated a set of 12 realizations of the $M_p-a$ plane
207: under different physical assumptions described in Table~\ref{tbl-1}.
208: One group of models result from disks with the characteristic bump in
209: gas surface density $\Sigma_g$ due to the coupling effect of the MRI
210: activity and the ice line and with an enhancement in solid surface
211: density $\Sigma_d$, another group of models has the bump in $\Sigma_g$
212: but not in $\Sigma_d$, and the last group ignores the effects the ice
213: line would have on $\Sigma_g$ and $\Sigma_d$. For each group of
214: models, four different parameterizations of the Type I migration rate
215: were used: 100\%, 30\%, 10\%, and 3\% of the full rate from linear
216: theory. For each group of models we marginalize the 2D $M_p-a$
217: distribution of the simulated exoplanet systems in the complete region
218: over planetary mass, leaving us with 1D distributions in semimajor
219: axis. We plot histograms for each group of models: Figure~\ref{fig2}
220: corresponds to the models that disregard the effects of the ice line
221: on $\Sigma_g$ and $\Sigma_d$, Figure~\ref{fig3} corresponds to the
222: models that includes the effect of the ice line on $\Sigma_g$ but not
223: $\Sigma_d$, and Figure~\ref{fig4} corresponds to the models that
224: include the effects of the ice line on $\Sigma_g$ and $\Sigma_d$.
225:
226: We determine the model that best matches the observed $M_p-a$
227: distribution in the complete region by computing for each model the
228: $p$-value distribution that results from comparing 1000 bootstrap
229: resamplings from that model in the complete region with 1000
230: resamplings of the observed data in the complete region. We include
231: the results of this calculation in Figure~\ref{fig5} and we also
232: report median $p$-values and 95\% intervals in the last three columns
233: of Table~\ref{tbl-1}. We find that only models that include the
234: effects of the ice line on both the gas and solid surface density of the disk
235: and apply a Type I migration rate an order-of-magnitude less than that
236: predicted by linear theory (models C01C and C003C) are consistent with the
237: observed data in the complete region. Models which neglect the presence of
238: the ice-line barrier generally do not yield the observed up-turn in the period
239: distribution of the known planets and are rejected at very high confidence.
240: Models with efficient Type I migration generally under predict the fraction of
241: stars with detectable gas giants, especially for those with $a$ outside the
242: ice line. Therefore, we argue that the population synthesis models
243: presented in \citetalias{ida08b} incorporating a Type I migration rate much
244: reduced from linear theory and the effects of the ice line on both the solid
245: and gas surface densities ($\Sigma_d$ and $\Sigma_g$ respectively) are at
246: least plausible deterministic models for giant planet formation. We also note
247: that the observed data suggests that the ad hoc prescription for the
248: location of the ice line used by \citetalias{ida08b} underestimates its radius
249: by perhaps even a factor of two.
250:
251: Furthermore, in the presentation of their population synthesis models
252: \citetalias{ida08b} illustrated the presence of a ``desert" in the $M_p-a$
253: distribution. This sparsely populated region is depleted due to both Type I
254: migration and runaway gas accretion. For Models C01C and C003C, this
255: ``desert" is centered around $M_p \sim 30~M_{\oplus}$ and $a <1$ AU.
256: We use the same detection strategy described above, only now we
257: model the number of radial velocity observations of each planetary
258: system over a period of about ten years as a Gaussian random variable
259: with mean $\mu_n$ and standard deviation $\sigma_n$; we round each
260: random deviate to the nearest whole number. We then fit a two
261: component Gaussian mixture model to the $M_p-a$ distribution of all
262: robustly detected planet with mass $M_p < 100~M_{\oplus}$, and we say
263: the ``desert" is detected if the mean vectors of the two Gaussians
264: are offset by more than $0.6$ in $\log{a}$ and the minor axis of the
265: Gaussian at smaller orbital radius is larger than the minor axis of the
266: Gaussian at larger orbital radius. In other words, if the two components of
267: the mixture model bracket the corner of a region devoid of extrasolar planets
268: -- a metaphorical ``desert" -- we say the ``desert" is resolved. We find that
269: when $\mu_n = 700$ and $\sigma_n = 50$, the two components of the mixture model
270: bracket a barren region and therefore the ``desert" is resolved more than 90\%
271: of the time. We illustrate results of our calculation in Figure~\ref{fig6}.
272: As a result, a radial velocity campaign with the parameters described above
273: will be able to confirm the fidelity of the models presented in
274: \citetalias{ida08b}.
275:
276: \section{Discussion}
277:
278: The key prediction of \citetalias{ida08b} is that for masses at which the
279: dominant migration mechanism is Type I migration -- $M_p \lesssim
280: 50~M_{\oplus}$ -- there will be a dearth of exoplanets within 1 AU of
281: their host stars, simply because the timescale for Type I migration
282: is so much shorter than the disk dispersal time. Since they are formed
283: interior to the ice line, these planets are likely to be rocky and
284: have mass a few $M_\oplus$. \citetalias{ida08b} also predict an
285: overdensity of gas giant planets at $\sim 2$ AU resulting from the ice
286: line. We show in Figure~\ref{fig6} that one can
287: quantitatively detect these features in the observed $M_p-a$ with a
288: radial velocity survey with 1 m s$^{-1}$ precision and about 700
289: radial velocity observations, or about ten years worth of data from the
290: Automated Planet Finder (APF). In future work, we will examine the
291: ability of missions like the Space Interferometry Mission (SIM) to
292: verify the same features in $M_p-a$ plane.
293:
294: In addition to the upper and lower $M_p$ bound, the ``desert" is also
295: surrounded by populated domains in the $a$ distribution. While the ice
296: line provides a strong up-turn at a few AU, the models by
297: \citetalias{ida08b} also imply a large population of short-period rocky
298: planets as a consequence of Type I migration. Despite an order-of-magnitude
299: decrement in the efficiency of Type I migration, the simulated
300: $M_p-a$ distribution of models C01C and C003C indicate that in the
301: proximity of their host stars, rocky planets with $M_p \sim$ a few
302: $M_{\oplus}$ are at least an order-of-magnitude more common than close-in gas
303: giants (see Figure 6 of \citetalias{ida08b}). Other authors have already
304: pointed out the observational difficulties inherent in the search for this
305: population of ``super-Earths" through radial velocity observations
306: \citep{nar05}. Finally, we note that the models of IL did not include
307: dynamical interactions between planets in multiple planetary systems, and
308: these interactions can broaden the simulated $a$ distribution and eccentricity
309: distribution. We will include these effects in future generations of the
310: population synthesis models.
311:
312: \section{Conclusion}
313:
314: We used the fact that existing exoplanet radial velocity surveys are
315: complete in the planetary mass-semimajor axis ($M_p-a$) plane where
316: 0.1 AU $< a <$ 2.0 AU and $M_p$ is in the range specified by
317: Equation~(\ref{eq2}) to show that the observed semimajor axis distribution in
318: the complete region cannot be explained by models of planet formation that use
319: the full Type I migration rate predicted by linear theory and that do not
320: include the effects of the ice line. Moreover, we also demonstrated that the
321: expected ``desert" in the $M_p-a$ plane at about $M_p \sim 30~M_{\oplus}$ and
322: $a <1$ AU predicted by \citetalias{ida08b} will be discernible by a radial
323: velocity survey with 1 m s$^{-1}$ precision and $n \sim 700$ radial velocity
324: observations of program stars. Such an observational campaign will also verify
325: the predicted inner boundary of the ``desert" where we expect a large
326: population of super-Earths have migrated to and halted in the proximity of
327: their host stars.
328:
329: \acknowledgments We thank A. Cumming, G. Laughlin, G. Marcy, and
330: Michel Mayor for useful conversation and the anonymous referee for some
331: insightful comments. This research has made use of NASA's Astrophysics Data
332: System Bibliographic Services. This material is based upon
333: work supported under a National Science Foundation Graduate Research
334: Fellowship, NASA (NAGS5-11779, NNG06-GF45G, NNX07A-L13G, NNX07AI88G),
335: JPL (1270927), NSF(AST-0507424), and JSPS.
336:
337: %\clearpage
338: \begin{thebibliography}{}
339: \bibitem[Butler et al.(2006)]{but06} Butler, R.~P., et al.\
340: 2006, \apj, 646, 505
341: \bibitem[Cumming(2004)]{cum04} Cumming, A.\ 2004, \mnras,
342: 354, 1165
343: \bibitem[Cumming et al.(2008)]{cum08} Cumming, A., Butler,
344: R.~P., Marcy, G.~W., Vogt, S.~S., Wright, J.~T.,
345: \& Fischer, D.~A.\ 2008, \pasp, 120, 531
346: \bibitem[Dobbs-Dixon(2007)]{dob07} Dobbs-Dixon, I.~M.\ 2007,
347: Ph.D.~Thesis
348: \bibitem[Goldreich \& Tremaine(1980)]{gol80} Goldreich, P., \&
349: Tremaine, S.\ 1980, \apj, 241, 425
350: \bibitem[Ida \& Lin(2004)]{ida04} Ida, S., \& Lin, D.~N.~C.\ 2004,
351: \apj, 604, 388
352: \bibitem[Ida \& Lin(2008a)]{ida08a} Ida, S., \& Lin, D.~N.~C.\
353: 2008, \apj, 673, 487
354: \bibitem[Ida \& Lin(2008b)]{ida08b} Ida, S., \& Lin, D.~N.~C.\ 2008,
355: ArXiv e-prints, 802, arXiv:0802.1114
356: \bibitem[Laughlin et al.(2004)]{lau04} Laughlin, G.,
357: Steinacker, A., \& Adams, F.~C.\ 2004, \apj, 608, 489
358: \bibitem[Li et al.(2005)]{li05} Li, H., Li, S., Koller, J.,
359: Wendroff, B.~B., Liska, R., Orban, C.~M., Liang, E.~P.~T.,
360: \& Lin, D.~N.~C.\ 2005, \apj, 624, 1003
361: \bibitem[Koller \& Li(2004)]{kol04} Koller, J., \& Li, H.\ 2004,
362: The Search for Other Worlds, 713, 63
363: \bibitem[Kretke \& Lin(2007)]{kre07} Kretke, K.~A., \&
364: Lin, D.~N.~C.\ 2007, \apjl, 664, L55
365: \bibitem[Masset et al.(2006a)]{mas06a} Masset, F.~S., D'Angelo,
366: G., \& Kley, W.\ 2006, \apj, 652, 730
367: \bibitem[Masset et al.(2006b)]{mas06b} Masset, F.~S.,
368: Morbidelli, A., Crida, A., \& Ferreira, J.\ 2006, \apj, 642, 478
369: \bibitem[Narayan et al.(2005)]{nar05} Narayan, R., Cumming,
370: A., \& Lin, D.~N.~C.\ 2005, \apj, 620, 1002
371: \bibitem[Nelson \& Papaloizou(2004)]{nel04} Nelson, R.~P., \&
372: Papaloizou, J.~C.~B.\ 2004, \mnras, 350, 849
373: \bibitem[Papaloizou \& Terquem(2006)]{pap06} Papaloizou, J.~C.~B., \&
374: Terquem, C.\ 2006, Reports of Progress in Physics, 69, 119
375: \bibitem[Tanaka et al.(2002)]{tan02} Tanaka, H., Takeuchi,
376: T., \& Ward, W.~R.\ 2002, \apj, 565, 1257
377: \bibitem[Ward(1986)]{war86} Ward, W.~R.\ 1986, Icarus, 67, 164
378: \end{thebibliography}
379:
380: \clearpage
381: \begin{deluxetable}{lrccccc}
382: \tablecaption{Model Descriptions from \citet{ida08b}\label{tbl-1}}
383: \tablewidth{0pt}
384: \tablehead{
385: \colhead{Name} & \colhead{$C_1$\tablenotemark{a}} &
386: \colhead{$\Sigma_g$ Enhanced} & \colhead{$\Sigma_d$ Enhanced} &
387: \colhead{$p_{L}$\tablenotemark{b}} & \colhead{$\bar{p}$\tablenotemark{c}} &
388: \colhead{$p_{U}$\tablenotemark{d}}}
389: \startdata
390: C1C & 1 & Yes & Yes & $6.9 \times 10^{-6}$ & $2.8 \times 10^{-3}$ &
391: $1.7 \times 10^{-1}$\\
392: C03C & 0.3 & Yes & Yes & $1.6 \times 10^{-6}$ & $1.7 \times 10^{-3}$ &
393: $1.2 \times 10^{-1}$\\
394: C01C & 0.1 & Yes & Yes & $9.9 \times 10^{-3}$ & $8.6 \times 10^{-2}$ &
395: $7.2 \times 10^{-1}$\\
396: C003C & 0.03 & Yes & Yes & $3.2 \times 10^{-3}$ & $6.0 \times 10^{-2}$ &
397: $6.0 \times 10^{-1}$\\
398: C1B & 1 & Yes & No & $2.9 \times 10^{-12}$ & $7.3 \times 10^{-8}$ &
399: $9.6 \times 10^{-5}$\\
400: C03B & 0.3 & Yes & No & $5.8 \times 10^{-9}$ & $1.4 \times 10^{-5}$ &
401: $2.8 \times 10^{-3}$\\
402: C01B & 0.1 & Yes & No & $3.4 \times 10^{-6}$ & $2.8 \times 10^{-3}$ &
403: $2.2 \times 10^{-1}$\\
404: C003B & 0.03 & Yes & No & $5.1 \times 10^{-5}$ & $1.2 \times 10^{-2}$ &
405: $3.8 \times 10^{-1}$\\
406: C1\_p4 & 1 & No & No & $1.0 \times 10^{-16}$ & $1.0 \times 10^{-16}$ &
407: $1.0 \times 10^{-16}$\\
408: C03\_p4 & 0.3 & No & No & $1.0 \times 10^{-16}$ & $1.0 \times 10^{-16}$ &
409: $1.0 \times 10^{-16}$\\
410: C01\_p4 & 0.1 & No & No & $1.0 \times 10^{-16}$ & $1.0 \times 10^{-16}$ &
411: $1.0 \times 10^{-16}$\\
412: C003\_p4 & 0.03 & No & No & $4.4 \times 10^{-16}$ & $2.2 \times 10^{-11}$ &
413: $3.6 \times 10^{-7}$\\
414: \enddata
415: \tablenotetext{a}{From \citet{ida08b} -- $C_1$ equivalent to the fraction of
416: the full Type I migration rate predicted from linear theory applied during
417: the simulation}
418: \tablenotetext{b}{Lower bound of an interval centered on median $p$-value
419: which contains 95\% of our bootstrap realizations}
420: \tablenotetext{c}{Median $p$-value of our bootstrap realizations}
421: \tablenotetext{d}{Upper bound of an interval centered on median $p$-value
422: which contains 95\% of our bootstrap realizations}
423: \end{deluxetable}
424:
425: \clearpage
426: \begin{figure}
427: %\plottwo{f1a.eps}{f1b.eps}
428: %\plotone{f1_color.eps}
429: \plotone{f1_color_small.eps}
430: \caption{Results of our Monte Carlo simulation. The open circles are
431: the simulated planets from Figure 3 of \citet{ida08b}; the model includes
432: the effects of the snow line on the surface density of gas $\Sigma_g$ and
433: dust $\Sigma_d$. The Type I migration rate in the simulation is 0.1 times
434: the prediction from linear theory. The filled circles are planets that are
435: robustly detected, that is, planets that would be detected at least 90\% of
436: the time by current radial velocity surveys. We also plot all known exoplanet
437: planets as filled squares. All simulated planets in the range
438: 0.1 AU $< a <$ 2.0 AU with $M_p$ as specified by Equation~(\ref{eq2}) denoted
439: by the heavy black lines are robustly detected, so existing radial velocity
440: surveys are complete in that range. See the electronic edition of the Journal
441: for a color version of this figure.\label{fig1}}
442: \end{figure}
443:
444: %\clearpage
445: %\begin{figure}
446: %\plotone{f1a.eps}
447: %\plotone{f1_color.eps}
448: %\caption{Results of our Monte Carlo simulation. The open circles are
449: %the simulated planets from Figure 3 of \citet{ida08b}; the model includes
450: %the effects of the snow line on the surface density of gas $\Sigma_g$ and
451: %dust $\Sigma_d$. The Type I migration rate in the simulation is 0.1 times
452: %the prediction from linear theory. The filled circles are planets that are
453: %robustly detected, that is, planets that would be detected at least 90\% of
454: %the time by current radial velocity surveys. We also plot all known exoplanet
455: %planets as filled squares. All simulated planets in the range
456: %0.1 AU $< a <$ 2.0 AU with $M_p$ as specified by Equation~(\ref{eq2}) denoted
457: %by the heavy black lines are robuslty detected, so existing radial velocity
458: %surveys are complete in that range. See the electronic edition of the Journal
459: %for a color version of this figure.\label{fig1}}
460: %\end{figure}
461:
462: \clearpage
463: \begin{figure}
464: %\plottwo{f2a.eps}{f2b.eps}
465: \plotone{f2_color.eps}
466: \caption{1D distribution in semimajor axis derived from the projection
467: of the 2D distribution in the complete region of the $M_p-a$ plane from
468: \citet{ida08b} for the disks without the bump in $\Sigma_g$ or $\Sigma_d$.
469: The upper left panel uses the full Type I migration rate from linear theory,
470: the upper right panel uses 30\% of the full rate, the bottom left panel
471: uses 10\% of the full rate, and the bottom right panel uses 3\% of the full
472: rate. See the electronic edition of the Journal for a color version of this
473: figure.\label{fig2}}
474: \end{figure}
475:
476: %\clearpage
477: %\begin{figure}
478: %\plotone{f2a.eps}
479: %\caption{1D distribution in semimajor axis derived from the projection
480: %of the 2D distribution in the complete region of the $M_p-a$ plane from
481: %\citet{ida08b} for the disks without the bump in $\Sigma_g$ or $\Sigma_d$.
482: %The upper left panel uses the full Type I migration rate from linear theory,
483: %the upper right panel uses 30\% of the full rate, the bottom left panel
484: %uses 10\% of the full rate, and the bottom right panel uses 3\% of the full
485: %rate. The error bars indicate the 2-$\sigma$ region. See the electronic
486: %edition of the Journal for a color version of this
487: %figure.\label{fig2}}
488: %\end{figure}
489:
490:
491: \clearpage
492: \begin{figure}
493: %\plottwo{f3a.eps}{f3b.eps}
494: \plotone{f3_color.eps}
495: \caption{1D distribution in semimajor axis derived from the projection
496: of the 2D distribution in the complete region of the $M_p-a$ plane from
497: \citet{ida08b} for the disks with the bump in $\Sigma_g$ but without
498: enhancement in $\Sigma_d$.
499: The upper left panel uses the full Type I migration rate from linear theory,
500: the upper right panel uses 30\% of the full rate, the bottom left panel
501: uses 10\% of the full rate, and the bottom right panel uses 3\% of the full
502: rate. The error bars indicate the 2-$\sigma$ region. See the electronic
503: edition of the Journal for a color version of this
504: figure.\label{fig3}}
505: \end{figure}
506:
507: %\clearpage
508: %\begin{figure}
509: %\plotone{f4a.eps}
510: %\caption{1D distribution in semimajor axis derived from the projection
511: %of the 2D distribution in the complete region of the $M_p-a$ plane from
512: %\citet{ida08b} for the disks with the bump in $\Sigma_g$ but without
513: %enhancement in $\Sigma_d$.
514: %The upper left panel uses the full Type I migration rate from linear theory,
515: %the upper right panel uses 30\% of the full rate, the bottom left panel
516: %uses 10\% of the full rate, and the bottom right panel uses 3\% of the full
517: %rate. The error bars indicate the 2-$\sigma$ region. See the electronic
518: %edition of the Journal for a color version of this figure.\label{fig3}}
519: %\end{figure}
520:
521: \clearpage
522: \begin{figure}
523: %\plottwo{f4a.eps}{f4b.eps}
524: \plotone{f4_color.eps}
525: \caption{1D distribution in semimajor axis derived from the projection
526: of the 2D distribution in the complete region of the $M_p-a$ plane from
527: \citet{ida08b} for the disks with the bump in both $\Sigma_g$ and $\Sigma_d$.
528: The upper left panel uses the full Type I migration rate from linear theory,
529: the upper right panel uses 30\% of the full rate, the bottom left panel
530: uses 10\% of the full rate, and the bottom right panel uses 3\% of the full
531: rate. The error bars indicate the 2-$\sigma$ region. See the electronic
532: edition of the Journal for a color version of this figure.\label{fig4}}
533: \end{figure}
534:
535: %\clearpage
536: %\begin{figure}
537: %\plotone{f4a.eps}
538: %\caption{1D distribution in semimajor axis derived from the projection
539: %of the 2D distribution in the complete region of the $M_p-a$ plane from
540: %\citet{ida08b} for the disks with the bump in both $\Sigma_g$ and $\Sigma_d$.
541: %The upper left panel uses the full Type I migration rate from linear theory,
542: %the upper right panel uses 30\% of the full rate, the bottom left panel
543: %uses 10\% of the full rate, and the bottom right panel uses 3\% of the full
544: %rate. The error bars indicate the 2-$\sigma$ region. See the electronic
545: %edition of the Journal for a color version of this figure.\label{fig4}}
546: %\end{figure}
547:
548: \clearpage
549: \begin{figure}
550: %\plottwo{f5a.eps}{f5b.eps}
551: \plotone{f5_color.eps}
552: \caption{Distribution of the Kolmogorov-Smirnov test $p$-values resulting
553: from 1000 bootstrap resamplings; similar distributions will have a
554: sharply-peaked $p$-value distributions with a maximum near
555: $p \sim 1 \Rightarrow \log{p} \sim 0$. The top row shows the results for
556: models including the bump in both $\Sigma_g$ and $\Sigma_d$, the middle
557: row shows models including just the bump in $\Sigma_g$, and the bottom
558: row shows models with no bump in $\Sigma_g$ or $\Sigma_d$. For all rows,
559: the first column shows models with the full Type I migration rate,
560: the second column shows models with 30\% of the full Type I migration rate,
561: the third column shows models with 10\% of the full Type I migration rate,
562: and the fourth column shows models with 3\% of the full Type I migration rate.
563: The $p$-value distribution of the model with the bump in both
564: $\Sigma_g$ and $\Sigma_d$ and 10\% of the full Type I migration rate is the
565: best match to the observed data. There are no histograms for models C1\_p4,
566: C03\_p4, or C01\_p4 because the $p$-values were vanishingly small.
567: See the electronic edition of the Journal for a color version of this
568: figure.\label{fig5}}
569: \end{figure}
570:
571: %\clearpage
572: %\begin{figure}
573: %\plotone{f5a.eps}
574: %\caption{Distribution of the Kolmogorov-Smirnov test $p$-values resulting
575: %from 1000 bootstrap resamplings; similar distributions will have a
576: %sharply-peaked $p$-value distributions with a maximum near
577: %$p \sim 1 \Rightarrow \log{p} \sim 0$. The top row shows the results for
578: %models including the bump in both $\Sigma_g$ and $\Sigma_d$, the middle
579: %row shows models including just the bump in $\Sigma_g$, and the bottom
580: %row shows models with no bump in $\Sigma_g$ or $\Sigma_d$. For all rows,
581: %the first column shows models with the full Type I migration rate,
582: %the second column shows models with 30\% of the full Type I migration rate,
583: %the third column shows models with 10\% of the full Type I migration rate,
584: %and the fourth column shows models with 3\% of the full Type I migration rate.
585: %The $p$-value distribution of the model with the bump in both
586: %$\Sigma_g$ and $\Sigma_d$ and 10\% of the full Type I migration rate is the
587: %best match to the observed data. There are no histograms for models C1\_p4,
588: %C03\_p4, or C01\_p4 because the $p$-values were vanishingly small.
589: %See the electronic edition of the Journal for a color version of this
590: %figure.\label{fig5}}
591: %\end{figure}
592:
593: \clearpage
594: \begin{figure}
595: %\plottwo{f6a.eps}{f6b.eps}
596: %\plotone{f6_color.eps}
597: \plotone{f6_color_small.eps}
598: \caption{Results of our Monte Carlo simulation. The open circles are
599: the simulated planets of model C01C. As in Figure~\ref{fig1}, the filled
600: circles are planets that are robustly detected by a radial velocity survey
601: with $n \sim 700$ observations per program star; we plot all known exoplanet
602: planets as filled squares. The centers of the best fit Gaussian mixture to
603: all robustly detected exoplanets below the dashed line at $100~M_{\oplus}$
604: models are marked by the two X's, while the characteristic ellipses
605: of each component are the solid lines. We say
606: the ``desert" is detected if the mean vectors of the two components
607: are offset by more than $0.6$ in $\log{a}$ and the minor axis of the
608: ellipse at small orbital radius is larger than the minor axis of the
609: ellipse at large orbital radius. We find that the ``desert" is detected more
610: than 90\% of the time when $\mu_n = 700$. See the electronic edition of the
611: Journal for a color version of this figure.\label{fig6}}
612: \end{figure}
613:
614: %\clearpage
615: %\begin{figure}
616: %\plotone{f6a.eps}
617: %\caption{Results of our Monte Carlo simulation. The open circles are
618: %the simulated planets of model C01C. As in Figure~\ref{fig1}, the filled
619: %circles are planets that are robustly detected by a radial velocity survey
620: %with $n \sim 700$ observations per program star; we plot all known exoplanet
621: %planets as filled squares. The centers of the best fit Gaussian mixture to
622: %all robustly detected exoplanets below the dashed line at $100~M_{\oplus}$
623: %models are marked by the two X's, while the characteristic ellipses
624: %of each component are the solid lines. We say
625: %the ``desert" is detected if the mean vectors of the two components
626: %are offset by more than $0.6$ in $\log{a}$ and the minor axis of the
627: %ellipse at small orbital radius is larger than the minor axis of the
628: %ellipse at large orbital radius. We find that the ``desert" is detected more
629: %than 90\% of the time when $\mu_n = 700$. See the electronic edition of the
630: %Journal for a color version of this figure.\label{fig6}}
631: %\end{figure}
632:
633: \end{document}
634: