Observations of exoplanetary systems provide clues about the intrinsic distribution of planetary systems, their architectures, and how they formed. We develop a forward modeling framework for generating populations of planetary systems and "observed" catalogs by simulating the Kepler detection pipeline (SysSim). We compare our simulated catalogs to the Kepler DR25 catalog of planet candidates, updated to include revised stellar radii from Gaia DR2. We constrain our model based on the observed 1-D marginal distributions of orbital periods, period ratios, transit depths, transit depth ratios, transit durations, transit duration ratios, and transit multiplicities. Models assuming planets with independent periods and sizes do not adequately account for the properties of the multi-planet systems. Instead, a clustered point process model for exoplanet periods and sizes provides a significantly better description of the Kepler population, particularly the observed multiplicity and period ratio distributions. We find that 0.56 +0.18 −0.15 of FGK stars have at least one planet larger than 0.5R ⊕ between 3 and 300 days. Most of these planetary systems (∼ 98%) consist of one or two clusters with a median of three planets per cluster. We find that the Kepler dichotomy is evidence for a population of highlyinclined planetary systems and is unlikely to be solely due to a population of intrinsically single planet systems. We provide a large ensemble of simulated physical and observed catalogs of planetary systems from our models, as well as publicly available code for generating similar catalogs given user-defined parameters.

The angular momentum deficit (AMD) of a planetary system is a measure of its orbital excitation and a predictor of long-term stability. We adopt the AMD-stability criterion to constrain the orbital architectures for exoplanetary systems. Previously, He et al. showed through forward modeling (SysSim) that the observed multiplicity distribution can be well reproduced by two populations consisting of a low and a high mutual inclination component. Here, we show that a broad distribution of mutual inclinations arising from systems at the AMD-stability limit can also match the observed Kepler population. We show that distributing a planetary system’s maximum AMD among its planets results in a multiplicity-dependent distribution of eccentricities and mutual inclinations. Systems with intrinsically more planets have lower median eccentricities and mutual inclinations, and this trend is well described by power-law functions of the intrinsic planet multiplicity (n): and , where and are the medians of the eccentricity and inclination distributions. We also find that intrinsic single planets have higher eccentricities (σ e,1 ∼ 0.25) than multiplanet systems and that the trends with multiplicity appear in the observed distributions of period-normalized transit duration ratios. We show that the observed preferences for planet-size orderings and uniform spacings are more extreme than what can be produced by the detection biases of the Kepler mission alone. Finally, we find that for systems with detected transiting planets between 5 and 10 days, there is another planet with a greater radial velocity signal ≃53% of the time.

Planet formation theories predict a large but still undetected population of short-period terrestrial planets orbiting brown dwarfs. Should specimens of this population be discovered transiting relatively bright and nearby brown dwarfs, the Jupiter-size and the low luminosity of their hosts would make them exquisite targets for detailed atmospheric characterisation with JWST and future ground-based facilities. The eventual discovery and detailed study of a significant sample of transiting terrestrial planets orbiting nearby brown dwarfs could prove to be useful not only for comparative exoplanetology but also for astrobiology, by bringing us key information on the physical requirements and timescale for the emergence of life.In this context, we present a search for transit-signals in archival time-series photometry acquired by the Spitzer Space Telescope for a sample of 44 nearby brown dwarfs. While these 44 targets were not particularly selected for their brightness, the high precision of their Spitzer light curves allows us to reach sensitivities below Earth-sized planets for 75% of the sample and down to Europa-sized planets on the brighter targets. We could not identify any unambiguous planetary signal. Instead, we could compute the first limits on the presence of planets on close-in orbits. We find that within a 1.28 day orbit, the occurrence rate of planets with a radius between 0.75 and 3.25 R ⊕ is η < 67 ± 1%. For planets with radii between 0.75 and 1.25 R ⊕ , we place a 95% confident upper limit of η < 87±3%. If we assume an occurrence rate of η = 27% for these planets with radii between 0.75 and 1.25 R ⊕ , as the discoveries of the Kepler-42b and TRAPPIST-1b systems would suggest, we estimate that 175 brown dwarfs need to be monitored in order to guarantee (95%) at least one detection.

A significant fraction of main sequence stars are part of a triple system. We study the long-term stability and dynamical outcomes of triple stellar systems using a large number of long-term direct N-body integrations with relativistic precession. We find that the previously proposed stability criteria by Eggleton & Kiseleva 1995 and Mardling & Aarseth 2001 predict the stability against ejections reasonably well for a wide range of parameters. Assuming that the triple stellar systems follow orbital and mass distributions from FGK binary stars in the field, we find that in ∼ 1% and ∼ 0.5% of the triple systems lead to a direct head-on collision (impact velocity ∼ escape velocity) between main sequence (MS) stars and between a MS star and a stellar-mass compact object, respectively. We conclude that triple interactions are the dominant channel for direct collisions involving a MS star in the field with a rate of one event every ∼ 100 years in the Milky Way. We estimate that the fraction of triple systems that forms short-period binaries is up to ∼ 23% with only up to ∼ 13% being the result of three-body interactions with tidal dissipation, which is consistent with previous work using a secular code.

The Kepler mission observed thousands of transiting exoplanet candidates around hundreds of thousands of FGK dwarf stars. He et al. applied forward modeling to infer the distribution of intrinsic architectures of planetary systems, developed a clustered Poisson point process model for exoplanetary systems (SysSim) to reproduce the marginal distributions of the observed Kepler population, and they showed that orbital periods and planet radii are clustered within a given planetary system. Here, we extend the clustered model to explore correlations between planetary systems and their host-star properties. We split the sample of Kepler FGK dwarfs into two halves and model the fraction of stars with planets (0.5–10R ⊕ and 3–300 days), f swpa, as a linear function of the Gaia DR2 color. We confirm previous findings that the occurrence of these planetary systems rises significantly toward later-type (redder) stars. The fraction of stars with planets increases from for F2V dwarfs to for mid-K dwarfs. About half ( ) of all solar-type (G2V) dwarfs harbor a planetary system between 3 and 300 days. This simple model can closely match the observed multiplicity distributions of both the bluer and redder halves in our sample, suggesting that the architectures of planetary systems around stars of different spectral types may be similar aside from a shift in the overall fraction of planet-hosting stars.

Early analyses of exoplanet statistics from the Kepler mission revealed that a model population of multiplanet systems with low mutual inclinations (∼1°–2°) adequately describes the multiple-transiting systems but underpredicts the number of single-transiting systems. This so-called “Kepler dichotomy” signals the existence of a subpopulation of multiplanet systems possessing larger mutual inclinations. However, the details of these inclinations remain uncertain. In this work, we derive constraints on the intrinsic mutual inclination distribution by statistically exploiting transit duration variations (TDVs) of the Kepler planet population. When planetary orbits are mutually inclined, planet–planet interactions cause orbital precession, which can lead to detectable long-term changes in transit durations. These TDV signals are inclination sensitive and have been detected for roughly two dozen Kepler planets. We compare the properties of the Kepler-observed TDV detections to TDV detections of simulated planetary systems constructed from two population models with differing assumptions about the mutual inclination distribution. We find strong evidence for a continuous distribution of relatively low mutual inclinations that is well characterized by a power-law relationship between the median mutual inclination ( μ ˜ i , n ) and the intrinsic multiplicity (n): μ ˜ i , n = μ ˜ i , 5 ( n / 5 ) α , where μ ˜ i , 5 = 1.10 − 0.11 + 0.15 and α = − 1.73 − 0.08 + 0.09 . These results suggest that late-stage planet assembly and possibly stellar oblateness are the dominant physical origins for the excitation of Kepler planet mutual inclinations.

Although the architectures of compact multiple-planet systems are well characterized, there has been little examination of their “outer edges,” or the locations of their outermost planets. Here we present evidence that the observed high-multiplicity Kepler systems truncate at smaller orbital periods than can be explained by geometric and detection biases alone. To show this, we considered the existence of hypothetical planets orbiting beyond the observed transiting planets with properties dictated by the “peas-in-a-pod” patterns of intrasystem radius and period ratio uniformity. We evaluated the detectability of these hypothetical planets using (1) a novel approach for estimating the mutual inclination dispersion of multitransiting systems based on transit chord length ratios, and (2) a model of transit probability and detection efficiency that accounts for the impacts of planet multiplicity on completeness. Under the assumption that the “peas-in-a-pod” patterns continue to larger orbital separations than observed, we find that ≳35% of Kepler compact multis should possess additional detected planets beyond the known planets, constituting a ∼7σ discrepancy with the lack of such detections. These results indicate that the outer (∼100–300 days) regions of compact multis experience a truncation (i.e., an “edge-of-the-multis”) or a significant breakdown of the “peas-in-a-pod” patterns, in the form of systematically smaller radii or larger period ratios. We outline future observations that can distinguish these possibilities, and we discuss implications for planet formation theories.

Population studies of Kepler's multiplanet systems have revealed a surprising degree of structure in their underlying architectures. Information from a detected transiting planet can be combined with a population model to make predictions about the presence and properties of additional planets in the system. Using a statistical model for the distribution of planetary systems, we compute the conditional occurrence of planets as a function of the period and radius of Kepler-detectable planets. About half (0.52 ± 0.03) of the time, the detected planet is not the planet with the largest semi-amplitude (K) in the system, so efforts to measure the mass of the transiting planet with radial velocity (RV) follow up will have to contend with additional planetary signals in the data. We simulate RV observations to show that assuming a single-planet model to measure the K of the transiting planet often requires significantly more observations than in the ideal case with no additional planets, due to systematic errors from unseen planet companions. Our results show that planets around 10 day periods with K close to the single-measurement RV precision (σ 1,obs) typically require ∼100 observations to measure their K to within 20% error. For a next generation RV instrument achieving σ 1,obs = 10 cm s−1, about ∼200 (600) observations are needed to measure the K of a transiting Venus in a Kepler-like system to better than 20% (10%) error, which is ∼2.3 times as many as would be necessary for a Venus without any planetary companions.