Research
Research
Our research focuses on the formation and dynamical evolution of planetary systems, the habitability of planets, and the detection of habitable planets around other stars. Machine learning is applied to all of these areas.
Research areas
Select an area to read more and to see related publications.
01Planet formation
The formation of planets, from the growth of dust particles in protoplanetary disks to planetesimals, planetary embryos and final planets. Our work focuses on the growth of planetesimals, giant impacts among planetary embryos, and the conditions that a model must satisfy to produce realistic and quantitative results.
Planetesimal growth · giant impacts · terrestrial planets
02Dynamical evolution
The long-term orbital evolution of planetary systems, including secular and mean-motion resonances, planet migration, and the stability of planets in and around binary star systems. We also study the origin and dynamics of small bodies such as asteroids and main-belt comets.
Orbital stability · secular resonances · small bodies
03Habitability
The formation of habitable planets and the origin of their water. We study how water-carrying bodies from the outer regions of a planetary system are delivered to the accretion zone of terrestrial planets, and how much of that water is retained, or lost, during collisions.
Water delivery · habitable planet formation
04Detecting habitable planets
The identification and detection of planets in habitable zones. This includes calculating the boundaries of the habitable zone in single and binary star systems, and the detection of super-Earths around low-mass stars.
Habitable zones · super-Earths · survey strategy
05Extrasolar planets
The detection, formation and dynamics of extrasolar planets, in particular circumbinary planets (planets that orbit both stars of a binary system). The Director has been part of the teams that discovered several of these planets with the Kepler space telescope.
Kepler-47 · Kepler-1647 · TOI-1338
06Machine learning
The application of machine learning to all of the above areas. Simulations of planet formation and planetary dynamics can take months to years to complete, and a large number of them is needed for statistically meaningful results. Trained on accurate simulations, machine learning models can be used as surrogates during numerical integrations to accelerate the computations, and as probabilistic estimators that predict the outcome of a simulation. They can also be used to explore large parameter spaces, and to process the large volume of data produced by ground- and space-based telescopes.
Surrogate models · probabilistic prediction · parameter-space exploration