Astrophysicists use artificial intelligence to determine exoplanets sizes
2019 October 09

This artist’s impression shows several of the planets orbiting the ultra-cool red dwarf star TRAPPIST-1. Credit: ESO/M. KornmesserTrue radii as a function of the predicted radii for the test set. Credit: Ulmer-Moll et al.
A team1 of Instituto de Astrofísica e Ciências do Espaço (IA2) researchers has published an article3, led by Solène Ulmer-Moll, which shows that by knowing an exoplanet’s mass and equilibrium temperature, it’s possible to constrain its radius, with higher accuracy than previous methods.

Solène Ulmer-Moll, a PhD student at the Science Faculty of the University of Porto (FCUP) explains this result was obtained by using knowledge from different fields: “This novel way to forecast exoplanet radius is a perfect example of the synergy between exoplanet science and machine learning techniques.”

To characterize a planet, both its mass and radius are needed, in order to find the planet’s density, and from that infer its composition. But both data are only available for a reduced number of exoplanets, since the mass is often determined by radial velocity4 measurements, while radius is measured with the transit method5.

The team developed an algorithm which accurately forecasts the radius of a wide range of exoplanets, if several other planetary and stellar parameters are known, mainly, the exoplanet's mass and equilibrium temperature. Solène Ulmer-Moll adds: “For the hundreds of planets discovered with the radial velocity method, we are now able to predict their radius. We can then understand if these exoplanets are potentially rocky worlds.

So far, only the mass of an exoplanet was used to predict its radius, but the team is working on changing this paradigm by using other planetary and stellar parameters to strengthen their predictions.

Nuno Cardoso Santos (IA & FCUP), leader of IA’s thematic line “Towards the detection and characterization of other Earths” adds: “This work beautifully puts together the expertise in our team, uniting the existing knowledge about exoplanet detection and characterization and the statistical analysis of the detected systems, using state-of-the-art mathematical tools. These are essentially the same mathematical tools that are now leading to the development of autonomous cars.

The result was published today in Astronomy & Astrophysics journal.

  1. The team is: Solène Ulmer-Moll (IA & FCUP), Nuno Cardoso Santos (IA & FCUP), Pedro Figueira (ESO & IA), Jarle Brinchmann (IA & Leiden Observatory, Leiden University) and João Faria (IA & FCUP).
  2. The Instituto de Astrofísica e Ciências do Espaço (Institute of Astrophysics and Space Sciences – IA) is the reference Portuguese research unit in space sciences and integrates researchers from the University of Porto and the University of Lisbon. The institute encompasses most of the field’s national scientific output and it was evaluated as Excellent in the last evaluation Fundação para a Ciência e Tecnologia (FCT) commissioned from the European Science Foundation (ESF). IA's activity is funded by national and international funds, including Fundação para a Ciência e a Tecnologia (UID/FIS/04434/2019).
  3. The article “Beyond the exoplanet mass-radius relation” was published in the journal Astronomy & Astrophysics Vol. 630 A135 (DOI: 10.1051/0004-6361/201936049)
  4. The Radial Velocity method detects exoplanets measuring tiny variations in the (radial) velocity of the star, due to the motion that an orbiting planet induces in the star. As an example, the speed variation that the Earth imprints in the Sun is of about 10 cm/s (about 0.36 km/h). With this method you can set a minimum value for the planets’ mass.
  5. The Transits method measures the dimming of starlight produced when an exoplanet crosses in front of its star (something similar to a “micro-eclipse”). A transit enables for the determination of the planets’ radius only. It’s also a difficult method to use, because it requires that both planet and star be exactly be in line with the observers’ line of sight.

Solène Ulmer-Moll
Nuno Cardoso Santos

Science Communication Group
Ricardo Cardoso Reis
Sérgio Pereira
Daniel Folha (Coordination, Porto)
João Retrê (Coordination, Lisboa)

Faculdade de Ciências da Universidade de Lisboa Universidade do Porto Faculdade de Ciências e Tecnologia da Universidade de Coimbra
Fundação para a Ciência e a Tecnologia COMPETE 2020 PORTUGAL 2020 União Europeia