EPSC-DPS2025: Artificial intelligence drives the discovery of new exoplanets
Researchers from the University of Bern have developed an Artificial Intelligence (AI) model capable of predicting the architecture of planetary systems and subsequently inferring the presence of yet-to-be-discovered planets. They use the so-called Transformer architecture which is the basis of the Large Language Models powering tools like the recently launched Swiss model Apertus or chatbots such as ChatGPT. The findings have been presented this week at the Joint Meeting of the Europlanet Science Congress and the Division for Planetary Sciences (EPSC-DPS) 2025 in Helsinki.
For more than two decades, researchers at the University of Bern have developed the so-called ‘Bern model’, a suite of computer programs that can numerically simulate the formation of planetary systems, thus shedding light on system architecture. These models are, however, very complex: each simulation from the Bern model can take a few days to a few weeks to be computed using modern super-computers.
Using modern AI techniques trained on the Bern model data, Prof. Yann Alibert and Sara Marques from the NCCR PlanetS and the Center for Space and Habitability of the University of Bern, and Dr. Jeanne Davoult, former PhD student of the University of Bern and now researcher at the DLR in Berlin, have developed an AI model capable of computing the formation of planetary systems in seconds, a million times faster than traditional computations. The study has just been published in the journal Astronomy and Astrophysics and was also presented last week at the ‘Fast Machine Learning for Science’ conference in Zurich.
Knowing where to observe
Present day and near future observational facilities will soon be able to observe and characterize extrasolar planets similar to the Earth, while they so far have been limited to planets closer to their host stars. “Earth-like planet detection requires large amount of observing time. In this context, knowing where to observe is very important to save very costly observation time”, explains Yann Alibert, first author of the study.
In order to prioritize between different possible targets, one can use the observations of easier-to-observe other planets in the same systems. This, however, requires a profound understanding of the so-called architecture of a system: how the properties (orbital position, mass, etc.) of one planet in a system relate to the properties of other planets in the same system.
Inspired by Large Language Models
The team trained its AI model on tens of thousands of numerical simulations of planetary system formation also developed at the University of Bern. “The new AI model can be used to predict the presence and properties of yet-to-be-discovered additional planets in already known extrasolar planetary systems”, as Sara Marques, PhD student at the University of Bern, points out.
In an experiment presented in the current study, the authors showed that in a real three-planet system, the properties of the second and third planet can be inferred from the properties of the innermost planet of the system. Alibert explains: “This approach can be used to generate new planetary systems: Knowing a single planet in a system, we can predict the rest of the planets for systems of three planets with our model.” Alibert continues: “The key in our study was to realize that planetary systems can be seen as sequences of planets, exactly as sentences are sequences of words. This triggered the idea of using the AI methods from Large Language Models, used for instance by chatbots such as ChatGPT, to build our AI model.”
The authors used the so-called ‘Transformer architecture’ introduced in the field in 2017 to create a generative model that can produce sequences of planets orbiting the same stars. “The Large Language Models predict the rest of a sentence based on the sequence created by the first few words. In our case, we predict the sequence of outer planets in a system, based on the first inner ones,” further explains Marques.
“This new study builds upon a previous AI model encouraging results,” points out Dr. Jeanne Davoult, former student in the NCCR PlanetS, now working at the DLR Berlin. “In the last model, based on the inner planet of a system, we were predicting the probability of an Earth-like planet to be in the system. Keeping the analogy with language models, it was like predicting the presence of a specific word in a sentence, based on its beginning. In this new study, we predict all the rest of the sentence and not only the probability of a single word.”
“The results of the generative AI model were so accurate that we were very skeptical at first,” remembers Marques. A large range of tests were made by the researchers, in which they used machine learning classifiers, and they submitted their results to other scientists. “In the end, they all concluded the same: generated planetary systems are virtually indistinguishable from numerical simulations,” continues Marques.
Preparing for the PLATO mission and others
Scheduled to be launched in 2026, the ESA PLATO mission will discover thousands of planetary systems, with the planet closest to the star being, in general, the first to be observed. Some of these systems could harbor planets like the Earth, yet these will likely be discovered by ground-based telescope using other observations later.
“Our new AI model could be used to prioritize the observations of these systems by telescope, enhancing the probability to find Earth twins”, says Davoult. In the coming years, the models will be extended to predict more properties of planets, such as their composition or habitability. “When I was hired as a postdoc in 2001, I initiated numerical simulations of planetary systems at the University of Bern. This new AI model is the natural continuation of this Bernese expertise”, says Alibert. “AI is now present in everyone’s life, I am convinced it will more and more be key in scientific discoveries, in planetary sciences and elsewhere”, he concludes.
Publication details:Alibert, Y, Davoult, J., Marques, S., 2025, A transformer-based generative model for planetary systems, Astronomy and Astrophysics.
URL: https://www.aanda.org/articles/aa/full_html/2025/09/aa52297-24/aa52297-24.html
DOI: 10.1051/0004-6361/202452297
Contacts
EPSC–DPS 2025 Press Office
press@europlanet.org
Image
Notes for Editors
About the Joint Meeting of the Europlanet Science Congress and the Division of Planetary Sciences (EPSC-DPS)
The Europlanet Science Congress (EPSC), established in 2006 as the European Planetary Science Congress, is the largest planetary science meeting in Europe. It covers the entire range of planetary sciences, with an extensive mix of talks, workshops and poster sessions, as well as providing a unique space for networking and exchanges of experiences.
EPSC joined forces for the first time with the American Astronomical Society’s Division for Planetary Sciences (DPS) for a joint meeting in Nantes, France, in 2011. This was followed by DPS-EPSC 2016 in Pasadena, EPSC-DPS 2019 in Geneva, and the return to the United States for the DPS-EPSC 2023 meeting in San Antonio. This year will mark the third iteration of a joint European-based meeting. The intent of the joint meetings is not only to connect the European and North American planetary science communities, but also to consolidate two major meetings and motivate planetary scientists from all over the globe to attend. With around 1800 participants expected to join in person and online, EPSC-DPS2025 will be the largest planetary science meeting held to date in Europe.
Follow on social media (Bluesky, X and LinkedIn) with the hashtag #EPSC-DPS2025 for updates on the meeting.
About Europlanet
Europlanet (europlanet.org) is a non-profit association and membership organisation that provides the planetary science community with access to research infrastructure, services and training. The Europlanet Association Sans But Lucratif (AISBL), established in 2023, builds on the heritage of a series of projects funded by the European Commission between 2005 and 2024 (Grant Numbers 871149, 654208, 228319 and RICA-CT-2004-001637) to support the planetary science community in Europe and around the world.
About the DPS
The Division for Planetary Sciences (DPS), founded in 1968, is the largest special-interest Division of the American Astronomical Society (AAS). Members of the DPS study the bodies of our own solar system, from planets and moons to comets and asteroids, and all other solar-system objects and processes. With the discovery that planets exist around other stars, the DPS has expanded its scope to include the study of extrasolar planetary systems as well. The American Astronomical Society (AAS), established in 1899, is the major organization of professional astronomers in North America. The mission of the AAS is to enhance and share humanity’s scientific understanding of the universe as a diverse and inclusive astronomical community, which it achieves through publishing, meeting organization, science advocacy, education and outreach, and training and professional development.
Bern Model of Planet Formation and Evolution
Statements can be made about how a planet was formed and how it has evolved using the “Bern Model of Planet Formation and Evolution”. The Bern model has been continuously developed at the University of Bern since 2001. Insights into the manifold processes involved in the formation and evolution of planets are integrated into the model. These are, for example, sub models of accretion (growth of a planet’s core) or of how planets interact gravitationally and influence each other, and of processes in the protoplanetary disks in which planets are formed. The model is also used to create so-called population syntheses, which show how often planets form in a protoplanetary disk under certain conditions.
Bernese space exploration
With the world’s elite since the first moon landingWhen the second man, “Buzz” Aldrin, stepped out of the lunar module on July 21, 1969, the first task he did was to set up the Bernese Solar Wind Composition experiment (SWC) also known as the “solar wind sail” by planting it in the ground of the moon, even before the American flag. This experiment, which was planned, built and the results analyzed by Prof. Dr. Johannes Geiss and his team from the Physics Institute of the University of Bern, was the first great highlight in the history of Bernese space exploration.Ever since Bernese space exploration has been among the world’s elite, and the University of Bern has been participating in space missions of the major space organizations, such as ESA, NASA, and JAXA. With CHEOPS the University of Bern shares responsibility with ESA for a whole mission. In addition, Bernese researchers are among the world leaders when it comes to models and simulations of the formation and development of planets.The successful work of the Space Research and Planetary Sciences Division (WP) from the Physics Institute of the University of Bern was consolidated by the foundation of a university competence center, the Center for Space and Habitability (CSH). The Swiss National Fund also awarded the University of Bern the National Center of Competence in Research (NCCR) PlanetS, which it manages together with the University of Geneva.
