When buying with links to our articles, it may gain the future and partnership partners in the commission.
A new AI algorithm can help discover a planet outside the planets suitable for housing. | Credit: ESO/L. Calsada
The algorithm learned the trained machine on artificial planetary systems-and in this process it has identified nearly four real stars that have a high possibility to host a rocky planet in their homeland.
“The 44 system selects a very likely unimpressed system land-He loves PlanetsJane Davott, an astronomer at the German Air Space Agency, said in A, in A. statement. “Another study confirmed the theoretical possibility of these systems to host an Earth -like planet.”
Often, “Earth-like” worlds-similar to the Earth in the sense that they have a similar block To our planet and reside in them star‘s A suitable area for housing They are found by chance, and they are often in huge surveys that see thousands of stars for transit planets. However, astronomers want even the risk of finding the planets of the Earth’s area, and thus require a more targeting way to find candidates.
This prompted Davott to develop the algorithm while she was at Bern University in Switzerland. Like all models based on the algorithms of learning the machine that learn to identify patterns and prepare predictions based on the place where the algorithm sees these patterns, they should have been trained in data. However, the problem is that although nearly 6000 external planets are discovered so far, the information we have in these worlds is incomplete. In general, up to 6000 worlds is not enough to train the algorithm.
Therefore, Davott and her colleagues at the University of Bern, Roman Elchengger and Yan Albert, have turned into another model capable of simulating worlds based on everything we know about planetary systems. Berne has been a planet formation model and development in continuous development at the University of Bern since 2003, and is constantly subject to improvements with more data and theoretical models.
“The Bern Model can be used to make data on how to form planets, and how the types and types of planets that develop under certain circumstances have evolved in a protoplaani disk,” Albert said in the statement. “Bern is one of the only models around the world that provides this wealth of interconnected physical processes and enables a study like the current result.”
Bern Model collects 53,882 Simulating planetary systems around three different types of stars: G-type stars like us sunand Red dwarfs With about half Sun blockAnd a second group of red dwarves with five solar mass.
The algorithm on the search for these simulator systems has identified for patterns or connections, and linking the presence or lack of a planet from the area suitable for the size of the Earth with various brown from the planetary systems.
Some of the links are more clear than others. For example, there is a relationship between the presence of an inner rocky planet that participates in the temptation of an external regime Giant. This is the same architecture Solar power system He has, with rocky planets closer to the sun than gas giants.
On the side of the face, there is a counter -connection between the hot planet, which are giants of gas near their sun, and the planets of the “peas in Bod”, which are chains of rocky planets with similar and oriental mass found around some of the stars of the red dwarfs such as Trappist-1 and Barnard star. Since the hot planet is a giant of the gas that is beyond its star and then migrated to the inside, expelling any planets on its way outside the road, we do not expect to find the hot planet alongside such organized rock planets.
But there are also deeper connections, which Davott identified in previous research. In particular, the mass and the half of the diameter and the tropical period of the detected planet seem to be a large sign of whether the system hosts a moderate planet of the size of the Earth or not.
For example, Davoult found that it is around the stars of the type G like our sun, the presence of a planet seems to the Earth’s size is more likely if the planet’s radius that can be detected is more deeper than 2.5 times the radius of the Earth’s diameter, or if it has a tropical period of more than 10 days.
Armed with the knowledge of these links, the algorithm was successfully trained in simulating data.
“The results are impressive: the algorithm achieves accurate values of up to 0.99, which means that 99 % of the systems that were identified by the machine learning model have at least one planet that resembles Earth,” said Davult.
I am confident of the ability of the algorithm to identify the links, then it was applied to real observations, providing 44 nominated planet systems in which there is a high possibility of the existence of a planet the size of the Earth in the area suitable for housing in its star. Astronomers can now follow these goals, instead of blindly searching for stars.
Related stories:
– How can climate change make an undesirable problem over the land area worse
– The large exotic planets may be born in chaos, and the discovery of retired external planets in NASA
– The nearby Exoplanet can provide evidence about the atmosphere about the hot, hot, rocky worlds
The algorithm will really prove its value in the future. European Space Agency Plato The task is expected to discover several thousand transit planets. By applying the algorithm to Plato’s discoveries, it should be able to narrow thousands of systems on the few who have a greater chance to support a Earth -like planet, allowing astronomers to find it more quickly and efficiently.
“This is an important step in searching for planets with favorable conditions, and in the end, to search for life in the universe,” Albert said.
The results are published in the April 2025 issue of the magazine Astronomy and astronomical physics.
