Thirty years have passed since the first scientific proof of the existence of planets outside the solar system was obtained. By the time of this publication, 3,767 objects had received official exoplanet status, with a total of more than 4,500 candidates.
Most of these planets are very harsh and absolutely unsuitable for life worlds, but some of them, according to scientists, may still have suitable conditions for its occurrence. At least they are not too hot and at the same time not too cold in order to maintain the presence of water on their surface in liquid form. And water, as you know, is one of the sources of life.
Of course, the main reason for the search for new exoplanets is the search for life outside the Earth. Why else spend huge sums of money on building new telescopes and creating new technologies for space exploration? Therefore, scientists from Columbia University (USA) have developed a new system that can simplify the "hunt" for potentially habitable worlds. Using machine learning algorithms, the researchers have created technology that makes it possible to more effectively determine the survival of a particular exoplanet in a stable orbit.
In this work, the researchers focused their attention on the so-called "Tatooines", or exoplanets orbiting double stars, just like the desert world of Luke Skywalker from "Star Wars". Formally known in scientific circles as circumbinary planets, they can undergo colossal orbital changes, since they are always in the gravitational pool of two stars at once. Being attracted to one star, then to another, they risk being thrown out of the system over time, and in the worst case - falling on one of their stars.
Scientists have developed an equation that helps determine the long-term stability of the orbit of circumbinary planets, however, according to Chris Lam, head of the development in question today, this equation cannot provide accurate data, taking into account all possible circumstances.
“The problem is that when there are three or more bodies in the system, the movement becomes 'chaotic', as physicists and mathematicians say,” comments Lam.
“Therefore, there are borderline cases where the equation predicts that the system is unstable when it is actually stable, and vice versa. We thought that a neural network would help us cope with this problem."
The ability to predict whether a planet will be thrown out of its system is not just a whim, it is an additional opportunity to determine the habitability potential of a particular world. In the end, it took several billion years for the emergence and development of life, at least the one that exists on Earth. In other words, there will be no chance for it if we are talking about a planet wandering in space and not tied to its luminary.
Promotional video:
For a more efficient method of determining the survivability of Tatooines, Lam and colleagues created a machine learning algorithm that scientists trained using 10 million simulated planets. After several hours of experimentation and tuning, Lam notes, the system was able to surpass the accuracy of the traditional equation "in all respects."
Scientists expect that NASA's new TESS space telescope, recently successfully launched into orbit, will be able to detect many new circumbinary planets, and the development of researchers from Columbia University, Lam said, could help in the study of these worlds.
“Our model will help astronomers understand which regions are best for finding planets around binary systems. This, I hope, will help us not only discover new exoplanets, but also better understand their features,”the scientist noted.
Nikolay Khizhnyak