Using AI to find the smallest and closest exoplanets around sun-like stars

Written by Copernical Team Wednesday, 16 October 2024 16:42

font size decrease font size increase font size
Print
Email

Write a comment

Sydney, Australia (SPX) Oct 16, 2024

An international research team led by Professor Jian Ge from the Shanghai Astronomical Observatory has announced the discovery of five new ultra-short-period planets (USPs), each smaller than Earth and orbiting their stars in less than one day. The discoveries were made using data from the Kepler space telescope, originally released in 2017. The breakthrough came through the application of an ad

Using AI to find the smallest and closest exoplanets around sun-like stars
by Simon Mansfield
Sydney, Australia (SPX) Oct 16, 2024

An international research team led by Professor Jian Ge from the Shanghai Astronomical Observatory has announced the discovery of five new ultra-short-period planets (USPs), each smaller than Earth and orbiting their stars in less than one day. The discoveries were made using data from the Kepler space telescope, originally released in 2017. The breakthrough came through the application of an advanced deep learning algorithm that combines GPU phase folding with convolutional neural networks. This marks the first time artificial intelligence has been used to both detect candidate signals and confirm true signals in a single process, and the first instance of neural networks trained with a large number of simulated transit signals to detect fainter signals than previously possible.

Among the newly found planets, four are the smallest and closest to their host stars discovered to date, with sizes comparable to that of Mars. The findings were published in the *Monthly Notices of the Royal Astronomical Society (MNRAS)*.

The team, led by Jian Ge, spent five years developing the deep learning algorithm, known as GPFC, which integrates GPU phase folding with convolutional neural networks. This new algorithm significantly improves the speed and accuracy of detecting transit signals, making the search approximately 15 times faster and boosting detection accuracy and completeness by about 7% over the widely used BLS method. The GPFC algorithm has already been employed to analyze Kepler's dataset, resulting in the identification of five new USPs: Kepler-158d, Kepler-963c, Kepler-879c, Kepler-1489c, and Kepler-2003b. Notably, Kepler-879c, Kepler-158d, Kepler-1489c, and Kepler-963c rank as some of the smallest ultra-short-period planets ever discovered, while Kepler-879c, Kepler-158d, Kepler-1489c, and Kepler-2003b are also among the closest to their stars, with orbital radii within five stellar radii.

Jian Ge explained, "The true start of this work was in 2015, when AlphaGo made a major breakthrough by defeating professional Go players. Inspired by my colleague Professor Xiaolin Li from the Computer Science Department at the University of Florida, I decided to apply deep learning in artificial intelligence to the data released by Kepler, in hopes of finding weak transit signals that traditional methods could not detect, including Earth 2.0 in the habitable zone around solar-like stars. Fortunately, after nearly 10 years of effort, we finally have our first harvest. Even more surprising is that this year's Nobel Prizes in Physics and Chemistry were awarded to advocates of deep learning and the AlphaFold team, who used deep learning to predict protein structures. It seems we have caught up with the new wave of technological advancements."

These discoveries offer important insights into the early evolution of planetary systems and the dynamics of planet-star interactions, such as tidal forces and atmospheric erosion. The findings are significant for advancing planetary formation theories and present a novel approach for efficiently identifying transit signals in high-precision photometric data, highlighting AI's potential in uncovering faint signals from vast astronomical datasets.

Professor Josh Winn of Princeton University remarked, "Ultrashort period planets, or 'lava worlds,' have such extreme and unexpected properties, and they might give us clues about how the orbits of planets can change over time - by spiraling inwards or being thrown around by interactions with other planets. I had assumed the data returned by the Kepler mission had already been 'mined out' and would contain no additional planets, so I'm delighted to know of these new possible planets."

Research Report:Discovery of small ultra-short-period planets orbiting Kepler KG dwarfs with GPU phase folding and deep learning