Since the days of Isaac Newton, scientists have tried to understand how to calculate the stability of a planetary system, and now artificial intelligence has become an ally in studies.
Artificial intelligence has become an important tool for astronomers to understand more about the functioning of planetary systems. After centuries of searching for formulas and mathematical models that help explain how the Solar System remains stable, researchers at Princeton University in the United States have adopted machine learning as a companion in their studies.
Since Isaac Newton, astronomers have tried to understand what makes planets not collide frequently, or how multi-planetary systems like ours organize themselves and maintain stability in their orbits. So far, no one has managed to come up with a way to predict theoretically stable configurations, but the Princeton group of scientists hopes to use artificial intelligence to come up with an answer.
The biggest obstacle faced by astronomers throughout history is that, in order to define the stability of a planetary system, it is necessary to measure all possible unstable configurations of the same system. Planets have their own orbits that last a specific time in each case.
In addition, countless other objects roam space and interact in different ways with other celestial bodies over billions and billions of years. Even the best supercomputer in the world unable to calculate all possible scenarios in a reasonable time
A new approach
Led by Professor Daniel Tamayo, the Princeton group of scientists found a kind of “shortcut”: to speed up the process, they will combine simplified models of the dynamic interactions of the planets with machine learning methods. This eliminates the need for numerous unstable orbit configurations quickly, and reduces the calculation time from tens of thousands of hours to just a few minutes.
The model was dubbed SPOCK (homage to the legendary Star Trek character, and acronym in English for the Planetrics Orbital Settings Stability Classifier). It determines planetary stability about 100 thousand times faster than previous methods.
Instead of calculating a particular configuration for billions of orbits, SPOCK simulates a model for 10,000 different orbits. Thus, the calculation time drops from 10 hours to a fraction of a second. The data is then used in machine learning algorithms to define whether the settings would maintain stability over a period of a billion orbits.
The aim of the study is to understand some of the most distant planetary systems that have been detected in recent times by the Kepler telescope. The study was published in the scientific journal Proceedings of the National Academy of Sciences.
space Solar system astronomers astrophysics Science & Space