AI will allow researchers to perform long and complex calculations in a fraction of a second
Have you ever wondered how long a planet will last? From our solar system? Or did it not occur to you for a second, like us, that the planets could also disappear or that their structure could weaken over the years? Okay, billions of years but still deteriorating? And if so, which planetary system was more likely to survive? Princeton researchers have dug up these questions.
ORBITAL STABILITY: A MYSTERY THAT HAS PASSED THROUGH THE AGES
Orbital stability is a subject that astronomers have long tried to understand. Isaac Newton was one of them and research on the subject has even led to several mathematical theories, reports Phys.org,including chaos theory. However, according to Daniel Tamayo, an astrophysics researcher at Princeton and a fellow in NASA’s Hubble Program, predicting and calculating spatial configurations that could lead to collisions is “a fascinating and difficult problem” because astronomers must take into account and calculate the movements of several planets over billions of years.
Over time, no scientist has yet discovered a way to theoretically predict stable configurations, and today’s astronomers still have to refer to calculations, although they now have at their disposal supercomputers to speed up their calculations.
MACHINE LEARNING TO MAKE IT EASIER TO CALCULATE
Tamayo discovered that he could accelerate these computational processes by “combining simplified models of dynamic planet interactions with machine learning methods.” This would mean that calculations that would have previously taken tens of thousands of hours can now be done in minutes.
According to the scientist, there are many possible orbital configurations but not all of them are stable. Theoretically possible configurations could thus deteriorate in a few million years. However, the researchers want to eliminate these “rapid instabilities”. ».
SPOCK COULD HELP SCIENTISTS GET A CLEARER PICTURE
In this study,published in Proceedings of the National Academy of Sciences,Tamayo, the lead author as well as co-authors Miles Cranmer, David Spergel and Charles Young, chose to simulate a given configuration for 10,000 orbits and resulted in a machine learning algorithm to predict stable configurations. Tamayo explains that:
WE CANNOT CATEGORICALLY SAY THAT “THIS SYSTEM WILL BE OK BUT THIS ONE WILL EXPLODE SOON”. RATHER, THE GOAL IS TO EXCLUDE ALL UNSTABLE POSSIBILITIES THAT WOULD HAVE ALREADY COLLIDED AND COULD NOT CURRENTLY EXIST FOR A GIVEN SYSTEM. WE CALLED THE SPOCK MODEL IN PART BECAUSE IT DETERMINES WHETHER THE SYSTEMS WILL LIVE LONG AND PROSPER. THIS NEW METHOD WILL PROVIDE A CLEARER WINDOW INTO THE ORBITAL ARCHITECTURES OF PLANETARY SYSTEMS BEYOND OURS.
Although Tamayo and his colleagues warn that they have not solved the problem of planetary stability, SPOCK – Stability of Planetary Orbital Configurations Klassifier – will already be able to identify rapid instabilities in compact systems and be able to determine the long-term stability of planetary configurations 100,000 times faster than previous methods.