Artificial intelligence (AI) algorithms trained on real astronomical observations are now outcompeting astronomers to aim through vast amounts of data to find new exploding stars, identify new types of galaxies, and detect mergers of massive stars, accelerating the speed of new discoveries in the world. eldest. science.
But AI, also called machine learning, can reveal something deeper, astronomers at the University of California, Berkeley found: Unimaginable connections hidden in the complex mathematics that stem from general relativity – especially how that theory is used to find new planets around other stars.
In a paper published in the magazine this week Nature astronomyscientists describe how an AI algorithm evolved to more quickly detect exoplanets when such planetary systems pass in front of a background star and briefly illuminate it – a process called gravitational microlensing – revealed that the decades-old theories now used to explain these observations are sadly incomplete.
In 1936, Albert Einstein himself used his new general theory of relativity to show how the light from a distant star can be bent by the gravity of a foreground star, not only making it brighter seen from Earth, but often dividing it into several points. light or distort it into a ring, now called an Einstein ring. This is similar to the way a hand lens can focus and intensify the light from the sun.
But when the foreground object is a star with a planet, the clearing over time – the light curve – is more complicated. What’s more, there are often several planetary orbits that can explain a given light curve just as well – so-called degenerations. It was there that people simplified the math and missed the bigger picture.
However, the AI algorithm pointed in a mathematical way to unite the two main types of degeneration in the interpretation of what telescopes detect during microlensing, and showed that the two “theories” are really special cases of a broader theory that scientists admit is probably still incomplete.
“A machine learning inference algorithm we have previously developed led us to discover something new and fundamental about the equations that govern the general relativistic effect of light bending of two massive bodies,” Joshua Bloom wrote in a blog post last year when he uploaded the newspaper . to a preprint server, arXiv. Bloom is a professor of astronomy at UC Berkeley and chairman of the department.
He compared the discovery of UC Berkeley graduate student Keming Zhang to connections that Google’s AI team, DeepMind, recently made between two different areas of mathematics. Taken together, these examples show that AI systems can reveal fundamental associations that humans miss.
“I claim that they are one of the first – if not the first – time[s] that AI has been used to directly provide new theoretical insights into mathematics and astronomy, “Bloom said.” Just as Steve Jobs suggested that computers could be the bikes of the mind, we have sought an AI framework to serve as an intellectual rocket ship for scientists. . “
“This is a kind of milestone in artificial intelligence and machine learning,” stressed co-author Scott Gaudi, a professor of astronomy at Ohio State University and one of the pioneers in using gravitational microlenses to detect exoplanets. “Keming’s machine learning algorithm revealed this degeneration, which had been missed by experts in the field who had been working with data for decades. This suggests how research is going to go in the future when it is aided by machine learning, which is really exciting.”
Discovery of exoplanets with microlensing
More than 5,000 exoplanets or extrasolar planets have been discovered around stars in the Milky Way, although few have actually been seen through a telescope – they are too faint. Most have been spotted because they create a Doppler wobble in the movements of their host stars, or because they slightly dim the light from the host star as they cross in front of it – transits that were the focus of NASA’s Kepler mission. Only a few more than 100 have been detected by a third technique, microlensing.
One of the main goals of NASA’s Nancy Grace Roman Space Telescope, scheduled to be launched in 2027, is to discover thousands more exoplanets via microlenses. The technique has an advantage over the Doppler and transit techniques in that it can detect lower-mass planets, including those the size of Earth, that are far from their stars, at a distance similar to Jupiter or Saturn in our solar system.
Bloom, Zhang, and their colleagues set out two years ago to develop an AI algorithm to analyze microlens data more quickly to determine the star and planetary masses of these planetary systems and the distances the planets orbit their stars. Such an algorithm would accelerate the analysis of the probable hundreds of thousands of events that the Roman telescope would detect to find the 1% or less caused by exoplanetary systems.
One problem astronomers encounter, however, is that the signal observed may be ambiguous. When a lone foreground star passes in front of a background star, the brightness of the background stars rises steadily to a peak and then decreases symmetrically to its original brightness. It is easy to understand mathematically and observationally.
But if the foreground star has a planet, the planet creates a separate brightness peak within the peak caused by the star. When attempting to reconstruct the orbital configuration of the exoplanet that produced the signal, the general theory of relativity often allows for two or more so-called degenerate solutions, all of which can explain the observations.
To date, astronomers have generally treated these degenerations in simplified and artificially different ways, Gaudi said. If the distant starlight passes close to the star, the observations can be interpreted as either a broad or a dense orbit of the planet – an ambiguity that astronomers can often solve with other data. Another type of degeneration occurs when the background starlight passes close to the planet. In this case, however, the two different solutions for the planetary orbit are generally only slightly different.
According to Gaudi, these two simplifications of two-body gravitational microlensing are usually sufficient to determine the true masses and orbital distances. In fact, in a paper released last year, Zhang, Bloom, Gaudi, and two other UC Berkeley co-authors, astronomy professor Jessica Lu and graduate student Casey Lam described a new AI algorithm that is not at all dependent on knowledge of these interpretations. The algorithm greatly speeds up the analysis of microlens observations, gives results in milliseconds instead of days, and drastically reduces computer crunch.
Zhang then tested the new AI algorithm on microlens light curves from hundreds of possible orbital configurations of star and exoplanet, noting something unusual: there were other ambiguities that the two interpretations did not take into account. He concluded that the commonly used interpretations of microlensing were in fact only special cases of a broader theory explaining the whole range of ambiguities in microlensing events.
“The two previous theories of degeneration deal with cases where the background star appears to pass close to the foreground star or the foreground planet,” Zhang said. “The AI algorithm showed us hundreds of examples from not only these two cases, but also situations where the star does not pass close to either the star or the planet and can not be explained by either previous theory. That was the key to us proposing the new unifying theory.”
Gaudi was skeptical at first, but came after Zhang produced many examples where the two previous theories did not fit with observations, and the new theory did. Zhang actually looked at the data from two dozen previous articles that reported the discovery of exoplanets through microlensing and found that in all cases, the new theory fit the data better than the previous theories.
“People saw these micro-lens events that actually exhibited this new degeneration, but were just not aware of it,” Gaudi said. “It was really just machine learning watching thousands of events that became impossible to miss.”
Zhang and Gaudi have submitted a new paper that carefully describes the new mathematics based on general relativity and explores the theory in microlens situations where more than one exoplanet orbits a star.
The new theory makes technical interpretation of microlens observations more ambiguous, as there are more degenerate solutions to describe the observations. But the theory also clearly demonstrates that observing the same microlens event from two perspectives – from Earth and from the orbits of the Roman space telescope, for example – will make it easier to adjust to the correct orbits and masses. That’s what astronomers are currently planning to do, Gaudi said.
“The AI suggested a way to look at the lens equation in a new light and uncover something really deep about the math in it,” Bloom said. “AI emerges in a way like not just this kind of blunt tool that’s in our toolbox, but like something that’s actually pretty smart. Together with an expert like Keming, the two were able to do something pretty fundamental.”
New sub-Jupiter mass exoplanet discovered by astronomers
Keming Zhang et al., A ubiquitous collective degeneration in two-body microlens systems, Nature astronomy (2022). DOI: 10.1038 / s41550-022-01671-6
Provided by the University of California – Berkeley
Citation: AI Reveals Independent Mathematics Underlying Search for Exoplanets (2022, May 24) Retrieved May 24, 2022 from https://phys.org/news/2022-05-ai-reveals-unsuspected-math-underlying.html
This document is subject to copyright. Apart from any reasonable trade for the purpose of private investigation or research, no part may be reproduced without written permission. The content is provided for informational purposes only.