Literally weighing the universe may sound like an impossible task, but it is possible, at least to some extent. For decades, astrophysicists haveCumulative electronic pressureAs a surrogate for measuring the mass of galaxy clusters, which includes, among many other complicating factors, the interaction of photons and gravitational forces. This is often an unreliable measurement. But now researchers believe they have developed a (relatively speaking) simple solution to this problem with the help of artificial intelligence.
As detailed in a paper published this month Proceedings of the National Academy of Scientists, the Institute for Advanced Study, the Flatiron Institute’s Center for Computational Astrophysics (CCA), Princeton University, and others used an AI tool called “symbolic regression” to measure the weight of galaxies. Did.As Statement from our collaborators As CCA explains, the tool “basically tries different combinations of mathematical operators like addition and subtraction and different variables to see which equation best fits the data. ”
[Related: We finally have a detailed map of water on the moon.]
The team first fed the tool with a state-of-the-art cosmic simulation featuring a large number of galaxy clusters, and then found variables that AI could improve the accuracy of mass estimates. From there, the AI generated a new equation featuring a single new term on top of the long-standing version focused on integrated electronic pressure. Working backwards, the researchers found that the gas concentration corresponds to regions of galaxy clusters where mass estimates are less reliable: supermassive black holes located within the galaxy’s core.
“In some ways, galaxy clusters are like spherical doughnuts,” explains the CCA announcement. “The new equation extracts the jelly in the center of the donut, which can lead to larger errors, and instead concentrates it on the outside of the dough for a more reliable mass estimate.”
In any case, the team plugged new AI-scripted equations into a digital suite containing thousands of simulated universes and found that they could generate mass estimates for galaxy clusters with 20-30% less variation. bottom. It’s that simple. That’s the beauty of this. in the announcementOf course, “simple” might be a bit of an exaggeration for those not in the business of weighing galaxies, but one thing is for sure.