MSGID: 1:317/3 641d276a   
   PID: hpt/lnx 1.9.0-cur 2019-01-08   
   TID: hpt/lnx 1.9.0-cur 2019-01-08   
    Artificial intelligence discovers secret equation for 'weighing' galaxy   
   clusters    
      
     Date:   
         March 23, 2023   
     Source:   
         Simons Foundation   
     Summary:   
         Astrophysicists have leveraged artificial intelligence to   
         uncover a better way to estimate the mass of colossal clusters   
         of galaxies. The AI discovered that by just adding a simple term   
         to an existing equation, scientists can produce far better mass   
         estimates than they previously had. The improved estimates will   
         enable scientists to calculate the fundamental properties of the   
         universe more accurately, the astrophysicists have reported.   
      
      
         Facebook Twitter Pinterest LinkedIN Email   
   FULL STORY   
   ==========================================================================   
   Astrophysicists at the Institute for Advanced Study, the Flatiron   
   Institute and their colleagues have leveraged artificial intelligence   
   to uncover a better way to estimate the mass of colossal clusters of   
   galaxies. The AI discovered that by just adding a simple term to an   
   existing equation, scientists can produce far better mass estimates than   
   they previously had.   
      
      
   ==========================================================================   
   The improved estimates will enable scientists to calculate the fundamental   
   properties of the universe more accurately, the astrophysicists reported   
   March 17, 2023, in the Proceedings of the National Academy of Sciences.   
      
   "It's such a simple thing; that's the beauty of this," says study   
   co-author Francisco Villaescusa-Navarro, a research scientist at the   
   Flatiron Institute's Center for Computational Astrophysics (CCA) in   
   New York City. "Even though it's so simple, nobody before found this   
   term. People have been working on this for decades, and still they   
   were not able to find this."  The work was led by Digvijay Wadekar of   
   the Institute for Advanced Study in Princeton, New Jersey, along with   
   researchers from the CCA, Princeton University, Cornell University and   
   the Center for Astrophysics | Harvard & Smithsonian.   
      
   Understanding the universe requires knowing where and how much stuff   
   there is.   
      
   Galaxy clusters are the most massive objects in the universe: A single   
   cluster can contain anything from hundreds to thousands of galaxies,   
   along with plasma, hot gas and dark matter. The cluster's gravity holds   
   these components together.   
      
   Understanding such galaxy clusters is crucial to pinning down the origin   
   and continuing evolution of the universe.   
      
   Perhaps the most crucial quantity determining the properties of a galaxy   
   cluster is its total mass. But measuring this quantity is difficult --   
   galaxies cannot be 'weighed' by placing them on a scale. The problem   
   is further complicated because the dark matter that makes up much of a   
   cluster's mass is invisible. Instead, scientists deduce the mass of a   
   cluster from other observable quantities.   
      
   In the early 1970s, Rashid Sunyaev, current distinguished visiting   
   professor at the Institute for Advanced Study's School of Natural   
   Sciences, and his collaborator Yakov B. Zel'dovich developed a new way to   
   estimate galaxy cluster masses. Their method relies on the fact that as   
   gravity squashes matter together, the matter's electrons push back. That   
   electron pressure alters how the electrons interact with particles of   
   light called photons. As photons left over from the Big Bang's afterglow   
   hit the squeezed material, the interaction creates new photons. The   
   properties of those photons depend on how strongly gravity is compressing   
   the material, which in turn depends on the galaxy cluster's heft. By   
   measuring the photons, astrophysicists can estimate the cluster's mass.   
      
   However, this 'integrated electron pressure' is not a perfect proxy for   
   mass, because the changes in the photon properties vary depending on   
   the galaxy cluster. Wadekar and his colleagues thought an artificial   
   intelligence tool called 'symbolic regression' might find a better   
   approach. The tool essentially tries out different combinations of   
   mathematical operators -- such as addition and subtraction -- with   
   various variables, to see what equation best matches the data.   
      
   Wadekar and his collaborators 'fed' their AI program a state-of-the-art   
   universe simulation containing many galaxy clusters. Next, their program,   
   written by CCA research fellow Miles Cranmer, searched for and identified   
   additional variables that might make the mass estimates more accurate.   
      
   AI is useful for identifying new parameter combinations that human   
   analysts might overlook. For example, while it is easy for human analysts   
   to identify two significant parameters in a dataset, AI can better parse   
   through high volumes, often revealing unexpected influencing factors.   
      
   "Right now, a lot of the machine-learning community focuses on deep   
   neural networks," Wadekar explained. "These are very powerful, but the   
   drawback is that they are almost like a black box. We cannot understand   
   what goes on in them. In physics, if something is giving good results,   
   we want to know why it is doing so. Symbolic regression is beneficial   
   because it searches a given dataset and generates simple mathematical   
   expressions in the form of simple equations that you can understand. It   
   provides an easily interpretable model."  The researchers' symbolic   
   regression program handed them a new equation, which was able to better   
   predict the mass of the galaxy cluster by adding a single new term to the   
   existing equation. Wadekar and his collaborators then worked backward   
   from this AI-generated equation and found a physical explanation. They   
   realized that gas concentration correlates with the regions of galaxy   
   clusters where mass inferences are less reliable, such as the cores of   
   galaxies where supermassive black holes lurk. Their new equation improved   
   mass inferences by downplaying the importance of those complex cores in   
   the calculations. In a sense, the galaxy cluster is like a spherical   
   doughnut. The new equation extracts the jelly at the center of the   
   doughnut that can introduce larger errors, and instead concentrates on   
   the doughy outskirts for more reliable mass inferences.   
      
   The researchers tested the AI-discovered equation on thousands of   
   simulated universes from the CCA's CAMELS suite. They found that the   
   equation reduced the variability in galaxy cluster mass estimates by   
   around 20 to 30 percent for large clusters compared with the currently   
   used equation.   
      
   The new equation can provide observational astronomers engaged in upcoming   
   galaxy cluster surveys with better insights into the mass of the objects   
   they observe. "There are quite a few surveys targeting galaxy clusters   
   [that] are planned in the near future," Wadekar noted. "Examples include   
   the Simons Observatory, the Stage 4 CMB experiment and an X-ray survey   
   called eROSITA. The new equations can help us in maximizing the scientific   
   return from these surveys."  Wadekar also hopes that this publication   
   will be just the tip of the iceberg when it comes to using symbolic   
   regression in astrophysics. "We think that symbolic regression is highly   
   applicable to answering many astrophysical questions," he said. "In a lot   
   of cases in astronomy, people make a linear fit between two parameters   
   and ignore everything else. But nowadays, with these tools, you can go   
   further. Symbolic regression and other artificial intelligence tools   
   can help us go beyond existing two-parameter power laws in a variety of   
   different ways, ranging from investigating small astrophysical systems   
   like exoplanets, to galaxy clusters, the biggest things in the universe."   
       * RELATED_TOPICS   
             o Space_&_Time   
                   # Galaxies # Astrophysics # Astronomy # Stars # Cosmology   
                   # Black_Holes # Solar_Flare # Big_Bang   
       * RELATED_TERMS   
             o Dark_matter o Galaxy o Globular_cluster o   
             Large-scale_structure_of_the_cosmos o Dark_energy o Supergiant   
             o Open_cluster o Galaxy_formation_and_evolution   
      
   ==========================================================================   
   Story Source: Materials provided by Simons_Foundation. Note: Content   
   may be edited for style and length.   
      
      
   ==========================================================================   
   Journal Reference:   
      1. Digvijay Wadekar, Leander Thiele, Francisco Villaescusa-Navarro,   
      J. Colin   
         Hill, Miles Cranmer, David N. Spergel, Nicholas Battaglia,   
         Daniel Angle's-Alca'zar, Lars Hernquist, Shirley Ho. Augmenting   
         astrophysical scaling relations with machine learning: Application   
         to reducing the Sunyaev-Zeldovich flux-mass scatter. Proceedings   
         of the National Academy of Sciences, 2023; 120 (12) DOI:   
         10.1073/pnas.2202074120   
   ==========================================================================   
      
   Link to news story:   
   https://www.sciencedaily.com/releases/2023/03/230323103405.htm   
      
   --- up 1 year, 3 weeks, 3 days, 10 hours, 50 minutes   
    * Origin: -=> Castle Rock BBS <=- Now Husky HPT Powered! (1:317/3)   
   SEEN-BY: 15/0 106/201 114/705 123/120 153/7715 226/30 227/114 229/110   
   SEEN-BY: 229/111 112 113 307 317 400 426 428 470 664 700 292/854 298/25   
   SEEN-BY: 305/3 317/3 320/219 396/45   
   PATH: 317/3 229/426