How Astrophysicists Will Avoid Drowning in Data
by Adam Hadhazy
More cosmic data will be collected in the near future than centuries'-worth of skygazing have gathered so far – an exciting development that requires innovative approaches
The Author
Astronomy and its close cousins astrophysics and cosmology are of course the "biggest" sciences, dealing with the universe in its vast totality. Appropriately, all are set to increasingly benefit from "big data." This now-familiar, though-still-buzzwordy field concerns the gathering of voluminous information and the extraction of unique, otherwise unobtainable insights from such unprecedented data hauls. To date, astronomers and astrophysicists have already done quite a lot in the big data space. To take but two examples, there have been multiple mammoth galaxy surveys, like the Dark Energy Survey that observed 300 million galaxies, as well as surveys of a billion Milky Way stars at a time with the Gaia mission. The numbers and kinds of objects and phenomena under study—some poorly or even never-before characterized—are set to expand dramatically, however. The power of astronomical instruments to hoover up observations, and just as importantly on the backend to digitally store and transmit that data, grows by the day, seemingly, just like information technology in general. Analyzing the copious data is an intertwined challenge that is concomitantly being met. So what's it all mean?
The hope is that heretofore intractably mysterious entities, namely dark matter and dark energy, will start to reveal themselves when submitted to the unsparing scrutiny of big data-driven analytics. Imaging billions of galaxies strewn through cosmic history will show how the gravitational behavior of dark matter helps form galaxies and galaxy clusters, constraining this dominant form of matter's enigmatic nature. At the same time, how the dark matter-enhanced masses of galaxies bend light from background objects will increasingly map out the growth of the universe and its expansion rate. That data will ferret out dark energy's influence and how the accelerative force has, as many researchers suspect, changed in intensity over the eons. On top of those anticipated discoveries, diligently tracking transient events, such as the explosions and outburst of stars, as never before will let us firm up models of stellar evolution and its connections to broader galactic behaviors. All in all, loads of astrophysical data will pour in in the coming years, quickly dwarfing everything that has cumulatively come before. This past month, Kavli Institute-affiliated researchers made some notable strides toward managing the coming data deluge. Thanks to their and their colleagues' efforts, we have big things to look forward to!
Empowering artificial intelligence to accelerate astrophysical discovery
Researchers at the Massachusetts Institute of Technology (MIT) are teaming up with scientists at other universities to bring artificial intelligence (AI) to bear in accelerating discoveries in colossal datasets. Erik Katsavounidis, a senior research scientist at MIT's Kavli Institute for Astrophysics and Space Research (MKI), is part of the effort centered on multi-messenger astrophysics. This nascent field combines observations from electromagnetic observatories—that is, telescopes that see light in its various forms, from radio waves to gamma rays—with other kinds of observatories that gather neutrinos, cosmic rays, and gravitational waves. These combined observations generate mountains of varied data that human scientists could never hope to pick through and correlate in any reasonable span of time. AI can quickly crunch the data, though, pulling out statistical relationships and new verities. Training AI to do the work as well—or better—than human scientists ever could is a challenge, but one worth pursuing to advance astrophysics.
Separating the astrophysical wheat from the chaff at the Vera C. Rubin Observatory
You want data? The Vera C. Rubin Observatory’s Legacy Survey of Space and Time (LSST) has got you covered. When the survey starts in a few years from a telescope in Chile, it will gather a whopping 20 terabytes of astronomical images nightly for a decade. Researchers deeply involved in the project at the Kavli Institute for Particle Astrophysics and Cosmology (KIPAC) at Stanford University have long thought about how best to manage this data tsunami. In a recent Q&A, KIPAC Senior Associate Member Richard Dubois explained many of the ins and outs. For example, a dedicated data facility will automatically parse the images obtained—at a rate of one every 37 seconds—to see if something changed, indicating an astrophysical event has started occurring (for instance, a supernova). The facility will then send out alerts to the astronomical community, who can decide whether to repurpose other observatories on the fly to immediately investigate. The data facility will also archive the voluminous data and allow researchers to readily access them. Dubois and colleagues are looking forward to the challenge of managing the data and for all the discoveries expected to be swept up in their gathering.
Exciting projects endorsed in the newest decadal survey
Members of the Kavli Institute for Cosmological Physics (KICP) at the University of Chicago have had projects they are involved in endorsed by the broader astronomical community in the newest so-called decadal survey. These priority- and funding-setting reports, issued by the National Academy of Sciences, have major sway in determining the direction of national astronomy and astrophysics efforts. Three areas highlighted for investment through the 2020s are finding and characterizing Earth-like exoplanets, learning more about extremely energetic astrophysical events (including black hole mergers and supernovae) for insights more broadly into cosmological mysteries, and strengthening our knowledge of galactic origins and evolution. KICP researchers have many connections to these fields of study. Examples include a next-generation ground-based telescope, the Giant Magellan Telescope, a next-generation ground-based cosmic microwave background (CMB) experiment dubbed CMB-S4, as well as the proposed Large UV/Optical/IR Surveyor (LUVOIR) and Habitable Exoplanet Observatory (HabEx) space telescopes. Overall, the report signals that mesmerizing new science is nigh.
Master astrophysicist chef cooking up galaxies from scratch
Mark Vogelsberger, a newly tenured professor and a member of MKI, was profiled in a recent piece from MIT News. The story spans his childhood interest in astronomy sparked by a book he received at age 10 and the pristine skies of his native west Germany, to his cutting-edge work today on ultra-detailed galaxy simulations. These massive simulations, which trace the evolution of myriad galaxies since their formation early in the universe's history, would take thousands of years for a regular computer to run. But by splitting the work across tens of thousands of computers, Vogelsberger and colleagues can run the sims in six months. The research has delivered key insights into how galaxies assume their classical shapes while still displaying the rich range of variations commonly observed.
Savoring the latest helping of gravitational waves
The international collaboration of the world's three gravitational wave-hunting instruments—LIGO in the United States, VIRGO in Italy, and KAGRA in Japan—have jointly reported on the discovery of as many as 35 new wave-spawning events, bringing the total known to around 90. LIGO made the first such detection of gravitational waves—ripples in spacetime—just in 2015. The breakthrough finding ended a century of speculation and earned its visionary creators, including MKI's Rainer Weiss, the 2016 Kavli Prize in Astrophysics. Many Kavli Institute-affiliated researchers are now involved in the revolutionary field of gravitational-wave astronomy. In the latest batch of detections, some monster black holes boasting more than 60 times the sun's mass were caught in the act of merging. How black holes this hulking could form is a matter of some debate, with there not being a plausible route via the death of a star, the standard way known to form black holes of more pedestrian single- to low-double-digit solar masses. The thinking is that these extra hefty black holes are actually the product of former mergers. If so, that would suggest that black holes can glom together repeatedly, going on to form progressively more massive objects. It's just one of the many advances in the science of extreme cosmic objects that gravitational wave observations are making possible.