What it took to capture a black hole

Last June, members of the Event Horizon Telescope (EHT) team convened in Cambridge, Massachusetts, to see if they could combine the data from eight telescopes into a single, clear image.

The researchers had their work cut out for them. Over the course of four days in April 2017, the EHT telescopes had stared at the supermassive black hole of Messier 87, an elliptical galaxy 55 million light-years away in the Virgo cluster. Three orders of magnitude as massive as the one at the center of our galaxy but also three orders as distant, M87’s central black hole (M87*) has an apparent diameter of 40 microarcseconds, roughly the size of the date on a quarter in Los Angeles as seen from Washington, DC. The telescopes—some single dishes, others multi-instrument arrays, and all susceptible to systematic noise—had viewed the tiny target from slightly different angles and had encountered varying degrees of atmospheric turbulence when collecting the 1.3-mm-wavelength photons. The technique of linking distant radio telescopes to form a virtual telescope the size of the distance between them, known as very long baseline interferometry, wasn’t new. But no one had ever tried to crunch the data from so many telescopes at such short wavelengths to view something just 40 μas across.

As we now know, the EHT team succeeded in “seeing the unseeable,” as project leader Sheperd Doeleman of MIT put it during a 10 April press conference in Washington. Just as he and his colleagues had hoped, the team was able to resolve superheated plasma streaking just outside the photon orbit radius, the distance from the center of a black hole at which any inward-moving photon no longer has a chance to escape (as opposed to the event horizon, where nothing, regardless of its motion, can escape). The details of how the image came to be, particularly the computational processing and the simulations that validated the derived image, appear in a series of papers published in Astrophysical Journal Letters.

The first step in imaging the black hole was to develop an enormous virtual radio telescope (see Physics Today, November 2008, page 14). Doeleman and his team achieved that feat by coordinating observations made at eight stations at six locations in Arizona, Hawaii, Mexico, Chile, Spain, and Antarctica. In multiple 3- to 7-minute scans on 5, 6, 10, and 11 April 2017, the telescopes observed M87*, along with other objects for calibration purposes. In the subsequent months, hard drives with the data, all of them stamped with times derived from the atomic clocks installed at each telescope station, were shipped to MIT and the Max Planck Institute for Radio Astronomy in Bonn, Germany. There the datasets were combined, calibrated, checked, and then recalibrated numerous times.

Once the data were calibrated, the researchers had to turn them into one static image. With no blueprint of a previous comparable project to guide its progress, the collaboration decided to divide the 50 or so scientists dedicated to imaging into four groups, each of which would crunch the M87* data in a different way. Two relied on a nearly half-century-old, tried-and-true computational imaging method called CLEAN; the other two used a more recently developed technique, regularized maximum likelihood (RML). Over the past few years, Katie Bouman (now at MIT, soon to join Caltech), Andrew Chael (Harvard), and colleagues have honed RML for the black-hole project by computationally cancelling out effects of the atmosphere and incrementally piecing together complete images. The work in each group was kept confidential. After seven weeks of analysis, each group submitted its end result to a web portal.

The imaging efforts came to a head at the weeklong June 2018 meeting in Cambridge, Massachusetts. Over the first few days, the four groups met separately but then began to share details about the intricacies of their analyses. Finally, on 25 June, the entire EHT imaging team viewed all four images. They weren’t identical, but they all shared a fundamental feature: a roughly 40 μas photon ring surrounding an orb of darkness, the long-sought silhouette of a black hole. “It was a remarkable moment,” says imaging team coleader Kazunori Akiyama of MIT.

The image of M87* presented this week is essentially a composite of the four group images, updated with higher-quality data. “Everyone can say, ‘It’s my image,'” says Ramesh Narayan of the Harvard-Smithsonian Center for Astrophysics in Cambridge. The final image contains plenty of catnip for black hole devotees: For example, the brighter southern region is the result of the Doppler effect; the plasma there is moving toward us. And deep within the dark central circle, with a radius no larger than 40% that of the visible photon ring, lies the event horizon, the true boundary of no return.

Not only did the images of the four groups agree with each other, but they also matched the output of models. Other members of the EHT team had developed general relativistic magnetohydrodynamic simulations to first predict what EHT might see and then validate what it did see. The researchers ran thousands of simulations, each one with slightly different values for properties such as plasma temperature and the black hole’s spin and magnetic flux.

As evident in the array of simulated images below, the modeled black holes look similar to the measured one. They also helped the EHT team pin down some of the details of its target. By comparing the image with simulations, the researchers were able to sharpen the relatively fuzzy boundary in the image marking the black hole’s photon orbit radius and identify a clear cutoff. The 42 ± 3 μas diameter of the ring, combined with the black hole’s distance from Earth, allowed the team to calculate a precise mass: 6.5 ± 0.7 billion solar masses. That resolves a discrepancy between previous calculations of M87* mass that were based on stellar motions and gas dynamics.

The image and the six papers that describe it are complete, but the analysis is not. EHT scientists are still analyzing polarization data from the observations, which will lend clues to the magnetic environment around M87* and its relativistic jet. Subsequent observing runs, the next scheduled for spring 2020 with 11 telescopes in the array, should allow researchers to determine the spin of M87*; that measurement can then be used to calculate the radius of the black hole’s event horizon. The scientists also will get a sense of how quickly things change in the neighborhood of a behemoth with a diameter of 38 billion kilometers; just over the course of the weeklong run in 2017, the researchers observed obvious changes in its profile.

Naturally, the EHT team also has its sights set on Sagittarius A*, the black hole at the center of our Milky Way galaxy (see the article by Dimitrios Psaltis and Feryal Özel, Physics Today, April 2018, page 70). The biggest obstacle is time variation: The relatively low mass of Sgr A*, 4 million Suns, means it has a smaller diameter, so the view over time changes far more quickly than it does around M87*. “We knew she’d be a problematic child,” says EHT scientist Feryal Özel of the University of Arizona. “But the data we’ve gotten should be enough.” Deciphering our nearest supermassive black hole will require further advances in computational analysis.

In the meantime, black hole enthusiasts aren’t the only ones who can join the excitement. The computational imaging advances that were required to offer humans’ first view of a black hole could also help improve the resolution of measurements of radio sources across the universe.