First Black Hole Image
TED:
At the heart of the Milky Way, there’s a supermassive black hole that feeds off a spinning disk of hot gas, sucking up anything that ventures too close -- even light. We can’t see it, but its event horizon casts a shadow, and an image of that shadow could help answer some important questions about the universe. Scientists used to think that making such an image would require a telescope the size of Earth -- until Katie Bouman and a team of astronomers came up with a clever alternative.
The solution adopted by the Event Horizon Telescope project is to coordinate measurements performed by radio telescopes at widely divergent locations. Currently, six observatories have signed up to join the project, with more likely to follow.
But even twice that many telescopes would leave large gaps in the data as they approximate a 10,000-kilometer-wide antenna. Filling in those gaps is the purpose of algorithms like Bouman’s.
[…]
Finally, Bouman used a machine-learning algorithm to identify visual patterns that tend to recur in 64-pixel patches of real-world images, and she used those features to further refine her algorithm’s image reconstructions. In separate experiments, she extracted patches from astronomical images and from snapshots of terrestrial scenes, but the choice of training data had little effect on the final reconstructions.
This particular algorithm was not used in the image reported this week; it sounds like it was a prototype that proved the approach.
On Wednesday, after 10 years of planning and scientific investments totaling over $50 million, researchers released the first-ever image of a black hole. The image is a feat of modern science — experts say it’s the equivalent of taking a photo of an orange on the moon with a smartphone — and international collaboration. Over 200 scientists across the globe contributed to the project.
[…]
“The team collected about five petabytes of data, and one petabyte is a thousand terabytes,” explains Bouman. “Your typical computer has maybe one terabyte or so. So that would be like 5,000 typical laptops of data.
[…]
“We spent years developing methods, many different types of methods — I don’t think any one method should be highlighted — because most of all, we were afraid of shared human bias,” says Bouman.
[…]
For this reason, the computer scientists broke into four teams and did not communicate while they were analyzing the data. After months of the teams working independently, they all converged in Cambridge, Massachusetts, and ran their algorithms in the same room, at the same time.
Using imaging algorithms like Bouman’s, researchers created three scripted code pipelines to piece together the picture.
They took the “sparse and noisy data” that the telescopes spit out and tried to make an image. For the past few years, Bouman directed the verification of images and selection of imaging parameters.
[…]
The result? A groundbreaking image of a lopsided, ring-like structure that Albert Einstein predicted more than a century ago in his theory of general relativity. In fact, the researchers had generated several photos and they all looked the same. The image of the black hole presented on Wednesday was not from any one method, but all the images from different algorithms that were blurred together.
Update (2019-04-16): Akash lists the Python code that was used.
The Astrophysical Journal paper is here (via Matplotlib).
The Physicist has some background information (via Hacker News).
1 Comment RSS · Twitter
The TED talk was from 2016, when they thought they were most likely to be able to see the 4 million solar mass black hole at the center of our milky way galaxy. In fact it turned out that the 6.5 billion solar mass black hole at the center of the M87 galaxy, 50 million light years away, was easier to see. That's what the picture is of.
