The PRIMO method was used to create a new image of the M2017 supermassive black hole using data from the 87 EHT (Event Horizon Telescope). With the help of machine learning, the well-known representation of the supermassive black hole at the center of M87 – sometimes called the “blurry, orange donut” – has gone through its first official transformation. The updated image further reveals a larger, darker central region surrounded by bright, accumulating gas with a "skinny donut" appearance. The team first reached full resolution of the series using data collected by the Event Horizon Telescope (EHT) partnership in 2017.
The EHT partnership achieved an “Earth-sized telescope” in 2017 by linking a network of seven existing telescopes globally to collect data on M87. But since it's not possible to cover the entire surface of the Earth with telescopes, there are gaps in the data, just like the missing pieces of a puzzle.
According to lead author Lia Medeiros of the Institute for Advanced Studies, “with our new machine learning technique, PRIMO, we were able to achieve the maximum resolution of the current array.” “Since we cannot study black holes closely, the level of detail in an image is crucial to our understanding of how they behave.
The width of the ring in the image has shrunk by nearly two times, which will serve as a strong limit for our theoretical theories and tests of gravity.
Principal component interferometric modeling or PRIMO, EHT members Feryal Special (Georgia Tech)Created by Tod Lauer (NOIRLab), Lia Medeiros (Institute for Advanced Study), and Dimitrios Psaltis. Their paper entitled “Image of Black Hole M87 Reconstructed with PRIMO” has been published in Astrophysical Journal Letters.
“PRIMO is a new approach to the challenging task of creating images from EHT observations,” said Lauer. It offers a way of compensating for the information about the object being viewed, necessary to produce the image that can be seen using a single gigantic radio telescope the size of the Earth.
Dictionary learning is the foundation of PRIMO, a subset of machine learning that enables computers to generate rules based on large amounts of training data. For example, if a computer is trained with enough examples of banana photos, it can tell whether an unknown image is a banana.
Beyond this simple example, the adaptability of machine learning has been demonstrated in a variety of contexts, from completing Beethoven's unfinished work to producing Renaissance-style artwork. So how can technology help scientists create an image of a black hole? This question has already been addressed by the study team.
More than 30.000 high-fidelity simulated images of gas-accumulating black holes were analyzed by computers using PRIMO. The simulation group looked for repetitive patterns in the visual structure in a wide range of scenarios for how the black hole absorbs matter.
To produce a very accurate depiction of the EHT observations and a high-fidelity approximation of the incomplete structure of the images, the various structure models were sorted by how often they appeared in the simulations and then blended. A publication on the method was published in The Astrophysical Journal on February 3, 2023.
Medeiros continued: “We are using physics to fill in areas where there is missing data in a way that has never been done before.” This method could have important implications for interferometry used in everything from astronomy to medicine.
To produce a very accurate depiction of the EHT observations and a high-fidelity approximation of the incomplete structure of the images, the various structure models were sorted by how often they appeared in the simulations and then blended. A publication on the method was published in The Astrophysical Journal on February 3, 2023.
“We achieved another milestone by creating an image that uses the full resolution of the array for the first time,” Psaltis said. “Nearly four years after the first horizon-scale image of a black hole was revealed by EHT in 2019, we have achieved another milestone. Thanks to the new machine learning tools we've created, we now have a great opportunity to understand black hole physics.
The new image should enable more precise estimates of the mass of the M87 black hole and the physical factors affecting its current appearance. The data also gives researchers a chance to perform more reliable gravity tests and place tighter constraints on the event horizon on possible alternatives (based on darker core luminosity drop).
“We achieved another milestone by creating an image that uses the full resolution of the array for the first time,” Psaltis said. “Nearly four years after the first horizon-scale image of a black hole was revealed by EHT in 2019, we have achieved another milestone. Thanks to the new machine learning tools we've created, we now have a great opportunity to understand black hole physics.
The new image should enable more precise estimates of the mass of the M87 black hole and the physical factors affecting its current appearance. The data also gives researchers a chance to perform more reliable gravity tests and place tighter constraints on the event horizon on possible alternatives (based on darker core luminosity drop).
PRIMO can also be used to analyze other EHT data, such as data from the galaxy's main black hole, Sgr A*.
The Virgo galaxy cluster contains a large and relatively nearby galaxy, M87. More than a century ago, a mysterious jet of hot plasma was seen erupting from its center. Beginning in the 1950s, the then-new field of radio astronomy revealed the bright compact radio emitter at the center of the galaxy. In the 1960s, it was thought that the massive black hole at the center of M87 was driving this activity.
Based on observations of fast-moving stars and gas at the center of M87, measurements from ground-based telescopes starting in the 1970s and then the Hubble Space Telescope starting in the 1990s show that M87 is actually a black hole that weighs several billion times the mass of the Sun. provided strong evidence that 87 EHT observations of M2017 were made over several days using various radio telescopes combined to achieve the highest resolution. The first effort to create an image from this data was represented by the now famous "orange donut" image of the M2019 black hole, published in 87.
“The 2019 image was just the beginning,” Medeiros said. “If a picture is worth a thousand words, the data underlying that picture has a lot more story to tell. PRIMO will continue to be a critical tool for gaining such insights.”
Source: ias.edu/news/
📩 14/04/2023 10:36