As reported by numerous scientific and general news outlets today, the image of Sagittarius A*, a supermassive black hole at the center of our galaxy, is a remarkable scientific achievement. However, one aspect that has not received much attention is the central role that simulation and synthetic data play in discovery.
If you haven’t read about this awesome science news yet, Event Horizon Telescope’s own post It’s a great place to get the point. Based on years of observations from around the world, a huge team from hundreds of institutions has been able to assemble images of the black hole in which our galaxy revolves. Despite the relatively close and light-years of dust, nebulae, nebula interference, and other whims of the Void.
But it wasn’t just a matter of making sure the telescope was pointing in the right direction at the right time. Black holes cannot be directly observed using things like Hubble or the still-warming web. Instead, all kinds of direct and indirect measurements of the object must be made. That is, how radiation and gravity bend around an object, and so on.
This means you have to collect and reconcile data from dozens of sources, which is a huge undertaking in itself and a big part of why our observations in 2017 are only now being published as final images. You can see it below. However, since this project was truly unprecedented (even the famous M87* image was superficially similar, but used a different process), we had to essentially test multiple possibilities of how the same observations were made.
For example, if the center is “dark”, is there something in the way (and about half the galaxy) or is it because the hole itself has a hole (and as it appears)? The lack of direct observational data makes it difficult to say. (The images shown here are not just images based on visible light, they are inferred shapes based on numerous radiation and other measurements.)
Imagine looking at an ordinary object from a distance. It looks like a circle when you look straight at it, but does that mean it’s a ball? Dish? Where is the end of the cylinder visible? Here on Earth, you can get more information by just moving your head or taking a few extra steps sideways. But try it on a cosmic scale! To achieve effective parallax in a black hole 27,000 light-years away, it would have to travel considerable distances, perhaps breaking the laws of physics in the process. Therefore, the researchers had to use different methods to determine which shapes and phenomena were best explained. could do Observe.
A series of synthetic data sets were generated to systematically explore and evaluate the design choices of imaging algorithms and their impact on the resulting image reconstructions. Synthetic data were carefully constructed to match the properties of Sgr A* EHT measurements. Synthetic data allow quantitative evaluation of image reconstructions against known ground truth. This allows you to evaluate your design choices and the performance of your imaging algorithms.
In other words, they generated an ocean of data related to various possible explanations for the observations and looked at how predictable these simulated black hole environments were.
Lisa Medeiros of the Institute for Advanced Study explained a bit about how and why she observed the rotation of black holes in her study in a very interesting Q&A that is worth seeing in its entirety if you have time. rotation of surrounding matter and throughout the galaxy.
“What is really exciting about these new results compared to what we did for the M87 in 2019 is that Paper 5 actually contains several simulations. [i.e. the spin relationships],” she said. “So there are simulations where the spin axis of a black hole does not coincide with the spin axis of matter swirling around the black hole, which is a really new and exciting simulation that was not included in the 2019 publication.”
Naturally, these simulations are incredibly complex tasks that supercomputers have to deal with, and there is the art and science of figuring out how many make sense and how close they should be. The alignment problem we’re looking at in this case has inherent scientific value, but it could also help us interpret interference from, for example, gas and dust swirling around a black hole. if the spin is the same thisThat gravity will affect the dust. thisThe reading means that it should read: this.
“Our simulations tend to favor the model that almost points us when we look at the simulation when compared to the data. It’s not pointing directly at us, it’s about 30 degrees away,” Medeiros said. “And that means the black hole’s axis of rotation is not entirely aligned with the galaxy’s axis of rotation, and if you believe what I said earlier, the disk would prefer to align with the black hole’s axis of rotation. Holes. Although the disk and black hole appear to be aligned, both are galaxies. is not aligned with .”
In addition to pursuing specific aspects like this, there was the more general question of which shape (or “primary source shape”) would generate a reading. It’s essentially a “ball-to-plate” question, but it’s much more complex. .
In one of today’s papers, the team describes building seven different potential forms for black holes, reflecting the diverse arrangement of matter, from rings to disks and even some sort of binary black hole. They simulated how these different shapes would produce different results on the device and compared them to “general relativistic magnetohydrodynamics” or GRMHD simulations, which are computationally (and linguistically) more demanding.
Here you can see a combination of two images on paper.
The idea was to find the simulations that produced the most similar results to what we actually saw during the simulations, and although there were no runaway winners, it produced the most consistent results in the Ring and GRMHD simulations (which I should say rather ring-like). How the data was interpreted for final interpretation of the data and the resulting image (note that I am broadly summarizing a very complex process here).
Considering that these observations were made about 5 years ago and a lot has happened since then, there is still much to investigate and more simulations to run. However, they had to hit “print” at some point, and the image at the top is the most informed interpretation of the data produced. There is no doubt that as observations and simulations pile up, better things can be expected.
In fact, as Richard Anantua of the University of Texas at San Antonio said in a Q&A session, you can try it yourself.
“If you’re in sixth grade and you have access to school computers, I think you have EHT imaging. We have all kinds of pipelines and tools to teach classes,” he said. Joke. “The data on some of these are public, so we can start working on them now, and by the time we get to college, we’ll have almost finished images.”