I bet you didn't know that this thread would have a part 2

1. A set of eight telescopes around the globe focused on the M87 galaxy's core on April 5, 6, 10, and 11 in 2017. This galaxy was chosen instead of closer ones like our own because it is very active/bright without being too far away and the central black hole was predicted to be of a very large size. In addition the central black hole was predicted to be relatively static. What this means is that unlike our own it doesn't yo-yo up and down in terms of brightness.
   
   
2. The light that was collected is synchotron radiation. Basically electrically charged particles throw off radiation as they are swung around the black hole. The radiation is way outside the human visual spectrum, so it's not a true colour image. These charged particles are all part of a highly magnetic stream of fluid coming from stars torn apart by the black hole.
   
   
3. Once the data was collected a bunch of common errors have to be filtered out. These are things like slight errors from the warping of the disks in the telescopic arrays, parts of the array being slightly unfocused, etc. The telescope in Mexico known as the LMT had a bit more of these than others as the galaxy was only accessible to it during the evening when the disks warp slightly at the millimeter scale from changes in temperature going from day to night and it is harder to focus in evening light.
   
   
4. All these errors have to be filtered out, but fortunately they are well understood so can be isolated easily. Each station has built up over years of study a statistical catalogue of how likely parts of their array are to be defocused or be slightly hot etc. Again the LMT in Mexico was the hardest to do this for because it's the youngest.
   
   
5. There's also well understood things like removing effects from the Earth's atmosphere and intervening stellar dust. In addition the slight differences between each telescope due to their different positions on the Earth
   
   
6. So after this we have the observational data with errors removed. All the data from all eight stations were collected, effectively giving you something like data from a telescope the size of the Earth. Now to reconstruct the image from that data.
   
   
7. There is an old piece of software called CLEAN used in Radio Astronomy that takes the amount of light obtained in each detector on the telescope, angles and distances between detectors and uses this information to build an image. However in this case half the teams used CLEAN and the others a newer algorithm called RML.
   
   
8. However these algorithms have a lot of settings. Basically they need to know what data is considered part of the black hole image (there will be data from other sources) what parts of the image to use as anchor points to detect warping in the picture and several others I don't understand. The details of imaging software is above my paygrade.
   
   
9. To determine what the settings should be they tested the algorithms against generated data. They basically simulated black holes and a few other astronomical objects, about 10,000 in total. With the "true" images from the simulations they fed the algorithms a sample of the simulation data to see if it could correctly rebuild the simulated image. And they did this for all 10,000 until they were satisfied.
   
   
10. Three teams did this coming up with very similar settings which were then given to the teams using CLEAN and RML. In the end every team obtained an image and the average was used to obtain the final image.
    
    
11. This gave an image for each of the four observation days.
    
    
12. The final step involved General Relativity itself. The images were fed into Einstein's field equations which can tell you how magnetic fluids like the shredded stars should move near a black hole. The equations basically predicted "If a black hole looked like this on April 5th it should look like this on April 6th". The image output by the equations was an extremely close match for the actual April 6th photo. This was done for every day and compared against others (e.g. using the equations on the April 10th data and predicting what April 5th must have been like). Also each image was checked against a simulation of a black hole of that size and spin using General Relativity.
    
    
13. This confirmed that these indeed seem to be images of a rotating black hole behaving exactly as General Relativity says it should.

New Home wrote: »

I heard someone say yesterday that the combined data collected for the photo of the black hole was too big to be transmitted by internet, and was physically brought to one place to be collated.

Too big for the internet!

Capt'n Midnight wrote: »

Never underestimate the bandwidth of a station wagon full of tapes hurtling down the highway.
- Andrew S. Tanenbaum