Finished with the first set of Illustris images!
We’re really excited to report that, with your help, the first batch of galaxy images from the Illustris simulation was finished last week! While we still have plenty of images still available to be classified (both from Illustris and DECaLS), I wanted to explain again how the images are being sorted in Galaxy Zoo and show some of the very early results we’re getting from your classifications.
The galaxies we selected from Illustris were a really big set – after eliminating galaxies which were likely to be too small or dim for accurate visual identification (we did this by filtering on the mass of the galaxies), we had over 110,000 images. In designing this phase of Galaxy Zoo, though, we wanted to try and prioritize the order of the images being shown so that we could do some early science projects along the way, rather than waiting many months for the entire data set to be finished before we started our analysis. This first set of Illustris data included 10,832 images, which were classified a total of more than 430,000 times by Galaxy Zoo volunteers.
One of the main questions we wanted to answer was: “How is the apparent morphology of a galaxy affected by the angle at which it’s viewed?” This is an important one – for observations of the real Universe, we can’t change the position of our telescope relative to the objects we’re looking at. If a galaxy is edge-on, for example, we’re really limited in being able to determine if there’s a bar, how many spiral arms there are, etc. In Illustris, though, we can change the viewing angle in the simulation! As a result, we might hypothesize that all edge-on disks should be identifiable as spiral or S0 galaxies at all the other viewpoints.
Here’s a quick test I’ve run of that. Using the set of collated classifications that just finished from Illustris, I looked for unique galaxies that were classified as “edge-on disks” from at least 1 of the 4 viewing angles that we have data for. Then I looked at the GZ classifications for the other viewing angles to see if they were still edge-on. Results:
Very close to what we expect! Only about 10% of galaxies had any edge-on classifications; of those, almost all of them are classified as face-on at every other angle (the big bump at N=1 in this plot). The exceptions are where the disk is aligned with two of the virtual cameras — then, we see it as edge-on twice and face-on twice. Since the cameras are oriented like they’re at points of a pyramid with the galaxy at the center, geometry tells us we should expect a typical disk to be edge-on for 0 cameras most of the time, 1 sometimes, 2 very rarely, and never 3 or 4. Just what we see!
We’re excited to be starting on the analysis phase and are, as always, extremely thankful for your help.