4 Sep 2016

Survey Visualiser


A small python project to visualise cosmological surveys. What started as a simple idea for a single plot for a poster turned into a score of plots, video animations, a student project, and an open-source GitHub repository. I regret nothing.

I was making a poster for the CAASTRO annual retreat on my BAO work, and realised what it really needed was a nice comparison of the WiggleZ survey compared to other more local surveys, to highlight the redshift depth of WiggleZ in comparison to other surveys of the same period, such as 2dF.

And because I am a stubborn fool, I decided to extend what I did to the general case. And then I thought, if I’ve done this, other people might be able to use it, so best put it on GitHub. Then I loaded in tons of data sets. And then I asked around to see if anyone else wanted figures, and soon was making figures for OzDES, 6dF, 2dFLenS and TAIPAN.

Here’s the figure I ended up creating with all datasets:

Each point on the plot is a galaxy. Blue is WiggleZ, yellow is 6dFGRS, red is 2dFGRS, purple is SDSS, green is GAMA and the thin gold bands are OzDES.

By far the most time was spent on getting the blending to happen nicely. Unfortunately, matplotlib has nothing equivalent to additive blending like you can find in Adobe AfterEffects, and so rendering multiple colours on top of each other does very little. I talk about the method used to fake additive blending in a separate blog post, complete with a small example.

Then, because I felt like that didn’t quite have the impact I wanted, I decided to make everything spin. I know that matplotlib now has an animation class that I could use to get video output, however the quality and more importantly, filesize, of the output render is completely outclassed by ffmpeg, such that the OzDES supernova video shown at the bottom of the page is only 2MB in size!

Twenty hours of work later, and here we are. Best to watch it on repeat, if that’s even possible. The top video is a comparison of all galaxy surveys I have, and the bottom video is an extension to my original work created by myself and an undergraduate student, Hugh McDougall. You can clearly see the three separate observing seasons in DES, and the bursts of light are the roughly thousand supernovae DES has detected in those three years.