I’m back from an exciting week at Stonybrook University (NY). I was there for the Polar High Performance Computing Workshop and Hackathon. The week kicked off with two days of training. A lot of the courses were from Software Carpentry : Unix Shell, version control with Git, 4 hours of Intro to Python… The hands-on coursework developed by this not-for-profit organization is great, and accessible online at no charge. Even after years of coding in Python and using Git, I still learned a couple of tricks - and finally switched to Python 3 !

The second half of the week was split between hacking time put aside for us to work on our projects and more advanced classes on high-performance computing with Open Science Grid and the Google Cloud Engine. Theses classes came with access codes and free resources for us to burn through with our projects.

It was super motivating to work alongside the other scientists there; I learned a lot and came back energized. As for the science part, well… I spent all week trying to make my java application truly portable. On the last night, when I finally got it running on a Google virtual machine jam-packed full of RAM, I realized that it was still as slow as on my good old 16 Go computer in the lab. Verdict : the SNAP code which my app is based on just isn’t sufficiently parallelized. I used the last two hours of the hackathon to hack a workflow using a parallelized k-means implementation in Python. Fun times ! It reminded me of my physics undergrad and I loved it.

Academic perk : I made a quick stop at the MoMa before flying back home. #WomenArtistsPostwarAbstraction