On Monday night I returned from my first ever trip to Germany where I had the pleasure of attending the first EuroSciPy conference which was held in Leipzig. Mike Mueller did an excellent job of organizing and hosting the conference (the venue was great and the catering was superb). About 45 people attended the conference resulting in very interesting talks and several productive discussions. I had a chance to give a talk on the history of NumPy/SciPy. (I hope the fact that I did not sleep the night before did not detract too much from the talk.) The slides from the talk are here, although it may be hard to glean the entire talk from just the slides. Travis Vaught took video of the entire conference and left it in the hands of Ondrej Certik. This video should find it’s way to the Net someday.
The talks were very interesting. I learned more about SimPy which is a mature discrete event simulation engine written in Python (which makes heavy use of generators). The author of the package, Dr. Klaus Mller, has had much experience in the area and was very interesting to listen to. I learned that you can use the sound card plus Python to measure the coefficient of restitution (which measures how elastic a collision is), and also to tune an instrument. I also learned that Python is being used to help search for neutrinos in the Ice of Antarctica.
There are several symbolic packages (Computer Algebra Systems) emerging for Python. Pearu Peterson is working on a very fast one: sympycore, while Ondrej Certik is trying to implement a lot more functionality more quickly in SymPy (not to be confused with the discrete simulation package mentioned earlier). Hopefully, the two projects will merge sometime soon.
Unfortunately, I missed the lightning talk session because I was absolutely exhausted after not sleeping for more than 36 hours (twice) on the trip. At that point I’d been driving on the German autobahn for more hours than I’d slept (going 210km/h on the autobahn is very exciting by the way — especially when you are tired). I heard there were some very good talks, however.
The next day, I missed the first two talks because I went to church in the morning (there was a very charming congregation in Leipzig there and a kind American missionary translated most of it for me).
Later, Mike Mueller presented on his abstraction called PyModelData for handling data for scientific simulations (it uses YAML). The talk by Klaus Zimmerman later in the day suggested using yet another markup language called FMF as part of his presentation on making sure meta-data accompanies all data for rapid information processing (he also showed us Pyphant at that time). Pyphant is a data-analysis framework from the University of Freiburg (that should be using more of the Enthought Tool Suite IMHO and hopefully will be in future versions 🙂 ). The University of Freiburg also gave a talk earlier in the day on machine learning.
I really enjoyed the talks by David Albanese on his machine learning toolbox (called mlpy) and by Robert Cimrman on his finite-element toolbox called sfepy because I love to see these kinds of libraries becoming available. It shows that the Scientific and Engineering community surrounding Python is really growing. Travis Vaught ended the day with a description/demonstration of the Enthought Tool Suite and how it can be used to build rich scientific applications. These are the kind of applications we build everyday, but rather than keep our special sauce a secret we are giving it away to grow the community and to get help in building these foundational tools in the best way possible.
The Enthought demo went on until about 7:30pm after which about 15 of the attendees including me went to dinner in Leipzig at a Greek (I think) restaurant. The food was great, but the company was even better. I didn’t want to leave, but I had to drive to Frankfurt to get at least a couple of hours of sleep before flying out the next morning.
There are a few photos from the event that you should check out here. Overall, EuroSciPy was a great event. If you are in Europe and can’t come to the SciPy conference in CalTech, make sure and attend EuroSciPy next year. You can even help to plan and organize it in a city of your choice by making contact with Mike Mueller and/or Travis Vaught early enough.