SciPy 2009 was a great success, and I want to congratulate everyone involved. Unfortunately, for the first time ever, I was unable to attend. I love meeting all the neat people that work on scientific tools for Python, and always find it gratifying to see how the community has each year, so it was a big disappointment to miss last week’s conference.
The event showcased some of the amazing open source tools being created: SciPy itself , Matplotlib , IPython (especially its use for parallel computing), SymPy, CorePy, Cython, FiPy, Sage, and so many more. It appears that Python is poised to become the lingua franca for scientific computing as it is increasingly utilized in technical, data-centric, and visualization applications.
Instead of soaking in the tutorials, talks, and sprints at SciPy in Pasadena, I spent last week in the slightly cooler city of Chicago (you can pick your own connotation of that statement). I was in the South Side at the University of Chicago, teaching a class on using Python for Scientific Computing to a very interesting group of scientists. These people are using text mining to discern complex relationships between self-discovered memes. That’s my explanation
and not theirs, and so I’m sure it doesn’t catch the true scope of their work.
I was invited to teach the class by Professor Andrey Rhetsky to help his group use Python better in their work. He is involved in an ambitious project to build an automatic map of all of the published connections between biomolecules (proteins, DNA, RNA, etc.). He works to build the map by mining abstracts from all the major scientific publications. There are many possible uses of such a map: drug discovery, treatment discovery, and automatic hypothesis generation. All of these have potentially significant implications for medicine as well as for general scientific practice.
Image from Andrey Rzhetsky.
Andrey’s work is an interesting application of the type of relationship-mining practiced by another University of Chicago professor, James A. Evans. James was also in the class with several of his colleagues, and believes this kind of mining can yield fruit in understanding social networks in general. In particular, Professor Evans sees potential in using text-mining to better understand the role of social networks in the production and dissemination of knowledge, especially scientific knowledge.
Their work is fascinating and quite different from my own experience. There was, however, significant overlap in the foundational areas of Bayesian analysis, string processing, and Python. If I couldn’t be at SciPy, then at least I had a chance to discuss at least some projection of meta-science with some very bright people.
My disappointment with not being able to attend SciPy was ameliorated a bit on Wednesday night, when I discovered that SciPy 2009 is available on video. I stayed up way too late several nights watching tutorials and talks. If you also missed SciPy, check out the recordings at archive.org and get addicted.
It was very fun to be able to watch Peter Norvig’s keynote talk on Thursday night, and then show the video to Andrey and James and Peter McMahan, who is working with James. Portions of the code displayed in the presentation seemed to have quite a bit of overlap to the code that they had just shown me the previous day.
Such an increase in information flow could surely lead to more and better progress… Maybe… Talking with Professor Evans also highlighted the potential for such communication speed, coupled with the limited amount of time modern humans are able to dedicate to fully comprehending new ideas, to have perilous effects. False ideas can become rigidly cemented in the minds of people reliant on trusted sources. It certainly makes me wonder which of my ideas are really wheat and which are just chaff waiting for the right wind to blow them away.
I’m really hoping to come to SciPy next year. Hopefully, I might have a few real things to actually talk about as well. In the meantime, our next Scientific Computing with Python webinar is scheduled for this Friday. If you haven’t already registered, we will be reviewing SciPy (Really just a good excuse for me to watch more of the videos!). If you didn’t make it to SciPy, this webinar is a chance to catch up. On the other hand, if you attended SciPy, the meeting will be a great opportunity to share your experience and provide feedback.
Scientific Computing with Python Webinar: SciPy 2009
Friday, August 28
1pm CDT/6pm UTC
