Sun is a member of SFI's Business Network, a set of roughly 50 businesses paying to be included in the SFI program as affiliate members. Among the benefits of the program is the opportunity to attend the summer school, which was held June 4 - 30 this year. After speaking with Jan Hauser and Emil Sarpa, Sun's primary contacts with SFI, I decided to apply. After the usual bureaucratic processes and delays, I was accepted, much to my horror. I faced this with no small amount of apprehension .. it had been years since grad school and it looked to be pretty tough.
I also spent some time getting more intensely involved with Java by writing a simple physical simulation program, Nodes, with springs and weights in a gravitational field. This proved to also be invaluable and even became part of a project during the course.
So, with June approaching, we (my wife got dragged along too) rented a house, filled up the station wagon with half our worldly goods, and headed to Santa Fe.
The school schedule began with a week-end Math Refresher course. The first week of classes was quite intense, giving a general overview of Chaos and Non Linear Dynamics in the morning and Statistical Mechanics, Adaptive Computing and Information Theory in the afternoon. This was a foundation week, aimed at general immersion in the field.
The next two weeks were half lecture, half project work. These lectures took on a different flavor: two morning sessions from actual practitioners in the field, taking on different and sometimes opposite, views. The first of these were on Economics. They were particularly interesting due to the radically different approaches of the two presenters. One was Doyne Farmer, of Prediction Company fame. His view was that traditional Equilibrium economics was quite flawed, not being able to expelling things like market crashes and the like. His fellow lecturer was a classical economist, John Geanakoplos, among who's achievements was a mortgage derivative hedge fund that helped to drive Los Angeles into ruin! Both had great stories of actually convincing financial institutions to use their schemes. And indeed, both have systems currently at play in the industry.
Selected evenings had talks by guest lecturers. Stuart Kauffman presented a talk on the minimal requirements for an agent in biology. Murray Gell-Mann discussed complexity from an information theoretical direction. Doug Erwin presented an ecology talk, which was open to the public to a standing room only crowd. Stephanie Forrest gave a talk on computer security and the immune system. There were also three informal gathering at the beautiful SFI site for afternoon wine, beer, and cheese.
The project portion of the school was for each attendee to do some sort of project around Chaos and Complexity. Mine was a three-part survey of the field, with this abstract, so as to get a broad perspective of its possibilities, and hopefully derive a project for us to do here in the labs. The list of projects can be seen during week four, at the end of the schedule page.
The last set of lectures was on scaling laws in biology. The two presenters were both physicists who had migrated into molecular biology. One set of talks was on the scaling laws of metabolism (the 3/4 power law) from micro-organisms to whales. The other was application of statistical mechanics and other probabilistic techniques to microbiology. Fascinating!
The final week was devoted to completing our projects, and delivering talks on them to the profs and students on the last two days of the school.
Secondly, I would like us to start taking a bit more advantage of the Business Network connection we have with SFI. This would include, for example, going to the general meetings and acquainting ourselves with the people at SFI. They are very approachable and would be delighted to work with us on a variety of topics. November is the next scheduled general meeting of the BN and I think a couple of us should go. It would also include working with some of the BN folks directly. I've teamed up with one of the BT attendees, for example, to get the BN more involved with SFI and to possibly include BN talks at next year's summer school. They would also like having some of the industry folks on site for as much as a month or two at a time to collaborate on investigations of shared interest.
Next, I'd like to talk to any of you who'd like to form a small "complexity interest group". The purpose would be to see if there are specific projects Sun Labs might want to engage in. Specifically, the third part of my survey was looking into applications of Complex Systems in the computer and networking area. There is not as much activity there as I expected, and I'd like to pursue possibilities there with anyone interested in doing so. The Small Worlds work, used to study the "6 degrees of separation" class of problem, could have exciting application in autonomous services such as Napster and Gnutella. Some of the Agent work looks applicable in our Java Auto project. Lots of possibilities.
So: anyone interested in pursuing this sort of thing, please get in touch.
