Summer Adventures in the Computational Universe
We’ve just finished the intense first week of our fifth NKS Summer School. Every year I get to spend three weeks playing professor. It’s not the same experience that most academics get, not least because our CEO, Stephen Wolfram is part of it, pushing to get science done, and to get the students to do great projects based on A New Kind of Science.
We get applicants for the summer school from all over the world, from all fields and all academic levels. Students are selected on their abilities, interests and enthusiasm. The center of the distribution is graduate students, but we always have some younger people, and some much more experienced people—both from academia and industry.
Here is a photo of this year’s class outside of the University of Vermont, our venue for this year’s summer school:
We’ve developed a pretty good system for the summer school. There’s a background of lectures, but the core of the summer school experience is for each student to do an original research project. Of course, it helps that NKS is a young and very energetic field, full of exciting problems to be solved.
But the most critical part is that Stephen is a great project generator. This last week he met with every student, defining a project for each of them to do. It was quite a process, jumping between linguistics, ecology, mathematics, computer systems, architecture and much more.
In each meeting, Stephen tries to understand what the students are interested in and what their bigger goals are. Then he tries to define a project that meets their interests and goals. I am always amazed at how he manages to find interesting questions in each area so quickly. More often than not, these questions lead to whole research programs which go well beyond the three weeks of the program and can last years.
These meetings are best described as brainstorming between the students and Stephen, which makes for a unique educational experience. The projects sometimes seem to arise out of the blue, and they quickly get developed into new, original research directions with spontaneously generated momentum.
It is the main idea of the summer school to get people started doing NKS research. For young students, this is often their first research experience. The more experienced students are often looking for a new direction in their research careers.
Most students are surprised at how down-to-earth Stephen is, and I like to think they are all encouraged by him to do top-tier science. Sometimes the project meetings go on for longer than planned, with lots of different possibilities being investigated. Since I try to keep the schedule on track, that can sometimes be nerve-wracking.
The summer school is a public service done by Stephen and Wolfram Research. He donates his time and the company provides instructors (this year there are eight instructors). The tuition is free, which helps us attract students from all over the world.
This past week, Stephen did three “live experiments” for the summer school. A live experiment is just that—an experiment done in real time, in front of an audience, with no preparation and no guarantees that things will work out as expected.
It is worth pointing out our modern tools are what make live experiments possible—in particular, Mathematica 6. A “pure NKS” experiment usually starts with a particular class of simple programs to explore—a corner of the computational universe. One quickly writes code to run these programs, perhaps using Mathematica functions like CellularAutomaton and TuringMachine. Next one needs visualization tools, which often involves only one line of code (say with ArrayPlot). There are many ways to go, but the dynamic visualizations available in Mathematica 6 let us write one-line functions to dynamically perform the live experiment (in particular with Manipulate).
It is worth pointing out that once a Manipulate is written, we are only a few more steps away from online publication in the Wolfram Demonstrations Project. By the time Stephen is done, we have basically gone from concept to publication.
In a sense, it is meta-education—a way to see someone like Stephen figure out things in real time, mistakes and all.
The summer school lasts for three weeks. In the first week, we had lectures covering the foundations and practical issues.
Now, in the second week, students turn in their homework and start work on their projects. This year the homework is to find their favorite three-color outer totalistic cellular automata. During the second week the lectures become slightly more specialized while students work on their projects. In the third week, their projects are completed, made into posters (also using Mathematica 6 technology) and presented at the end.
This year, using Demonstrations, there is a new way to supplement the normal publishing of results. So expect to see a sequence of Demonstrations from the summer school arriving soon… as well as some exciting final results.