Schloss Dagstuhl


Dagstuhl is a small place in the western part of Germany, near Trier and Saarbrucken. Schloss Dagstuhl is the name of a castle that serves as a research center for computer science. Every week, a seminar on a specific topic in computer science is held. The castle can have about 45 guests, and if you want to go to Dagstuhl you need to receive an invitation in order to get there.

I attended Dagstuhl February 15-20, 2004, for the third "Theory of Evolutionary Algorithms" seminar (


In the castle, everything is available for a fantastic stay. There's a billiards room, a music room, a wine cellar, and most of the things you expect to find in a castle. Next to the 17th century castle, a new building has been constructed that has a lot of guest rooms, a library, a computer room and some lecture halls. It also has a fitness and sauna. Everything is available 24 hours a day, so you can also work 24 hours a day with everyone who is attending the conference. None of the rooms in the two buildings have keys, not even the guest rooms. There is a cabinet in each room that can be locked to store expensive stuuf, and you can lock your door from the inside for some privacy, but otherwise, you are free to go wherever you find it necessary to go. As an example, you can access the library and its great collection of books at any moment during your stay. There's a numerical key lock on the outside doors, and everyone attending is sharing that same code. As such, all attendees are treated as a family in order to bring about a nice and free environment to get some work done.

The rooms, guest rooms, cafetarias, ... are all very clean and filled with all available modern comfort. The guest rooms are small, but you don't see that room too often, thus big enough for sleeping. The only problem I noticed were that the blankets provided in the rooms were too small for me, but you can easily collect an extra blanket from the storage.


I think that the Dagstuhl idea and research center is something unique, at least in the field of theoretical computer science. I've heard about a similar idea for mathematicians, but that story was vague. Dagstuhl seems like one of the best kept secrets in computer science.

As the castle is somewhat in the middle of nowhere, it is really relaxing to work at the seminar center. All of the facilities make sure that you can work under excellent conditions. At night, the center is very quiet and you can get a good night's sleep.


A trip to Dagstuhl is cheap (you have to pay a 150euro share in the costs, including food and your room), everyone is there. It really is a must to come over. Personally, I like the idea that I can get there by car, but I would like to convince everyone else to go through the trouble of travelling here, as the environment and atmosphere for working with all of your colleagues is really good in Dagstuhl.


Not only is Dagstuhl a pleasant place to stay and walk around, the food offered for the attendees is also very good. Most of the dishes are original european dishes, with attention to detail. I liked the fact that I knew most dishes, and I thought of them as a german "implementation" of european cuisine.

You don't pay meals served in the main restaurant. There's always coffee, cake and fresh fruit available for free. You can buy candy, beers and wines for cheap prices, but you have to keep track of what you spend yourselves. At the end of a Dagstuhl seminar, you have to pay for what you've been using.

Before every meal, seat and table assignments are randomly distributed over the available tables. You always sit at a new table and have the opportunity to meet and talk to other people.

More information

You can find Dagstuhl's web site at


Saturday February 21, 2004

I'm home again after a great week. I was allowed to attend an amazingly unique experience, and I really anjoyed it.

Every night, I made a little walk away from the castle, to look at the schloss from a couple of 100s of meters away, to enjoy the looks of it. It's a small space with a lot of atmosphere, facilities to work, and opportunities to share ideas and find friends among colleagues.

Friday February 20, 2004

During the morning session, I really enjoyed Ken Dejong's talk on his interpretation of Price's theorem on the changes of population characteristics during strict periods of selection and variation. By seperating these components, Ken was able to give some insights in the derivatives of mean fitness during these processes, and actually show us the powers of selection and variation over a number of generations. I am really looking forward for a paper with these ideas written down. My personal interest is in the balance between selection and Genetic Drift, and I think that Ken presented a method whose predictions can be explored to understand this balance.

I left Dagstuhl right after lunch. We all said goodbye and I hope I can see most of these people again at a next conference.

Thursday February 19, 2004

The morning session had a couple of different subjects. I skipped out of the afternoon session in order to get some work done, as I'll be in the hospital next week and have some deadlines that needed to be finished. I also found some time to browse through some books, made some copies of articles I won't be able to find at the library at our university, and finished some similar things that had to be ready before I can start my hospital week in a couple of days.

There's going to be a cheese and wine evening tonight, which is supposed to be good. I haven't tasted much of the available wines here, so I might try a couple of them tonight. It's going to be our last night at Dagstuhl, and I think I'm going to relax and enjoy this last evening.

Wednesday February 18, 2004

I missed breakfast, but was in time to get to the morning session, which had a couple of differing subjects.

In the afternoon, we went on a hike with the whole group. There's a couple of great forests and tracks around the schloss.

In the evening, we managed to get a lot of people together in the lounge, in the attic of the schloss. It's pretty comfortable up there, and allows discussions of severalk topics. We had some lengthy discussions about philosophy, religion and scientific methods. It wasn't really realted to the work we do in our daily lives, but I think it's still a good part of why we're working towards PhDs. (the "Ph" stands for filosophy, and the "D" stands for doctor, so you at least need some notion of filosophy in order to earn the PhD title)

Tuesday February 17, 2004

The morning session was on co-evolution, and my talk was scheduled as the last talk. I think it all went pretty well, and I received some nice comments afterwards. Hopefully some people will walk up to me in order to discuss the subject. I also received a couple of references I should check out and on which I can concentrate my future work on finite models for evolutionary game theory.

Lunch was very good. Today's menu told us we were going to get a soup and some salad. This was obviously an understatement. I very much liked the soup, and the salad was more like a full option cold food buffet, with excellent food!

We took a little walk up to the ruins of the old castle, sitting on top of a mountain just next to the Schloss.

In the afternoon session, landscape structures and some techniques in order to do landscape analyses have been presented. I'm really impressed by the barrier trees. I wasn't familiar with the ideas, but there have been three talks discussing the trees and what you can do with them in optimization. I'm planning on looking this stuff up, and I think I have a couple of ideas on how we can approach the construction of barrier trees with scale space theory.

After dinner, we had a lengthy meeting with the coevolution people in order to discuss the several aspects that have been pointed out in the talks during the morning session. We've been talking for about 6 hours. It is really good to have the library nearby to look stuff up, even if it is the middle of the night. I returned the books I lent at 1:30 in the morning.

Note: You don't have to register the books you are borrowing, you can just take them out of the library without any questions asked. The library is only manned during the day, but it's open 24 hours a day, excellent! The library is pretty good, although we didn't find any good books on evolutionary game theory. I'll fill out a form tomorrow requesting the library to buy the Hofbauer/Sigmund book in the near future. The library is fully stashed. You can find almost all CS journals and conference proceedings. The book collection is very complete. Not only academic books can be found, but also the complete O'Reilly collection is available. Great stuff.

Monday February 16, 2004

I slept well for the first night in my room in the new Dagstuhl building. The first talks are about to start in about 30 minutes, everybody seems ready to go. I'll be exploring some parts of the castle I haven't seen yet (like the computer science library) during the day, and maybe get some work done. I brought some papers that need to be reviewed during this week, and I'll try to find some time for that too today.

Most of today's schedule is filled with talks on Evolution Strategies and optimization in continuous search spaces.

In the evening, I found my way to a very special room, in the attic of the castle. There's some nice chairs, a radio, and, of course a whiteboard. Combine that with a couple of german beers and a couple of colleagues, and discussion starts flowing. We've been working up there until 00.30 at night. I think this is what Dagstuhl is all about: sit down, relax, and work. Nobody's going to bug you and you're free to do whatever work you find interesting.

Sunday February 15, 2004

I came to Dagstuhl by car as I live about 350 kilometers away from the seminar center. (Note to self: it was 732km to go there and return) The place looks really great. After my arrival, I explored the castle and came across some old friends during the dinner. Among other topics, we talked Emergent Behavior. I managed to get to bed early and am hopeful for a constructive workshop.

comments powered by Disqus