Sunday, March 25, 2007

LSD 2007

So I went to London again, this time for the LSD 2007...

(1) Obviously, you can't expect stringology people to be interested in online scheduling algorithms. Not much happened in the talk.

(2) It was announced that I made an important discovery. No, I didn't prove any important theorems. It was only about "discovering" a new toilet...

(3) We went to Chinatown again. Most stuff there are more expensive than those in here (the only problem here is that brands and choices are rather limited), but it turns out that instant noodles are cheaper.

Sunday, March 18, 2007

Matchings, algorithms and problems

Some time ago I heard a presentation of a final year project. Somehow the student believed the project involves stable matchings. It is good that they attempt these things, and at first sight it appears to make sense. But when you think about it for a bit more you will realize that it actually does not make any sense to use stable matching in that problem. I won't go into the details here, but basically, the input arrives online rather than a batch, and only perhaps one side has preferences (and even that is arguable).

But this is not the point I want to make. The student did not understand the difference between a problem and the algorithms that solve the problem. His presentation showed a stable matching algorithm, but did not say anything about stable matching itself. So I asked, "So what is stable matching?" Then he showed that slide with the algorithm and is about to explain, when I interrupted and asked the matching itself, not the algorithm. He couldn't answer, and at the end said "a matching that everyone is happy". Other students there, even though they also don't know what a stable matching is, can at least appreciate that this answer cannot be correct.

And of course, he is not the only student with similar problems. Last year I had a project student whose project involve matching papers to referees, and the problem is to find a maximum matching. He just proceeded directly to invent some algorithm to find a "large" matching. I attempted to explain but he couldn't understand that there is a "maximum" matching independent of what algorithm you might what to use.

More recently, a student is doing a project on timetabling. I have nothing to do with his project, but anyway he came to ask me how to find the "optimal solution". I asked, "So what is the optimal solution you want to find?" and he wanted me to tell him what is the optimal solution. He has no idea of different "objective functions".

And I wonder, is all these because of me teaching too badly, or are they simply impossible to be understood by some people?

Tuesday, March 06, 2007

Almost an issue 3

Within less than 6 months I lost it again! Except this time I discover it within 10 minutes, so rushing back to the supermarket to get it back...

I will now seriously consider not to use it in supermarkets, at least not in Sainsbury's.