Reaction for: Interactive Illustrations by Andrew Schulak

I have personally tried to create an interactive illustration before so i will focus on how the experiene was for me. As part of my work for Tom Dean last summer ('97) I was supposed to come up with some labs for the students to do. My friend Brandon, who was visiting, happened upon an old paper about Simulated Annealing (a form of combinatorial search) that was being thrown away. Being involved in search for my work I decided to procure it from him.

being interested in it I decided to try and code it up. I decided to use GP because it visualization aspects were easy and familiar to use. I could just focus in on the algorithm. After coding it up I then decided to add an Genetic Algorithm to the applet. now students could play with a tweak two different combinatorial optimizers and see how they work differently.

At least that was the theory. The problem the students are using these algorithms on is the all too familar TSP problem. All my algorithm does is illustrate how the tours are changing -- it isn't displaying how the internal state of the algorithm is being changed. And I think, in retrospect, that that is what would be most helpful.

In general I would like to agree with what was said in the paper. That is that planning and being able to prototype is very important. I didn't do any of this and it resulted in an inferior albeit useful tool.


Reactions


[BACK]