I figured this was how it worked. What ways did you use to get an accurate model of the rules? Did you just use wikis and a stopwatch to figure out how long builds take, what can be built when, etc. or did you actually memory watch or decompile the game itself to figure out the algorithms and loops used?
I made a Tic-Tac-Toe GA in undergrad as my senior project. It was horribly stupid, but I didn't reduce the problem space down, and I could have easily implemented rotation and flip mapping to reduce the search space by a significant amount (just over 7% of the nieve search space), but the algorithm still worked pretty well and served as a great learning experience.
The genetic part is ridiculously simple, IMO. The hard part is figuring out how to define the rules of the game, break them out, and figure out what maps to the genome and how it maps. Evolution from that point is simply modifying the genome in different ways.
No decompiling, just observing the game with a timer. I used a lot of community wiki/forum information as well. There has been some really amazing research done by the community [1][2], to figure out optimal resource harvesting strategies, and I used quite a bit of data from that.
Thanks for the info! It really helps knowing how others have solved problems like that. I am interested to know if there is an open source finite state machine for building in SC2.
I made a Tic-Tac-Toe GA in undergrad as my senior project. It was horribly stupid, but I didn't reduce the problem space down, and I could have easily implemented rotation and flip mapping to reduce the search space by a significant amount (just over 7% of the nieve search space), but the algorithm still worked pretty well and served as a great learning experience.
The genetic part is ridiculously simple, IMO. The hard part is figuring out how to define the rules of the game, break them out, and figure out what maps to the genome and how it maps. Evolution from that point is simply modifying the genome in different ways.