There isn't much to say about today. I fixed a few problems with my code. One of them was to prioritize winning moves that take fewer steps over ones that take more. This required a surprisingly trivial modification to my negamax algorithm. That's great when you don't make any typos. It cost me at least half an hour of frantic searching this evening to figure out how I could have improperly done negamax. In the end, it was because of an equals sign in a function call.
I also completed a test to run all possible player moves and make sure my ai is unbeatable. I'm pretty confident that it works. It worries me that it only executes about 1400 moves. I guess that's right though because the ai should never lose and that cuts out a ton of possible outcomes. As I said, I'm confident but not too confident.
No comments:
Post a Comment