I think we should talk about strategy and see if some teams naturally form. Language is secondary. The thing that’s holding me back atm is finding a good way to come up with moves in local fights. (I believe I do have a good way to figure out where the local fights are, though.) Local fights between just a few ants ought to be able to be alpha-beta’d, but mass battle is a different story completely.
Pathfinding to food and exploring are not too hard to do decently. I think good local fighting is what will distinguish the top bots.
As for me, I know C++ and Go (google’s newish language) quite well. Last year my bot was in the top 10% but not quite the top 5%. I’m 2d at Go (the game), if that counts for anything.
Considering that I’m new at the game, I’m a bit fuzzy about which strategies work and which ones are dead ends.
My first instinct would be to try simple strategies (cluster with my ants, follow walls, avoid my ants, just charge straight forwards, etc.) and then use a mixed strategy whose parameter I can vary, possibly function of simple environment metrics (so I can systematically explore one bit of the space of possibilities).
But I haven’t read through all the site yet, and didn’t even get my local version to run correctly either (I need to use the right version of Python or something, the bots all crash). So my understanding of the kind of strategies involved is still very detached from reality.
Watch some of the games of the top bots. One thing I haven’t seen a top bot do yet is identify situations where they have an impasse (e.g., three ants holding off an entire horde) and try another route.
I’ve been doing that, it’s interesting indeed, thanks : I probably won’t have the time to make a good enough bot to compete with the top of the top, but I’ll do what I can.
I think we should talk about strategy and see if some teams naturally form. Language is secondary. The thing that’s holding me back atm is finding a good way to come up with moves in local fights. (I believe I do have a good way to figure out where the local fights are, though.) Local fights between just a few ants ought to be able to be alpha-beta’d, but mass battle is a different story completely.
Pathfinding to food and exploring are not too hard to do decently. I think good local fighting is what will distinguish the top bots.
As for me, I know C++ and Go (google’s newish language) quite well. Last year my bot was in the top 10% but not quite the top 5%. I’m 2d at Go (the game), if that counts for anything.
Considering that I’m new at the game, I’m a bit fuzzy about which strategies work and which ones are dead ends.
My first instinct would be to try simple strategies (cluster with my ants, follow walls, avoid my ants, just charge straight forwards, etc.) and then use a mixed strategy whose parameter I can vary, possibly function of simple environment metrics (so I can systematically explore one bit of the space of possibilities).
But I haven’t read through all the site yet, and didn’t even get my local version to run correctly either (I need to use the right version of Python or something, the bots all crash). So my understanding of the kind of strategies involved is still very detached from reality.
Watch some of the games of the top bots. One thing I haven’t seen a top bot do yet is identify situations where they have an impasse (e.g., three ants holding off an entire horde) and try another route.
I’ve been doing that, it’s interesting indeed, thanks : I probably won’t have the time to make a good enough bot to compete with the top of the top, but I’ll do what I can.