|
|
Post by Andrew Rose on Jan 1, 2014 7:19:55 GMT -8
For Christmas, I received a copy of Mancala (specifically the 4-seed Kalah variant, as far as I can tell by Googling). It's a game I haven't met before. I also notice that it's mentioned in the MCTS Survey paper than I've mentioned in a couple of places.
Does anybody have a kif for this (or any other) variant of Mancala?
Thanks,
Andrew
|
|
|
|
Mancala
Jan 6, 2014 23:02:13 GMT -8
Post by alandau on Jan 6, 2014 23:02:13 GMT -8
|
|
|
|
Mancala
Jan 7, 2014 10:59:44 GMT -8
Post by Andrew Rose on Jan 7, 2014 10:59:44 GMT -8
Thanks.
That's a 3 seed 2x5 version and it produces absolutely enormous propnets (way bigger than anything we played in the finals). Fortunately most of that is because it allows addition up to 201, when only 30 is needed. However, even removing the step count (which isn't needed, because the game is guaranteed to finish all by itself) and reducing the numbering scheme to 30, still leaves a huge propnet (although ~1/3 of the original size). Off the top of my head, I think I was only getting 130 iterations / second on a quad-core PC.
I'll revisit it once I've got some more important things sorted out.
|
|
