My program StoneGrid calculates unconditional life and death at every move, in the UCT Tree and in the random playout. I think it helps on its strength a little bit, especially in the end game. In the begining of the game, seems to be completely useless. It is slow. But it makes the random playout slightly shorter. The random playout is about 80 moves per game. I do not have a comparison if the unconditional life and death is disabled, since it is not easy to do in the current data structure.
On Dec 13, 2007 3:20 PM, Jason House <[EMAIL PROTECTED]> wrote: > On Dec 13, 2007 2:17 PM, <[EMAIL PROTECTED]> wrote: > > > I'd like to start a more specific discussion about ways to combine > > tactical information with MC-UCT. Here's the scenario. > > > > It's the bot's turn and, prior to starting any playouts, it runs a > > tactical analyzer (for want of a better name) that labels each string as > > unconditionally alive, unconditionally dead, conditionally alive, uncertain, > > or too strong to bother testing. For each conditionally alive string, it > > finds the move(s) to kill/rescue it. Each possible move, has a critical > > score indicating the net number of stones whose life/death status would be > > changed for the better if the bot moved there. Moves with a non-zero > > critical score are "critical moves." > > > > As worded, I don't know if I agree with that formalization of what > information will be useful. It's in the right ballpark for the discussion, > but will likely shift a bit with any final design. > > > > > * Connectivity. I confess to having no clue, but wise people on this > > list doubtless will. > > > Connectivity of stones and L&D for stones are intimately tied together. > If I'm alive, I don't need a connection. If I connect to another living > group, I don't need to live. > > I've been tempted to insert forced responses into my random playouts to > preserve connections, but I didn't get that far since I don't have a > tactical analyzer. One day I will... > > > > > > * Collect information from all those local searches and somehow use it > > in the global search. This is my long term goal. Maybe n-grams... > > > What's an n-gram? > > > > > I think that there is a crossover point where, once the cpu time > > needed to do a decent tactical analysis at the root becomes a small fraction > > of the total available for a move, it becomes worthwhile to do so. My > > program/hardware isn't there yet, but it's just a matter of time. > > > Don't forget that local tactical analysis can be reused many moves later > if the local area has remained unaffected. In a multi-core system, it may > become increasingly valuable to dedicate a core to tactical analysis. > > _______________________________________________ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/ >
_______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
