On Tue, 2008-10-28 at 12:22 -0400, Michael Williams wrote: > I ran a quick test of the check in question combined with the decaying > playouts. The one WITH the check won 70% at 2000 playouts over a 200 game > match.
I just queried my results at 4096 playouts and get very different numbers. It has only played 257 games so far, but the stats are 56.4% - far from 70%. It could be that: 1. I have a buggy implementation. 2. You have a buggy implementation of the non-enhanced version. 3. Diminishing returns (you played 2000, I played 4096 4. Statistical noise. (you high, me low, etc.) It could be some combination of the above or something else. I'm going to try a specific test of 2000 playouts. - Don irb(main):004:0> irb(main):005:0* 145 / 257.0 => 0.56420233463035 irb(main):006:0> > > Don Dailey wrote: > > On Tue, 2008-10-28 at 14:08 -0200, Mark Boon wrote: > >> Don, > >> > >> Don't get me wrong, I understand very well where you're coming from. > >> It's very handy to have a concise defintion of a minimalist playing > >> program that can be used to check correctness. The main reason I > >> thought Michael's idea was worth considering was that it was both > >> simpler AND played better AND scaled better. But if in order to get > >> a > >> meaningful boost in strength you have to maintain some sort of check > >> whether a move was played by one side before, then I agree it may be > >> better to leave his improvement out of the reference specification. > > > > The important idea here of course is the decaying weights, not whether > > to do the "already played" check. It's clearly simpler not to do the > > check. And with decaying weights ignoring the check would have less > > impact anyway. So don't get me wrong either, I definitely like the > > idea and find it a useful enhancement. > > > > I think we will accumulate a number of useful enhancement to the simple > > reference bot and I want to track them on this web site (the web site > > that doesn't yet exist :-) > > > > - Don > > > > > > > > > >> Mark > >> > >> ------------------------------------------------------------------------ > >> > >> _______________________________________________ > >> 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/
signature.asc
Description: This is a digitally signed message part
_______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
