[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [gnugo-devel] DFA results

From: Dave Denholm
Subject: Re: [gnugo-devel] DFA results
Date: 14 Jan 2002 15:02:32 +0000

Tanguy Urvoy <address@hidden> writes:

> Hi,
> Another difference between dfa enabled and not enabled should
> be the "-m" option in mkpat.
> With this option mkpat try to place the anchor in the center of the
> pattern
> to reduce the dfa size. This may change the match order on the full
> board.
> I will try the same test without dfa but with "-m".

FWIW I enabled use of -m without dfa in mkpat when testing the hypothesis
that anchoring the patterns in the middle would make the
"grid optimisation" in the standard matcher work better. But because of the
random changes to the reading noted already, I didn't get identical
results before and after, and didn't have time to verify that
the unexpected passes and failures were due to this.

My recollection is that grid test was able to reject slightly
more patterns, but it wasn't all that significant.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]