Philippe,

Many thanks again for being so helpful and quick to respond.

I think there is a bug in your pubEvalVal() function in gnubg-nn/gnubg/eval.c as explained in the attached PDF. If there are men on the bar, these are not subtracted from the number of men off and the total number of checkers is >15.

Best,
Robert.

On 1/26/2019 2:36 PM, Philippe Michel wrote:
On Sat, Jan 26, 2019 at 10:42:10AM +0100, Øystein Schønning-Johansen wrote:
Thanks for your effort Philippe. Your numbers looks correct.

However, I think it is important to state some more details.

First: Are the games played to completion? Or are the games terminated at
race or bearoff or ...
Second: Does the pubeval evaluate all the position classes? I once did the
mistake in a similar experiment where the pubeval player actually used a
full bearoff look up table.
And then: These are cubeless moneygames I assume. These are not one-point
matches.

(Another potential bug is the opening roll. I guess that it is taken care
of.)
I didn't use gnubg + some changes, but the gnubg-nn tools from Joseph
Heled, so that should be cubeless games without bearoff databases and,
at least, directly comparable to gnubg's 2012 numbers.


Attachment: gnu_pubeval_bug.pdf
Description: Adobe PDF document

_______________________________________________
Bug-gnubg mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/bug-gnubg

Reply via email to