On Thu, 12 Jan 2017, Øystein Schønning-Johansen wrote:

Really strange!
I have built a one sided database using GMP with fractions with my own
piece of code. Since I use fractions from GMP there should be no rounding
error. Here are my numbers for position index 47:

Index: 47
0: 0
1: 0
2: 1/18
3: 445/648
4: 5939/23328
5: 73/23328

These numbers agrees to the numbers GNU Backgammon. Can it really play this
incorrectly? How does that happen?

I now think GNU gets it right. By default race4 reads a race4.hash file where it stored the results of prior uses. Mine was apparently somehow corrupted. Starting with the lingering hash file present, I get :

% race4 1 0 0 0 1 1
initial hash stats: entries 15227 overwrites 0
Eval <12 | 1 0 0 0 1 1 | >...

Expected number of rolls = 2.117155; sigma = 0.575819
hash stats: queries: 1 hits 1 entries 15227 overwrites 0


Starting from scratch :

% race4 -i /dev/null -o /dev/null 1 0 0 0 1 1
initial hash stats: entries 0 overwrites 0
Eval <12 | 1 0 0 0 1 1 | >...
.....................
Expected number of rolls = 2.205290; sigma = 0.529637
hash stats: queries: 1565 hits 1526 entries 39 overwrites 0
_______________________________________________
Bug-gnubg mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/bug-gnubg

Reply via email to