I cannot reproduce this problem. However, notice that 21/15 falls outside the normal movefilter, so unless you are careful it won't be evaluated on 2ply. Try choosing worldclass but change the movefilter to huge. Then save settings and see what you get.
Christian. On Jan 18, 2008 7:09 PM, bar tomas <[EMAIL PROTECTED]> wrote: > Hi, > > thanks very much for your reply. I've attached to this email an example of > a match (in tmg format) in which I observed the incident I mentioned. Before > starting this test I reinstalled gnubg, so there would be no customized > saved settings. These are the exact steps I took: > > TEST 1 > > - Start Gnubg (GUI version) > > - I set evaluation and analysis parameters to 'world class', by: > > Settings/Analysis/Chequer play 'world class' > Settings/Analysis/Cube decisions 'world class' > Settings/Evaluation/Chequer play 'world class' > Settings/Evaluation/Cube decisions 'world class' > > - I imported the tmg match (file test.tmg attached to this email). > > - In the navigator panel which contains all the moves of the game (at the > right of the GUI) I selected the 12th move of player 'p1' in game 1: > > move : 21/15 > dice : 51 > position ID : hbcdAFC2bUgKBA > match ID : cImmAAAAAAAA > > When selecting this move, I press the Hint button and, this move is ranked > 7th (with MWC 60.33%) > > TEST 2 > > > -Start Gnubg (GUI version) > > - I set evaluation and analysis parameters to 'world class', by: > > > Settings/Analysis/Chequer play 'world class' > Settings/Analysis/Cube decisions 'world class' > Settings/Evaluation/Chequer play 'world class' > Settings/Evaluation/Cube decisions 'world class' > > - I start a new match and enter the position ID hbcdAFC2bUgKBA and the > match ID cImmAAAAAAAA. > > - I press the Hint button but the move 21/15 is ranked 9th (with MWC > 61.33%) > > > TEST 3 > > > - Start Gnubg (CLI version) > > - I set evaluation and analysis parameters to 'world class', by the > following commands (maybe some unnecessary because default): > > set evaluation chequerplay type evaluation > set evaluation chequerplay evaluation plies 2 > set evaluation chequerplay evaluation cubeful on > set evaluation chequerplay evaluation prune on > set evaluation movefilter 1 0 0 8 0.160000 > set evaluation movefilter 2 0 0 8 0.160000 > set evaluation movefilter 2 1 -1 0 0.000000 > set evaluation movefilter 3 0 0 8 0.160000 > set evaluation movefilter 3 1 -1 0 0.000000 > set evaluation movefilter 3 2 0 2 0.040000 > set evaluation movefilter 4 0 0 8 0.160000 > set evaluation movefilter 4 1 -1 0 0.000000 > set evaluation movefilter 4 2 0 2 0.040000 > set evaluation movefilter 4 3 -1 0 0.000000 > set evaluation cubedecision type evaluation > set evaluation cubedecision evaluation plies 2 > set evaluation cubedecision eval prune on > set analysis chequerplay type evaluation > set analysis chequerplay eval plies 2 > set analysis chequerplay evaluation cubeful on > set analysis chequerplay evaluation prune on > set analysis movefilter 1 0 0 8 0.160000 > set analysis movefilter 2 0 0 8 0.160000 > set analysis movefilter 2 1 -1 0 0.000000 > set analysis movefilter 3 0 0 8 0.160000 > set analysis movefilter 3 1 -1 0 0.000000 > set analysis movefilter 3 2 0 2 0.040000 > set analysis movefilter 4 0 0 8 0.160000 > set analysis movefilter 4 1 -1 0 0.000000 > set analysis movefilter 4 2 0 2 0.040000 > set analysis movefilter 4 3 -1 0 0.000000 > set analysis cubedecision type evaluation > set analysis cubedecision evaluation plies 2 > set analysis cubedecision eval prune on > > -Then I entered positionId and matchID: > > set matchid cIlmAAAAAAAA > set board hbcdAFC2bUgKBA > > - Then when I ask for a hint: > > hint 15 > > In this case, the move 21/15 is ranked 12th with MWC 66.09% > > > Thankyou very much if you have an idea why I get different scores for the > same move with apparently the same evaluation settings. > > tomas > > > On 1/18/08, Christian Anthon <[EMAIL PROTECTED]> wrote: > > > > If you could create a reproducible example it would make things a lot > > easier for us to track down. > > > > Christian. > > > > On Jan 18, 2008 12:58 PM, bar tomas < [EMAIL PROTECTED]> wrote: > > > > > Hi, > > > When using Gnubg, sometimes I get slightly different results when > > > using the 'Hint' button. > > > What I mean is if I load a match I have already played, and select a > > > move (in the right hand panel) and ask for a Hint, sometimes the result is > > > not the same as if I create a new match and enter the match id and the > > > position id for that move (in the text boxes at the top of the screen), > > > and > > > then ask for a Hint. The differences aren't huge, for instance one of my > > > moves might be ranked 4th in a Hint when selecting a move of a loaded > > > match. > > > And then the same move is ranked 6th when entering the match id and > > > position > > > id of the move in a new match. > > > I have exactly the same settings for analysis and evaluation (2-ply). > > > I have checked every setting in the .gnubgautorc file (movefilter, etc) > > > and > > > they are identical for analysis and evaluation. > > > Is there anything else I could set in order to get the same result for > > > Hint used in both modes. > > > Thank you very much. > > > _______________________________________________ > > > Bug-gnubg mailing list > > > [email protected] > > > http://lists.gnu.org/mailman/listinfo/bug-gnubg > > > > > > > > >
_______________________________________________ Bug-gnubg mailing list [email protected] http://lists.gnu.org/mailman/listinfo/bug-gnubg
