Re: [Bug-gnubg] Possible evaluation bug

[Øystein Schønning-Johansen]

Yes! Things looks correct.

I browse the code history from your release to the one I'm using. The only change I can see that may matter, is the use of pruning neural networks in move selections.

Can you try also with pruning turned turned off. Does that change anything?

-Øystein

On Sat, Feb 17, 2018 at 12:38 PM, Terje Pedersen <address@hidden> wrote:

Hi!

Initially it triggered on an ubuntu server: 4.4.0-67-generic
#88-Ubuntu that runs the same 1.05.000 version of gnu bg then I tried
it on my Windows 10 machine and got the same result.

(No game) show matchequitytable
Match equity table: Kazaross XG2 25 point MET
(/usr/local/share/gnubg/met/Kazaross-XG2.xml)

looks good.

I'll give 1.06 version a try and see if it will handle this position
differently.

Thanks for your reply!

Best regards,
TP

On Sat, Feb 17, 2018 at 11:44 AM, Øystein Schønning-Johansen
<address@hidden> wrote:
> Really strange. However it is a heisenbug (of course). I cannot recreate the
> bug on my system. (GNU Backgammon 1.06.000  Dec 13 2017, Arch Linux)
>
> Here is how the same thing looks at my system:
>
>  GNU Backgammon  Position ID: PgAAALYLBDMMAA
>                  Match ID   : cImmADAAAAAE
>  +13-14-15-16-17-18------19-20-21-22-23-24-+     O: gnubg
>  | X              X |   |    X        O  X | OO  3 points
>  |                X |   |    X        O  X | OO
>  |                  |   |             O    | OO
>  |                  |   |             O    | OO
>  |                  |   |             O    | OO
> v|                  |BAR|                  |     5 point match (Cube: 1)
>  |                  |   |                  |
>  |                  |   |                  |
>  |                  |   | X                |
>  |                  |   | X  X  X          |     Rolled 51
>  |                X |   | X  X  X          |     0 points
>  +12-11-10--9--8--7-------6--5--4--3--2--1-+     X: oystein
>
>
> Cube analysis
> 2-ply cubeless equity -2.2349 (Money: -2.0391)
>   0.0860 0.0000 0.0000 - 0.9140 0.7715 0.4396
> Cubeful equities:
> 1. No double           -2.0550
> 2. Double, pass        +1.0000  (+3.0550)
> 3. Double, take        -2.4258  (-0.3708)
> Proper cube action: No double, take (10.8%)
>
> Rolled 51 (-0.066):
> *    1. Cubeful 2-ply    24/19 18/17                  Eq.: -2.0615
>        0.0892 0.0000 0.0000 - 0.9108 0.7435 0.6447
>         2-ply cubeful prune [world class]
>      2. Cubeful 2-ply    13/7                         Eq.: -2.1614 (-0.0999)
>        0.0707 0.0000 0.0000 - 0.9293 0.7800 0.6404
>         2-ply cubeful prune [world class]
>      3. Cubeful 2-ply    13/8 6/5                     Eq.: -2.1766 (-0.1151)
>        0.0701 0.0000 0.0000 - 0.9299 0.7866 0.6634
>         2-ply cubeful prune [world class]
>      4. Cubeful 2-ply    18/13 6/5                    Eq.: -2.1963 (-0.1348)
>        0.0603 0.0000 0.0000 - 0.9397 0.7831 0.6918
>         2-ply cubeful prune [world class]
>      5. Cubeful 2-ply    18/17 18/13                  Eq.: -2.2086 (-0.1470)
>        0.0567 0.0000 0.0000 - 0.9433 0.7815 0.7003
>         2-ply cubeful prune [world class]
>      6. Cubeful 2-ply    18/12                        Eq.: -2.2291 (-0.1676)
>        0.0544 0.0000 0.0000 - 0.9456 0.7887 0.6961
>         2-ply cubeful prune [world class]
>      7. Cubeful 2-ply    18/17 13/8                   Eq.: -2.2481 (-0.1866)
>        0.0524 0.0000 0.0000 - 0.9476 0.7953 0.6954
>         2-ply cubeful prune [world class]
>      8. Cubeful 2-ply    18/17 7/2                    Eq.: -2.2952 (-0.2337)
>        0.0476 0.0000 0.0000 - 0.9524 0.8135 0.7098
>         2-ply cubeful prune [world class]
>      9. Cubeful 0-ply    24/19 6/5                    Eq.: -2.1277 (-0.0661)
>        0.0653 0.0000 0.0000 - 0.9347 0.7599 0.4187
>         0-ply cubeful prune [expert]
>     10. Cubeful 0-ply    18/13 7/6                    Eq.: -2.1289 (-0.0674)
>        0.0648 0.0000 0.0000 - 0.9352 0.7593 0.5707
>         0-ply cubeful prune [expert]
>
> I'm on this system:
> address@hidden ~]$ gnubg -t < terjebug.txt
> GNU Backgammon 1.06.000  Dec 13 2017
>
> Can you supply which OS, version/build of GNU Backgammon (well that looks
> like you got "GNU Backgammon 1.05.000  Nov 27 2016"). Can you also state
> which match equity table you are using?
>
> I have a theory that it get's the wrong answer because it calculates the
> gammonvalue (or actually the backgammon value in this case) incorrectly.
>
> can you append
>
> show matchequitytable
> show marketwindow
> show gammonvalues
>
> to your command file, and submit the output?
>
> -Øystein
>
>
> On Fri, Feb 16, 2018 at 5:43 PM, Terje Pedersen <address@hidden> wrote:
>>
>> Hi!
>>
>> I just ran into what looks like a gnu evaluation bug where gnu seems
>> to suggest that I should try to avoid backgammon when a gammon will
>> lose the match anyway. command file:
>>
>> set player 0 human
>> new match 0
>> set output rawboard off
>> set xgid XGID=----BBCA-----A----B-B--eB-:0:0:1:51:0:3:0:5:10
>> move 24/19 18/17
>> next
>> previous
>> analyze move
>> show board
>>
>> $ ./gnubg.exe -t -c commands.txt
>> GNU Backgammon 1.05.000  Nov 27 2016
>> Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004 by Gary Wong.
>> Copyright (C) 2015 by Gary Wong and the AUTHORS; for details type
>> `show version'.
>> This program comes with ABSOLUTELY NO WARRANTY; for details type `show
>> warranty'.
>> This is free software, and you are welcome to redistribute it under
>> certain conditions; type `show copying' for details.
>> Moves for gnubg must now be entered manually.
>> A new session has been started.
>> TTY boards will be given in ASCII.
>> The dice have been set to 5 and 1.
>>  GNU Backgammon  Position ID: PgAAALYLBDMMAA
>>                  Match ID   : cImmADAAAAAE
>>  +13-14-15-16-17-18------19-20-21-22-23-24-+     O: gnubg
>>  | X              X |   |    X        O  X | OO  3 points
>>  |                X |   |    X        O  X | OO
>>  |                  |   |             O    | OO
>>  |                  |   |             O    | OO
>>  |                  |   |             O    | OO
>> v|                  |BAR|                  |     5 point match (Cube: 1)
>>  |                  |   |                  |
>>  |                  |   |                  |
>>  |                  |   | X                |
>>  |                  |   | X  X  X          |     Rolled 51
>>  |                X |   | X  X  X          |     0 points
>>  +12-11-10--9--8--7-------6--5--4--3--2--1-+     X: terje
>>
>>
>> Cube analysis
>> 2-ply cubeless equity  -2.238 (Money:  -2.039)
>>   0.086 0.000 0.000 - 0.914 0.771 0.440
>> Cubeful equities:
>> 1. No double            -2.058
>> 2. Double, pass         +1.000  ( +3.058)
>> 3. Double, take         -2.430  ( -0.372)
>> Proper cube action: No double, take (10.9%)
>>
>> Rolled 51 (-0.066):
>>      1. Cubeful 2-ply    24/18                        Eq.:  -1.850
>>        0.144 0.000 0.000 - 0.856 0.716 0.430
>>         2-ply cubeful prune [world class]
>>      2. Cubeful 2-ply    24/19 6/5                    Eq.:  -1.994 (
>> -0.143)
>>        0.106 0.000 0.000 - 0.894 0.735 0.627
>>         2-ply cubeful prune [world class]
>>      3. Cubeful 2-ply    24/19 7/6                    Eq.:  -2.015 (
>> -0.164)
>>        0.102 0.000 0.000 - 0.898 0.739 0.626
>>         2-ply cubeful prune [world class]
>> *    4. Cubeful 2-ply    24/19 18/17                  Eq.:  -2.065 (
>> -0.214)
>>        0.089 0.000 0.000 - 0.911 0.743 0.647
>>         2-ply cubeful prune [world class]
>>      5. Cubeful 2-ply    13/7                         Eq.:  -2.165 (
>> -0.314)
>>        0.071 0.000 0.000 - 0.929 0.780 0.641
>>         2-ply cubeful prune [world class]
>>      6. Cubeful 2-ply    13/8 6/5                     Eq.:  -2.180 (
>> -0.329)
>>        0.070 0.000 0.000 - 0.930 0.787 0.663
>>         2-ply cubeful prune [world class]
>>      7. Cubeful 2-ply    18/13 6/5                    Eq.:  -2.200 (
>> -0.349)
>>        0.060 0.000 0.000 - 0.940 0.783 0.692
>>         2-ply cubeful prune [world class]
>>      8. Cubeful 2-ply    7/2 6/5                      Eq.:  -2.206 (
>> -0.355)
>>        0.068 0.000 0.000 - 0.932 0.799 0.669
>>         2-ply cubeful prune [world class]
>>      9. Cubeful 2-ply    18/17 18/13                  Eq.:  -2.212 (
>> -0.361)
>>        0.057 0.000 0.000 - 0.943 0.781 0.700
>>         2-ply cubeful prune [world class]
>>     10. Cubeful 2-ply    18/13 7/6                    Eq.:  -2.229 (
>> -0.378)
>>        0.056 0.000 0.000 - 0.944 0.789 0.702
>>         2-ply cubeful prune [world class]
>>
>> Any idea what is going on here? I am using gnu bg for evaluating moves
>> on Backgammon Studio when playing online matches and was hit in the
>> face with this double blunder which I was confident couldn't possibly
>> be correct. XG doesn't think it is much of an error:
>>
>>     2. XG Roller++ 24/19 18/17                  eq:-2.2250 (-0.0057)
>>       Player:   7.58% (G:0.00% B:0.00%)
>>       Opponent: 92.42% (G:76.45% B:64.53%)
>>
>>
>> Best regards,
>> TP
>>
>> _______________________________________________
>> Bug-gnubg mailing list
>> address@hidden
>> https://lists.gnu.org/mailman/listinfo/bug-gnubg
>
>