[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Bug-gnubg] Wrong move in external interface
From: |
Jim Segrave |
Subject: |
Re: [Bug-gnubg] Wrong move in external interface |
Date: |
Wed, 16 Mar 2005 10:29:13 +0100 |
User-agent: |
Mutt/1.4.2.1i |
On Wed 16 Mar 2005 (07:54 +0100), amarganth wrote:
> Hi all
>
> I compared both "boards": the one of Joachim, the other from Jim:
>
> board:you:her:3:2:1:0:1:2:0:1:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:-4:0:1:1
> :2:0:0:1:1:1:0:1:-1:0:25:11:11:0:0:2:2:0:0
> board:you:her:3:2:1:0:1:2:0:1:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:0:-4:0:1:1
> :2:0:0:1:0:1:0:1:-1:0:25:11:11:0:0:2:2:0:0
>
> The difference is field number 38 (beginning by 0), which is "1" in the
> first, "0" in the second board structure.
> Field 38 is "may double" of player 0. The "was doubled" flag is number 40.
Yes - you are right - when parsing the sscanf's in drawboard.c, I got
out of step.
My variant command causes the cube ownership to be set rather than
centred.
> I cannot understand, what gives back another move, if that "may double" flag
> is other. It's not a doubling situation. And it's crawford (1-away, 2-away).
And that's the clue - it's not Crawford it's post Crawford.
Try simply setting up the match up as follows in gnubg:
set dice 2 1
set board DwAAoAkAAAAAAA
set score 1 2
set player 1 gnubg
set player 0 human
play
analyse match
Which makes it non-Crawford, but otherwise the same as the original
posting. Gnubg moves 4/1 for me. And the analysis of gnubg's move gives
the following strange numbers
1. Cubeful 2-ply 4/1 Eq.: -1.00000
0.11574 0.00000 0.00000 - 0.88426 0.00000 0.00000
2-ply cubeful prune [world class]
2. Cubeful 2-ply 4/2 1/off Eq.: -1.00000 (+0.00000)
0.11574 0.00000 0.00000 - 0.88426 0.00000 0.00000
2-ply cubeful prune [world class]
3. Cubeful 2-ply 4/3 2/off Eq.: -1.00000 (+0.00000)
0.11574 0.00000 0.00000 - 0.88426 0.00000 0.00000
2-ply cubeful prune [world class]
4. Cubeful 2-ply 2/1 2/off Eq.: -1.00000 (+0.00000)
0.11574 0.00000 0.00000 - 0.88426 0.00000 0.00000
2-ply cubeful prune [world class]
5. Cubeful 0-ply 2/off 1/off Eq.: -4.94842 (-3.94842)
0.53239 0.00000 0.00000 - 0.46761 0.00000 0.00000
0-ply cubeful prune [expert]
Then do:
set dice 2 1
set board DwAAoAkAAAAAAA
set score 1 2
set player 1 gnubg
set player 0 human
set crawford on
play
analyse match
Now gnubg does something sensible and plays 2/off 1/off
1. Cubeful 0-ply 2/off 1/off Eq.: +0.06477
0.53239 0.00000 0.00000 - 0.46761 0.00000 0.00000
0-ply cubeful prune [expert]
2. Cubeful 0-ply 4/1 Eq.: -0.76849 (-0.83326)
0.11575 0.00000 0.00000 - 0.88425 0.00000 0.00000
0-ply cubeful prune [expert]
3. Cubeful 0-ply 4/2 1/off Eq.: -0.76849 (-0.83326)
0.11575 0.00000 0.00000 - 0.88425 0.00000 0.00000
0-ply cubeful prune [expert]
4. Cubeful 0-ply 4/3 2/off Eq.: -0.76849 (-0.83326)
0.11575 0.00000 0.00000 - 0.88425 0.00000 0.00000
0-ply cubeful prune [expert]
5. Cubeful 0-ply 2/1 2/off Eq.: -0.76849 (-0.83326)
0.11575 0.00000 0.00000 - 0.88425 0.00000 0.00000
0-ply cubeful prune [expert]
The problem isn't in external.c or the FIBs processing - it's doing a
very strange move selection when the cube is live.
(BTW, when setting player 1 to gnubg above, it deosen't help having it
set to Supremo chequer/World Class cube)
Another anomoly - with the position set up as shown - try a zero ply
evaluation:
1. Cubeful 0-ply 2/off 1/off Eq.: +5.07797
0.53239 0.00000 0.00000 - 0.46761 0.00000 0.00000
0-ply cubeful prune [expert]
2. Cubeful 0-ply 4/1 Eq.: -0.76849 (-5.84646)
0.11575 0.00000 0.00000 - 0.88425 0.00000 0.00000
0-ply cubeful prune [expert]
3. Cubeful 0-ply 4/2 1/off Eq.: -0.76849 (-5.84646)
0.11575 0.00000 0.00000 - 0.88425 0.00000 0.00000
0-ply cubeful prune [expert]
4. Cubeful 0-ply 4/3 2/off Eq.: -0.76849 (-5.84646)
0.11575 0.00000 0.00000 - 0.88425 0.00000 0.00000
0-ply cubeful prune [expert]
5. Cubeful 0-ply 2/1 2/off Eq.: -0.76849 (-5.84646)
0.11575 0.00000 0.00000 - 0.88425 0.00000 0.00000
0-ply cubeful prune [expert]
An equity of +5.07797 seems slightly on the high side to me
--
Jim Segrave address@hidden
- [Bug-gnubg] Wrong move in external interface, Joachim Matussek, 2005/03/15
- Re: [Bug-gnubg] Wrong move in external interface, Øystein O Johansen, 2005/03/15
- Re: [Bug-gnubg] Wrong move in external interface, Joachim Matussek, 2005/03/15
- Re: [Bug-gnubg] Wrong move in external interface, Øystein O Johansen, 2005/03/15
- Re: [Bug-gnubg] Wrong move in external interface, Joachim Matussek, 2005/03/15
- Re: [Bug-gnubg] Wrong move in external interface, Joachim Matussek, 2005/03/15
- [Bug-gnubg] Wrong move in external interface, amarganth, 2005/03/16
- Re: [Bug-gnubg] Wrong move in external interface,
Jim Segrave <=
- Re: [Bug-gnubg] Wrong move in external interface, Joachim Matussek, 2005/03/16