BGonline.org Forums

3 more rollouts

Posted By: Philippe Michel
Date: Monday, 17 December 2012, at 8:29 p.m.

In Response To: Evaluations on all levels (mtuhtan)

It sure looks possible for Blue to play this for a pass, and I'd be inclined to think a strong human player would do even better :

The score (after 0 games) is: GNUbg 0, You 0

Move number 4: You on roll, cube decision?

GNUbg	68
You	154
Position ID: BgABFNsOIlAJAA Match ID: UQkAAAAAAAAA

• You doubles

Cube decision
Rollout cubeless equity		+0.557
Cubeful equities:
1.	Double, pass	+1.000
2.	Double, take	+1.017	+0.017
3.	No double	+0.983	-0.017
Proper cube action:		Redouble, pass
Rollout details
Win W g W bg Lose L g L bg Cubeless Cubeful Player You owns 2-cube 0.792 0.000 0.000 - 0.208 0.024 0.004 +0.557 +0.983 Standard error 0.002 0.000 0.000 - 0.002 0.001 0.000 0.004 0.009 Player GNUbg owns 4-cube 0.799 0.000 0.000 - 0.201 0.022 0.003 +1.143 +1.017 Standard error 0.002 0.000 0.000 - 0.002 0.001 0.000 0.009 0.011
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 703063414 and quasi-random dice
Play: world class 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]

The score (after 0 games) is: GNUbg 0, You 0

Move number 4: You on roll, cube decision?

GNUbg	68
You	154
Position ID: BgABFNsOIlAJAA Match ID: UQkAAAAAAAAA

• You doubles

Cube decision
Rollout cubeless equity		+0.562
Cubeful equities:
1.	Double, pass	+1.000
2.	Double, take	+1.033	+0.033
3.	No double	+0.997	-0.003
Proper cube action:		Redouble, pass
Rollout details
Win W g W bg Lose L g L bg Cubeless Cubeful Player You owns 2-cube 0.794 0.000 0.000 - 0.206 0.023 0.004 +0.562 +0.997 Standard error 0.002 0.000 0.000 - 0.002 0.001 0.000 0.004 0.009 Player GNUbg owns 4-cube 0.802 0.000 0.000 - 0.198 0.022 0.004 +1.157 +1.033 Standard error 0.002 0.000 0.000 - 0.002 0.001 0.000 0.008 0.011
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 702780209 and quasi-random dice
Play: supremo 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 16 more moves within equity 0.32
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]

The score (after 0 games) is: GNUbg 0, You 0

Move number 4: You on roll, cube decision?

GNUbg	68
You	154
Position ID: BgABFNsOIlAJAA Match ID: UQkAAAAAAAAA

• You doubles

Cube decision
Rollout cubeless equity		+0.565
Cubeful equities:
1.	Double, pass	+1.000
2.	Double, take	+1.057	+0.057
3.	No double	+0.975	-0.025
Proper cube action:		Redouble, pass
Rollout details
Win W g W bg Lose L g L bg Cubeless Cubeful Player You owns 2-cube 0.795 0.000 0.000 - 0.205 0.021 0.004 +0.565 +0.975 Standard error 0.001 0.000 0.000 - 0.001 0.001 0.000 0.003 0.006 Player GNUbg owns 4-cube 0.805 0.000 0.000 - 0.195 0.019 0.003 +1.177 +1.057 Standard error 0.002 0.000 0.000 - 0.002 0.001 0.000 0.006 0.008
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 703584244 and quasi-random dice
Play: grandmaster 3-ply cubeful prune [grandmaster]
keep the first 0 0-ply moves and up to 16 more moves within equity 0.32
Skip pruning for 1-ply moves.
keep the first 0 2-ply moves and up to 4 more moves within equity 0.08
Cube: 3-ply cubeful prune [grandmaster]

The No Double equities are suspect since there must be bad takes later, but I think the Double numbers are about right.

Please note that this is exported from gnubg, but with retrained neural nets. Stock gnubg 0.90 surely would do worse.

Regarding immediate checker play as pointed in Michael Depreli's follow-up, it seems the bot tends to play somewhat purely here, for instance 31: 11/10*/7 (43: 14/10*/7 as well) ; not the awful 6/3, but no random shuffling of back checkers either.

Messages In This Thread

• Containment + XG question
  mtuhtan -- Saturday, 15 December 2012, at 5:38 p.m.
  • Containment + XG question
    rambiz -- Saturday, 15 December 2012, at 7:09 p.m.
  • Containment + XG question
    Daniel Murphy -- Saturday, 15 December 2012, at 11:14 p.m.
    • Containment + XG question
      Timothy Chow -- Saturday, 15 December 2012, at 11:32 p.m.
    • Containment + XG question
      Casper van der Tak -- Sunday, 16 December 2012, at 8:26 a.m.
  • Containment + XG question
    neilkaz -- Sunday, 16 December 2012, at 3:23 a.m.
  • Containment + XG question
    Maik Stiebler -- Sunday, 16 December 2012, at 7:41 a.m.
  • Evaluations on all levels
    mtuhtan -- Sunday, 16 December 2012, at 3:50 p.m.
    • 5 rollouts and 1 question
      mtuhtan -- Sunday, 16 December 2012, at 10:38 p.m.
      • correction
        mtuhtan -- Sunday, 16 December 2012, at 10:40 p.m.
      • 5 rollouts and 1 question
        Timothy Chow -- Monday, 17 December 2012, at 12:51 a.m.
        • 5 rollouts and 1 question
          Casper van der Tak -- Monday, 17 December 2012, at 8:11 a.m.
        • cubeless RO
          Maik Stiebler -- Monday, 17 December 2012, at 3:55 p.m.
          • cubeless RO -- Janowski takepoint
            Chuck Bower -- Monday, 17 December 2012, at 5:53 p.m.
            • cubeless RO -- Janowski takepoint
              Maik Stiebler -- Monday, 17 December 2012, at 8:22 p.m.
              • cubeless RO -- Janowski takepoint
                Chuck Bower -- Tuesday, 18 December 2012, at 12:27 a.m.
                • cubeless RO -- Janowski takepoint
                  Maik Stiebler -- Tuesday, 18 December 2012, at 8:34 a.m.
      • 5 rollouts and 1 question
        Michael Depreli -- Monday, 17 December 2012, at 3:10 p.m.
    • 3 more rollouts
      Philippe Michel -- Monday, 17 December 2012, at 8:29 p.m.
      • 3 more rollouts
        neilkaz -- Monday, 17 December 2012, at 8:44 p.m.
      • 3 more rollouts - and one more
        Daniel Murphy -- Tuesday, 18 December 2012, at 8:03 a.m.