[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

WTH@GNU

Posted By: Daniel Murphy
Date: Wednesday, 30 July 2008, at 7:36 p.m.

In Response To: WTH@GNU (Mislav Radica)


Maybe I wrote so before, but I want to clarify that the bug with extending Gnubg v0.90 rollouts occurs whenever the rollout is stopped and then restarted, whether or not the number of trials is changed in the parameter window. Example:

Position st8MAGB32w0AAA:cAluAAAAAAAA, Play = Supremo, Cube = World Class, Trials = 72. Seed = 834497255.

Start the rollout but stop it manually after 36 trials. The result: 

 

    1. Rollout          6/3 6/2                      Eq.:  +1.549320 

       0.951548 0.442207 0.002564 - 0.048452 0.000000 0.000000 CL  +1.567880 CF  +1.549320 

      [0.004177 0.009453 0.000863 - 0.004177 0.000000 0.000000 CL   0.022639 CF   0.038221] 

        Full cubeful rollout with var.redn. 

        36 games, Mersenne Twister dice gen. with seed 834497255 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]


Then restart the rollout and let it run for the whole 72 trials. The result: 

 

    1. Rollout          6/3 6/2                      Eq.:  +1.549320 

       0.951548 0.442207 0.002564 - 0.048452 0.000000 0.000000 CL  +1.567880 CF  +1.549320 

      [0.002933 0.006637 0.000606 - 0.002933 0.000000 0.000000 CL   0.015895 CF   0.026835] 

        Full cubeful rollout with var.redn. 

        72 games, Mersenne Twister dice gen. with seed 834497255 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]


Identical! The program didn't add 36 more trials to the first 36. It merely repeated the first 36 trials with the same dice rolls.

The Standard Errors reported after 72 trials are smaller, but they're erroneous.


Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.