Re: [Bug-gnubg] Bug: pruning problem (solved or not?)

[Neil Robins]

There is nothing wrong with the 2-ply evaluation or rollout result. All three
plays are equally good, of course. You should never get to throw again.
Either the opponent throws an immediate winning number or the correct cube
action is Double/Pass at this score whatever you play now.


boomslang wrote:
> 
> Hi team,
>  
> I am not sure if this has been tackled yet but I get strange results in
> the following position:  
> 
>      Position ID: LQAAQAoAAAAAAA
>      Match ID: cIlmAAAACAAA     
> 
> On 0-ply, it gives the correct move (5/off 2/1) but on 1-ply it rates
> '5/off 4/3' equally good as '5/off 2/1'... During a full rollout it rates
> all three possible moves equally good. See tables below.
> 
> Does this have something to do with pruning? Am I running an old version
> perhaps?  I have "Version GNU Backgammon 0.90-mingw 20091102" and use a
> large cahce and one eval thread, Windows XP HE 2002 SP3.
> 
> Thx in advance
> 
> 
> --boomslang
>  
>  
>     1. Cubeful 0-ply    5/off 2/1                    Eq.:  +0.929
>        0.732 0.000 0.000 - 0.268 0.000 0.000
>         0-ply cubeful prune [expert]
>     2. Cubeful 0-ply    5/off 4/3                    Eq.:  +0.734 (
> -0.196)
>        0.642 0.000 0.000 - 0.358 0.000 0.000
>         0-ply cubeful prune [expert]
>     3. Cubeful 0-ply    5/off                        Eq.:  +0.479 (
> -0.450)
>        0.597 0.000 0.000 - 0.403 0.000 0.000
>         0-ply cubeful prune [expert]
> 
> 
>     1. Cubeful 1-ply    5/off 2/1                    Eq.:  +0.778
>        0.732 0.000 0.000 - 0.268 0.000 0.000
>         1-ply cubeful prune
>     2. Cubeful 1-ply    5/off 4/3                    Eq.:  +0.778 (
> +0.000)
>        0.642 0.000 0.000 - 0.358 0.000 0.000
>         1-ply cubeful prune
>     3. Cubeful 1-ply    5/off                        Eq.:  +0.712 (
> -0.066)
>        0.599 0.000 0.000 - 0.401 0.000 0.000
>         1-ply cubeful prune
> 
> 
> 
> 
> 
>     1. Cubeful 2-ply    5/off 2/1                    Eq.:  +0.778
>        0.732 0.000 0.000 - 0.268 0.000 0.000
>         2-ply cubeful prune [world class]
>     2. Cubeful 2-ply    5/off 4/3                    Eq.:  +0.778 (
> +0.000)
>        0.642 0.000 0.000 - 0.358 0.000 0.000
>         2-ply cubeful prune [world class]
>     3. Cubeful 2-ply    5/off                        Eq.:  +0.778 (
> +0.000)
>        0.599 0.000 0.000 - 0.401 0.000 0.000
>         2-ply cubeful prune [world class]
> 
> 
> 
> 
> 
> 
>     1. Rollout          5/off                        Eq.:  +0.785
>        0.599 0.000 0.000 - 0.401 0.000 0.000 CL  +0.198 CF  +0.785
>       [0.000 0.000 0.000 - 0.000 0.000 0.000 CL   0.000 CF   0.003]
>         Full cubeful rollout with var.redn.
>         1296 games, Mersenne Twister dice gen. with seed 845114611 and
> quasi-random dice
>         Play: 0-ply cubeful prune [expert]
>         Cube: 0-ply cubeful prune [expert]
>     2. Rollout          5/off 2/1                    Eq.:  +0.778 (
> -0.008)
>        0.732 0.000 0.000 - 0.268 0.000 0.000 CL  +0.463 CF  +0.778
>       [0.000 0.000 0.000 - 0.000 0.000 0.000 CL   0.000 CF   0.000]
>         Full cubeful rollout with var.redn.
>         1296 games, Mersenne Twister dice gen. with seed 845114611 and
> quasi-random dice
>         Play: 0-ply cubeful prune [expert]
>         Cube: 0-ply cubeful prune [expert]
>     3. Rollout          5/off 4/3                    Eq.:  +0.778 (
> -0.008)
>        0.642 0.000 0.000 - 0.358 0.000 0.000 CL  +0.284 CF  +0.778
>       [0.000 0.000 0.000 - 0.000 0.000 0.000 CL   0.000 CF   0.000]
>         Full cubeful rollout with var.redn.
>         1296 games, Mersenne Twister dice gen. with seed 845114611 and
> quasi-random dice
>         Play: 0-ply cubeful prune [expert]
>         Cube: 0-ply cubeful prune [expert]
> 
> 
> 
> 
> 
> 
>     1. Rollout          5/off 2/1                    Eq.:  +0.778
>        0.732 0.000 0.000 - 0.268 0.000 0.000 CL  +0.463 CF  +0.778
>       [0.000 0.000 0.000 - 0.000 0.000 0.000 CL   0.000 CF   0.000]
>         Full cubeful rollout with var.redn.
>         1296 games, Mersenne Twister dice gen. with seed 845114611 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]
>     2. Rollout          5/off 4/3                    Eq.:  +0.778 (
> +0.000)
>        0.642 0.000 0.000 - 0.358 0.000 0.000 CL  +0.284 CF  +0.778
>       [0.000 0.000 0.000 - 0.000 0.000 0.000 CL   0.000 CF   0.000]
>         Full cubeful rollout with var.redn.
>         1296 games, Mersenne Twister dice gen. with seed 845114611 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]
>     3. Rollout          5/off                        Eq.:  +0.778 (
> +0.000)
>        0.597 0.000 0.000 - 0.403 0.000 0.000 CL  +0.194 CF  +0.778
>       [0.000 0.000 0.000 - 0.000 0.000 0.000 CL   0.001 CF   0.000]
>         Full cubeful rollout with var.redn.
>         1296 games, Mersenne Twister dice gen. with seed 845114611 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]
> 
> 
> 
> 
> 
> _______________________________________________
> Bug-gnubg mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/bug-gnubg
> 
> 

-- 
View this message in context: 
http://old.nabble.com/Bug%3A-pruning-problem-%28solved-or-not-%29-tp27529821p27555172.html
Sent from the Gnu - Backgammon mailing list archive at Nabble.com.