Re: [Bug-gnubg] table stakes beaver

[Christian Anthon]

Hi Thomas,

To evaluate a money beaver you only need to do two rollouts

a) a rollout for the cube in it's initial position
b) a rollout with the cube on 2 the takers side.

However, match beavers are fundamentally different from money beavers
because two extra factors comes in to play:

a) changes in values of cube ownership
b) changes in gammon values

Assuming that raccoons are not allowed you would need to roll out
three different positions to solve the presented problem:

a) cube centered
b) cube at 2 on takers side
c) cube at 4 on takers side

and if raccoons are allowed a fourth one

d) a dead cube at 8

So as you can see the internal logic that gnubg uses for deciding on
beavers in money games cannot directly be transferred to match or
table stake beavers. And likewise, the normal double rollout that you
have shown will not be enough to decide the correct cube action when
beavers are allowed.

Christian.

On Sun, Nov 23, 2008 at 12:13 PM, Thomas Koch <address@hidden> wrote:
> In order to analyse positions with table stakes using gnubg, I put together
> a special met (http://page.mi.fu-berlin.de/tkoch/Backgammon/limit8.xml). Now
> I was wondering if this position
>
> X on roll, cube decision?
>
>     GNU Backgammon  Position ID: bNsIADa8HQAAAA
>                     Match ID   : cAkAAAAAAAAA
>     +24-23-22-21-20-19------18-17-
> 16-15-14-13-+  O: Alice
>     |       O  O  O  O |   | O        O       |  0 points
>     |       O  O  O  O |   | O                |
>     |                  |   |                  |
>     |                  |   |                  |
>     |                  |   |                  |
>     |                  |BAR|                  |v (Cube: 1)
>   X |                  |   |                  |
>   X |       X          |   |                  |
>   X |       X     X    |   |                  |
>   X | O  O  X  X  X    |   |                  |  On roll
>  XX | O  O  X  X  X    |   |                  |  0 points
>     +-1--2--3--4--5--6-------7--8--9-10-11-12-+  X: Bob
> Pip counts: O 154, X 35
>
> Output generated Sat Nov 22 19:34:37 2008
> by GNU Backgammon 0.16-devel (Text Export version 1.76)
>
> is a correct double/beaver even without Jacobyrule if the table stakes are
> set to 8 times the initial stakes (it is on snowie 3ply)!? But of course
> using a met makes gnubg work in match mode so beavers are not allowed. Is
> there a way to allow for beavers even in a match? If not, how can I deduce
> from the results of the rollout if a beaver would be correct?
>
> Cube analysis
> Rollout cubeless equity +0.498 (Money: +0.499)
>
> Cubeful equities:
> 1. Double, take +0.594
> 2. Double, pass +1.000 ( +0.406)
> 3. No double +0.520 ( -0.073)
> Proper cube action: Double, take
> Rollout details:
> Centered 1-cube:
> 0.538 0.295 0.129 - 0.462 0.000 0.000 CL +0.498 CF +0.520
> [0.003 0.002 0.002 - 0.003 0.000 0.000 CL 0.008 CF 0.017]
> Player Alice owns 2-cube:
> 0.553 0.295 0.136 - 0.447 0.000 0.000 CL +1.067 CF +0.594
> [0.004 0.003 0.002 - 0.004 0.000 0.000 CL 0.018 CF 0.027]
>
> Thanks!
>
> Thomas Koch
>
> --
> Backgammon in Berlin: http://berlinbg.wordpress.com
> Der Backgammon Blog: http://graswurzelbackgammon.wordpress.com
>
> _______________________________________________
> Bug-gnubg mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/bug-gnubg
>
>