bug-gnubg
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Bug-gnubg] Use of Two-Sided DB in matches


From: Neil Robins
Subject: Re: [Bug-gnubg] Use of Two-Sided DB in matches
Date: Sat, 5 Sep 2009 13:07:25 +0100

That would seem to imply to me that GNU is useless for match play. Which seems to be exactly the case here.

As I see it the bottom line is GNU 1-ply gives

   GNU Backgammon  Position ID: gwEAQCoAAAAAAA
                   Match ID   : cAmgACAAGAAA
   +24-23-22-21-20-19------18-17-16-15-14-13-+  O: gnubg
OOO | O              O |   |                  |  2 points
OO | O              O |   |                  |
OO |                  |   |                  |
OO |                  |   |                  |
OO |                  |   |                  |
   |                  |BAR|                  |v 5 point match (Cube: 1)
XX |                  |   |                  |
XX |                  |   |                  |
XX |                  |   |                  |
XX |                  |   |                  |  On roll
XXX |    X     X  X  X |   |                  |  3 points
   +-1--2--3--4--5--6-------7--8--9-10-11-12-+  X: Neil

Cube analysis
1-ply cubeless equity  +0.434 (Money:  +0.434)
 0.717 0.000 0.000 - 0.283 0.000 0.000
Cubeful equities:
1. Double, take         +0.731
2. Double, pass         +1.000  ( +0.269)
3. No double            +0.725  ( -0.006)
Proper cube action: Double, take

Its equivalent XG 2-ply gives:

Analyzed in 2 ply
Player Winning Chances:    73.54% (G:  0.00% B:  0.00%)
Opponent Winning Chances:  26.46% (G:  0.00% B:  0.00%)

Cubeless Equities: No Double=+0.471, Double=+1.549

Cubeful Equities:
     No Double:   +0.566 (-0.311)
     Double/Take: +0.877
     Double/Drop: +1.000 (+0.123)

Best Cube action: Double / Take

XG can give a reliable rollout in no time, showing that any lookahead can give the correct figure for D/T after factoring in the automatic recube:

Analyzed in Rollout
Player Winning Chances:    73.54% (G:  0.00% B:  0.00%)
Opponent Winning Chances:  26.46% (G:  0.00% B:  0.00%)

Cubeless Equities: No Double=+0.471, Double=+1.549

Cubeful Equities:
     No Double:   +0.626 (-0.251)
     Double/Take: +0.877
     Double/Drop: +1.000 (+0.123)

Best Cube action: Double / Take

Rollout
 20736 Games rolled with Variance Reduction.
 Moves and cube decisions: 3 ply
 Confidence No Double: ± 0.003 (+0.623<E<+0.630)
 Confidence Double:    ± 0.000 (+0.877<E<+0.877)

 Double Decision confidence: 100.0%
 Take Decision confidence: 100.0%

 Duration: 1.4 seconds

And yet GNU 7-ply is even more useless:

Cube analysis
7-ply cubeless equity  +0.387 (Money:  +0.387)
 0.694 0.000 0.000 - 0.306 0.000 0.000
Cubeful equities:
1. No double            +0.600
2. Double, pass         +1.000  ( +0.400)
3. Double, take         +0.544  ( -0.056)
Proper cube action: No double, take (12.3%)

----- Original Message ----- From: "Christian Anthon" <address@hidden>
To: "Neil Robins" <address@hidden>
Cc: "Michael Petch" <address@hidden>; <address@hidden>
Sent: Saturday, September 05, 2009 12:29 PM
Subject: Re: [Bug-gnubg] Use of Two-Sided DB in matches


On Sat, Sep 5, 2009 at 8:11 AM, Neil Robins<address@hidden> wrote:
Surely, therefore, at least 1-ply or higher should see the recube and the
then dead cube situation and go to the DB for an exact valid value.


The plays are cubeful as well and therefore only valid for moneygame.

Christian.






reply via email to

[Prev in Thread] Current Thread [Next in Thread]