Re: [Bug-gnubg] address@hidden: GNUBG (23062002)]

[Joern Thyssen]

On Thu, Jun 27, 2002 at 04:16:56PM +0000, Øystein Johansen wrote
> Hi,
> 
> Anyone care to look at this?
> 
> -Øystein
> 
> ----- Forwarded message from address@hidden -----
> 
> From: address@hidden
> Date: Thu, 27 Jun 2002 09:31:25 EDT
> Subject: GNUBG (23062002)
> To: address@hidden
> X-Mailer: AOL 6.0 for Windows DE sub 10512
> 
> Hi Oeystein,
> 
> thanks for alyways replying so fast - and for all your other efforts
> concerning GNUBG, of course!
> 
> When I analysed a FIBS-match played by me against iceblot I was
> astonished that according to GNUBG, version 23.06.2002 I committed
> 4 wrong passes (iceblot1.txt).
> Strange - I only did pass 2 times in the whole match (as you can see
> easily: iceblot1.mat).
> 
> Apparently, there is something wrong with the import tool,
> in Game 1 the game record of GNUBG looks like this:
> 12 Double to 2      Drop
> 13                  Drop
> 
> The same procedure as well in Game 2:
> 22 Double to 2      Drop
> 23                  Drop
> 
> Why these double drop lines ?

Looks like there is a bug in the import of Jellyfish files. I'll look
into it.

> 
> By the way: I do not agree with the judgement of GNUBG concerning the
> second pass (correct response:take):
> Take 31,864%
> Drop 31,780%
> -0,084%
> In my opinion this pass was justified. What do the experts say ?

Oeystein, have you made a new build with the fix from yesterday
regarding reading of match equity tables?

I believe this is the problem here.

Blotman, can you try with a build dated 2002-06-27 or later, please?

> One further question: in the take analysis windows stands "cubeless 2 ply"
> although I activated cubeful chequer evaluation !? How about that ?

I've had quite a few questions regarding this. The cubeless equity is
just given for your information:

Cube analysis

 Cubeless 2-ply equity: -0.068

 0.61% 13.12% 46.62% 53.38% 15.50% 0.58%

 1.   No double                -0.128
 2.   Double, pass             +1.000  (+1.128)
 3.   Double, take             -0.545  (-0.417)

Proper cube action:   No double, beaver   (27.0%)

This tells you: your cubeless equity for this position is -0.068 with
cubeless winning chance 46.62%, and 13.12% gammons etc.

The cubeful equity for "no double" is -0.128 and the cubeful equity for
"double, take" is -0.545. 

You should only double this position if 27% of your opponents would drop
the cube.

> Just one little point: if I export a position into html-file, the position
> is not exactly the same I wanted to export (the colours of the stones
> were mixed).

Can you send a screenshot of the resulting html-file (or even better:
post it on a web-page)?

Joern

-- 
Joern Thyssen, PhD
Vendsysselgade 3, 3., 9000 Aalborg
+45 9813 2791 (private) / +45 2077 2689 (mobile) / +45 9633 7010 (work)