Re: [Bug-gnubg] Rollouts in annotation window

[Joern Thyssen]

On Mon, Jul 22, 2002 at 12:52:40PM +0100, Ian Shaw wrote
> Windows build 020719.
> Starting from a 2-ply analysis, I rolled out the 1st and 5th plays.
> 
> position yAQPoD3YzsIBQQ match 8IlmAAAAEAAA
> 
>                               Win    W(g)   W(bg)  L(g)   L(bg) Equity
> Cubeful
>                  bar/24 8/3* 0.7902 0.6408 0.0885 0.0111 0.0009 +1.0275
> +1.0275
>               Standard error 0.1247 0.1280 0.0611 0.0065 0.0006  0.3529
> 0.3529
>                   bar/20 9/8 0.6487 0.4623 0.0250 0.0372 0.0071 +0.2990
> +0.2990
>               Standard error 0.0455 0.0741 0.0383 0.0418 0.0106  0.0905
> 0.0905
> 
> The equities posted in the annotation window are -1.096 and -1.825
> respectively. This does not match up with the rolled out figures. 

OK, I'll look into it. I found a similar bug yesterday -- maybe it is
the same. I think the cube info is inverted (i.e., inverted cube owner).

> They
> become listed as the top two plays even though they have lower equities than
> the remaining 2-ply evals.

Currently it sorts primarily on the evaluation strength (any rollout >
any ply;  all rollouts are equal and (n)-ply is better than (n-1)-ply).
The equity is used as the secondary sort parameter.

Isn't this want we want? 

rollouts > 3-ply > 2-ply > 1-ply > 0-ply, i.e., we do not mix different
evaluations.

> I then selected the sixth play which had been evaluated at one-ply, and hit
> the eval button (trying to get a 2-ply eval).
> 
> The annotation window flickered for a while, then crashed. I think the sort
> order changed. Can't repeat this bit though.

I'll see if I can reproduce it.

Jørn

-- 
Joern Thyssen, PhD
Vendsysselgade 3, 3., DK-9000 Aalborg, Denmark
+45 9813 2791 (private) / +45 2077 2689 (mobile) / +45 9633 7036 (work)