|
From: | Nis |
Subject: | Re: [Bug-gnubg] Why is odd ply equity always lower? |
Date: | Thu, 19 Jun 2003 10:54:42 +0200 |
--On 19 June 2003 07:20 +0000 Joern Thyssen <address@hidden> wrote:
On Wed, Jun 18, 2003 at 10:24:38PM +0200, Nis wrote [snip]Only relevant for cube decisions? The wrong estimations of equity is mainly a problem of absolute equities. As Joseph has pointed out, chequerplay relies almost solely on the relative equities of positions. Thus a fix should only "need" to be applied for cube decisions (and cubeful chequerplay? Someone help me here) So we might want to start using reduced evaluation settings for cube decisionsHaven't seen your patch yet, I think it may be much easier if we add an option to evalcontext that enables this. The evaluation routines doesn't really know if it's evaluating a cube decision or chequer play decision.
My idea was to set this up in the predefined evaluation settings. TODO: allow named configurations to be set up by command: set evaluiation HUGE
BTW, I'd be happy to see your patch! If it's not too large you can probably post it on the list.
I've attached it. Please note that the combination of your "deterministic reduction" is that 1-ply reduction is always using the first of the roll lists. But reduced 1-ply is probably a bad idea anyway. The option is there, however - I implemented a fix.
My first unscientific tests show a much higher correlation between 1- 2- and 3-ply when using reduced evaluations. For instance, after an opening 5-4 played 24/20, 13/8, this is the eval on 100%
Win W(g) W(bg) L(g) L(bg) Equity (cubeful) static: 0.484 0.125 0.003 0.136 0.006 (-0.046 (-0.064)) 1 ply: 0.501 0.134 0.005 0.136 0.005 (-0.001 (-0.001)) 2 ply: 0.481 0.128 0.004 0.141 0.007 (-0.053 (-0.073)) 3 ply: 0.499 0.133 0.004 0.136 0.005 (-0.006 (-0.008)) and this is the eval on 50% Win W(g) W(bg) L(g) L(bg) Equity (cubeful) static: 0.484 0.125 0.003 0.136 0.006 (-0.046 (-0.064)) 1 ply: 0.488 0.123 0.004 0.143 0.006 (-0.046 (-0.062)) 2 ply: 0.492 0.131 0.004 0.138 0.006 (-0.024 (-0.032)) 3 ply: 0.491 0.130 0.004 0.138 0.006 (-0.029 (-0.040)) -- Nis Jorgensen Greenpeace Amsterdam
eval.patch
Description: Text document
[Prev in Thread] | Current Thread | [Next in Thread] |