Re: [gnugo-devel] Bug in combinaison attacks

[SP LEE]

> 
> There is probably a problem in the combinaison attack module.
> 
> In the following 2 stones game at move 127, gnugo correctly
> sees that Black threatens a combinaison attack in the left
> territory begining with ataris at J12, G10, E8, B9, C9, but
> gives 33 defense points, some of them irrelevant.
> 
> Stephane
> 
> Top moves:
> 1. Q10 11.02
> 2. H12 10.82
> 3. S10 10.78
> 4. R12 10.44
> 5. Q11 10.44
> 6. J12 10.43
> 7. G11 10.43
> 8. B9  10.38
> 9. K16 10.32
> 10. E9  10.30
> Checking safety of a white move at Q10
> Move generation likes Q10 with value 11.02
> genmove() recommends Q10 with value 11.02
> white (O) move Q10
> 

I agree that H12 worth about 10 points. When black play at J12, white can 
answer at H12, sacrificing 2 stones. The problem is why gnugo thinks Q10 having 
11.02 points. Maybe the attack/defense there was read wrong.

SP Lee