[Computer-go] Semeais

[fastgo]

KO cannot really be ignored during playouts. There are some close end games
dependent upon proper handling of KO.

I think in the playout phase it needs to remember to take the ko back after
the opponent make a non-threat move.

One example was given in the following post.

http://www.mail-archive.com/computer-go@computer-go.org/msg12293.html


On Fri, Jan 14, 2011 at 1:30 PM, Kahn Jonas <jonas.kahn at math.u-psud.fr>wrote:

> On Fri, 14 Jan 2011, terry mcintyre wrote:
>
>  Any solution must balance competing concerns. What happens during a ko
>> fight? Threats and counterthreats and sacrifices may disturb all neat
>> solutions.
>>
>
> I thought that up to now ko was just ignored during playouts (except
> for strict legality). Given the level at which they are played during
> this phase, I would not expect ko to change that much. I might be wrong,
> of course.
> And if you mean ko in the semeai, then I guess the automaton should just
> win the ko without answering the threat. OK, there might be a problem if
> there are two semeais on the board. Now if we must arrive to this corner
> case (two semeais, one with ko) to really have problems, this would
> already be huge progress.
> Moreover, even that could be solved by the correction I had made
> earlier, that is that semeai should also be calculated for the losing
> side, in case it wins it with ko. I thought this could not happen in the
> playouts but only in the tree, but I was wrong, obviously.
>
> Naturally, these automata might get very complicated, so it's probably
> not a silver bullet, once again...
>
>
> Jonas
> _______________________________________________
> Computer-go mailing list
> Computer-go at dvandva.org
> http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://computer-go.org/pipermail/computer-go/attachments/20110114/22e447e2/attachment.html>