RE: computer-go: Programming Go in Java!

[David Fotland <fotland@xxxxxxxxxxxxxxx>]

I don't think that very many people use Benson's algorithm, since it 
requires too much searching.  My life and death searches terminate
much sooner, since I recognize much larger eyes than Benson does.

Also, in a go playing program, rather than a tsume-go program, you
have to evaluate the strength of groups that don't have clear-cut 
eyes yet.  Benson's algorithm is no help there.

David Fotland (Many Faces of Go)


At 07:58 PM 5/1/00 -0700, you wrote:
>At 06:08 PM 5/1/2000 -0400, you wrote:
>>At 1:01 PM -0700 5/1/00, Ray Tayek wrote:
>>>...
>>>i can probably answer more detailed questions by looking at the code, but 
>>>maybe we ought to do this off line as perhaps not all on the list are 
>>>interested in benson.
>>
>>I, for one, am much *more* interested in "bensonistics" than in the 
>>various peripheral things that usually appear on this list.  This little 
>>piece of compiled go-reasoning allows significant chunks of tree pruning 
>>in life-and-death situations.  I think this is exactly the sort of thing 
>>this list should be discussing.
>
>i did this as more of an exercise in java than benson. i guess i assumed 
>that since benson is pretty old (25years?) that everyone here would have 
>this sort of thing in their code to terminate searches. i paid absolutely 
>no attention to efficiency when i was coding this up so my implementation 
>may be pretty slow.
>
>there is a lot of computation involved in doing benson, so i suspect that 
>people have invented some kind of incremental approach. i suspect that no 
>more than 4 groups of one color and however-many-you-just-captured of the 
>opponents groups (plus a few more of his maybe) can change status after 
>each move, so maybe there is some kind of incremental approach to the 
>computation?
>
>hth
>
>---
>ray  http://home.pacbell.net/rtayek/
>vice chair orange county java users group http://www.ocjug.org/
>hate spam? http://samspade.org/ssw/
>
>