Computer Chess Club Archives




Subject: Re: And here's another problem for you (Knightshead4mate Revisited)!

Author: leonid

Date: 09:32:02 01/22/01

Go up one level in this thread

On January 22, 2001 at 08:09:23, Paul wrote:

>On January 21, 2001 at 20:57:16, leonid wrote:
>>I still went to search the position by curiosity. Mine 13 moves deepest
>>selective search said, after thinking for 17 sec., that it find nothing. Tried
>>brute force. Ended search after 8 moves. No mate in sight. Branching factor was
>>usual for mate positions but their was no possibility to search 30 moves deep.
>>4 moves - 0.055 sec
>>5 moves - 4.4   sec
>>6 moves - 3.6   sec
>>7 moves - 35    sec
>>8 moves - 389   sec
>>This way, even with all the hash possible I must stay with my program up to the
>>end of Universe.
>Well, there seems to be a restaurant there/then; I'll buy you a drink, ok? :)
>>My program can extend its search as much as I would like. For now I don't need
>>much deeper search and with DOS I am limited somehow with one segment problem.
>>When I will rewrite my program for Linux I will make search much deeper. In my
>>program plys are not reusable but each ply is specific.
>Tim Foden posted the solution in another subthread, so you can have a look at
>that. It is funny! And it is easy to find if you extend the right way.
>Every white move is check, and every time black has only one possible reply.
>And you don't need hash for this, I solve it without.

Now I see how somebody can solve mate so far. When one side have all the time
move that lead to check and other only few responses to it. Then speedy response
can be found and mainly by selective search. Brute force search could lead, even
in very favorable conditions, to very long time. 30 moves sound like for ever. I
speak about search that is done only by logic, not look up tables... I am not
sure, for instance, that previous position that one selective search found in
two seconds, brute force will find even in few hours.



This page took 0 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.