Author: Rolf Tueschen
Date: 14:35:18 01/05/06
Go up one level in this thread
On January 05, 2006 at 17:00:40, Eelco de Groot wrote: >On January 05, 2006 at 16:34:52, Rolf Tueschen wrote: > >>On January 05, 2006 at 12:18:52, Vasik Rajlich wrote: >> >>>On January 05, 2006 at 09:25:15, Eduard Nemeth wrote: >>> >>>>On January 05, 2006 at 08:49:15, Eduard Nemeth wrote: >>>> >>>>> >>>>>>Errh, I dont read the emails of Vas, so could you please post it here? >>>>> >>>>>A little bit from the Readme by the Programmer: >>>>> >>>>>"As its name implies, Rybka WinFinder can be used to look for wins. It can spot >>>>>and evaluate long, complex tactical variations much faster than the normal Rybka >>>>>engine. If you have some complex position and just want to know if one side can >>>>>somehow win material or get a clear advantage, this is the tool you want." >>>> >>>>A little example see here: >>>> >>>>http://f27.parsimony.net/forum67838/messages/1383.htm >>>> >>>>ED. >>> >>>Yes, this is the idea. On average it should find tactics like this much faster. >>> >>>Vas >> >> >>What would be the difficulty to include it into the "normal" program so that a >>sort of always present tactical speedy is looking for some food? I dont >>understand the meaning of a "seperate" tool. Isnt it something which screws the >>general strength of the program? > >If it was part of the general program it would eat up CPU time in the >background, that would screw with the general strenghth of Rybka. I think the >real use of this tool is if it can avoid the "false positives". It is easy to >tune for finding combinations, but weeding out the AM moves with deep tactics is >the hard part for Vas. So, I'd suggest letting loose a enormous bunch of avoid >moves on the winfinder and see how it does! > >Actually such testposions I would be interested in too, especially if they are >checked. I know the Swiss test, but it is only 64 positions that are more for >testing general programstrength. I think with a couple of hundred positions, >each tested for a minute or so, such a test would be more useful. > > Eelco Eelco, you do me a great favor. In special that you remind me of an old idea from the years 1982 or such. I always knew that a chess machine that could be tuned exactly for the specific task on the board - that engine would be stronger than World Champions. But in reality it's weaker because it's impossible to tune for infinite positions unless you dont program some invisible moves with your mouse as an operator. :) I dont know if it's generally known. A chess machine can't decide on its own what engine tool it should employ for the actual position. Aint I a genius? Because I, not even a GM player, I know what to do in specific situations. If I must search for a mate or a draw or material. I mean, I'm not that strong because else I were a GM...
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.