Author: Stuart Cracraft
Date: 19:06:40 08/22/04
Go up one level in this thread
On August 22, 2004 at 16:03:54, Uri Blass wrote: >On August 22, 2004 at 15:54:09, Stuart Cracraft wrote: > >>On August 22, 2004 at 04:00:53, Uri Blass wrote: >> >>>On August 21, 2004 at 23:56:48, Stuart Cracraft wrote: >>> >>>>On August 21, 2004 at 23:39:17, Stuart Cracraft wrote: >>>> >>>>>There was an error. The right position is: >>>>> >>>>>[D]5r1k/6p1/1n2Q2p/4p3/8/7P/PP4PK/R1B1q3 w >>>>> >>>> >>>>It takes my program 142M nodes to get the right move for the >>>>right reason in 10 minutes. This would be about 1.5 minutes >>>>on a more modern PC. >>>> >>>>I am wondering, specifically, what tricks there are to get it >>>>solved faster than this on non-parallel. >>>> >>>>Stuart >>> >>>What about check extensions and check in the qsearch? >>> >>>What is your main problem >>>seeing that Qxb6 has negative evaluation or seeing that Bxh6 has better >>>evaluation than Qxb6 >>> >>>In order to see that Qxb6 is losing >>>You need to see the following line >>> >>>1.Qxb6 Rf1 2.Qd8+ Kh7 3.Qd3+ e4 4.Qxx Qh1# >>> >>>4...Qh1# is move that my qsearch detects so even without check extensions it >>>seems to be only 7 plies of search >>>2.Qd8+ and 3.Qd3+ are extended by 1 ply so I can see that 1.Qxb6 is losing the >>>queen in 5 plies(the evaluation is only slightly behind 0 at this point) >>> >>>If you want to see that Bxh6 is at least a draw for white then check extensions >>>may also help because white has a perpetual check but a lot of checks. >>> >>>You can use the botvinik extension that is doing partial extensions for >>>threating the same piece move after move when the same piece may be also the >>>king. >>> >>>I use the botvinik extension today only for checks. >>> >>>Uri >> >>Haven't heard of this one. Sounds interesting. So how do you tell that >>you're threatening the same piece move after move? At what point do you >>make that decision and grant the partial extension? (fractional same >>thing?) >> >>Stuart > >In my case I simply extend partial extension when there is more than one >consecutive check. > >If I have check at ply x and x+2 then I extend partial extension(one of the >factor that I use to decide how much to extend is the number of legal moves of >the side that is in check). > >I did not write a special code to try it for repeated threat against another >piece. > >Uri Ah yes, I see. Sorry -- how did it do for you?
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.