Author: Sune Fischer
Date: 16:05:13 01/20/04
Go up one level in this thread
On January 20, 2004 at 14:56:19, Robert Hyatt wrote: >On January 20, 2004 at 13:22:18, Sune Fischer wrote: > >>On January 20, 2004 at 12:01:59, Robert Hyatt wrote: >> >>>Then Grandmasters are producing tons of "eye candy" every day as they >>>merrily annotate games and give variations. Are you really a chess player? >>>IE I want analysis, not just "white is better" myself. >> >>Yes I am really a chess player, and I know how to analyze using an engine. >> >>You need to go back and forth down the lines that look interesting. >> >>Particularly big sacrifices can take a while for engines to see, they fail high >>10 times faster if you just execute the move so nullmove and other pruning >>devices gets a better root window to work with. > >I rarely use that approach, as it doesn't prove the sacrifice was the best >move at the root position if you force the game a few moves down... In a real game there may be only one really interesting move, say Bxh7+, with a follow up by a king attack. If you want the engine to see these 20 plies ahead you can wait for half an hour or you can play out the first few moves and the score will either go up pretty fast or the engine will show you the proper defending moves, either way it takes only a few minutes. Some lines are too deep even for engines today, other times the engine just "doesn't get it" and wants to play some boring line, then I take over and play out the lines I want to see. I do this often with the lines I thought about while playing a game for instance. >>If you have lost the first half of the pv how are you going to track it? > > >you search down to the end of the PV, and then go forward. The engine is >going to follow the same path again... I don't understand what you mean. The problem is the move that was interesting is gone, it's been replaced by the new line. How do you get that line back without some sort of backup gizmo? >>Looks to me like Shredder is a bit of an extreme example. >>You have the same problem though if you can't resolve a fail high, no pv :) > >I can _always_ resolve a fail-high. The question is, do I want to take the >time or can I take the time? But if someone wants to see the score or PV, >they can wait and get it eventually... > >Here you are mixing the concepts of playing a game and analyzing an old >game. When playing a game, you play the best move. When analyzing you want >more than just a best move. IE why print the score either? It not the same. The PV is a tiny fraction of a big search tree, the score is the score. >>>In tactical positions I trust the PV to the end. IE mate in 15 is not going >>>to be flakey near the end. Nor is winning a pawn or piece, or promoting >>>something... >>> >>>So it is not as useless as you seem to think. >> >>The hash shows the same behavior on tactical positions. >> >>-S. > > >again, not always. The PV is quite meaningless, just take the mate in 15 as an example. Unless all the moves by the opponent are "one replies", it won't tell you anything about what happens if he does this or if he does that. The PV is just an example, even in this extremely tactical case it only shows you a fraction of the tree. You might not even be able to guess the right ponder move! :) -S.
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.