Author: Uri Blass
Date: 07:42:03 01/06/01
Go up one level in this thread
On January 06, 2001 at 10:32:31, Robert Hyatt wrote: >On January 05, 2001 at 14:37:49, Uri Blass wrote: > >>On January 05, 2001 at 14:09:24, Robert Hyatt wrote: >> >>>On January 05, 2001 at 08:50:43, José Carlos wrote: >>> >>>>On January 05, 2001 at 08:38:07, Uri Blass wrote: >>>> >>>>>On January 05, 2001 at 08:04:38, Ulrich Tuerke wrote: >>>>> >>>>>>On January 05, 2001 at 07:50:42, Mark Schreiber wrote: >>>>>> >>>>>>>In the match with v/d Wiel, Rebel is running on P3 866 MHz. Using a faster >>>>>>>computer would be an improvemnt. Maybe a P4 1.5 GHz. They could also improve >>>>>>>Rebel to run on dual or multi processor like Junior. The Junior that ran on an 8 >>>>>>>processor at Dortmund would clobber v/d Wiel. At Dortmund, Junior performed at >>>>>>>Fide 2700. >>>>>> >>>>>>You are wrong. Van der Wiel enforces games which are highly non-tactically. A >>>>>>high node/sec won't help here. Deep Junior would have the same trouble. >>>>>> >>>>>>Uli >>>>> >>>>>I disagree. >>>>>Programs can find better positional moves when they search deeper. >>>> >>>> But the curve strength/speed in non-tactical positions is almost flat. >>>> For examples, if a program doesn't understand weak pawns, a speed improvment >>>>won't help it unless it can find the loss of the pawn, which turns the position >>>>into tactics. >>>> I understand speed can help sometimes in strategical positions, very few IMO. >>>> >>>> José C. >>>> >>>>>Uri >>> >>> >>>Speed is absolutely _not_ going to repair holes in a program's evaluation. If >>>it is missing something important (say how to handle blocked pawn positions) >>>then making it run 10x faster won't help one iota... >> >>Deeper search can help to repair holes in program evaluation. >> >>It is possible that the program does not understand that a move is not good >>because the opponent has a positional advantage at small depth but at big depth >>it can see that the opponent can get a positional advantage that it knows to >>evaluate. >> >>Uri > > >OK.. let's be clear here. If your program doesn't understand forks, a 3 ply >search will fix that hole. If your program doesn't understand blocked >positions, a 50 ply search will repair that hole. But I don't think a 50 >ply search is going to happen. Some kinds of missing knowledge can be covered >by a search (don't realize a passed pawn on the 7th is strong? a few plies >will show this although sometimes it takes 10-20 plies to see that the pawn >is going to cause problems. Don't realize two connected passers on the 6th are >strong? again sometimes 8-10 plies will show this, sometimes it will take >20+). > >The kind of hole being exploited vs Rebel isn't going to be helped by another >10 plies. Ed found that Rebel could save the game that it lost against Van dar wiel by deeper search. It seems that 1 or 2 more plies than the game are enough to save game 3. I do not say that it is enough to win the match but it is enough to get better result. Uri
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.