Author: Serge Desmarais
Date: 16:49:42 09/27/98
Go up one level in this thread
On September 27, 1998 at 17:51:48, John Coffey wrote: >On September 27, 1998 at 17:45:17, Serge Desmarais wrote: > >>On September 20, 1998 at 18:47:20, John Coffey wrote: >> >>>I found this on http://www.xs4all.nl/~verhelst/chess/search.html ........... >>> >>>"Null move heuristic >>> >>>The null move heuristic is a method of skipping searches in parts of the tree >>>where the position is good enoigh. This is tested by doing a null move (i.e. >>>passing, doing no move at all) and then seraching with reduced depth. If the >>>result of this is higher than beta, no further search is done; if the result is >>>lower than beta we do a normal search. >>> >>>The null move heuristic has big dangers because it can fail to detect deep >>>combinations. On the other hand it can save a lot of time by skipping large >>>parts of the search tree. " >>> >>>John Coffey >> >> >> So I assume the "null move technique" was invented at a time when computers >>were fairly slow and so the programmers had to find some means of bypassing this >>by all sorts of techniques. Now that computers are faster than "Mach 3", it is >>no more necessary to take the chance of missing deep combinations? They will >>reach these with normal search and selective extensions? I had a Super Nine from >>Fidelity Electronics in the 80's and it needed days to reach a depth of 9-11 >>plies, even in very simple ending positions! >> >>Serge Desmarais > >Actually null moves pruning is one of the reasons why modern programs don't >take days to get to 11 plys. > >John coffey But when a program reaches 11 plies with 35 ply extensions, it is supposed to have examined all positions up to ply 11? I mean, it examined all moves at ply 1, then started again at ply 2, 3 and so on, up to ply 11? Of course, looking at a program like Fritz 5, it mostly spend time with the first 10 moves and goes very quickly over the rest, then starting again for the next ply. But if the move order is quite accurate, most legal moves in about any position are quite stupid/weak. No need to spend much time at examining every queen move which would put her en prise to every ennemy pawns and pieces... So is it not the pruning of the lines that allow the program to reach a depth of 11 plies (if not even more)? Serge Desmarais
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.