Author: Robert Hyatt
Date: 11:17:27 08/22/02
Go up one level in this thread
On August 22, 2002 at 12:43:47, Gian-Carlo Pascutto wrote: >On August 22, 2002 at 11:13:46, Robert Hyatt wrote: > >>If you don't see the difference, pick up any good AI book. Most all cover >>it. Reducing the depth is reducing the depth. It is not the same thing >>as saying "I choose to throw this move out at this ply and not even consider >>it at all". razoring says "this move looks suspiciously bad. Rather than >>just throwing it out, I am going to search it to a 1-ply reduced depth, so that >>if it _does_ do something good, I will have a chance to see it with the reduced >>search." > >Which is exactly the same thing. > >Pruning cuts the move you apply it to. Razoring cuts the frontier moves below >the subtree your razor. > >>Forward pruning will result in outright blunders. By pruning a single move >>you miss the fact that that move kills you. Or kills your opponent. Even >>if the threat is very shallow. Razoring doesn't hide the shallow threat at >>all, which is the point... > >Razoring does hide the 'shallow threats' at the moves it *prunes* (which are not >the same as the ones you apply the razoring to). > Razoring doesn't hide "shallow threats". It hides threats that are deeper than the remaining depth minus 1 ply. Note that we are talking about a specific move X here. Razoring searches X to a shallower-than-normal depth. Forward pruning throws X out without any consideration at all. _clearly_ searching X with any sort of search misses fewer threats than not searching X at all, wouldn't you say??? And that is the point of the definition of forward pruning. Forward pruning hides _everything_ because the move is not even looked at whatsoever. >-- >GCP
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.