Author: Uri Blass
Date: 01:31:29 08/25/02
Go up one level in this thread
On August 25, 2002 at 00:11:45, Robert Hyatt wrote: >On August 24, 2002 at 17:59:37, Uri Blass wrote: > >>On August 24, 2002 at 17:46:18, Robert Hyatt wrote: >> >>>On August 23, 2002 at 14:48:19, Brian Richardson wrote: >>> >>>>On August 23, 2002 at 14:41:11, Dann Corbit wrote: >>>> >>>>>http://www.talkchess.com/forums/2/message.html?41900 >>>> >>>>This looks like an interesting variation on internal iterative deepening IID, >>>>but with generally deeper sub-searching. >>> >>> >>>Don't yo umean _shallower_??? >>> >>>IE IID does a depth-2 (recursive) search, while Ed proposed depth/2, which >>>is significantly less... >> >>It is dependent on the qusetion if depth<4 or depth>4. >>If depth=4 both searches are the same. >> >>Uri > >Not depth< 4 but only depth==1,3,4... > >ie for depth=1, depth/2=0, depth-2=0 so they are the same You assume that 1/2 ply and 0 ply are the same. I am not sure about it because there may be partial extensions. > > for depth=2, depth/2=1, depth-2=0 so my method does a deeper search. Ed's method does a deeper search here. > > for depth=3, depth/2=1, depth-2=1 so they are the same. > > for depth=4, depth/2=2, depth-2=2 so they are the same. > > for depth=5 and above, mine is always deeper. > >However, I think the point is that at positions closer to the root, move >ordering is more critical. A deeper search will produce a better move. The question is how much nodes do you waste on searching. Another possible difference: I am also not sure if Ed searches for the best move at reduced depth or for moves that are good enough to get a score above beta but may not be the best at reduced depth. You do not search for the best move at reduced depth by your internal iterative deepening because you do not use alpha=-infinite,beta=+infinite. 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.