Author: leonid
Date: 14:28:12 03/06/00
Go up one level in this thread
On March 06, 2000 at 16:18:57, Tom Kerrigan wrote: >On March 06, 2000 at 14:28:58, leonid wrote: > >>Here not everything is clear to me. Are you putting all the time checking moves >>at the head of the moves ligne or not? In my mate solving logic it is all the >>time with exception of two plies. For now I don't remember exectly advantage > >Yes, I did it for every ply. > >>that this alignment of moves give but it was done only when statistics said to >>me to do so. I never did anything in my game only after my personal taste. > >Like I said, I'm sure the heuristic works great for mate finders. Just not >regular chess programs. > >>>(In other words, there is a good reason why nobody does this in their chess >>>programs.) >>Maybe everybody is right here as well as everybody is wrong. Even to write mate >>solver for the chess game sound so new and wrong to everybody. For me it is very > >But why should programs spend their time searching for mate in positions where >there's no mate to be found? And why have a special search for mate when the >regular search can find mates perfectly fine? > >-Tom Hello, Tom! It was real fun. I thought that I spoke to the Heiner and was astounded when he said that putting the checking moves at the head of the line had no sense whatsoever. This is why I have asked before if you use in your mate solver my kind of alignment. Now about regular chess program, and speaking no more about alignement of moves in mate solver. I just recently remembed you when I found (used special tool for seeing the numbers) that actual number of nodes that logic see in each ply is not like I expected before (between 5 and 7) but only somewhere around 2 or 4. Mistake from my side that I have forgotten that in alpha-beta we see each ply two times. Exception only some positions that have only one response in the frist search. And if in each ply we see only around 2, 3 or 4 moves, your logic became at once very solid. Why even use legal moves inside of each ply and spend some time for alignement checking moves first. Using legal moves inside of each ply had some sense only when you use at least 4 or 5 moves from each chain of moves. When I started producing the illegal moves and where no alignement of checking moves first was executed, I found just this strange result that I metioned before. In more positions have the advantage of time but general time disadvantage. This is probably derive from the logic that checking moves must go first. In some positions advantage is dramatic when checking moves goes first in the chain of moves. It is still possible that I did not tried enough good number of positions in order to reach the secure result. One interesting conclusion also from the fact that inside of each ply logic see only around 2, 3 or 4 moves (very often just one move). Probably in order of reducing actual branching factor to more good one it must be done in second revision. There practically you must see what moves between the "good moves" is the best one. In this case some limitation of the moves seeing inside of the final revision could be done without putting that much the final result to big danger. Leonid.
This page took 0.01 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.