Author: leonid
Date: 18:50:55 03/08/00
Go up one level in this thread
On March 08, 2000 at 11:19:04, Tom Kerrigan wrote: >On March 07, 2000 at 14:03:33, leonid wrote: > >>In alpha-beta I am curious just about one moment. Attempt to find this tiny >>explantion in very general description could take too long time. Will try >>shortly say what is the question. Maybe you, or somebody else will grasp my >>exlantion and will privide me with expected answer. > >If you're searching the moves at the root of the tree, and you find a score >between alpha and beta, then that's your best move. No need to search the tree >twice. Actually, I can't think of any reason why searching the tree twice would >make the program go faster. You puzzled me here. Still it is possible that something very simple I can't see or can't say in the way that it could sound familiar. It is very nice that you indicated me this point. Even if I listen what people speak in this place I do my writing completely alone. Some entirely wrong or new decisions are possible. At the beginning, when I only found alpha-beta for positional logic, my logic searched upper ply in once search. It is from where I thought for a long time that my logic search in each ply as many nodes as it was in previous single search alpha-beta. >>I think that separate mate solving unit is more that useful. Reason for this I >>see in the fact that mate solver (at least mine) don't go after material >>exchange at all. Its speed. In this logic only the fact that king is under the >>fire, or not, is taken in consideration. This permit to eliminate all possible >>"allergic reaction" that material exchange logics are so prone. Many good chess > >But still, if you spend one minute looking for a mate and you don't find one, >then that's one minute that you've absolutely wasted. Tom, you know the address of my program on the Net. Take the last version (mate solver was not changed for the last 4 years anyway) and try any position. You will see on the right side two lines of time that two logics spend, one after another, to find its solution. Mate solver say its time below. Compare time spent by positional logic and mate solving logic. This will give you some idea what the real speed of this logic is all about. Mate solving logic (default level) even make double search. First 6 plies deep brute force, and second mixed brute force and through speedy logic 14 plies deep. All the time that mate solving logic think on the 400Mhz it represent less that 0.055 sec. Since the speediest "positional logic" can see 6 plies by brute force only in 1 or 2 seconds, 0.055 spent by two search in "mate solving logic" is not that much. Leonid. >-Tom
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.