Author: Robert Hyatt
Date: 12:37:29 04/03/00
Go up one level in this thread
On April 03, 2000 at 15:28:51, Jeremiah Penery wrote: >On April 03, 2000 at 08:57:27, Robert Hyatt wrote: > >>On April 02, 2000 at 22:39:06, Tina Long wrote: >> >>>I am suggesting that "If the "score" of the best 3 moves is not too dissimilar" >>>increasing the analysis time from about 40min to 4 hours for 2nd & 3rd best May >>>find one of them to be actually Best. This is the Benefit to weigh against the >>>Cost of the 1st best move being analysed to 19 ply rather than 20 ply. >> >> >>This is the same thing as adding 2+2, but then going over and over the result, >>checking yourself, for the next hour. Is your answer any more accurate? Or >>did you just spend a lot of time? >> >>If the second move was searched deeper, that would be different. But _all_ >>moves are searched to the same base depth. > >One thing (probably of many) that can mess this up a bit is null-move pruning. >Take WAC #141 for example (the Queen sac for mate) - Pretend that Qxf4 is the >second move on your list, and so would show up that way in k-best mode. If you >do a normal search, it might take 10 plies to see that Qxf4 is best. However, >if you do K-best, you'd see at ply 8 that Qxf4 actually has a higher score. > >I've seen similar things happen tons of times with Crafty. It finds move X >after a normal search to some depth, but when you force it to search some other >move, it can find a higher score at the same depth - Why didn't it find the >'better' move the first time? But K-best will find it, most likely. > >I'm not arguing for K-best to be used in any serious game(s), or that it is any >better than normal search. I'm just pointing out that in some cases, it can >actually produce something better, due to the imperfections in pruning methods >and such. > >Jeremiah No doubt. But how many real positions do you search, vs positions where a queen sac wins? Best to put your money on the cases you expect to see most often?
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.