Author: Heiner Marxen
Date: 15:08:45 01/28/01
Go up one level in this thread
On January 28, 2001 at 16:05:49, Paul wrote: >On January 28, 2001 at 15:13:01, Heiner Marxen wrote: > >>You should not compare a playing program with Chest. >>They perform quite different jobs, even when the result from both >>is a mate PV. > >Yeah ... I hope we both know that I know that you know that I wasn't serious >about that, don't we? :/ Ok, Ok, now I got it 8-} >>That is, because another job with 350 MB is still running all the time. >>[The mate in 17 twin. My first guess of 8 hours turns out to be too small: >> it is at 32 hours, now, and will need 1-2 days more... just another guess] >>There is still enough memory left to add another 60MB Chest running, >>but not much more (I tried 120MB, and the system started to page as mad). > >Great ... you just did it again! Boy ... *now* I'm jealous ... :) >But actually they are triplets! José Antônio later added this one: > >[D]8/2p4Q/8/8/p7/K6p/2r5/1k6 w > >Mine is currently thinking about this mate in 20. Changed my pawnpush extension >for it, but don't have the patience to let it run for too long, so I'm pretty >certain I'm not going crack this one. Yes, I did notice the third one. But with the current version of Chest I'm not going to attack it. 20 moves is too deep, at least in this case. I will reconsider to attack such deep jobs, when I have started to implement functions that can cut off complete branches, independant from the requested depth. That requires a proof that one side cannot loose. Without that it does not make much sense, for me. >What I do find surprising is that even with 1 or 2 ply extensions my search >doesn't really explode! Well on some positions it does, but in this one it's not >happening. Probably hash and repetition check control it somehow. Pretty funny >... > >Paul Since Chest has just two values (solved/not solved) it is not as vulnerable to explosions as others, but it can happen also. Hashing and ETC helps a lot, of course. With the mate in 17 I see increasing factors to reach the next depth: until depth 14 the factors decrease down to around 2, but then are followed by 3.4 and 5.5. That is a clear indication that many shallow mates are seen near the leaves, which cause alternate black moves to be searched. We'll see. But I will not try such a beast yet another 3 moves deeper. That you manage to have 1 and 2 ply extensions is another story. I am also surprised, but from your followup I see, that some luck is involved :-( Heiner
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.