Author: Amir Ban
Date: 16:14:37 02/23/00
Go up one level in this thread
On February 23, 2000 at 10:22:04, Randy Schmidt wrote: >I have always been troubled by the fact that Junior does not display ply depth >in the same manner as every other program out there! When you run J6a in it's >own interface or in Fritz6s' instead of seeing 9/21 in the play depth and 2/35 >in the move being considered, you see depth = 20 . > >9/21 means that a program is to 9 ply in the main search, and in some lines has >gone to ply=21. > >2/35 means that the program is now considering (at depth=9), the second of 35 >possible moves in the given position. > >depth=20 as much as I can tell means very little at all. I had asked this >question about J5 a long time ago, and Amir said that if you divide by 2, >you get the actual depth. I can tell you that it is very doubtful that this is >the case, and I have run dozens and dozens of examples. My own opinion is that >you should multiply the depth by 2/3. > >I guess I have 2 questions. What DOES depth mean in J6 and why is it not the >case that J6 does not follow every other program out there? I never accept the >analysis of a computer move that is not at ply=9, second choice. With Junior >you never know when you are there. When preparing openings, I cannot use Junior >for that reason. ( I use CAP now ) Thanks Dann Corbit! Junior counts line depths in "half-plies". Approximately this means that "regular" moves count for a "ply" (i.e. two "depths"), while "special" moves count for a "half-ply" (i.e. a single "depth"). If you insist on translating the Junior depth into a ply depth, I would suggest to divide the Junior depth by 2 and round up. For a maximum ply depth, you may take Junior depth. Both numbers are not strictly true: lines are sometimes shorter than the minimum ply depth, or longer than the maximum ply depth as I specified here. The reason that I don't give the exact minimum/maximum ply depth is not because they are secret, but because the algorithm is constructed along lines where these limits are not explicit and of no real significance. Besides, because many program display search depths as min/max ply, it is wrong to assume that these numbers have any objective meaning. A null-mover that displays minimum 10 plies is in fact not guaranteed to see even some 2-ply variations. If you believe that Fritz ply 9 = Rebel ply 9 = Hiarcs ply 9, you are falling to an illusion. None of them are equal to either Junior depth 9 or depth 18. The depths given by various programs condense something quite complicated into a single number. This number can be used for comparison within a single program, i.e. there's no doubt Fritz 11 ply is more than Fritz 10 ply, and Junior depth 14 is more than Junior depth 13. I would not give this number an objective meaning as in "9 ply is needed for reliable analysis when preparing openings". This is IMO of course, and you should follow what works best for you. Amir
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.