Author: Robert Hyatt
Date: 13:28:09 12/01/03
Go up one level in this thread
On December 01, 2003 at 13:03:47, Les Fernandez wrote:
>On December 01, 2003 at 10:56:51, Robert Hyatt wrote:
>
>>On December 01, 2003 at 00:35:26, Les Fernandez wrote:
>>
>>>Anyone have any idea what the percentage is for a chess engine after say 90 sec
>>>of analysis that one of the top 5 suggested moves is in fact the best move? And
>>>what percent of the time would you guess that after an additional 90 sec that an
>>>alternate move is found and that it was not one of the top 5 choices after the
>>>first 90 seconds? And finally can crafty when it is done with its analysis
>>>report the top 5 highest rated moves and if not what engine can do that?
>>>
>>>Thanks,
>>>
>>>Les
>>
>>
>>1. If you use the annotate command, Crafty can show you the N best
>>moves, scores, and PVs.
>
>Hi Bob,
>
>Can you show me how to use the annotate command so that I can have the top 5
>best moves sent to a file?
>
Say you have a file test.pgn, and you are only interested in move
40, and want to see the 7 best moves after a 60 second search on each:
annotate test.pgn w 40-40 -1 60 7
For the paramaters, "help annotate" will explain, but here is the
excerpt from "the authority" crafty itself:
annotate[h] filename b|w|bw moves margin time [n]
where filename is the input file with game moves, while the
output will be written to filename.can. the input file is
PGN-compatible with one addition, the ability to request that
alternative moves also be analyzed at any point. to do this
at the point where you have alternative moves, simply include
them in braces {move1, move2}, and Crafty will then search
them also. b/w/bw indicates whether to annotate only the white
side (w), the black side (b) or both (bw). moves indicates
which moves to annotate. a single value says start at the
indicated move and go through the entire game. a range (20-30)
annoates the given range only. margin is the difference between
the search value for the move played in the game, and the best move
crafty found, before a comment is generated (pawn=1.0). time is
the time limit per move in seconds. if the optional "n" is
appended, this produces N best moves/scores/PV's, rather than
just the very best move. it won't display any move that is worse
than the actual game move played, but you can use -N to force
Crafty to produce N PV's regardless of how bad they get.
using 'annotateh' produces an HTML file with bitmapped
board displays where analysis was displayed.
>>
>>2. Lots of evidence has shown that in roughly 1/6 of the cases, going
>>a ply deeper will cause a program to change its mind and choose a different
>>move from the last iteration.
>
>Ok I was estimating that 1/6. What I would like to know is lets say the top 5
>choices are A,B,C,D,E after ~90 seconds. What do you think the likelyhood of
>crafty finding a different best move after an additional 90 sec that is NOT one
>of the top 5 choices we found in the first 90 sec considering all 5 choices?
Harder to answer. You might look at Ernst Heinze's JICCA paper. I believe
he measured that.
>
>Thanks
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.