Author: Bob Durrett
Date: 09:10:11 12/18/03
Go up one level in this thread
On December 16, 2003 at 19:56:21, Mike S. wrote: >On December 16, 2003 at 18:10:01, m.d.hurd wrote: > >>Is there any way to analyse or blunder check an small opening book I have >>created within Fritz gui ? > >Run an engine match with very many games using that book, but set the move limit >in the match dialogue to "1" only. This will produce short "games" where only >one "out of book" move will be calculated by the engine each. I think, for real >blunders to spot only a few seconds per move will be sufficient, i.e. just >enough to reach ~8 plies depth. If high evals occur, you'd have to investigate >further anyway. > >This doesn't check every move though, but only the resulting positions at the >end of the book variants. Maybe it's recommendable to repeat that test with >different F4 book options. > >The F1 help says about this: > >"Move limit: This limits the length of individual games. The number gives the >number of ply (half-moves) after leaving the openings book. In the database, the >games are stored without a result, but with an evaluation symbol. > >Tip: If you want to check an openings book, set the move limit to “1”. The >program will go to the ends of different lines, calculate one move and save them >with an evaluation symbol." > >Regards, >M.Scheidl This is a great idea due to it's simplicity and speed. You might wish to do it using several different engines to get added confidence. In a bulletin I posted here awhile back, I noted that the engines are not consistent in this regard. Comparison of engine results is therefore useful. For example, Deep Junior running on a single processor machine will typically give position evaluation values quite different from those of other engines. Junior is also an oddball. Note that it is also possible to close or disable the chess program's opening book. Then the engine would analyze all moves in the opening to be examined including the initial position of each opening line. One could do "overnight analysis" of the entire collection of lines to be examined. [To make the findings useful, it would be necessary to set the computer to take a long time for each position examined.] Maybe it would take a month but the results might be worth the time expended. One problem with this is that "overnight analysis" does not return numerical position evaluation scores. To get them, additional work would be necessary. Whether or not this much effort is justified depends on the purpose of the exercise. If only a "quickie" blundercheck is needed, then the effort would not be worth the trouble. However, one must wonder whether or not doing a blundercheck is really necessary. Bob D.
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.