Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Aquarium / Bug when combining IDeA analysis with IA
- - By kirill57 (**) [us] Date 2017-10-19 14:28
Usually I feed my IA analysis to the IDeA analysis to generate alternatives and prolong variations. Recently I begin to see  that some ridiculous evaluations (like -300 ,+5.43 and so on) are popping out of nowhere screwing  tree evaluations. If I want to keep to rely on the IDeA tree evaluations I had to spend a lot of time to check all the lines and reevaluate moves with screwed evaluations. I do not recall such problems in Aq2016 (I am using Aq2017 now). Not sure what is going on...
Parent - By kirill57 (**) [us] Date 2017-10-19 16:20
And another question. Some of my "infinite analysis" does not show the "f" icon in the stage status window. How can I fix that?
- - By kirill57 (**) Date 2017-10-19 16:01 Edited 2017-10-19 16:07
Here is an example: I see a HUGE difference in IDeA evaluation compare with IA evaluation. I track it down the line to record a new evaluation and hit "Minimax now". See what happens:




First image is where I begin, attachment what I see after minimax.
Parent - By pawnslinger (***) [us] Date 2017-10-19 16:17
I have been complaining about this for awhile now.  I had to stop feeding analysis from IA into the IDeA tree.  The problem stems from the fact that many engines generate these spurious evaluations, why I do not know.  They do not do it in a pattern that I am able to predict.  But once fed into the tree, it messes the tree up for a long time... as you have found out.

Until this bug is fixed, you must change your IDeA parameters so that only the moves are accepted from the IA engine... not the evaluations.  I realize this is a big penalty, but I know of no way to fix the problem otherwise.
Up Topic Rybka Support & Discussion / Aquarium / Bug when combining IDeA analysis with IA

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill