Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Aquarium / Finding the good line in IDeA trees
- - By Riskakor Date 2020-12-12 16:18
Hey all, I have an issue with IDeA: it includes repetitions, without indicating this. I realize this sounds unclear, so as an example: in a certain position 32. Rab1 would be the correct move. IDeA will indicate this, but it might also include a variation that goes 32. R1a2 Kh8 33. R2a1 Kg8 34. Rab1

In this example, it is easy to figure out what the repetition is and what the correct 32nd should be. But if your project has reached multiple thousands of positions, it can become completely unclear what the correct move sequence is on which the engine bases its evaluation.

So my question is: is there a way to clear this up in the IDeA tree? Many thanks for your insights!!
Parent - - By mattchess (**) Date 2020-12-15 23:38 Edited 2020-12-15 23:47
The thing to remember is that IDEA is analyzing positions, not games. So when IDEA evaluates a position it has no consideration of how you got there (whether directly or by transposition).  In your example from IDEA's perspective move 32 Rab1 and move 34 Rab1 are identical. 

I think your only option is to manually color the move red to avoid spending further time on it in the future.  Here is a similar discussion:

http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?pid=581130;hl=repetition

BUT - there IS a way I think you can get what you are looking for in the notation window rather than the tree.

If you use the TREE --> Best Line from Tree function, it will write the best lines from that position based on your specified tree(s)....and THAT will indicate where there are repetitions that can be avoided by more direct moves.  It writes the lines into the notation window.  I use this all the time to see what lines are "best" based on my current tree and I extend them from interesting positions myself and add those new positions for IDEA analysis.  I *think* this gets you at least part of what you are asking, although not in the tree itself.  You will need to color the moves in the tree.
Parent - - By Riskakor Date 2020-12-16 10:00
Hi Mattchess, many thanks for your explanation! The 'Best Line from Tree' function I didn't know and it would help me tremendously. Only problem: I can't find it... Could this be a version issue? I'm using Aquarium 2019. Was the function introduced in a later version? In that case, I'd be happy to upgrade... Just want to make sure this is indeed bc of the version... :)

Thanks!!
Parent - By mattchess (**) Date 2020-12-16 22:29
In the menu ribbon select Tree and then it is one of the options in the ribbon on the right

Parent - - By Dadi Jonsson (Silver) Date 2020-12-16 20:14

> when IDEA evaluates a position it has no consideration of how you got there (whether directly or by transposition)


This is not correct.

1) IDeA sends the moves leading to the position to the engine (not the EPD/FEN for the position)
2) IDeA never analyzes the same position twice and each position is only stored once in the tree

This can lead to transpositions in the tree, but the illusion of endless repetition is just that -- an illusion. It's like circling a roundabout in your car. Even though you can circle it 10 times (as you can loop 10 times through a transposition in a tree), it's still just a single physical circle. To make progress you need to exit the roundabout. Same with chess position trees.
Parent - - By mattchess (**) Date 2020-12-16 23:06 Edited 2020-12-16 23:10
Dadi -

I am a bit confused by (1) but understand (2). 

I think the confusion is over what "the moves leading to the position" actually are :P  When you say "the moves" what moves are you referring to.  Are you referring to the best line in the tree leading to that position?  I think that is what you mean....but that is different than IDEA considering how someone actually got to the position in their specific game or circumstances.  So if someone does unforced repetition then IA will show 0.00.  But IDEA will still show the correct eval for the position reached by a better line (not the unforced repetition) assuming they do not go on to repeat.  Which is consistent with your (2).  That is what I was trying to say when I said IDEA evaluates a position without considering how the position is reached (whether directly or by transposition).  It will not recognize repetitions that are not forced.  Instead of "IDEA evaluates" I probably should have said "IDEA reports evaluations for."

Is that correct?  I am trying to understand what you mean by (1).

I think what this means is that IDEA evaluates positions, including consideration of lines leading to positions, when it evaluates the position but once it has been evaluated the minimaxed evaluation reported in the tree doesn't consider how you actually got to the position in the notation window.

Thanks,

- Matt
Parent - By Uwak (***) Date 2020-12-20 18:09
I think only positions are evaluated.  The different lines leading to the position are not evaluated. Line evaluations have no meaning.
Up Topic Rybka Support & Discussion / Aquarium / Finding the good line in IDeA trees

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill