My first attempt:
I assigned via Advanced engine Options a tree to each engine i want to use.
Then starting Idea with 5 engines, engine trees added as columns to IdeA tree configuration.
Analysis adjusted to ridiculous 1s and 3 to get fast results.
But columns of engine evals remained empty (settings of Add analysis to Idea tree or Add moves and evaluation to IdeA tree)
IdeA column was filled with results (of what engine?? all? the first?...)
Trying to analyse with IA (game analysis, 5 engines), no filling of my columns, even with only 1 engine.
If I analyse manually with IA, each engine generates an output to it´s tree. But the result isn´t shown immediatedly, only after going foreward or back a move the evaluation can be seen in tree.
What did I make wrong?
Any useful hints appreciated.
regards
max
back in the tree and then forward for the evaluation to appear or change engine.
Andrew
> A user in the german part of this forum asked for an output of each engine in IdeA tree.
there is already such a file. For every project in {AquariumData}\idea\Results\projectname.epd
If you mean output an epd file
> Then starting Idea with 5 engines, engine trees added as columns to IdeA tree configuration.
first of all i cannot understand why would anyone want a separate file for every instance of the SAME engine.
Second, since those engines are not REAL, but they are instanced on the fly, i do not expect for THOSE engines to work (Game analysis uses the same technique)
> first of all i cannot understand why would anyone want a separate file for every instance of the SAME engine.
I think he meant 5 DIFFERENT engines.
I don´t want a set of EPD but a tree with the results of different engines.
In Dadi´s article at chesscafe.com I saw this picture:
I hope this picture can clarify what my poor knowledge of english language failed.
max
Previously, users used to make mistakes in the settings leading to lot of problems, so in recent versions, IdA tree configuration changes have been restricted. So you have to go to database view, and then create / select the appropriate tree configuration and columns.
The picture shows the result of multiple engines with IA (!), but it don´t (!) work for IDeA, because no 4 different engines runs in the same time through all (!) variations in IDeA (they solve only a part of variations) and no separate output pro engine is displaying in the tree. Is that the point?!
Yes, I also want to see 4 engine-specific analysis with the same single IDeA project in the tree. I possess a quad computer. Okay, I could organise 4 projects one after another to achieve the same ...
Greetings, Frank
As Radost sayed, it works in IA but not in IdeA. Have you tried it??
The feature of comparison of engines is new with Aquarium 2011 (as described in Dadi´s article), so I tested it..
Let me show you Pic1 after (short analysis with IdeA) and Pic2 after short analysis with IA. Can you or someone else solve the puzzle??
Regards
Max
You may be using 1 as an engine multiplier (so you are saying hey... i am not copying engines on the fly) but i am pretty sure the algorithm is general for 1 to 100000 engines.
Just a guess.
Maybe it would be nice to have some subscript which shows the depth or analysis time under the eval of the IA tree on the tree configuration, a tool tip which shows the pv when mouse is moved on the IA's eval cell and a right click function which puts the IA pv on the notation.
we don´t know the statistical power of the results shown in the picture above.
With this topic I wanted to point out that (only for me?) this new feature is yet not useful.
1. you can´t see the depth and PV (in my example i used very short analysis time and poor depth!); in IdeA the number of positions and the statistical functions provide a better view of the evalution
2. it should be a feature of IdeA as suggested by Dadi´s article but it is only a feature of manually IA, no automatic generation of results.
If Dadi can´t clear the mystery we have to hope for an update
max
A little knowledge of, or control over depth would transform the usefullness of this feature.
> A little knowledge of, or control over depth would transform the usefullness of this feature.
You will be able to see the depth in an upcoming (free) update of Aquarium, although it won't be included in the first Aquarium 2011 update which will probably be available as an unofficial release tomorrow.
In spite of not being able to see the depth yet, I already find this feature invaluable and I think that anyone who tries it will agree that it's a big step forward compared to not having the evaluations so easily accessible. Of course you can't know anything about the depth when you see analysis done by someone else, but you should have some knowledge about your own analysis. There are also various methods you can use to keep track of your analysis when it's important to know the time or depth.
>Of course you can't know anything about the depth when you see analysis done by someone else, but you should have some knowledge about your own analysis.
There will be times that there will be some bogus on eval, so it really needs depth for reliability check and pv for confirmation. And I really don't see the point of using .hsh for seeing IA evaluations on tree config. Why not just use epds? It stores the eval, depth, time, pvs, and it is easily readable.
Columns showing Ho15: 1840 and Ho15: 0.29 seem to show positions analysed and Houdini's result;
but do you use IDeA or IA, the position count from epd, from tree config categories or whatever??
If you don´t tell me in which way that tree is build up, i will have no clue what your answer reveal...
But I think buffos answer has pointed out that "due to the way engines are initialized (and copied on the fly in memory) when used with IDEA and GameAnalysis" that it can´t be expected to get different results from different engines in IDeA while performing parallel analysis (in IDeA) .
After re-reading of Dadi´s article I see that I have to change part of my use of IdeA to :
Linking games to an IdeA project, running IDeA all the time, but analysing with IA the game(s) same time (>= 2 cores required for that procedure), setting "Add moves and evaluation to IDeA tree"; and I want to examine an option (that should resolve the problem of depth of analysis written in the tree) I have overlooked all the time: see picture
Thx for all help and hints,
Yours sincerely
max
> Columns showing Ho15: 1840 and Ho15: 0.29 seem to show positions analysed and Houdini's result;
> but do you use IDeA or IA, the position count from epd, from tree config categories or whatever??
> If you don´t tell me in which way that tree is build up, i will have no clue what your answer reveal...
The pic shows comparative IdeA evaluations of 2 different engines: Ho15 and SF20. The numbers are the number of positions in each tree (Using a different tree for each engine during IdeA) and the corresponding evaluations. In IA, you do not get position counts and there is no easy way to obtain the position count from epd.
Of course on can run IdeA projects and obtain position count of the resulting tree, but you have to run it twice or more (one for Ho15, one SF20, and so on..)
My question was based on the idea of getting evaluations from different engines by only one IdeA project.
Buffos answer made clear what was my misunderstanding of generating Idea tree by engines.
The new article (Part two) of Dadi Jonsson in Chesscafe.com shows analysing like I mentioned above (derived by reading of his first article; I wish he had published it earlier but on the other hand I learned more by own thinking).
Kind regards
max
Where do you find 'Advanced Engine Options' i want to assign a tree for each engine, but cannot find the place to enter the name of the tree. This must be looking extremely stupid I tried my best but just could not find the place in Aquarium 2011
> Where do you find 'Advanced Engine Options
engines mode (the place where you install the engines). If the windows (panel) is not shown, just open it, by using the little arrow down left (where you find layouts)
Read here
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill
