The only other interaction usable by the non-advanced IDeA users seems to be "Good Move" or "Bad Move." However, I rarely uses those, since IDeA by default spends 90% of its time on the top 1-2 moves and ignores the lower scored moves anyway.
Presumably top users like Moz and Dadi know how to interact with IDeA, but the average IDeA user simply runs it without interaction to build a tree and very occasionally gets new lines to examine in IA.
> what interactions to use? I have never found really useful interactions.
There are many methods to choose from and you shouldn't be afraid to experiment a little. That's the best way to develop your interaction skills. I could give you a long list of things to try, but I suggest you experiment first with the Infinite button in the Interactive group. It's a method for running infinite analysis in any position you choose and it even allows you to analyze many positions simultaneously.
For a description of Infinite, see Sending Games and Position to IDeA.
> IDeA by default spends 90% of its time on the top 1-2 moves and ignores the lower scored moves anyway.
This was a common complaint in previous versions, but it is greatly improved in the free update which will be released soon (and of course also in Houdini 2 Aquarium). I still haven't seen this problem come up in the new version.
> how would one get that into the IDeA analysis: by the Current Position method?
Insert and all positions.
> I assume that that button stops IDeA while it is being used.
No, IDeA keeps analyzing while infinite analysis runs. Therefore it's best to select <IDeA engine> after you click the Infinite button. This means that Aquarium will wait until there is a free IDeA engine and assign it to the task. This happens as soon as the next IDeA task is finished.
If you want to use a different engine for the analysis, you must make sure that there are enough free cores free to run it. I wouldn't try that until I understood how the Infinite button works.
> Suppose one finds a good line with the Infinite Button, how would one get that into the IDeA analysis: by the Current Position method?
You don't have to do anything. It happens automatically!
(Click the Engines button in the Home - IDeA tab and make sure that "Add analysis to IDeA tree and generate alternatives" is selected)
> there is no way to stop IA in IDeA
First select the infinite analysis task in the queue (it's the one with the white background; see the image below) and then click Finish Task.
We need to know what the different colors mean in the queue.
White: infinite task
Green: active tasks
Red/Pink: next tasks in queue
This method is half successful. It does make the 5th engine (used for IA) inactive, but I still have 5 engines loaded, which is incorrect, as I have only 4 cores. And using "unload inactive" would then leave me with only 1 core for IDeA.
So one still has to stop IDeA and start all over again after finishing IA within IDeA.
Thanks, Dadi, I will make sure to select <IDeA engine>.

This setting applies to all your projects. If you did that, then the next thing to check is that the project you are running accepts positions from infinite analysis. Click the Edit button to bring up the IDeA Project Properties and select "Insert infinite analysis from linked games" as shown here.

Set "Min. time" to 20 seconds and "Min. depth" to 15. This means that an infinite analysis task will start sending positions to IDeA when it has run for 20 seconds and reached depth 15 or higher.
After you have done this, start IDeA and make sure the project is active. Go to the end of a branch in the project tree and click Infinite to create an infinite analysis task. You should see red squares being added to the task queue after 20-40 seconds.
If you want to browse the variation that the engine is displaying in the Stage Status window, you can click the Insert link to add it to the notation.
I have never had this last option set before. I did have "Add analysis to IDeA tree and generate alternative tasks" in the IDeA Engines Setup based on a comment that it was needed to add IA analysis to the IDeA tree.
So it seems that all my IA in DB mode has not been added to the IDeA tree?
And that one needs to check both of these options to add IA analysis in DB mode to the IDeA tree?
Very confusing! I have to believe that not 1 on 20 IDeA users knows about BOTH of these options.
"This means that an infinite analysis task will start sending positions to IDeA when it has run for 20 seconds and reached depth 15 or higher.
After you have done this, start IDeA and make sure the project is active. Go to the end of a branch in the project tree..."
I don't understand this at all. I thought one could run Infinite at any point of in the IDeA tree.
And what is "the end of a branch in the project tree.."?
Is that the last ply with an IDeA score or the last ply in the tree?
"and click Infinite to create an infinite analysis task. You should see red squares being added to the task queue after 20-40 seconds."
If you want to browse the variation that the engine is displaying in the Stage Status window, you can click the Insert link to add it to the notation.
> I thought one could run Infinite at any point of in the IDeA tree.
You can.
> And what is "the end of a branch in the project tree.."?
> Is that the last ply with an IDeA score or the last ply in the tree?
I only suggested that you go to the end of a branch to see the results more quickly. If you run infinite analysis somewhere in the middle of the tree, you may not see any positions being added (if the PV moves already exist in the tree).
an active task into an infinite analysis task by clicking it in the IDeA queue, as
shown below.
There are two active tasks in the IDeA queue shown in this image. I have
selected the second from the left by clicking it with the mouse, resulting in
four actions being displayed below the queue (underlined blue links). One of
them is Infinite and when I click it, the task is turned into an infinite analysis
task and the background color turns to white. When the analysis time and
depth reaches the limits specified in the IDeA Project Properties (Min. time
and Min. depth), positions from the analysis line will start to be added to the
tree and alternatives will be generated. The engine will continue analyzing
until it is manually stopped by clicking the Finish link, which is displayed for
all active infinite analysis tasks."
This 2nd method of using IA within IDeA is quite interesting. Still, it is confusing on how to use it. When one sees a green active task, one does not know what line that is until one clicks it, after which it may turn out that one has no interest in I.A. for that line.
Is there a way to know what line is being examined by an active (green) task before clicking it into an IA task?
If you click a green "a" the task is shown (as Dadi showed), but it is not turned into IA, that happens only if you click the blue underlined "infinite", otherwise the task stays in IDeA until finished (time and/or depth).
regards,
max
However, clicking on "IDeA1-Rybka 41. SS E42 x64" does result in adding a 5th engine (even though I have only 4 cores) to the IDeA analsys and also adding an "f" task with a white background (IA task.) So that is an improvement, but even here, I do not have the 4 underlined operations offered.
setup your IdeA project with e.g. 60s and depth 20 (so tasks stay longer, just for testing)run IDeA, maximize IDeA window and click on a green field "a", now you must see the blue underlined options.
Kind regards,
max
Thanks, CMA
> I am unclear how Infinite is supposed to work in IDeA: I see the usual IA display, as in the Notation pane/Database mode. However, when I tried to make a move, that was impossible? Is that the way it is supposed to be when using Infinite in IDeA mode
display the idea project you want to run infinite on. click infinite. a task will be generated. click on the white square when it shows. then you should have the display I'll attach. From there you can use the menu at the bottom..
Sorry didn't know Dadi was answering you. You should have enough details now.
Dadi, I have been using the Infinite button and Current Position button from time to time. However, I often encounter the situation where IDeA is down to only 1 task to do, and is using only 1 of 4 engines.
Are there other methods than the above two to create tasks for IDeA?
Would two or more nodes of the same project be as good a solution as two or projects?
Thanks,
CMA
> Would two or more nodes of the same project be as good a solution as two or projects?
No, that wouldn't solve the problem you are referring to. Regardless of how many root positions you define, there is only one queue for all the project tasks.
Thanks, much appreciated.
Build 490 is a really superb achievement for IDeA and a tribute to the Chess OK Team (Victor, Dadi, and the others.)
CMA
Most of your processor's time will be spent on the primary project. When the cores are not busy with the primary project, IDeA will work on the secondary project. When you stop IDeA, the primary project may have 500 "analyzed tasks"; while the secondary projects will have only 40 "analyzed tasks."
I have, however, had the problem that sometimes the secondary project will show an absurdly large # of "analyzed tasks", e.g., 62000 for the session. I never know if I should delete that project and start all over again.
On the first part: I would like to know more generally how IDeA distributes its resources when one has several active projects. Assume that CPU usage is 'active' in all projects - will IDeA first perform one task in project A, then one in project B, ...? And is there a way to tell IDeA something like 'First analyze all (manually added) tasks in project A, then all (manually added) tasks in project B, ...?
The second part of what you write leads me to the question of whether there are additional risks involved. Does this only happen when you have more than one active project? And what does the tree show? Does it also give much higher numbers than there are actually positions in the tree?
When I have the errors I mentioned, so far it has happened only when I have more than one active project. The #s in the "N" column for the tree are wildly inflated.
> will IDeA first perform one task in project A, then one in project B
The only guarantee is that it will distribute that analysis time equally between the projects.
> is there a way to tell IDeA something like 'First analyze all (manually added) tasks in project A, then all (manually added) tasks in project B, ...?
No.
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill