Rybka Chess Community Forum
Maybe, you guys could try to check this: If running Deep Positional Analysis in Fritz9 GUI it seems that the PH file is not or not reliably updated, although it should be. I will try that again tomorrow to perhaps find out more about it, but maybe this can be confirmed by someone else, too? Or perhaps guys with Fritz10 GUI try it, too?
Well, it really seems there is something, because when I tried today she couldn't remember anything from yesterday....
Okay, two things seem to be sure: PH is in analysis mode not updated on the fly, while Rybka is calculating, but only when she is unloaded afterwards. Also, she does not remember the Deep Positional Analysis stuff, even though there is an update here, too, when she is unloaded. This is valid for FritzGUI 9, but might apply for other FritzGUIs, too, perhaps even GUI 11. Can someone confirm?
I haven't done much testing yet, but I was under the impression that you needed to click Save Hash to save data to the PH file.
No, save hash is for the preserve analysis part.
No, save hash is just for saving a copy of the hash.
Uh yes, that's what I meant but did not write down ;)
Saving hash is a different function which I haven't used yet. It's aimed at users who just want to save their hash after analysis for further use tomorrow - and then forget about it. Persistent hash is being stored for ever in a different file so it can be used if the position comes up again next Christmas. And as far as I understand it, PH is also stored in a different way, which doesn't take much space.
Thanks for looking into it, I'll wait for further comment. In my tests (in command-line mode :)), this worked.
This issue seems to be the more hazy, the longer I fight with it. Meanwhile it seems that, yes, there is being saved something to the PH file. But my problem is: If you let normal infinite analysis run with persistent hash, then exit the GUI and later get back to it, the analysis instantly jumps to the point where you left, be it depth 15 or 20 or whatever. Now one should think the same would happen with Deep Positional Analysis. If you start it again at the same position the results should come more or less instantly. But they don't...
This should work as you describe. Maybe the DPA search depths are below your write depth ?!
Nope, search depth 17 - definitely above write depth, which is default = 10. Since DPA actually is a whole load of analyzes on changing positions, instead of just one on one position, maybe this could at least be part of the answer? Perhaps not everything is stored?
> Nope, search depth 17 - definitely above write depth, which is default = 10. Since DPA actually is a whole load of analyzes on changing positions, instead of just one on one position, maybe this could at least be part of the answer? Perhaps not everything is stored
Ok, thanks. The PH should store everything.
You use the two terms "normal infinite analysis" and "deep positional analysis". Do you know if there is any difference between the two?
Well, normal analysis is just analysis: let it run and see what comes out. DPA is a FritzGUI feature, actually it is sort of an analysis tree with predefined branches. Not unlike to the IDeA feature of Aquarium, but less flexible. You use it to deeply analyse a position and find out if there are hidden trapdoors or hidden hits.
Kapaun, thanks for the explanation. Has your experience been that DPA is any more useful or efficient than infinite analysis?
Yes, definitely. It's time consuming but especially in unclear positions you get quite a good idea about what's possible.
what is the position you are doing DPA on?
Any. Take the one you like.
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill