Page 1 of 1

akmc script - kdbquery

Posted: Wed May 02, 2012 4:28 pm
by pr-dub
Hello.

If anyone is able to help me out with a few questions, it would very much be appreciated.

I've been attempting to run the provided example for the the adaptive Kinetic Monte Carlo script, however I've run into a few problem. I'm not certain what the problem is but currently I think it might be to do with the kdbquery.pl script.

When running akmc.pl, after the performing the initial minimization and creating the folder st0001, the output reads:

Create kdbmatches from kdbdatbase ... ...
Use kdbmatches to prepare saddle point searches... ...
Warning: In UseKdbMatches: the number of saddle guesses Done. is not a digital number.
0
In the current step 2, state st0001, we just prepared 0 saddle point searches from kdbmatches in st0001
Preparing saddle point searches by recycling good saddles from the previous state (if any).
Previous state: st0001
In the current step 2, state st0001, 0 good saddle points were recycled from state st0001.

I'm guessing that the problem is with the line 'Warning: In UseKdbMatches: the number of saddle guesses Done. is not a digital number.'

I've tried running kdbquery.pl by itself, and in these instances the output will read:

Calculating Car atomic distance table... done.
0, 0, 0
Done.

It would appear from this that akmc.pl is reading the third line of the output expecting something else to be there.

The main question I have therefore is what is the exact nature of the kdbquery.pl script? Is there a specific file format that the CAR needs to be in so that it is read properly? Also, is there possibly an example of the output that kdbquery.pl should produce so I'd know how it should look?

If anyone could provide some information about this that would be fantastic. Let me know if you need more information from me.

Thanks in advance.

Re: akmc script - kdbquery

Posted: Wed May 02, 2012 5:58 pm
by ryeterrell
Hi pr-dub, let me see if there is a bug in my code. I may have committed some debugging statements that have confused the akmc.pl script.

Have you looked into our new Eon akmc (http://theory.cm.utexas.edu/henkelman/c ... index.html) code? It can also interface with VASP, and is perhaps under more active development.

Re: akmc script - kdbquery

Posted: Wed May 23, 2012 9:13 am
by pr-dub
Hello ryeterrell, thank you for the suggestion of the EOM akmc script, however I feel that the original akmc script would work better with my previous work.

I have been looking at kdbquery.pl script a bit more. It seems to be failing at line 169, which is the main loop over all directories. It never enters the loop and instead goes straight to line 1166 at the end of the script. Would you have any ideas why this is? If you had any suggestions about where to go from here I would be very thankful.


EDIT - I've looked a bit closer at what was happening, and now it seems the problem is with the subroutine 'getProcessDirs' in the kdbutil.pm script. At the moment, it isn't returning a value. Should it always return a value, i.e. will it only return an answer when the akmc.pl script has advanced enough steps? Or is it normal for it to not do so.

If it the case that getProcessDirs should be returning something, what prerequisites/inputs are necessary.

Thanks again for your time.