Forum: Forums topas RSS
Saving results using parametric refinement
Florian Frank #1
User title: Florian Frank
Member since Jul 2019 · 4 posts · Location: Graz
Group memberships: Members
Show profile · Link to this post
Subject: Saving results using parametric refinement
Hi everybody,

I recently performed a parametric Rietveld refinement of in situ powder measurements (about 440 patterns).

I followed this tutorial https://community.dur.ac.uk/john.evans/topas_workshop/tuto… and adapted the input file(s) according to my data.

The refinement works fine for a reduced number of patterns and gives reasonable results.
However, the time it takes TOPAS to save the "results.txt" output is very long (depending on the number of patterns used). After the parametric refinement for all patterns is finished, I had to cancel the saving process after 1.5 hours.

For a reduced number of patterns (about 120) it takes about 10 min to save the results file.
With the data provided in the tutorial it also takes about 10 min to save the results file.

TOPAS is in "not responding" mode after the refinement process. It still uses about 1.2 GB of RAM.

Is there a way to speed up the saving process?

Best Regards,
Florian

P.S.: I'm using TOPAS-64 v6. I didn't attach any of my input files, since the same problem arises with the files provided by the tutorial.
johnsoevans (Administrator) #2
User title: John Evans
Member since Aug 2009 · 219 posts
Group memberships: Administrators, Members
Show profile · Link to this post
Dear Florian,

It shouldn't take that long.  Do you have do_errors turned on?  You might try without errors.

John
AlanCoelho #3
Member since Aug 2011 · 288 posts
Group memberships: Members
Show profile · Link to this post
John is probably correct; how many independent parameters are there and is do_errors ON.

If there are more than about 5000 parameters then the singular value decomposition matrix inversion used (when calculating errors) may well be taking the time.

In such cases use the boot strap method of error determination.
Florian Frank #4
User title: Florian Frank
Member since Jul 2019 · 4 posts · Location: Graz
Group memberships: Members
Show profile · Link to this post
Thank you both John and Alan for your answers.
Indeed, the error calculation takes it's time (for all patterns I have 10573 parameters). I will try and implement the boot strap method.

Cheers,
Florian
Close Smaller – Larger + Reply to this post:
Verification code: VeriCode Please enter the word from the image into the text field below. (Type the letters only, lower case is okay.)
Smileys: :-) ;-) :-D :-p :blush: :cool: :rolleyes: :huh: :-/ <_< :-( :'( :#: :scared: 8-( :nuts: :-O
Special characters:
Go to forum
Not logged in. · Lost password · Register
This board is powered by the Unclassified NewsBoard software, 20120620-dev, © 2003-2011 by Yves Goergen
Current time: 2019-11-22, 08:51:39 (UTC +00:00)