[Ifeffit] Pseudo Voigt crash Athena

[Lawley Craig Richard (PSI)]

Dear Matt,

I will take the time to look into this. I saw the previous thread from yourself with the link to the installer and relative information etc.

Many thanks.
Best regards,
Craig

__________________________________________
Paul Scherrer Institut
Craig Lawley
OFLB/U109
Forschungsstrasse 111
5232 Villigen PSI
Schweiz


Telefon: +41 56 310 5054
E-Mail: craig.lawley at psi.ch<mailto:craig.lawley at psi.ch>
www.psi.ch<http://www.psi.ch/>

From: Ifeffit [mailto:ifeffit-bounces at millenia.cars.aps.anl.gov] On Behalf Of Matt Newville
Sent: Donnerstag, 2. Mai 2019 04:58
To: XAFS Analysis using Ifeffit
Subject: Re: [Ifeffit] Pseudo Voigt crash Athena

Hi Craig,

On Wed, May 1, 2019 at 10:51 AM Lawley Craig Richard (PSI) <craig.lawley at psi.ch<mailto:craig.lawley at psi.ch>> wrote:
Dear Mailing list,

I have run into an issue with the peak fitting section included in Athena. If I include a peak that is a Pseudo-Voigt and then save the project file. Upon trying to reopen the same project file, Athena becomes non responsive with the error message “PV is not yet implemented”.

This may sound like a broken record, but I strongly encourage you to try Larch and the XAS Viewer application for peak fitting of XAS spectra.  The ifeffit library is really old (has not been updated in 5 years), not supported, and its interface to peak fitting is primitive.

Larch (and the lmfit library it uses) and the XAS Viewer GUI are all supported. Peak-fitting of XANES is one feature which I am confident to say is definitely better than Athena. I'm willing to take 100% of the blame for this, but peak-fitting does not work well in Athena. XAS Viewer has more lineshapes including both PseudoVoigt and (true!) Voigt lineshape functions.  It also supports better mechanisms for applying mathematical constraints and for applying bounds on parameters which can be important for modeling XANES with multiple peaks.    lmfit/Larch also provides more and better fit statistics, such as reporting FWHM (and its uncertainty) for each peak, and (for XANES) the centroid of the pre-edge peaks.  Though not supported directly in the GUI, lmfit/Larch does allow one to do a more thorough estimate of uncertainties and to use minimization methods other than Levenberg-Marquardt.  My experience is that these seem to be not so important for peak-fitting of XANES data, but I don't know that this is always the case (and to be clear, there are certain classes of "line shape fitting" where these are important).  Like Athena, XAS Viewer records all the code actually run making it easy to turn a GUI session into a Larch script or Python program to do more detailed analysis or to help write analysis scripts for batch processing.

Hope that helps,

--Matt

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://millenia.cars.aps.anl.gov/pipermail/ifeffit/attachments/20190502/1b7b96e0/attachment.html>