[Ifeffit] Question on (Combo-) LCF plot output

[t.zahoransky at mineralogie.uni-hannover.de]

Dear Mailing list,

I am a relatively new user in terms of XAS data handling and I  
encountered a strange problem during LCF in Athena.
I am working on LCF-combo fittings for Mn minerals as described in  
Manceau et al. 2012*.

Fitting f.ex. a cryptomelane sample of myself with the open source  
monovalent manganese mineral standards of Manceau et al. 2012 in a fit  
range of -20 to +30 k, gives the following result shown below (I´ll  
attach the Athena project):

" LCF fit of Cryptomelane_T_20K_ecd as flattened mu(E) from 6536.79 to 6586.79
Fit included 100 data points and 6 variables, and approximately 41.333  
measurements
Weights sum to 1: no
Weights forced between 0 and 1: no
Overall e0 shift used: yes
Noise added to data: 0
R-factor = 0.0003537
Chi-square = 0.00815
Reduced chi-square = 0.0000858

.    standard             weight           e0
========================================================
.   Manganosite.dat       0.042 (0.006)    1.631 (0.044)
.   Groutite.dat          0.030 (0.029)    1.631 (0.044)
.   Ca2Mn3O8.dat          0.288 (0.026)    1.631 (0.044)
.   Ramsdellite.dat       0.584 (0.022)    1.631 (0.044)
.   Mn2O3.dat             0.044 (0.025)    1.631 (0.044)

.         sum ........... 0.988  "

Note that the box “all standards share an E0” is checked before  
clicking on “Fit this group”. Optically, the fit itself “looks” good  
compared to the data, yet, it is somewhat shifted on the x-axis  
compared to the data. I was wondering why, as this was not the first  
sample where this was the case.

When I now uncheck the “All standards share an E0” box, leave  
everything else just the way it is and just click “Fit this group”  
again, the fit suddenly is shifted perfectly on my data. The strange  
thing is: Nothing else in the output-results changes – weights,  
R-factor, chi-square etc. just stay the same. Only slight changes in  
the numbers in brackets after the reported weights. The "new" result is:

" LCF fit of Cryptomelane_Mikon_T_20K_ecd_reb-a-m as flattened mu(E)  
from 6536.79 to 6586.79
Fit included 100 data points and 5 variables, and approximately 41.333  
measurements
Weights sum to 1: no
Weights forced between 0 and 1: no
Overall e0 shift used: no
Noise added to data: 0
R-factor = 0.0003537
Chi-square = 0.00815
Reduced chi-square = 0.0000849

.    standard             weight           e0
========================================================
.   Mn2O3.dat             0.044 (0.024)    1.631 (0.000)
.   Manganosite.dat       0.042 (0.006)    1.631 (0.000)
.   Ca2Mn3O8.dat          0.288 (0.024)    1.631 (0.000)
.   Ramsdellite.dat       0.584 (0.019)    1.631 (0.000)
.   Groutite.dat          0.030 (0.026)    1.631 (0.000)

.         sum ........... 0.988 "

Can anybody tell me, what is actually happening? I exported both data  
to excel to replot the fits to exclude a simple plot-bug, but the data  
behind show that this is real. So, I am changing the data behind the  
plot, but actually not the output data … what is going on here?

You find the project attached – I hope someone can tell me, how to  
cope with that!

Thank you very much,
Teresa


*Manceau, A., Marcus, M. A., & Grangeon, S. (2012). Determination of  
Mn valence states in mixed-valent manganates by XANES spectroscopy.  
American Mineralogist, 97(5-6), 816-827.

-- 
Teresa Zahoransky

Soil Mineralogy

Gottfried Wilhelm Leibniz Universität Hannover

Institute of Mineralogy

Callinstr. 3, Room 325

D-30167 Hannover, Germany



Phone: +49 (0)511 762-8058

Email: t.zahoransky at mineralogie.uni-hannover.de

-------------- next part --------------
A non-text attachment was scrubbed...
Name: LCF-combo_question.prj
Type: application/octet-stream
Size: 43940 bytes
Desc: not available
URL: <http://millenia.cars.aps.anl.gov/pipermail/ifeffit/attachments/20200330/4eff7041/attachment-0001.obj>