Multi component fitting: % composition from weights

[vanst]

Hello,

Briefly, I am curious how what the basis weighting scheme used in multicomponent fitting is comprised of.

Background: I have been performing some weighted fits for a multicomponent spectrum arising from multiple species with different spins, and was interested in seeing the % composition of each component. I looked at what the compositions would be if I simply took percentages from the weighting scheme (ex. A spectrum has S=3/2 component (a) with a weight of 0.5 and S=1/2 component (b) with a weight of 1, therefore the spectrum is 33% component (a) and 67% component (b)).

I then checked this by running the individual components (with no variation) to acquire their spectra, and rescaled them to match the proportions of the original fit. This was followed by double integration of each component, and application of the Aasa-Vånngård correction for the conversion from field to frequency domain. The resulting % composition of the spectrum arising from each resulting component was drastically different, particularly that the contribution from the S=3/2 component increased considerably (~20%). Therefore, I am curious how the weighting scheme functions in easyspin, and whether there is a way to perform such a composition analysis in the program. Thanks!

[joscha_nehrkorn]

It is very simple. Try the following example:

Code: Select all

Exp.mwFreq =9.6;
Exp.Range = [300 380];
Sys1.g = 2;
Sys1.lw = 1;
Sys1.weigth =1;
Sys2 =Sys1;
Sys2.g =2.1;

[b,s] = pepper({Sys1,Sys2},Exp);
s1 = pepper(Sys1,Exp);
s2 = pepper(Sys2,Exp);
plot(b,[s;s1;s2]);
sum(abs(s-(s1+s2)))

Basically, your idea how it should work is completely right.
Your try to check was probably wrong.
I don't get why you want to do a conversion to frequency domain after double integration.

One point which was discussed here a few times:
weigth can be used. If you don't normalize you will get different double-integrals for 0.33/0.67 and 0.5/1. However, fitting uses normalization, so you can only fit relative weigths.