Difficulty with goodness of fitting spectral features for multicomponent slow-motion spectra
Hello,
I am relatively new to CW-EPR and have been playing around with nitroxide spin probes (specifically TEMPO maleimide) labeled on mobile loops of various proteins. When fitting the data using EasySpin, I have run into two issues.
1.) When fitting the data with chili, it is not able to correctly grab certain spectra features (specifically at the high field) leading me to not trust the fitting. I have attached a image at the bottom to show this. I have tried to watch the EasySpin academy online lectures to figure out other parameters that I may be missing to no avail. Is there something I am missing within my fitting parameters that might solve this issue?
2.) When I find a local minimum using the Nelder-Mead simplex and it spits out the fit and error, if I refit starting at those same parameters it no longer converges at the same minima. Is this just a part of the error in fitting using this algorithm or is this an issue with the algorithm stopping after reaching a certain number of iterations and not a local minima?
Code :
%% Data import
[B,spc] = eprload('datafile.DTA');
b = B/10;
plot(b,spc);
%Data Variables
Exp.mwFreq = 9.405711; %GHz
Exp.Range = [min(b) max(b)]; % mT
Exp.nPoints = numel(b);
%Spin system
Nx.g = [ 2.00595 2.0086 2.0021]; % full g-tensor
Nx.Nucs = '14N'; % Nucleus
Nx.A = [26 11.6 103.64]; % Hyper-fine coupling (MHz) Gaussian
Nx1 = Nx;
Nx2 = Nx;
% Starting parameters
Nx1.logtcorr = -8.5 ; % log(tr)
Nx2.logtcorr = -8.8 ; % log(tr)
Nx1.weight = 0.5;
Nx2.weight = 0.5;
Nx1.lwpp = 0.26 ; % mT Gaussian
Nx2.lwpp = 0.25 ; % mT Gaussian
%Vary
Vary1.weight = 0.3;
Vary2.weight = 0.3;
Vary1.logtcorr = 1;
Vary2.logtcorr = 1;
Vary1.lwpp = 0.05 ; %Gaussian
Vary2.lwpp = 0.05 ; %Gaussian
% Least-squares fitting
esfit(spc,@chili,{{Nx1,Nx2}, Exp},{{Vary1 Vary2}});
Any help would be greatly appreciated!