Dear All, dear Stefan,
I am trying to run an esfit:
esfit('pepper',experimental,Sys0,Vary,Exp,Opt);
where
Sys0.cAm = cAm;
Vary.cAm = 0.1;
Whatever value I set for cAm at the top of the script, the esfit gives the answer that it is the optimal one. For example, when cAm is set to 0, the output:
-- esfit ------------------------------------------------
Simulation function: pepper
Problem size: 1 spectra, 1 components, 1 parameters
Minimization method: Nelder/Mead simplex
Residuals computed from: integral
Scaling mode: lsq0
---------------------------------------------------------
initial simplex...
1: 2.06721e+001 1.00000e-001 initial simplex
2: 2.06721e+001 5.00000e-002 reduction
3: 2.06721e+001 2.50000e-002 reduction
4: 2.06721e+001 1.25000e-002 reduction
5: 2.06721e+001 6.25000e-003 reduction
6: 2.06721e+001 3.12500e-003 reduction
7: 2.06721e+001 1.56250e-003 reduction
8: 2.06721e+001 7.81250e-004 reduction
9: 2.06721e+001 3.90625e-004 reduction
10: 2.06721e+001 1.95313e-004 reduction
11: 2.06721e+001 9.76563e-005 reduction
---------------------------------------------------------
Best-fit parameters:
cAm: 0
Residuals of best fit:
rmsd 20.6721
=========================================================
And if I set cAm to, say, 2, the outcome is:
-- esfit ------------------------------------------------
Simulation function: pepper
Problem size: 1 spectra, 1 components, 1 parameters
Minimization method: Nelder/Mead simplex
Residuals computed from: integral
Scaling mode: lsq0
---------------------------------------------------------
initial simplex...
1: 2.06721e+001 1.00000e-001 initial simplex
2: 2.06721e+001 5.00000e-002 reduction
3: 2.06721e+001 2.50000e-002 reduction
4: 2.06721e+001 1.25000e-002 reduction
5: 2.06721e+001 6.25000e-003 reduction
6: 2.06721e+001 3.12500e-003 reduction
7: 2.06721e+001 1.56250e-003 reduction
8: 2.06721e+001 7.81250e-004 reduction
9: 2.06721e+001 3.90625e-004 reduction
10: 2.06721e+001 1.95313e-004 reduction
11: 2.06721e+001 9.76563e-005 reduction
---------------------------------------------------------
Best-fit parameters:
cAm: 2
Residuals of best fit:
rmsd 20.6721
=========================================================
It is quite clear from the visual that cAm = 0 is much better than cAm = 2.
I know, it might be too much to ask but maybe somebody could straightaway point to an obvious reason for that? It looks that the fitting procedure is going on without knowing anything about the experimental spectrum (which could be called allright from the workspace).
I am a novice in EasySpin…
Thanks,
Dima Svistunenko