Uncertainties in Spectral Parameters

[ERClark]

I've modelled a nice set of well behaved radicals and managed good fits with three parameters - two hyperfine couplings and a linewidth. However, the values come out at e.g. 15.662 MHz and 13.132 MHz for the two hyperfine couplings, with 0.2696 MHz linewidth. This seems to be unreasonable precision to me given the linewidth is an appreciable fraction of the peak separations, especially since I know the resolution of the recorded spectrum is 0.002 mT which in this case is about 0.05 MHz.

Are there any sensible guidelines on estimating uncertainties or sensible rounding for this type of case, other than repeatedly modelling a spectrum, changing parameters in a stepwise fashion, and seeing when residuals change appreciably?

(Sorry - I'm very much a synthetic chemist, not a spectroscopist, so I know that is probably an horrible question for all sorts of reasons)

Ewan

[thanasis]

For the moment this is not implemented:
viewtopic.php?f=3&t=608

[Stefan Stoll]

Uncertainty estimation for `esfit` will be introduced in the next version. For now, I suggest manually varying the parameters and determining visually when the fits deteriorates.