set kind of shape to lorentzian by default for peaks

[jobo322]

@lpatiny @jliermann, I think the shape looks better as gaussian than lorentzian by default just after auto-peakpicking.

It is really improved after the optimization step

so I propose to keep gaussian as default for peaks out of the box from auto-peak-picking

[jliermann]

Thanks for the comment and the examples! However, I would disagree. If you look at the sum spectrum, the optimized Lorentzian is better.

Gaussian:

Lorentzian:

And that is actually expected from FT theory which gives Lorentz-distributed lines (J. Keeler, Understanding NMR Spectroscopy, 2nd edition, p. 84):

[lpatiny]

@jobo322 I'm wondering if GSD is done with Lorentzian. The difference with and without optimization is too big. There must be a bug somewhere.

[jobo322]

GSD gets the width from the inflection points of the shape (experimental data), but the relation between width and FWHM depends of the kind of shape. Based on the wolfram pages of lorentzian and gaussian I found that:

for Lorentzian shape, the relation between width and FWHM is:

width = FWHM/sqrt(3)

for Gaussian shape, the relation between width and FWHM is:

width = FWHM/sqrt(2 * Ln(2))

So, for a gaussian the difference between width and FWHM is an 17.7 %. But for a Lorentzian shape the difference is around 73 %. That is why, the gaussian shape looks better than lorentzian for peaks without optimization.

Let me know if I am wrong.

[jobo322]

I did test the shape with the simulated spectra. It is drawn with a gaussian shape but the idea is the same.

[jobo322]

@jliermann @lpatiny I propose to use pseudoVoigt shape as default for peaks,

[jliermann]

@jliermann @lpatiny I propose to use pseudoVoigt shape as default for peaks,

Sorry for not answering earlier! I appreciate that the Lorentzian shape causes optimisation issues but I still feel that the theoretically correct shape should be the default as this will have an educational impact. Other shapes should only be used by people how choose so intentionally.