Lineshapes

Here are the lineshapes used by peakipy. Use the --lineshape option in peakipy fit to select a lineshape for fitting.

For example,

peakipy fit test.csv test1.ft2 fits.csv --lineshape G

Would fit to a Gaussian lineshape in both dimensions. Other options are V, L or PV for Voigt Lorentzian or Pseudo-Voigt in both dimensions, respectively. If you want to fit a seperate lineshape for the indirect dimension then PV_PV allows you to fit a Pseudo-Voigt with seperate X and Y fraction parameters

Gaussian

\(\frac{1}{\sigma_g\sqrt{2\pi}}\exp \frac{-(x - center)^2 } { 2 \sigma_g^2}\)

Lorentzian

\(\frac{1}{\pi} \left( \frac{\sigma}{(x - center)^2 + \sigma^2}\right)\)

Pseudo-Voigt

\(\frac{(1-fraction)}{\sigma_g\sqrt{2\pi}}\exp \frac{-(x - center)^2 }{ 2 \sigma_g^2} + \frac{fraction}{\pi} \left( \frac{\sigma}{(x - center)^2 + \sigma^2}\right)\)

This is the default lineshape used for fitting and the fraction of G or L is assumed to be the same for both dimensions. The --lineshape PV_PV option will fit a seperate pseudo-voigt lineshape in each dimension (i.e. fraction_x and fraction_y parameters).

Fit quality

Fit quality can be evaluated by inspecting the contour plot of residuals that is generated when viewing fits interactively. \(\chi^2\) and \(\chi_{red}^2\) are calculated using the noise estimate from --noise or the threshold value calculated from threshold_otsu if --noise is not set explicitly. Peakipy does calculate the linear correlation between the NMR data and the simulated data from the fit. If the slope deviates by more than 0.05 from 1.0 then it is advised that you check the fit. However, this is not totally robust and it is best to check fit quality by plotting the data using the peakipy check command.