As we talk about plotting in Q-space (just for information in Maud is
available from few months thanks to Klaus-Dieter advocating for it),
I would advocate another "plot option" that I would rather see as a
default way of plotting.
Looking at the other axis (the intensity) I am asking why we don't
introduce the practice to plot in a more useful scale as the root
square of the intensity (instead of the usual linear scale). This has
several benefits:
- the plot will be at iso-error (I recall for who may have forgotten
the noise is proportional to the root square of the intensity), so in
the residuals you may better evaluate which are the peaks or part
well fitted or poor fitted. Otherwise with the linear scale you just
see only the bad residuals of the more intense peaks and you may
think these are the peaks poorly fitted. Instead most of the time in
the true statistical meaning they may be well fitted compare to
other. In the square root intensity mode you can evaluate them more
unbiased
- you see also the small peaks and it is not necessary to enlarge the
intensity scale to check them
In this regard some people are using the log10 scale plot (normally
used for reflectivity measurements). It may enhance more the small
peaks but I don't favor it as again you do not compare the fitting of
different peak on an equal statistical base.
As an image is better than thousand words, as they say, I put
together one web page with the comparison of linear/sqrt/log10 scale,
Q and 2theta so everyone may take its own conclusion.
http://www.ing.unitn.it/~maud/plotoptions/
I would encourage the list to propose a standard way (or advised way)
to plot, as it would be not too difficult for the different program
to provide a standard way to present the results for the benefit of
comparisons.
Best regards,
Luca Lutterotti
