Question: units for Q?

[toqduj]

Quick question, since I can't seem to find it in the documentation (https://xraypy.github.io/XrayDB/python.html#elastic-scattering-factors): what are the units of q used here? 1/Ångström?

[newville]

@toqduj Yes, 1/Å. We should probably make that clear in the docs, huh?

[toqduj]

Thanks! Yes, just a small mention would help... either that or accept pint Quantities 😄

Cheers, and have a nice weekend!

[toqduj]

Oh.. and Q is normally defined as $4 \pi \sin(\theta) / \lambda$.. so there's an additional factor $4 \pi$ to be aware of when using this method.

[newville]

@toqduj The data are taken from D. Waasmaier & A. Kirfel Acta Cryst. (1995). A51, 416-431.
They tabulate coefficients to calculate
f0[k] = c + Sum_i [a_i * exp(-b_ikk)]

where k = sin(theta) / lambda and c, a_i and b_i are stored coefficients. (see https://github.com/xraypy/XrayDB/blob/master/data_sources/waasmaeir_kirfel.dat)

.... so maybe we should use k here and be clear what the inverse-length the function actually expects?

[newville]

@toqduj handling pint quantities would be interesting, as would values with uncertainties... sounds like a good project for someone with some free time... ;)

[toqduj]

@toqduj The data are taken from D. Waasmaier & A. Kirfel Acta Cryst. (1995). A51, 416-431. They tabulate coefficients to calculate f0[k] = c + Sum_i [a_i * exp(-b_i_k_k)]

where k = sin(theta) / lambda and c, a_i and b_i are stored coefficients. (see https://github.com/xraypy/XrayDB/blob/master/data_sources/waasmaeir_kirfel.dat)

.... so maybe we should use k here and be clear what the inverse-length the function actually expects?

That would be an excellent solution, and would avoid the careless user (which was me a few days ago) making assumptions on q without fully reading the documentation. If it is named k, the aforementioned mythical person with free time can add an optional keyword argument q which is the standard definition q with Pint units and uncertainties that maps to k 😄

But the low-threshold, quick and minimal-effort solution here is to change the name to $k$ (to closer match the literature and avoid assumptions), and to indicate the units in the documentation.

[newville]

@toqduj agreed -- changing to k will help clarify that. Over in the EXAFS world we use k to also mean a completely different "momentum measured in 1/A" (of an electron) which is probably why I used q there, but there is less chance of confusion there. I hope ;).

... and now I am checking my other calculations for whether the value send to f0() is really sin(theta)/lambda or scattering q....