A script for calculating non-Maxwellianity

[vetarvus]

A pyCalculations script to calculate the non-Maxwellianity of a VDF, which has been previously available as a Vlasiator datareducer.

The output was tested using a 2D Kelvin-Helmholtz run from Tarvus et al. 2024.
Difference between the Vlasiator datareducer and this script is of the order of 1E-4 (significant values are typically > 1E-2).

[alhom]

Great to get this in as well!

But, comments:

• please remove all the .pyc files from the PR.
• We could wrap this with a datareducer, to get direct plotting support (even if it will be slow to plot). Gotta check the VDF reducers, not exactly sure if those worked straight out of the box. Let's see if that is easy to do, and if yes, let's add that. Otherwise we'll include the routine and defer the reducer to an issue.

[alhom]

More comments!

Difference between the Vlasiator datareducer and this script is of the order of 1E-4 (significant values are typically > 1E-2).

It would be nice to understand this discrepancy. It might be from this hard-coded threshold, so please check if syncing that with your test case fixes it:

analysator/pyCalculations/non_maxwellianity.py

Lines 54 to 55 in 3c0406b

	threshold = 1e-21
	distribution[distribution<threshold] = 0

Worth adding a kword for the threshold, and a sensible default threshold value (or no threshold as default?).
Vlasiator reducer just loops over everything with no threshold, no?
https://github.com/fmihpc/vlasiator/blame/62ac639d96eba005b66ba589bd69c2093663db92/datareduction/datareductionoperator.cpp#L2005
Edit: Threshold kword added with default of 0

Suggest adding a feature to calculate the non-maxwellianity for nearest existing VDF, using get_cellid_with_vdf (

analysator/pyVlsv/vlsvreader.py

Line 2118 in b6e96c4

def get_cellid_with_vdf(self, coords, pop = 'proton'):

) as in plot_vdf (

analysator/pyPlots/plot_vdf.py

Line 754 in b6e96c4

cidNearestVspace = vlsvReader.get_cellid_with_vdf(np.array([xReq[ii],yReq[ii],zReq[ii]]), pop=pop) # deprecated getNearestCellWithVspace(). needs testing

)

Multipop-Vspace reducers needs to be implemented for the reducer to work with modern data. Would you Vertti like to have a go at it? If not, we'll defer.

analysator/pyVlsv/vlsvreader.py

Line 810 in 3c0406b

if reducer.useVspace: