reflectors param in EBSDDetector.get_indexer() method (kp v0.9, pyebsdindex v0.2)

[Erlendos12]

Hi there (sorry about the long post)!
It has been a while since I last worked with kikuchipy and PyEBSDIndex, and I would like to take advantage of the new capabilities in pyebsdindex v0.2. For example, I would like to be able to hough index a trippel-phased sample which contains austenite, ferrite and sigma phase. To do this, I understand that I now have to take advantage of the new reflectors parameter in the EBSDDetector.get_indexer() method. I am a little confused about this parameter (probably due to my background), and how to properly derive a reflector list from my list of phases. I have tried the following:

...
reflectors = []
for indx, ph in self.phases:
    if indx != -1:
        lattice = ph.structure.lattice
        lattice.setLatPar(lattice.a * 10, lattice.b * 10, lattice.c * 10)

        rlv = ReciprocalLatticeVector.from_min_dspacing(ph, 0.7)
        rlv.sanitise_phase()
        rlv = rlv.unique(use_symmetry=True)
        rlv.calculate_structure_factor("lobato")
        structure_factor = abs(rlv.structure_factor)
        order = np.argsort(structure_factor)[::-1]
        rlv = rlv[order[0:7]]    # I workaround: slice the array because it contains values giving ValueError: negative dimensions are not allowed
        rlv = rlv.symmetrise()
        rlv.print_table()
        reflectors.append(rlv)

indexer = det.get_indexer(
    phase_list=self.phases, 
    reflectors=reflectors, 
    rhoMaskFrac=self.rho_mask, 
    nBands=self.number_bands,
    tSigma = 2,
    rSigma = 2
)
...

After passing the indexer to kikuchipy's EBSD.hough_indexing method, I get the following phase map!

This looks good, and somewhat similar to the phase map produced from Dictionary indexing.

After saving the crystal map to an .h5 format, I do however encounter some issues. Opening the crystal map and deriving ReciprocalLatticeVectors from the phases in the file, suddenly produces an extreme number of vectors (instead of just a couple hundred):

...
for i, ph in self.crystal_map.phases:
    phase_lattice = ph.structure.lattice
    phase_lattice.setLatPar(
        phase_lattice.a * 10, phase_lattice.b * 10, phase_lattice.c * 10
    )  # diffsim uses ångstrøm and not nm for lattice parameters
    if i != -1:
        rlv = ReciprocalLatticeVector.from_min_dspacing(ph, 0.7)
        print(rlv)

# Output
ReciprocalLatticeVector (567706,), austenite (m-3m)
[[ 51.   6.   0.]
 [ 51.   5.   3.]
 [ 51.   5.   2.]
 ...
 [-51.  -5.  -2.]
 [-51.  -5.  -3.]
 [-51.  -6.   0.]]
ReciprocalLatticeVector (287584,), ferrite (m-3m)
[[ 40.   8.   3.]
 [ 40.   8.   2.]
 [ 40.   8.   1.]
 ...
 [-40.  -8.  -1.]
 [-40.  -8.  -2.]
 [-40.  -8.  -3.]]
ReciprocalLatticeVector (4309214,), steel_sigma (4/mmm)
[[ 125.   12.    2.]
 [ 125.   12.    1.]
 [ 125.   12.    0.]
 ...
 [-125.  -12.    0.]
 [-125.  -12.   -1.]
 [-125.  -12.   -2.]]

Doing anything with such a number of vectors ends in an infinite freeze.

Could these number of vectors be due to how I dervied the list of reflectors before indexing?
I would anyway like to avoid the workaround that includes slicing the array, done to avoid the ValueError: negative dimensions are not allowed which appears in line 390 in ...\pyebsdindex\tripletvote.py libANG = np.zeros((nlib, 3)).
Is there a better and more generalized way to get this list of reflectors, given a list of phases?

Thanks for any pointers that can get me back on track again 👍

[hakonanes]

Hi @Erlendos12! Sorry for the late response.

I'm glad you're initial indexing is working out with multiple reflector lists. We're discussing the general approach of building reflector lists for PyEBSDIndex here: USNavalResearchLaboratory/PyEBSDIndex#50. Feel free to pitch in!

I believe the higher number of reflectors comes from you multiplying your lattice parameters one times too many. EMsoft stores lattice parameters in nanometers, which is why we multiply them to get Ångströms. orix, on the other hand, stores lattice parameters as they are given, because their unit is not stored anywhere.

Regarding the ValueError you're seeing, are you able to create minimal working example? If so, it would be great if you could open an issue!

[Erlendos12]

Hi, thank you so much for the response Håkon!
I will make sure to follow the discussion, and add anything of interest that may come up.

Thanks to you I now see (painfully obvious) that the parameters are indeed multiplied one times too many. Avoiding the second multiplication, and using the approach described in the discussion in USNavalResearchLaboratory/PyEBSDIndex#50, solved it!

Regarding the ValueError; Would you still like me to send you a minimal example?

[hakonanes]

You may have moved on from this @Erlendos12. Still, if you could send me a minimal example of the ValueError, that would be great.

[Erlendos12]

Hi, I have sent you a minimal example written in a jupyter notebook to your email. Let me know if there is anything else you need. 👍