Add new functions/methods to featurize module

[naik-aakash]

Closes #329

Changes

• Add ultiliy functions to compute reduced masses, electronegativity and sort dict by values
• Add the possibility to modify default lobsterpy kwargs in featurizers
• Add ICOXX featurizer with BWDF functionality
• Add tests

[naik-aakash]

Hi @JaGeo, there seems to be some issue with coverage artificats not being generated and uploaded in CI, I have not been able to pinpoint the cause yet. Will check again in detail tomorrow. Leaving this as a comment only for my reference here.

[naik-aakash]

Could be simply bug associated with a new release of pytest today. Will have to check downgrading it

[JaGeo]

@naik-aakash sure. Don't worry!

[naik-aakash]

@naik-aakash sure. Don't worry!

issue was with upload artificats update ignoring hidden files from being uploaded. It should be fixed now in PR #331

[naik-aakash]

Hi @JaGeo , posting here results of some tests I did to verify the implementation for BWDF

Note: In all the tests, a supercell of 2x2x2 is constructed

ZnSb Structure : mp-753

NaCl

Si3N4 : mp-11607

AlN : mp-661

[naik-aakash]

Just for NaCl and AlN, I do not have exactly overlapping peaks. But seems to work in other cases. Can't seem to figure out what could be the reason.

[JaGeo]

@naik-aakash can you provide the structures for both AlN cases and both NaCl cases here? And icohps for all Al-Al and cl-cl.interactions?

[naik-aakash]

@naik-aakash can you provide the structures for both AlN cases and both NaCl cases here? And icohps for all Al-Al and cl-cl.interactions?

cannot upload compressed files here, will share the calc files on email.

[naik-aakash]

Hi @JaGeo , Below you can find the outputs for the same structures with updated implementation, which now works fine for primitive and supercells

NaCl

AlN

Si3N4

ZnSb

[JaGeo]

Very good. I think you can test some more cases and see if your implementation can cover all of those

[QuantumChemist]

Hi @JaGeo , posting here results of some tests I did to verify the implementation for BWDF

Note: In all the tests, a supercell of 2x2x2 is constructed

ZnSb Structure : mp-753

NaCl

Si3N4 : mp-11607

AlN : mp-661

Looks like you only checked cubic and tetragonal crystal symmetries so far. You could try some less symmetry structures as well, like monoclinic, triclinic. And also hexagonal and orthorhombic.

[QuantumChemist]

Oh, I replied to the wrong comment

[naik-aakash]

Hi @JaGeo and @QuantumChemist , I did some more tests on different kinds of structures for supercell and primitive comparisons. Just posting here some results of summed BWDF I got from these comparisons for this structures. It seems to work in these cases as well.

orthorhombic : mp-10381

monoclinic : mp-10408

hexagonal : mp-10614

triclinic : mp-14983

[JaGeo]

Good, I would recommend testing your implementation on your complete dataset and check if there are any failures (does the algorithm find all ICOHPs in the ICOHP list?)

[naik-aakash]

Good, I would recommend testing your implementation on your complete dataset and check if there are any failures (does the algorithm find all ICOHPs in the ICOHP list?)

Will check this soon 😄

[QuantumChemist]

Hi @JaGeo and @QuantumChemist , I did some more tests on different kinds of structures for supercell and primitive comparisons. Just posting here some results of summed BWDF I got from these comparisons for this structures. It seems to work in these cases as well.

orthorhombic : mp-10381

monoclinic : mp-10408

hexagonal : mp-10614

triclinic : mp-14983

Fantastic!