FIX: Running PDOS for spin-polarised system

[mbercx]

Fixes #202

Currently it is not possible to calculate the PDOS for a spin-polarised
system (i.e. setting Magnetism to Ferromagnetic), since the
QeAppWorkChain expects the projections output from the
PdosWorkChain. However, for spin-polarised systems the
ProjwfcCalculation returns two outputs: projections_up and
projections_down.

Here we adapt the QeAppWorkChain to check for the projections_up
output, and in case it is present it will add both projections_up and
projections_down to the outputs. Otherwise it will only add the
projections output (both only in case the workchain_pdos was
actually run).

To deal with plotting the corresponding (P)DOS, we simply add the total
DOS up and down states together, and separately add the PDOS orbitals
for the up and down states to the list passed to the widget-bandsplot.
Since the widget currently adds all the PDOS of the same orbital types
(s, p, d, f) together, there will currently not be any differentiation
between the up and down states in the visualization. This is something
that will need to be improved on the widget side.

[unkcpz]

Hi @mbercx, thanks for this fix. I find another issue that when running the test on an oxygen structure with the setting ElectronicType=insulator, Magnetism=ferromagnetic, the workflow failed at a very early stage.

It is caused by the occupation=fixed while no tot_magnetization is set in the input file. But can be in another PR.

[unkcpz]

@mbercx thanks. I still running some tests on your branch. all looks good to me but some minor questions.

[mbercx]

It is caused by the occupation=fixed while no tot_magnetization is set in the input file. But can be in another PR

Yeah, this is a known issue, but needs to be fixed in the plugin. Thought I already opened an issue for this, but apparently not. Will do so now.

[unkcpz]

@mbercx Thanks! I still have some minor questions. Bit almost there I think.

[mbercx]

Hmm, it seems to fail to initialize the pre-commit env. I'm also having this issue locally...

[unkcpz]

I think it relates to the issue ContinuumIO/anaconda-issues#12094 (comment). I encountered it also in my aiida-sssp-workflow repo, add python -m pip install virtualenv==20.0.33 in pre-commit CI fixes it, but not sure why it happened.

[unkcpz]

@csadorf do you know which cause this pre-commit issue?

[csadorf]

@csadorf do you know which cause this pre-commit issue?

@unkcpz Yes, I provided some summary here.

[unkcpz]

Hi @mbercx, the pre-commit issue is temporarily fixed, can you update the branch and try it again. I think there is a minor issue with code which cause other pre-commit fail that is,

aiidalab_qe/node_view.py:147:54: F821 undefined name 'ProjectionData'
        for projections, suffix in projection_list:  # type: ProjectionData, str
                                                     ^
1     F821 undefined name 'ProjectionData'

[unkcpz]

Hi @mbercx, could you have a look and we finish this PR these days? should be nothing to add and block it. I force push with rebase the latest master branch, hope you don't mind.

[unkcpz]

Thanks @mbercx, looks all good.