smtg-bham/aiida-user-addons

Additonal code and plugins for working with AiiDA. VASP workflows for batteries and more.

aiida-user-addons's Issues

automatic phonopy primitive cell for magnetic systems

For some reasons, the primitive cell matrix for a magnetic system must be pass explicitly. This should be investigated and allow automatic generation.
Because the NAC calculations use the primitive structure, settings passed at the builder time may not be valid, such as the LDAU settings. This should be checked and warned if necessary.

pipline based definition for inputs

Setting builder for AiiDA input can be interactive but can be very expliticy. At the moment, pre-defined default inputs for VASP can be prepared using VASPInputset but the user still have to link it up to each input port.

An alternative approach is to define the input as a pipline, encourages code resuse.
So a workflow can be defined as:

pipeline = BuilderPipline(process_class).default().override_incar(...).set_computer(...)
pipeline.set_resources(...)

def use_pbesol(pipeline):
    pipeline = pipeline.override_incar(...).set_pseudos_family(...)
    return pipeline 

def use_myriad(pipeline):
    pipeline = pipeline.set_resources(...).set_code(...)
    return pipeline 

pipeline = input_for_relax(pipeline)
submit(pipline.builder)

pull remote data during `export_calculation`

Allow pulling data from the remote folder/stash folder, as some of the files (WAVECAR/CHGCAR) are not retrieved by default. If they are still present on the remote, we may want to pull them down ata per-case basis for process processsing.
Note that this can already be achieved with verdi data addons remotepull.

Add support for hybrid DFT band structure

The current bandstructure workchain in new_bands.py only works for GGA calculations.

For hybrid DFT, there is no need to perform the SCF calculations then non-SCF calculations. Instead, the k points path should be split and a series of SCF calculations with paths that contains zero-weighted points should be performed. The fully k path needs to be reconstructed in the end.

Add additional check at the end of relaxation

Sometimes VASP can converge into the different electronic state for the final singlepoint calculation after a geometry optimisation, giving increased forces and energy as a results.

The relaxation workchain should add additional checks of the final force output.

Note that this does not apply to ISMEAR=-5 used in the final singlepoint calculation, since the forces are wrong already in those calculations, and hence there is no way to validate the result reliably.

make restarts using CHGCAR/WAVECAR easier

workflow engine should workout the needed type of restart and set the INCAR tags accordingly. This may introduce another input settings for VaspWorkChain though.

Choose from:

• restart with WAVECAR for wavefunction
• restart with CHGCAR for the charge density without WAVECAR
• no action - leave the INCAR tags as they are

The default should be "use as much as possible". The content of the remote folder can be obtained during the workchain operation to check which files are present for restarts before setting the parameters.