Immigrations¶
Sometimes it is usefull or necessary to import VASP calculations that have not been executed in AiiDA or with the AiiDA-VASP plugin. For these cases, existing VASP calculations can be imported, or immigrated into the AiiDA-VASP realm. To make this process simple, we have designed am particular calculation that performs the data immigration task called VaspImmigrant.
Working principles¶
The immigration process is straightforward and utilizes the underlying calculation framework in AiiDA.
First, we take the input files from the previous calculation and construct a new calculation
The calculation is executed, for instance through a workchain, but VASP is never run.
The output files are parsed similar to a normal execution of VASP using AiiDA-VASP.
This strategy is very convenient as it gives full data provenance.
Basic process¶
The followings are assumed:
AiiDA and AiiDA-VASP are installed with a profile set-up.
A local
<computer>is in the profile. Remote<computer>can not be used currently. The local<computer>setup may have the following properties:* Hostname: localhost * Transport type: local * Scheduler type: direct
A VASP
<code>(does not have to be runnable in principle) is associated with an existing<computer>in the profile.How to run calculations in AiiDA is understood.
A directory where VASP was executed exists on the remote (or local)
<computer>, for which any readable direction that ssh can reach is possible.
Assuming you ran VASP in /scratch/<user>/<runfolder>/ on the remote (or local) <computer>, the following python snippet will help you to run into your AiiDA profile database:
from aiida.orm import Code
from aiida.plugins import CalculationFactory
from aiida.engine import submit
# the folder you want to import
calculation_folder = '/scratch/<user>/<runfolder>/'
# first, get the code which the calculation will use
code = Code.get_from_string('<code>@<computer>') # use the name of your code and computer
# then comes the information which can not be read from input files
resources = {'num_machines': 1, 'num_mpiprocs_per_machine': 16} # whatever is appropriate for you
_, builder = CalculationFactory('vasp.vasp').immigrant(
code, calculation_folder,
metadata={'options': {'resources': resources}})
node = submit(builder)
After running the above snippet, node will be the node of the created calculation.
Parsing additional information¶
In order to parse additional quantities, first, please see Parsing. For the, this follows the same mechanism as for normal calculations. Here we take as an example, that the user want to also parse and include the electronic band structure (not enabled by default). We then make this available by passing the settings kwarg to the immigrant method:
settings_dict = {'parser_settings': {'add_bands': True}}
process, inputs = CalculationFactory('vasp.vasp').immigrant(..., settings=settings_dict)
And running the job as described above should produce a result containing the electronic band structure.
Missing POTCAR file¶
Requires:
You have the POTCAR potentials to be used uploaded to your profile (see Potentials).
In case you have removed the POTCAR files in your VASP calculation directories (e.g. to save disk space), you need to specify suitable potentials from the ones uploaded to your database:
potential_mapping = {'Si': 'Si'}
potential_family = 'PBE.54'
process, inputs = CalculationFactory('vasp.vasp').immigrant(..., potential_family=potential_family, potential_mapping=potential_mapping)
This will use the chosen Si potential. It will fail if the POSCAR does not specify element names.
Additional input files¶
For CHGCAR and/or WAVECAR it can not always be inferred from the INCAR, whether they are input or output files. If they are not automatically recognized as inputs, you may pass additional arguments to the immigrant method which supports those inputs:
process, inputs = CalculationFactory('vasp.vasp').immigrant(..., use_chgcar=True, use_wavecar=True)
which will create input nodes from these files without checking the INCAR file.