Running a simple VASP calculation#

In this example, we will run a simple VASP calculation using the aiida-vasp plugin. The VaspWorkChain will be used to run a calculation on a sample structure.

It is assumed that the [installation][#installation] has been completed, the InstalledCode for the VASP executable and the PBE.54 pseduopotentials has been [configured][#configuration].

The following steps can be run with an ipython shell launched by the verdi shell command.

Setting up the inputs#

First, we create the sample structure using the ase package (assuming the ase package is installed):

from ase.build import bulk
from aiida import orm
si = bulk('Si', 'diamond', 5.4)
si_structure = orm.StructureData(ase=si)

Note that the orm.StructureData class is used to store the structure data in the AiiDA database.

Next, we create necessary input node and set the required parameters:

from aiida.plugins import WorkflowFactory
builder = WorkflowFactory('vasp.vasp').get_builder()

builder.parameters = orm.Dict(
    dict={'incar':
    {
        'encut': 300,
        'ismear': 0,
        'ispin': 1,
    }
    }
)
builder.options = orm.Dict(dict={
    'resources': {'num_machines': 1,
    'tot_num_mpiprocs': 1}}
    )
builder.structure = si_structure
builder.code = orm.load_code('vasp@localhost')   # Assuming VASP is installed locally and the code is configured
builder.potential_family = 'PBE.54'     # Note that we can use python type - they are converted into AiiDA types automatically
builder.potential_mapping = {'Si': 'Si'}
builder.kpoints_spacing = 0.05
builder.metadata.label = 'test-calculation'
builder

It is a good practice to inspect the ProcessBuilder before submitting the calculation.

Run and inspect the results#

Finally, the calculation can be run with the run_get_node method:

from aiida.engine import run_get_node
out_dict, process_node = run_get_node(builder)

If everything runs fine, you should see something like

Report: [1615|VaspWorkChain|run_process]: launching VaspCalculation<1618> iteration #1
Report: [1615|VaspWorkChain|results]: work chain completed after 1 iterations
Report: [1615|VaspWorkChain|on_terminated]: cleaned remote folders of calculations: 1618

The out_dict is dictionary containing the output nodes of the VaspWorkChain and the process_node is a WorkChainNode representing the work chain that has been executed.

>>> out_dict
{'remote_folder': <RemoteData: uuid: c9351bfb-237c-4a80-9710-8f3136f44e4a (pk: 1619)>,
 'retrieved': <FolderData: uuid: 8b84eeb2-b58f-474d-9d33-eb8a0f48d61d (pk: 1620)>,
 'misc': <Dict: uuid: a5c71be6-0a6d-4ef3-b2f2-ba91fbc27a67 (pk: 1621)>,
 'dielectrics': <ArrayData: uuid: e24207ea-cf2e-4459-b2b4-66feef172aa9 (pk: 1622)>}
>>> process_node
<WorkChainNode: uuid: a84d5dfb-03be-4b49-86d0-d81b3b9f31e4 (pk: 1615) (aiida.workflows:vasp.v2.vasp)>

Note

The actual uuid and pk displayed on your compute will be different.

The total energy, forces and stress can be found in the misc output node:

>>> out_dict['misc']['total_energies']
{'energy_extrapolated': -10.81676131,
 'energy_extrapolated_electronic': -10.81676131}
>>> out_dict['misc']['forces']
{'final': [[0.0, 0.0, 0.0], [0.0, 0.0, 0.0]]}
>>> out_dict['misc']['stress']
{'final': [[35.63282046, 0.0, 0.0],
  [0.0, 35.63282046, 0.0],
  [0.0, 0.0, 35.63282046]]}

The forces are zero as the structure has high symmetry. The stresses are not zero since we did not put the equilibrium lattice constant of Si in the first place. Geometry optimization is needed to fully relax the structure.

We also can access the same output through the WorkChainNode using the outputs attribute:

>>> process_node.outputs.misc['run_stats']
{'nsw': 0,
 'nelm': 60,
 'nbands': 8,
 'finished': True,
 'ionic_converged': None,
 'contains_nelm_breach': False,
 'electronic_converged': True,
 'last_iteration_index': [1, 10],
 'consistent_nelm_breach': False}