Running workflows#

AiiDA can help you run individual calculations, but it is really designed to help you run workflows that involve several sub-processes, while automatically keeping track of the provenance for full reproducibility.

To see all currently available workflows in your installation, you can run the following command:

$ verdi plugin list aiida.workflows

We are going to run the PwBandsWorkChain workflow of the aiida-quantumespresso plugin. You can see it on the list as quantumespresso.pw.bands, which is the entry point of this work chain. This is a fully automated workflow that will:

  1. Run a calculation on the cell to relax both the cell and the atomic positions (vc-relax).

  2. Refine the symmetry of the relaxed structure, and find a standardized cell using SeeK-path.

  3. Run a self-consistent field calculation on the refined structure.

  4. Run a band structure calculation at a fixed Kohn-Sham potential along a standard path between high-symmetry k-points determined by SeeK-path.

Submitting a work chain#

In order to run this work chain, open the verdi shell and load the work chain using its entry point and the WorkflowFactory:

In [1]: PwBandsWorkChain = WorkflowFactory('quantumespresso.pw.bands')

Setting up the inputs one by one as we did for the pw.x calculation in the previous section can be quite tedious. Instead, we are going to use one of the protocols that has been set up for the workflow. To do this, all we need to provide is the code and initial structure we are going to run:

In [2]: code = load_code(<CODE_PK>)
   ...: structure = load_node(<STRUCTURE_PK>)

Tip

Forgot the PK of the Code or StructureData nodes? Remember that you can use verdi code list and verdi data core.structure list to retrieve them!

Next, we use the get_builder_from_protocol() method to obtain a prepopulated builder for the workflow:

In [3]: builder = PwBandsWorkChain.get_builder_from_protocol(code=code, structure=structure)

The default protocol uses the PBE exchange-correlation functional with suitable pseudopotentials and energy cutoffs from the SSSP library version 1.1 we installed earlier. Finally, we just need to submit the builder in the same way as we did for the calculation job:

In [4]: from aiida.engine import submit
   ...: workchain_node = submit(builder)

And done! Just like that, we have prepared and submitted an automated process to obtain the band structure of silicon. We will now see how to follow its execution and check the results.

Monitoring the process#

If you want to check the status of the workflow, you can exit the verdi shell and run verdi process list just like you did for calculations:

$ verdi process list
  PK  Created    Process label     Process State    Process status
----  ---------  ----------------  ---------------  ---------------------------------------
 113  19s ago    PwBandsWorkChain  ⏵ Waiting        Waiting for child processes: 115
 115  15s ago    PwRelaxWorkChain  ⏵ Waiting        Waiting for child processes: 118
 118  13s ago    PwBaseWorkChain   ⏵ Waiting        Waiting for child processes: 123
 123  11s ago    PwCalculation     ⏵ Waiting        Monitoring scheduler: job state RUNNING

Total results: 4

Info: last time an entry changed state: 8s ago (at 23:32:21 on 2021-02-09)

You may notice that verdi process list now shows more than one entry: indeed, there are a couple of calculations and sub-workflows that need to be run. The total workflow should take about 5-10 minutes to finish.

Another way of getting information on the processes of a workchain in a way that shows the hierarchical overview of the calls is by running verdi process status. If your workchain is still running, you will get a partial output compared to the one shown below:

$ verdi process status <PK>
PwBandsWorkChain<113> Finished [0] [7:results]
    ├── PwRelaxWorkChain<115> Finished [0] [3:results]
    │   ├── PwBaseWorkChain<118> Finished [0] [7:results]
    │   │   ├── create_kpoints_from_distance<119> Finished [0]
    │   │   └── PwCalculation<123> Finished [0]
    │   └── PwBaseWorkChain<132> Finished [0] [7:results]
    │       ├── create_kpoints_from_distance<133> Finished [0]
    │       └── PwCalculation<137> Finished [0]
    ├── seekpath_structure_analysis<144> Finished [0]
    ├── PwBaseWorkChain<151> Finished [0] [7:results]
    │   ├── create_kpoints_from_distance<152> Finished [0]
    │   └── PwCalculation<156> Finished [0]
    └── PwBaseWorkChain<164> Finished [0] [7:results]
        └── PwCalculation<167> Finished [0]

The bracket [7:result] indicates the current step in the outline of the PwBandsWorkChain (step 7, with name result). The process status is particularly useful for debugging complex work chains, since it helps pinpoint where a problem occurred.

Tip

You can combine verdi process status with the watch command to continuously monitor the work chain status:

$ watch verdi process status <PK>

The work chain might take a couple minutes to complete. While you wait, you can try uploading your .cif file to the SeeK-path tool on Materials Cloud.

Important

The protocols and the get_builder_from_protocol() have only been fairly recently implemented in aiida-quantumespresso! Be sure to check the open issues on GitHub related to this feature and run some tests before starting calculations in production.

Displaying the results#

Once the work chain has finished running, use verdi process show <PK> to inspect the PwBandsWorkChain and find the PK of its band_structure output:

$ verdi process show <PK>
(...)
Outputs                PK  Type
-------------------  ----  -------------
band_parameters       352  Dict
band_structure        350  BandsData
primitive_structure   327  StructureData
scf_parameters        341  Dict
seekpath_parameters   325  Dict
(...)

Look for the PK of the output with link band_structure of type BandsData. You can then plot the band structure using the verdi shell:

$ verdi data core.bands export --format mpl_pdf --output band_structure.pdf <PK>

Open the band_structure.pdf file with a viewer at your disposal. It should look similar to the one shown here:

../../_images/si_bands.png

Fig. 7 Band structure computed by the PwBandsWorkChain.#

Customizing the inputs#

Sometimes you may want to take advantage of the convenience of getting a prepopulated builder from the get_builder_from_protocol() method while still being able to customize some of the inputs.

The straightforward way to do this is by first obtaining the prepopulated builder and overriding its inputs with the desired values. First use the get_builder_from_protocol() method to obtain the builder:

In [1]: PwBandsWorkChain = WorkflowFactory('quantumespresso.pw.bands')
   ...: code = load_code(<CODE_PK>)
   ...: structure = load_node(<STRUCTURE_PK>)
   ...: builder = PwBandsWorkChain.get_builder_from_protocol(code=code, structure=structure)

Next, have a look at the top-level bands_kpoints_distance input, used by the work chain to determine the linear density along the k-point path constructed using SeeK-path:

In [2]: builder.bands_kpoints_distance
Out[2]: <Float: uuid: ab9eb3f6-7c61-4a6d-964c-5b55964e306c (unstored) value: 0.025>

Say you’re only doing a test calculation, and want to increase the distance between the k-points. Simply adapt the Float input to the desired value:

In [3]: builder.bands_kpoints_distance = Float(0.2)

and once again submit the builder to the AiiDA engine:

In [3]: from aiida.engine import submit
   ...: workchain_node = submit(builder)

Once the work chain is complete, you can have a look at the newly calculated band structure and compare it with the old one.

Using the overrides argument#

Adapting the builder might be a good option for small modifications, but for more complex and powerful customizations, the get_builder_from_protocol() method also offers the option to use the overrides argument.

Important

For the following demonstration you will need the pseudo-dojo pseudopotentials you used for the exercises of the module on running calculations. You can check if it is available and its label by running the following command in the terminal:

$ aiida-pseudo list
Label                                Type string                Count
-----------------------------------  -------------------------  -------
PseudoDojo/0.4/PBE/SR/standard/upf   pseudo.family.pseudo_dojo  72
SSSP/1.1/PBEsol/precision            pseudo.family.sssp         85
SSSP/1.1/PBEsol/efficiency           pseudo.family.sssp         85
SSSP/1.1/PBE/precision               pseudo.family.sssp         85
SSSP/1.1/PBE/efficiency              pseudo.family.sssp         85

If you see no pseudo-dojo entry, you can check the dropdown box below to see how to install them.

Overrides allow access to some of the underlying construction features of internal methods, such as the pseudo_family input of the PwBaseWorkChains launched by the PwBandsWorkChain:

In [2]: pseudo_family = "PseudoDojo/0.4/PBE/SR/standard/upf"
   ...: overrides = {
   ...:     'relax': {'base': {'pseudo_family': pseudo_family}},
   ...:     'scf': {'pseudo_family': pseudo_family},
   ...:     'bands': {'pseudo_family': pseudo_family},
   ...: }
   ...: builder2 = PwBandsWorkChain.get_builder_from_protocol(code=code, structure=structure, overrides=overrides)

By investigating the inputs, you will notice that not only the pseudos inputs in builder['bands']['pw'], builder['scf']['pw'] and builder['relax']['base']['pw'] are different, but also the ecutwfc and ecutrho have been adapted to the hints provided by the Pseudo-dojo library:

In [3]: builder['scf']['pw']['parameters']['SYSTEM']
Out[3]:
{'nosym': False,
 'occupations': 'smearing',
 'smearing': 'cold',
 'degauss': 0.01,
 'ecutwfc': 30.0,
 'ecutrho': 240.0}

In [4]: builder2['scf']['pw']['parameters']['SYSTEM']
Out[4]:
{'nosym': False,
 'occupations': 'smearing',
 'smearing': 'cold',
 'degauss': 0.01,
 'ecutwfc': 36.0,
 'ecutrho': 144.0}

As a final exercise, run the PwBandsWorkChain with the pseudo-dojo pseudos, and compare the resulting band structure with the one obtained using the SSSP library version 1.1.

Important

Key takeaways

  • Work chains can be prepared from a builder and submitted similar to calculations.

  • The method get_builder_from_protocol() can return a pre-populated builder from minimal inputs.

  • In addition to verdi process list, you can see a hierarchical overview of a workflow by using verdi process status.

  • You can customize the builder returned by get_builder_from_protocol() by providing the overrides optional argument.

Even after only running a couple of work chains, you can already see that it becomes more and more difficult to find the data that you are looking for. In the “Organising your data” module, you will learn how to use Groups to keep your database more tidy.