How to create custom workflows with NOMAD entries

This how-to will walk you through how to use nomad-utility-workflows to generate the yaml file required to define a custom workflow.

Prerequisites

• A very basic understanding of NOMAD terminology: entry, mainfile
• Python environment with this utility module
• Example simulation data files (provided below)

Example Overview

To demonstrate, we will use the following 3 step molecular dynamics equilibration workflow:

1. Geometry optimization (energy minimization)
2. Equilibration MD simulation in NPT ensemble
3. Production MD simulation in NVT ensemble

The final result will be a workflow graph visualization in NOMAD that looks like this:

Example Data Structure

Each task in this workflow represents a supported entry in NOMAD. All three simulations will be uploaded together with the workflow.archive.yaml file, within the following structure:

upload.zip
├── workflow.archive.yaml  # The file we'll create in this guide
├── Emin
│   ├── mdrun_Emin.log     # Geometry Optimization mainfile
│   └── ...other raw simulation files
├── Equil_NPT
│   ├── mdrun_Equil-NPT.log  # NPT equilibration mainfile
│   └── ...other raw simulation files
└── Prod_NVT
    ├── mdrun_Prod-NVT.log   # NVT production mainfile
    └── ...other raw simulation files

Download Example Data

Complete Workflow Creation Example

Let's create the workflow from start to finish. First, import the necessary packages:

import gravis as gv
import networkx as nx

from nomad_utility_workflows.utils.workflows import (
    NodeAttributes,
    NodeAttributesUniverse,
    build_nomad_workflow,
    nodes_to_graph,
)

Step 1: Define the workflow structure

We'll define our workflow structure using a dictionary of NodeAttributes:

node_attributes = {
    0: NodeAttributes(
        name='input system',
        type='input',
        path_info={
            'mainfile_path': 'Emin/mdrun_Emin.log',
            'supersection_index': 0,
            'section_index': 0,
            'section_type': 'system',
        },
        out_edge_nodes=[1],
    ),
    1: NodeAttributes(
        name='Geometry Optimization',
        type='workflow',
        entry_type='simulation',
        path_info={'mainfile_path': 'Emin/mdrun_Emin.log'},
    ),
    2: NodeAttributes(
        name='Equilibration NPT Molecular Dynamics',
        type='workflow',
        entry_type='simulation',
        path_info={'mainfile_path': 'Equil_NPT/mdrun_Equil-NPT.log'},
        in_edge_nodes=[1],
    ),
    3: NodeAttributes(
        name='Production NVT Molecular Dynamics',
        type='workflow',
        entry_type='simulation',
        path_info={'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log'},
        in_edge_nodes=[2],
    ),
    4: NodeAttributes(
        name='output system',
        type='output',
        path_info={
            'section_type': 'system',
            'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log',
        },
        in_edge_nodes=[3],
    ),
    5: NodeAttributes(
        name='output properties',
        type='output',
        path_info={
            'section_type': 'calculation',
            'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log',
        },
        in_edge_nodes=[3],
    ),
}

node_attributes_universe = NodeAttributesUniverse(nodes=node_attributes)

IMPORTANT

To ensure that all functionalities work correctly, the node keys must be unique integers that index the nodes. For example, node_keys = [0, 1, 2, 3, 4, 5] for a graph with 6 nodes.

Step 2: Create the workflow graph

Now, convert the node attributes dictionary to a graph using nodes_to_graph() and display the resulting workflow graph with gravis (gv):

workflow_graph_input_minimal = nodes_to_graph(node_attributes_universe)

gv.d3(
    workflow_graph_input_minimal,
    node_label_data_source='name',
    edge_label_data_source='name',
    zoom_factor=1.5,
    node_hover_tooltip=True,
)

The visualization of the input graph should look like this:

Step 3: Generate the workflow YAML

Finally, define workflow_metadata (i.e., the filename of the output yaml and the name of the workflow) and generate the workflow YAML file with build_nomad_workflow():

workflow_metadata = {
    'destination_filename': './workflow_minimal.archive.yaml',
    'workflow_name': 'Equilibration Procedure',
}

workflow_graph_output_minimal = build_nomad_workflow(
    workflow_metadata=workflow_metadata,
    workflow_graph=nx.DiGraph(workflow_graph_input_minimal),
    write_to_yaml=True,
)

The resulting workflow_minimal.archive.yaml file will look like this:

'workflow2':
  'name': 'Equilibration Procedure'
  'inputs':
  - 'name': 'input system'
    'section': '../upload/archive/mainfile/Emin/mdrun_Emin.log#/run/0/system/0'
  'outputs':
  - 'name': 'output system'
    'section': '../upload/archive/mainfile/Prod_NVT/mdrun_Prod-NVT.log#/run/0/system/-1'
  - 'name': 'output properties'
    'section': '../upload/archive/mainfile/Prod_NVT/mdrun_Prod-NVT.log#/run/0/calculation/-1'
  'tasks':
  - 'm_def': 'nomad.datamodel.metainfo.workflow.TaskReference'
    'name': 'Geometry Optimization'
    'task': '../upload/archive/mainfile/Emin/mdrun_Emin.log#/workflow2'
    'inputs':
    - 'name': 'input run/0/system/0 from input system'
      'section': '../upload/archive/mainfile/Emin/mdrun_Emin.log#/run/0/system/0'
    'outputs':
    - 'name': 'output system from Geometry Optimization'
      'section': '../upload/archive/mainfile/Emin/mdrun_Emin.log#/run/0/system/-1'
    - 'name': 'output calculation from Geometry Optimization'
      'section': '../upload/archive/mainfile/Emin/mdrun_Emin.log#/run/0/calculation/-1'
  - 'm_def': 'nomad.datamodel.metainfo.workflow.TaskReference'
    'name': 'Equilibration NPT Molecular Dynamics'
    'task': '../upload/archive/mainfile/Equil_NPT/mdrun_Equil-NPT.log#/workflow2'
    'inputs':
    - 'name': 'input system from Geometry Optimization'
      'section': '../upload/archive/mainfile/Emin/mdrun_Emin.log#/run/0/system/-1'
    'outputs':
    - 'name': 'output system from Equilibration NPT Molecular Dynamics'
      'section': '../upload/archive/mainfile/Equil_NPT/mdrun_Equil-NPT.log#/run/0/system/-1'
    - 'name': 'output calculation from Equilibration NPT Molecular Dynamics'
      'section': '../upload/archive/mainfile/Equil_NPT/mdrun_Equil-NPT.log#/run/0/calculation/-1'
  - 'm_def': 'nomad.datamodel.metainfo.workflow.TaskReference'
    'name': 'Production NVT Molecular Dynamics'
    'task': '../upload/archive/mainfile/Prod_NVT/mdrun_Prod-NVT.log#/workflow2'
    'inputs':
    - 'name': 'input system from Equilibration NPT Molecular Dynamics'
      'section': '../upload/archive/mainfile/Equil_NPT/mdrun_Equil-NPT.log#/run/0/system/-1'
    'outputs':
    - 'name': 'output run/0/system/-1 from output system'
      'section': '../upload/archive/mainfile/Prod_NVT/mdrun_Prod-NVT.log#/run/0/system/-1'
    - 'name': 'output system from Production NVT Molecular Dynamics'
      'section': '../upload/archive/mainfile/Prod_NVT/mdrun_Prod-NVT.log#/run/0/system/-1'
    - 'name': 'output calculation from Production NVT Molecular Dynamics'
      'section': '../upload/archive/mainfile/Prod_NVT/mdrun_Prod-NVT.log#/run/0/calculation/-1'

Uploading and Viewing Your Workflow

After generating the workflow.archive.yaml file:

1. Place it in the root directory of your example data, matching the Example Data Structure above
2. Upload to NOMAD via the API or Drag-and-Drop
3. Navigate to the entry overview page associated with the workflow.archive.yaml file.

You should see a workflow visualization identical to the one in Example Overview.

Understanding the Workflow Graph

The visualization of the output graph should look like this:

gv.d3(
    workflow_graph_output_minimal,
    node_label_data_source='name',
    edge_label_data_source='name',
    zoom_factor=1.5,
    node_hover_tooltip=True,
)

The workflow graph has two representations:

1. Input graph: The simplified graph you define with your node attributes
2. Output graph: The expanded graph generated by build_nomad_workflow() that includes additional nodes and connections

What happens during graph transformation?

When you run build_nomad_workflow(), the function:

1. Adds default input/output nodes for each workflow node
2. Connects these nodes appropriately
3. Generates the YAML representation

For nodes with entry_type = 'simulation', the automatically generated outputs include:

• The System section
• The Calculation section

Alternative Approach: Creating a Graph Manually

If you prefer to create the graph structure directly with NetworkX instead of using nodes_to_graph(), you can do so:

# Create an empty directed graph
workflow_graph_input = nx.DiGraph()

# Add nodes with attributes
workflow_graph_input.add_node(0,
    name='input system',
    type='input',
    path_info={
        'mainfile_path': 'Emin/mdrun_Emin.log',
        'supersection_index': 0,
        'section_index': 0,
        'section_type': 'system'
    }
)

workflow_graph_input.add_node(1,
    name='Geometry Optimization',
    type='workflow',
    entry_type='simulation',
    path_info={
        'mainfile_path': 'Emin/mdrun_Emin.log'
    }
)

# Add more nodes...

# Add edges to connect the nodes
workflow_graph_input.add_edge(0, 1)
workflow_graph_input.add_edge(1, 2)
# Add more edges...

Adding additional inputs/outputs

You can add inputs/outputs beyond the defaults set by the utility by simply adding them to the node attributes dictionary:

node_attributes = {
    0: NodeAttributes(
        name='input system',
        type='input',
        path_info={
            'mainfile_path': 'Emin/mdrun_Emin.log',
            'supersection_index': 0,
            'section_index': 0,
            'section_type': 'system',
        },
        out_edge_nodes=[1],
    ),
    1: NodeAttributes(
        name='Geometry Optimization',
        type='workflow',
        entry_type='simulation',
        path_info={'mainfile_path': 'Emin/mdrun_Emin.log'},
        outputs=[
            {
                'name': 'energies of the relaxed system',
                'path_info': {
                    'section_type': 'energy',
                    'supersection_path': 'run/0/calculation',  # this can be done,
                    # but at this point it's safer / easier to just use archive_path
                    'supersection_index': -1,
                },
            }
        ],
    ),
    2: NodeAttributes(
        name='Equilibration NPT Molecular Dynamics',
        type='workflow',
        entry_type='simulation',
        path_info={'mainfile_path': 'Equil_NPT/mdrun_Equil-NPT.log'},
        in_edge_nodes=[1],
        outputs=[
            {
                'name': 'MD workflow properties (structural and dynamical)',
                'path_info': {
                    'section_type': 'results',
                },
            }
        ],
    ),
    3: NodeAttributes(
        name='Production NVT Molecular Dynamics',
        type='workflow',
        entry_type='simulation',
        path_info={'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log'},
        in_edge_nodes=[2],
        outputs=[
            {
                'name': 'MD workflow properties (structural and dynamical)',
                'path_info': {
                    'section_type': 'results',
                },
            }
        ],
    ),
    4: NodeAttributes(
        name='output system',
        type='output',
        path_info={
            'section_type': 'system',
            'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log',
        },
        in_edge_nodes=[3],
    ),
    5: NodeAttributes(
        name='output properties',
        type='output',
        path_info={
            'section_type': 'calculation',
            'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log',
        },
        in_edge_nodes=[3],
    ),
}

node_attributes_universe = NodeAttributesUniverse(nodes=node_attributes)

workflow_graph_input = nodes_to_graph(node_attributes_universe)

gv.d3(
    workflow_graph_input,
    node_label_data_source='name',
    edge_label_data_source='name',
    zoom_factor=1.5,
    node_hover_tooltip=True,
)

You can explore the details of the graph nodes using:

for node_key, node_attributes in workflow_graph_input.nodes(data=True):
    print(node_key, node_attributes)

output

0 {'name': 'input system', 'type': 'input', 'entry_type': None, 'path_info': {'mainfile_path': 'Emin/mdrun_Emin.log', 'supersection_index': 0, 'section_type': 'system', 'section_index': 0}, 'in_edge_nodes': [], 'out_edge_nodes': [1]}
1 {'name': 'Geometry Optimization', 'type': 'workflow', 'entry_type': 'simulation', 'path_info': {'mainfile_path': 'Emin/mdrun_Emin.log', 'archive_path': 'workflow2'}, 'in_edge_nodes': [], 'out_edge_nodes': []}
2 {'name': 'Equilibration NPT Molecular Dynamics', 'type': 'workflow', 'entry_type': 'simulation', 'path_info': {'mainfile_path': 'Equil_NPT/mdrun_Equil-NPT.log', 'archive_path': 'workflow2'}, 'in_edge_nodes': [1], 'out_edge_nodes': []}
3 {'name': 'Production NVT Molecular Dynamics', 'type': 'workflow', 'entry_type': 'simulation', 'path_info': {'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log', 'archive_path': 'workflow2'}, 'in_edge_nodes': [2], 'out_edge_nodes': []}
4 {'name': 'output system', 'type': 'output', 'entry_type': None, 'path_info': {'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log', 'section_type': 'system'}, 'in_edge_nodes': [3], 'out_edge_nodes': []}
5 {'name': 'output properties', 'type': 'output', 'entry_type': None, 'path_info': {'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log', 'section_type': 'calculation'}, 'in_edge_nodes': [3], 'out_edge_nodes': []}
6 {'type': 'output', 'name': 'energies of the relaxed system', 'path_info': {'section_type': 'energy', 'supersection_path': 'run/0/calculation', 'supersection_index': -1}}
7 {'type': 'output', 'name': 'MD workflow properties (structural and dynamical)', 'path_info': {'section_type': 'results'}}
8 {'type': 'output', 'name': 'MD workflow properties (structural and dynamical)', 'path_info': {'section_type': 'results'}}

and the details of the graph edges using:

for edge_1, edge_2, edge_attributes in workflow_graph_input.edges(data=True):
    print(edge_1, edge_2, edge_attributes)

output

0 1 {}
1 6 {'inputs': [{'name': 'energies of the relaxed system', 'path_info': {'section_type': 'energy', 'supersection_path': 'run/0/calculation', 'supersection_index': -1}}]}
1 2 {}
2 7 {'inputs': [{'name': 'MD workflow properties (structural and dynamical)', 'path_info': {'section_type': 'results'}}]}
2 3 {}
3 8 {'inputs': [{'name': 'MD workflow properties (structural and dynamical)', 'path_info': {'section_type': 'results'}}]}
3 4 {}
3 5 {}

And again we generate the final workflow graph and yaml:

workflow_metadata = {
    'destination_filename': './workflow.archive.yaml',
    'workflow_name': 'Equilibration Procedure',
}

workflow_graph_output = build_nomad_workflow(
    workflow_metadata=workflow_metadata,
    workflow_graph=nx.DiGraph(workflow_graph_input),
    write_to_yaml=True,
)

gv.d3(
    workflow_graph_output,
    node_label_data_source='name',
    edge_label_data_source='name',
    zoom_factor=1.5,
    node_hover_tooltip=True,
)

For a better understanding of the workflow graph generation, you can compare the output graph nodes and edgess with the input graph details above:

for node_key, node_attributes in workflow_graph_output.nodes(data=True):
    print(node_key, node_attributes)

output

0 {'name': 'input system', 'type': 'input', 'entry_type': None, 'path_info': {'mainfile_path': 'Emin/mdrun_Emin.log', 'supersection_index': 0, 'section_type': 'system', 'section_index': 0, 'archive_path': 'run/0/system/0'}, 'in_edge_nodes': [], 'out_edge_nodes': [1]}
1 {'name': 'Geometry Optimization', 'type': 'workflow', 'entry_type': 'simulation', 'path_info': {'mainfile_path': 'Emin/mdrun_Emin.log', 'archive_path': 'workflow2'}, 'in_edge_nodes': [], 'out_edge_nodes': []}
2 {'name': 'Equilibration NPT Molecular Dynamics', 'type': 'workflow', 'entry_type': 'simulation', 'path_info': {'mainfile_path': 'Equil_NPT/mdrun_Equil-NPT.log', 'archive_path': 'workflow2'}, 'in_edge_nodes': [1], 'out_edge_nodes': []}
3 {'name': 'Production NVT Molecular Dynamics', 'type': 'workflow', 'entry_type': 'simulation', 'path_info': {'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log', 'archive_path': 'workflow2'}, 'in_edge_nodes': [2], 'out_edge_nodes': []}
4 {'name': 'output system', 'type': 'output', 'entry_type': None, 'path_info': {'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log', 'section_type': 'system', 'archive_path': 'run/0/system/-1'}, 'in_edge_nodes': [3], 'out_edge_nodes': []}
5 {'name': 'output properties', 'type': 'output', 'entry_type': None, 'path_info': {'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log', 'section_type': 'calculation', 'archive_path': 'run/0/calculation/-1'}, 'in_edge_nodes': [3], 'out_edge_nodes': []}
6 {'type': 'output', 'name': 'energies of the relaxed system', 'path_info': {'section_type': 'energy', 'supersection_path': 'run/0/calculation', 'supersection_index': -1, 'mainfile_path': 'Emin/mdrun_Emin.log', 'archive_path': 'run/0/calculation/-1/energy/-1'}}
7 {'type': 'output', 'name': 'MD workflow properties (structural and dynamical)', 'path_info': {'section_type': 'results', 'mainfile_path': 'Equil_NPT/mdrun_Equil-NPT.log', 'archive_path': 'workflow2/results/-1'}}
8 {'type': 'output', 'name': 'MD workflow properties (structural and dynamical)', 'path_info': {'section_type': 'results', 'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log', 'archive_path': 'workflow2/results/-1'}}
9 {'type': 'input', 'name': 'input run/0/system/0 from input system', 'path_info': {'mainfile_path': 'Emin/mdrun_Emin.log', 'supersection_index': 0, 'section_type': 'system', 'section_index': 0, 'archive_path': 'run/0/system/0'}, 'is_default': True}
10 {'type': 'output', 'name': 'output system from Geometry Optimization', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'system', 'mainfile_path': 'Emin/mdrun_Emin.log'}}
11 {'type': 'output', 'name': 'output calculation from Geometry Optimization', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'calculation', 'mainfile_path': 'Emin/mdrun_Emin.log'}}
12 {'type': 'input', 'name': 'input system from Geometry Optimization', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'system', 'mainfile_path': 'Emin/mdrun_Emin.log'}}
13 {'type': 'output', 'name': 'output system from Equilibration NPT Molecular Dynamics', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'system', 'mainfile_path': 'Equil_NPT/mdrun_Equil-NPT.log'}}
14 {'type': 'output', 'name': 'output calculation from Equilibration NPT Molecular Dynamics', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'calculation', 'mainfile_path': 'Equil_NPT/mdrun_Equil-NPT.log'}}
15 {'type': 'input', 'name': 'input system from Equilibration NPT Molecular Dynamics', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'system', 'mainfile_path': 'Equil_NPT/mdrun_Equil-NPT.log'}}
16 {'type': 'output', 'name': 'output system from Production NVT Molecular Dynamics', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'system', 'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log'}}
17 {'type': 'output', 'name': 'output calculation from Production NVT Molecular Dynamics', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'calculation', 'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log'}}

for edge_1, edge_2, edge_attributes in workflow_graph_output.edges(data=True):
    print(edge_1, edge_2, edge_attributes)

output

0 1 {'inputs': [], 'outputs': [{'name': 'input run/0/system/0 from input system', 'path_info': {'mainfile_path': 'Emin/mdrun_Emin.log', 'supersection_index': 0, 'section_type': 'system', 'section_index': 0, 'archive_path': 'run/0/system/0'}, 'is_default': True}]}
1 6 {'inputs': [{'name': 'energies of the relaxed system', 'path_info': {'section_type': 'energy', 'supersection_path': 'run/0/calculation', 'supersection_index': -1, 'mainfile_path': 'Emin/mdrun_Emin.log', 'archive_path': 'run/0/calculation/-1/energy/-1'}}, {'name': 'output system from Geometry Optimization', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'system', 'mainfile_path': 'Emin/mdrun_Emin.log'}}, {'name': 'output calculation from Geometry Optimization', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'calculation', 'mainfile_path': 'Emin/mdrun_Emin.log'}}], 'outputs': []}
1 2 {'inputs': [], 'outputs': [{'name': 'input system from Geometry Optimization', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'system', 'mainfile_path': 'Emin/mdrun_Emin.log'}}]}
1 10 {}
1 11 {}
2 7 {'inputs': [{'name': 'MD workflow properties (structural and dynamical)', 'path_info': {'section_type': 'results', 'mainfile_path': 'Equil_NPT/mdrun_Equil-NPT.log', 'archive_path': 'workflow2/results/-1'}}, {'name': 'output system from Equilibration NPT Molecular Dynamics', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'system', 'mainfile_path': 'Equil_NPT/mdrun_Equil-NPT.log'}}, {'name': 'output calculation from Equilibration NPT Molecular Dynamics', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'calculation', 'mainfile_path': 'Equil_NPT/mdrun_Equil-NPT.log'}}], 'outputs': []}
2 3 {'inputs': [], 'outputs': [{'name': 'input system from Equilibration NPT Molecular Dynamics', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'system', 'mainfile_path': 'Equil_NPT/mdrun_Equil-NPT.log'}}]}
2 13 {}
2 14 {}
3 8 {'inputs': [{'name': 'MD workflow properties (structural and dynamical)', 'path_info': {'section_type': 'results', 'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log', 'archive_path': 'workflow2/results/-1'}}, {'name': 'output system from Production NVT Molecular Dynamics', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'system', 'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log'}}, {'name': 'output calculation from Production NVT Molecular Dynamics', 'path_info': {'entry_id': None, 'upload_id': None, 'section_type': 'calculation', 'mainfile_path': 'Prod_NVT/mdrun_Prod-NVT.log'}}], 'outputs': []}
3 4 {'inputs': [], 'outputs': []}
3 5 {'inputs': [], 'outputs': []}
3 16 {}
3 17 {}
9 1 {}
12 2 {}
15 3 {}

The resulting workflow.archive.yaml file will look like:

'workflow2':
  'name': 'Equilibration Procedure'
  'inputs':
  - 'name': 'input system'
    'section': '../upload/archive/mainfile/Emin/mdrun_Emin.log#/run/0/system/0'
  'outputs':
  - 'name': 'MD workflow properties (structural and dynamical)'
    'section': '../upload/archive/mainfile/Prod_NVT/mdrun_Prod-NVT.log#/workflow2/results/-1'
  - 'name': 'output system'
    'section': '../upload/archive/mainfile/Prod_NVT/mdrun_Prod-NVT.log#/run/0/system/-1'
  - 'name': 'output properties'
    'section': '../upload/archive/mainfile/Prod_NVT/mdrun_Prod-NVT.log#/run/0/calculation/-1'
  'tasks':
  - 'm_def': 'nomad.datamodel.metainfo.workflow.TaskReference'
    'name': 'Geometry Optimization'
    'task': '../upload/archive/mainfile/Emin/mdrun_Emin.log#/workflow2'
    'inputs':
    - 'name': 'input run/0/system/0 from input system'
      'section': '../upload/archive/mainfile/Emin/mdrun_Emin.log#/run/0/system/0'
    'outputs':
    - 'name': 'energies of the relaxed system'
      'section': '../upload/archive/mainfile/Emin/mdrun_Emin.log#/run/0/calculation/-1/energy/-1'
    - 'name': 'output system from Geometry Optimization'
      'section': '../upload/archive/mainfile/Emin/mdrun_Emin.log#/run/0/system/-1'
    - 'name': 'output calculation from Geometry Optimization'
      'section': '../upload/archive/mainfile/Emin/mdrun_Emin.log#/run/0/calculation/-1'
  - 'm_def': 'nomad.datamodel.metainfo.workflow.TaskReference'
    'name': 'Equilibration NPT Molecular Dynamics'
    'task': '../upload/archive/mainfile/Equil_NPT/mdrun_Equil-NPT.log#/workflow2'
    'inputs':
    - 'name': 'input system from Geometry Optimization'
      'section': '../upload/archive/mainfile/Emin/mdrun_Emin.log#/run/0/system/-1'
    'outputs':
    - 'name': 'MD workflow properties (structural and dynamical)'
      'section': '../upload/archive/mainfile/Equil_NPT/mdrun_Equil-NPT.log#/workflow2/results/-1'
    - 'name': 'output system from Equilibration NPT Molecular Dynamics'
      'section': '../upload/archive/mainfile/Equil_NPT/mdrun_Equil-NPT.log#/run/0/system/-1'
    - 'name': 'output calculation from Equilibration NPT Molecular Dynamics'
      'section': '../upload/archive/mainfile/Equil_NPT/mdrun_Equil-NPT.log#/run/0/calculation/-1'
  - 'm_def': 'nomad.datamodel.metainfo.workflow.TaskReference'
    'name': 'Production NVT Molecular Dynamics'
    'task': '../upload/archive/mainfile/Prod_NVT/mdrun_Prod-NVT.log#/workflow2'
    'inputs':
    - 'name': 'input system from Equilibration NPT Molecular Dynamics'
      'section': '../upload/archive/mainfile/Equil_NPT/mdrun_Equil-NPT.log#/run/0/system/-1'
    'outputs':
    - 'name': 'MD workflow properties (structural and dynamical)'
      'section': '../upload/archive/mainfile/Prod_NVT/mdrun_Prod-NVT.log#/workflow2/results/-1'
    - 'name': 'output system from Production NVT Molecular Dynamics'
      'section': '../upload/archive/mainfile/Prod_NVT/mdrun_Prod-NVT.log#/run/0/system/-1'
    - 'name': 'output calculation from Production NVT Molecular Dynamics'
      'section': '../upload/archive/mainfile/Prod_NVT/mdrun_Prod-NVT.log#/run/0/calculation/-1'

And after uploading to NOMAD with the simulation files as before, you should get a workflow entry with the visualization graph:

By clicking on the middle task box "Equilibration NPT Mole...", you will open the sub-workflow where the additional inputs and outputs added to this task are clearly visible:

References

For more details on node attributes and other options, see:

• Explanation > Workflow > Node Attributes
• Reference > Workflows