Discovering graph usage

The usage of a graph is described using the parameters() and outputs() methods. The methods will return the variables and outputs for all the actions in the graph.

Retrieving the graph variables

The parameters() method returns the set of Variable objects that a graph or action expects as inputs.

For actions that read parameterised input files (DynaAnalysis, RadiossAnalysis, OpenFOAMAnalysis, JinjaReplace), the input file must be present in the current working directory when parameters() is called. WorkArea and SimulationIterator both handle this automatically: they copy all required files to a work directory before delegating to the graph.

Using WorkArea:

from kunene.graph_actions import WorkFlow, WorkArea
from kunene.dyna_actions import DynaAnalysis
from kunene.d3plot_actions import d3plot_File

wf = WorkFlow(name='SpringWorkFlow')
wf.add_action(DynaAnalysis(name='RunSpring', input_path='spring.k'))
wf.add_action(d3plot_File(name='field'))

wa = WorkArea(wf, copy_paths=['path/to/spring.k'])
for v in wa.parameters():
    print(v)

Using SimulationIterator:

from kunene.graph_actions import WorkFlow, SimulationIterator
from kunene.dyna_actions import DynaAnalysis

wf = WorkFlow(name='SpringWorkFlow')
wf.add_action(DynaAnalysis(name='RunSpring', input_path='spring.k'))

itr = SimulationIterator(wf, copy_paths=['path/to/spring.k'])
for v in itr.parameters():
    print(v)

Neither reads the deck in place, so the source files are never modified: WorkArea copies them into its work directory, and SimulationIterator into a variables_discovery subdirectory of the results root. Each variable carries its name, type, default value and a description:

Variable Name: floatpar1, Data Type: float, Value: 1.23, Description: 'From 'spring.k''
Variable Name: intpar2, Data Type: int, Value: 789, Description: 'From 'spring.k''

The value is the default read from the deck. The description is generated by the action, not taken from the deck: every solver action records the file the variable was found in, so it identifies the source rather than describing the quantity.

Retrieving the graph outputs

The outputs() method returns information about what an action produces. For a single action it returns a (eval_type, description) tuple. For a WorkFlow or DirectedGraph it returns a dictionary {action_name: (eval_type, description)} covering all child actions.

from kunene.graph_actions import WorkFlow, WorkArea
from kunene.dyna_actions import DynaAnalysis
from kunene.d3plot_actions import d3plot_File

wf = WorkFlow(name='SpringWorkFlow')
wf.add_action(DynaAnalysis(name='RunSpring', input_path='spring.k'))
wf.add_action(d3plot_File(name='field'))

for name, (eval_type, description) in wf.outputs().items():
    print(f'{name}: {eval_type} — {description}')

Visualising the action graph

Any action can print itself as a tree, rooted at that action. Call it on a top-level action (such as a SimulationIterator or WorkArea) to see the whole workflow. The tree shows each wrapper, graph, and leaf action.

from kunene.graph_actions import WorkFlow, SimulationIterator
from kunene.dyna_actions import DynaAnalysis
from kunene.d3plot_actions import d3plot_File

wf = WorkFlow(name='SpringWorkFlow')
wf.add_action(DynaAnalysis(name='RunSpring', input_path='spring.k'))
wf.add_action(d3plot_File(name='field'))

itr = SimulationIterator(wf, work_area_path='results', copy_paths=['spring.k'])
itr.print_tree()

SimulationIterator 'SpringWorkFlow_Iter'
└── WorkFlow 'SpringWorkFlow'
    ├── DynaAnalysis 'RunSpring'
    └── d3plot_File 'field'

Pass describe=True to append each action’s description to its node.

Visualising the work directory

The print_work_dir() method shows the directory layout that running the workflow creates on disk. It is a predicted layout built from the actions’ metadata, so it works before anything is run and does not require the solvers to be installed. A SimulationIterator shows a representative job_0/ directory (one is created per design evaluation), while a WorkArea shows the single directory it reuses.

itr.print_work_dir()

results/   (results root)
├── status.json   (run progress: current job, jobs done; see kunene.progress)
├── jobs_index.json   (job -> variable values and group labels; ...)
├── job_0/   (one directory per design evaluation)
│   ├── iter_variables.json   (this design's variable values)
│   ├── actions_output.pkl   (this design's action outputs)
│   ├── spring.k   (copied in)
│   ├── status.json   (live action states; see kunene.progress)
│   ├── dyna_variables.json
│   ├── dyna_action_inp.k
│   ├── run_file.stdout
│   ├── run_file.stderr
│   ├── d3plot*
│   └── d3hsp
└── job_1/ … job_N/

When several sub-workflows run in their own directories — for example a DirectedGraph whose children are WorkArea or SimulationIterator wrappers — each child’s directory appears as a nested subtree:

from kunene.graph_actions import DirectedGraph, WorkFlow, WorkArea
from kunene.dyna_actions import DynaAnalysis

def make_wa(i):
    wf = WorkFlow(name=f'WF{i}')
    wf.add_action(DynaAnalysis(name=f'run{i}', input_path='spring.k'))
    return WorkArea(wf, work_area_path=f'WF{i}', copy_paths=['spring.k'])

graph = DirectedGraph('aGraph')
graph.add_action(make_wa(1))
graph.add_action(make_wa(2))
graph.print_work_dir()

./   (current working directory)
├── status.json   (live action states; see kunene.progress)
├── WF1/   (work area, overwritten each run)
│   ├── spring.k   (copied in)
│   ├── status.json   (live action states; see kunene.progress)
│   ├── dyna_variables.json
│   ├── dyna_action_inp.k
│   ├── run_file.stdout
│   ├── run_file.stderr
│   ├── d3plot*
│   └── d3hsp
└── WF2/   (work area, overwritten each run)
    ├── spring.k   (copied in)
    └── ...   (the same files, written by run2)

Call describe_workflow() to print both the action tree and the work directory structure together. The formatting helpers format_tree() and format_work_dir() return the same output as strings instead of printing it.