External workflow jobs¶

External workflow jobs invoke programs and scripts that are not bundled with ERT, which makes them similar to jobs defined as part of forward models. The difference lies in the way they are run. While workflow jobs are run on the workstation running ERT and go through all the realizations in one loop, forward model jobs run in parallel on HPC clusters.

The executable invoked by the workflow job can be an executable you have written yourself - in any language, or it can be an existing Linux command like e.g. cp or mv.

Internal workflow jobs¶

Warning

Internal workflow jobs are under development and the API is subject to changes

Internal workflow jobs is a way to call custom python scripts as workflows. In order to use this, create a class which inherits from ErtScript:

from ert import ErtScript

class MyJob(ErtScript):
    def run(self):
        print("Hello World")

ERT will initialize this class and call the run function when the workflow is called, either through hooks, or through the gui/cli.

The run function can be called with a number of arguments, depending on the context the workflow is called. There are three distinct ways to call the run function:

1. If the run function is using *args in the run function, only the arguments from the user configuration is passed to the workflow:

class MyJob(ErtScript):
    def run(self, *args):
        print(f"Provided user arguments: {args}")

Note

The name of the argument is not required to be args, that is just convention.

2. There are a number of specially named arguments the user can call which gives access to internal state of the experiment that is running:

class ert.config.workflow_fixtures.WorkflowFixtures¶

ensemble: Ensemble | None¶

es_settings: ESSettings¶

storage: Storage¶

workflow_args: list[Any]¶

Note

The current ensemble will depend on the context. For hooked workflows the ensemble will be: PRE_SIMULATION: parameters and no reponses in ensemble POST_SIMULATION: parameters and responses in ensemble PRE_FIRST_UPDATE/PRE_UPDATE: parameters and responses in ensemble POST_UPDATE: parameters and responses in ensemble The ensemble will switch after the POST_UPDATE hook, and will move from prior -> posterior

class MyJob(ErtScript):
    def run(
        self,
        workflow_args: List,
        ert_config: ErtConfig,
        ensemble: Ensemble,
        storage: Storage,
    ):
        print(f"Provided user arguments: {workflow_args}")

For how to load internal workflow jobs into ERT, see: installing workflows