# Key Concepts

## Task Manager

A task manager is a wrapper around one BlackDynamite study. A task is one pair
of jobs and runs.

The Motoko task manager creates job/run pairs and can select subsets of runs
using constraints. For example, it can select only finished runs:

```python
finished = workflow.mult.select("state = FINISHED")
```

## Workflow

The workflow is the top-level object. It owns all task managers and coordinates
their creation and execution.

```python
from motoko.workflow import Workflow


workflow = Workflow("motoko.yaml")
workflow.create()
workflow.start_launcher_daemons()
workflow.execute(inputs=[2.1, 3.1])
```

The workflow also stores persistent variables in `workflow.vars` and exposes
task managers as attributes, such as `workflow.mult` or `workflow.norm`.

## Orchestrator

An orchestrator is a Python module attached to a workflow. It registers
asynchronous routines and event actions that decide when tasks should be
created.

The workflow configuration points to an orchestrator function with
`module.function` syntax:

```yaml
orchestrator: orchestrator.main
```

Motoko loads `orchestrator.py` from the workflow directory and calls
`main(workflow, **params)`.