Workflows with Dragon

Dragon supports a range of workflow patterns, from tightly coupled AI/simulation loops to task-parallel parameter sweeps and MPI application orchestration. The examples below cover some common use cases.

• AI-in-the-loop: Couple inference and simulation in a single managed workflow.

• Dragon MPI Workflow: Orchestrate MPI applications as part of a larger Python workflow.

• Dragon Parsl MPI App: Integrate with Parsl for flexible task-parallel workflow management.

• Dragon joblib: Use Dragon as a backend for joblib-based parallel computation.

• Batch DAG Execution: Build and execute directed acyclic graphs of tasks with the Batch API.

• AI-in-the-loop Workflow
• An example MPI workflow using Dragon ProcessGroup
• Example Parsl @mpi_app Using Dragon Based Executor
• Dragon JobLib Examples
• Graph-based Execution with Batch