The article introduces MaRDIFlow, a computational workflow framework developed by the Max Planck Institute, aimed at improving reproducibility in scientific software. The framework automates metadata abstraction from mathematical objects, reducing dependencies and creating structured descriptions of workflow components. This approach allows for interchangeable and redundant use of components while ensuring detailed output to facilitate experiment reproduction. Various use cases, including CO2 conversion rates in a methanization reactor model, demonstrate the tool's effectiveness and versatility.
MaRDIFlow's core functionality is centered on automating metadata abstraction in computational workflows, facilitating reproducibility through well-defined input-output behavior of its components.
The prototype of MaRDIFlow was evaluated through various use cases, showcasing its capability to streamline the workflow process in computational experiments.
Collection
[
|
...
]