Celebi / Concepts

Data

Data objects represent datasets in a project — both raw inputs (e.g. ntuples from the experiment) and derived products produced by tasks.

Why register data

• Provenance — Celebi knows which task produced which dataset, and which tasks consume it. This is what makes the dependency graph complete.
• Re-execution — when an upstream dataset changes, exactly the affected downstream tasks can be identified and re-run.
• Traceability — every result can be walked back to the raw inputs it came from.

Data, algorithms, and tasks together form the project’s directed acyclic graph (DAG), which you can inspect with trace or in the desktop UI’s DAG viewer.