Provenance and Data Lineage
Provenance records the full history of a result: which codes, versions, inputs, and machines produced it, so any number can be traced to its origin.
Tracing a number to its source
In a large workflow, a final result passes through many codes and transformations. Provenance, also called data lineage, is the recorded metadata that captures this history: which input datasets, which code versions, which parameters, which machine, and which intermediate outputs led to a given number. With provenance, any result can be traced backward to everything that produced it.
This answers the questions that matter when a result is questioned: what exactly was run, in what order, with what inputs, and can it be regenerated. Without provenance, a plot in a report is an orphan.
Automatic capture
Provenance is most reliable when captured automatically by the orchestration engine rather than recorded by hand. As each task runs, the engine logs its inputs, code version, parameters, environment, and output identifiers, building a queryable graph of the whole computation.
Immutable identifiers
Provenance relies on stable identifiers, content hashes for data and version tags for code, so that references cannot silently drift. Archived deposits with persistent identifiers make results citable and their inputs recoverable years later, complementing the reproducibility toolchain.
Design relevance
For the Hyperion breeder and the burner concepts, provenance lets every simulated design value be traced to the exact run and inputs behind it. This auditable lineage supports the honest framing that the design is a documented body of simulation, distinct from measured hardware performance, before construction begins Q2 2027.
- Records the full history behind a result
- Captured automatically during orchestration
- Relies on immutable content and version identifiers
- Makes every number traceable and auditable