Examples

Runnable Jupyter notebooks, rendered directly in these docs.

• Basic Usage

  Create a store, build a small lineage, visualise it. Start here.

• Machine Learning Pipeline

  A multi-step workflow: the iris dataset fanned out across scalers, embeddings and clusterers, then queried after the fact.

Worked pipelines

Four longer examples from different fields. Each takes a different graph shape, so together they cover the whole API.

• Genomics Cohort

  A fan-in. Eight samples down identical paths, joined into one cohort. Rules, a sample that fails QC and is kept as evidence, and a multi-parent join.

• Design Optimisation

  A chain. Simulated annealing over a heat sink, thirty designs deep. Deep lineage, an abandoned branch, and prune reclaiming its space.

• Survey Reprocessing

  A campaign. A telescope survey reduced twice after a calibration fault. Generations, identical-node reuse, and scoring two runs against each other.

• Quant Backtest Grid

  A lattice. Strategy configurations across walk-forward folds, with a look-ahead bug planted, traced and withdrawn. Contamination tracing, reuse_identical=False for audit, and counting trials for a multiple-testing correction.