Paperlab Notebook Regression Baselines

PaperLab Notebook Regression Baselines

When to Use

USE WHEN: a book has notebooks whose outputs support chapter figures, tables, worked examples, or quantitative claims.

Pair with:

• neqsim-notebook-patterns for notebook execution structure,
• neqsim-regression-baselines for numerical drift principles,
• paperlab_scientific_traceability_audit for claim provenance.

Baseline Types

Classification

• pass: outputs are within tolerance and figures are fresh.
• stale: notebook likely runs, but figure or claim output is outdated.
• broken: execution or import fails.
• missing-baseline: notebook has important outputs but no baseline yet.
• expensive-skip: execution is known expensive and requires manual review.

Output Schema

{
  "notebook": "chapters/ch10/notebooks/ch10_s02_teg_contactor.ipynb",
  "status": "pass",
  "engine": "neqsim_dev_setup",
  "baselines": [
    {
      "name": "dry_gas_water_ppm",
      "value": 32.4,
      "unit": "ppm",
      "abs_tol": 0.5,
      "rel_tol": 0.02
    }
  ],
  "figures": [
    {"path": "figures/ch10_teg_sensitivity.png", "status": "fresh"}
  ]
}

Safety Rules

• Do not overwrite accepted baselines without explicit approval.
• Record environment metadata whenever execution occurs.
• Prefer stable engineering outputs over fragile full-notebook diffs.