Reproducible Experiments

For papers, reports, and benchmark claims, store reproducibility artifacts under reproducibility/.

A strong artifact contains:

• the PyRecEst version or commit hash;
• a lock file or exported environment;
• scenario configuration and random seeds;
• expected metrics with tolerances;
• a single command that regenerates results;
• notes about backend-specific behavior.

Start from reproducibility/templates/paper-artifact/ and keep generated files small enough to inspect in a pull request.

Executable Documentation

Treat public documentation as part of the reproducibility surface. Python code blocks in README and tutorial pages should either be executable or explicitly marked as skipped:

# pyrecest: skip

Use scripts/run_doc_examples.py before release to catch stale snippets.