simulation-reproduction

To convert evidence into runnable scripts, you provide the recovered evidence bundle to execute a reproduction workflow that generates executable code, numerical outputs, and reproducible figures. This process applies equations and parameters within a task workspace.

Provenance recording works by tracing outputs back to the evidence bundle across planning, implementation, and verification stages. This allows reviewers to verify that numerical outputs and reproducible figures faithfully match the original scientific computation parameters.

The best way to generate reproducible figures is by executing deterministic scripts within an isolated task workspace. This workspace captures stdout, stderr, and runtime, then persists the resulting figures, raw data, and metadata alongside provenance records.

Yes, you can execute numerical reproduction workflows within an isolated task workspace. This environment satisfies deterministic scripting and workspace isolation requirements, independently capturing stdout, stderr, and runtime while persisting outputs.

Yes, the reproduction workflow supports tracking plot semantics and metadata across planning, implementation, and verification stages. It records provenance to ensure reviewers can trace numerical outputs and figures back to the original evidence bundle.

A key limitation is the requirement for deterministic scripting and a recovered evidence bundle containing sufficient equations, parameters, and plot semantics. Without complete evidence, the workspace cannot properly generate runnable scripts or persist accurate provenance.