Plot Reproduction
A generated plot is a scientific claim in visual form. It should be possible to identify what was plotted, how it was computed, which units and conventions were used, and what validation supports it.
Plot Record
Section titled “Plot Record”Every notebook-generated plot should record:
- figure path,
- source notebook or script,
- canonical page,
- plotted quantity,
- units or dimensionless variables,
- parameters,
- numerical method,
- benchmark or validation check,
- environment,
- license status.
For analytic plots, record the formula and sampling range. For numerical plots, record convergence or error evidence.
Quantity Labels
Section titled “Quantity Labels”Captions and axes should distinguish:
- amplitude ,
- probability density ,
- radial probability density,
- energy,
- phase,
- current,
- expectation value,
- entropy or information measure,
- convergence error.
Do not rely on readers to infer the quantity from the plot shape.
Reproduction Workflow
Section titled “Reproduction Workflow”A good reproduction workflow is:
- run the notebook or script from a clean environment,
- execute validation cells,
- export the figure to the stable path,
- confirm the page caption matches the generated quantity,
- run the full site build.
Generated plots that cannot pass validation should stay out of the committed figure set or be labeled explicitly as exploratory.
Static TikZ Diagrams
Section titled “Static TikZ Diagrams”The current committed figures are source-tracked TikZ diagrams. For these, reproduction means:
- source
.texfile is present, - generated
.svgis present, - labels do not overlap,
- caption and alt text are meaningful,
- build passes with the image path.
TikZ diagrams can be schematic; they should not pretend to be numerical data.
Common Mistakes
Section titled “Common Mistakes”- Plotting a rescaled curve without saying it was rescaled.
- Reusing a caption after changing parameters.
- Exporting from an old notebook run after source changes.
- Showing a convergence plot without a reference value.
- Using color alone to distinguish physical cases.
- Cropping away boundary effects that are part of the numerical result.
Cross-Links
Section titled “Cross-Links”References
Section titled “References”- Matplotlib Developers, Matplotlib Documentation.
- Project Jupyter, Jupyter Documentation.
- W3C Web Accessibility Initiative, Images Tutorial.
- E. Tufte, The Visual Display of Quantitative Information, 2nd ed., Graphics Press, 2001.