# Scientific visualization profile

Use for scientific measurements, experimental data, and dense analytical figures, even inside a product. Also load common visualization fundamentals and the relevant chart-type guide.

## Mandatory colourmap protocol

Scientific visualizations must use an approved map from [`colourmaps.json`](colourmaps.json) as the source of all data-mark colors. The catalogue contains ordered primitive source `tokens`, optional exact sRGB `lightColors` overrides, and exact sRGB `darkColors`; it is canonical. Do not assemble an ad hoc palette from unrelated hues.

Choose by data structure:

- `sequential` — ordered magnitude from low to high.
- `diverging` — values around a meaningful midpoint, baseline, or neutral state.
- `multi_sequential` — separate ordered regimes that require related ramps.
- `cyclic` — periodic values whose beginning and end meet.
- Categorical data — distinct, contrast-safe stops sampled from one selected map.

`grayC` is `explicit_request_only`. Never select it autonomously. A user's explicit greyscale request activates it.

Choose the map once for a coordinated artifact or working session and retain it across related charts. Change it when the user asks or when the selected structure cannot truthfully encode a later chart; if a change is necessary, state why.

Independent charts may use different stops from the selected map. Facets of one measurement should retain the same mapping. Within one plot, every color difference must correspond to a series, category, magnitude, status, threshold, selection, or other documented meaning.

Do not import product success/warning/error colors into scientific data marks. Select colors by the scientific variable and encoding, including dense multi-series data. Only a domain-established convention or explicit user request changes that mapping; document it and retain non-color cues.

For light charts, use `lightColors` when present; otherwise resolve `tokens`. Source tokens remain provenance when an explicit variant overrides them.

For dark charts, use the selected map’s `darkColors` in the same stop order, not its light-theme `tokens`. These opaque colors meet 3:1 on dark `--bg-canvas` and `--bg-surface-1`; recheck other backgrounds and opacity. Sequential dark variants increase CIELAB lightness in approximately 5 L* steps. Diverging and cyclic maps intentionally reuse lightness across opposing arms; they cannot distinguish every category in grayscale. Keep labels or patterns for categorical distinctions. These adaptations are not a claim of Crameri perceptual uniformity or validated color-vision-deficiency safety. Matplotlib callers pass `theme="dark"` to the helper APIs; CSS/JS callers select by the rendered chart theme.
