pyplotcontour_examples

pyplotcontour_examples — Enhancement-218: pyplot -contour

pyplot -contour <z> <x> <y> renders a 2-D contour map of a quantity z over the (x, y) plane — the natural view of a 2-D parameter sweep:

pyplot -contour i(vd) vgs vds          # a device current over a (Vgs, Vds) grid
pyplot -contour gain rval cval         # a gain surface over an (R, C) grid
pyplot mymap -contour p x y            # named output (mymap.py/.data/.png)

The three arguments are the height/colour (z) and the two axes (x, y), each a flattened, equal-length sweep vector. matplotlib triangulates the (x, y) points (tricontourf), so gridded or scattered sweep data plots with no grid-dimension metadata needed. Where do x, y, z come from? Any 2-D sweep that leaves three equal-length vectors — a nested .dc, the sweep command family (E-146, E-190), a .step grid, or vectors you build in .control.

It reuses the entire pyplot pipeline (E-94 onward) — the matplotlib back end, set pyplot_terminal=png|svg|pdf for a headless render, styles, figure size, backend. Three contour-specific knobs are added:

• set pyplot_contour_levels=<N> — number of contour levels (default: let matplotlib choose);
• set pyplot_contour_lines — overlay labelled black contour lines on the filled map;
• set pyplot_contour_cmap=<name> — the colormap (default viridis).

The demos

(A) pyplotcontour_demo.cir — an analytic surface. A grid built in .control with a known surface so the contour can be checked, not just rendered:

• z = x² + y² over x, y ∈ [-2, 2] → concentric circular contours (a paraboloid: z = 0 at the centre, z = 8 at the corners).

(B) bridge_dc_demo.cir — a real nested .dc sweep. The everyday use: a diode-OR bridge whose output V(c) follows whichever of its two inputs is higher. A nested .dc sweeps both inputs; since v1 drives node a and v2 drives node b, V(a)/V(b) are the two swept values at every point, so pyplot -contour v(c) v(a) v(b) maps the output over the (V(a), V(b)) plane — a max-like corner surface rising toward the top and right edges. This shows the feature on genuine simulation output (not just .control math), with the turbo colormap and labelled contour lines.

What is verified

verify_pyplotcontour.py (19 checks, both solvers) runs both decks and parses the generated .data/.py and PNGs.

For (A) the analytic surface:

1. the -contour path is taken (tricontourf, not plot/hist; a colorbar labelled z; axes labelled x/y);
2. the data table has three columns (x, y, z), all N rows;
3. the column mapping is correct — z reconstructs x² + y² from the .data;
4. the sweep is genuinely 2-D — x and y each span a real range, and z runs from ~0 at the grid centre to ~8 at a corner;
5. a valid, non-trivial PNG is rendered.

For (B) the real nested .dc:

1. the -contour path is taken with the requested knobs (tricontourf, cmap=turbo, overlaid lines; colorbar v(c), axes v(a)/v(b));
2. the nested .dc produced the flattened 51×51 = 2601-row, 3-column grid;
3. the axes are the two real swept sources (V(a), V(b) each span [-1, 1]);
4. the output is the diode-OR surface — ~0 when both inputs are low, rising when either input is high, maximal at the both-high corner (confirms the columns map correctly onto a genuine circuit result);
5. a valid PNG is rendered.

Run

python3 verify_pyplotcontour.py