Result: exactly 6 distinct shapes.
The 16-cell has 8 vertices ±e_i and 24 edges (between every pair of
non-antipodal vertices). Edges are NOT axis-parallel (each has 2 nonzero
components), so the alignment constraint is much tighter than for the
tesseract — and the search saturates.
| rng | candidate directions | hits | distinct shapes |
|---|---|---|---|
| 2 | 1,819 | 46 | 6 |
| 3 | 20,474 | 92 | 6 |
| 4 | 135,750 | 154 | 6 |
Same 6 shapes at every range tested — confidently saturated empirically.
(See ..\docs\METHODOLOGY.md for caveats about empirical-vs-formal
saturation.)
| # | Balls | Edge signature | Geometry | Example kernel n |
|---|---|---|---|---|
| 1 | 8 | B:6, R:12, Y:6 | square antiprism A | (√5, 1, 1, 1) |
| 2 | 7 | B:12, G:12 | octahedron + center (vertex-first) | (1, 0, 0, 0) |
| 3 | 6 | B:6, Y:16, (2 collapsed) | octahedron | (1, 1, 0, 0) |
| 4 | 8 | B:6, R:12, Y:6 | square antiprism B | (φ²·1.618, φ²·1.618, φ²·1.618, φ⁻¹) |
| 5 | 8 | B:12, G:12 | cube (cell-first) | (1, 1, 1, 1) |
| 6 | 8 | B:6, R:12, Y:6 | square antiprism C | (φ³·1.382·…, 1.382, 1.382, 1.382) |
Notes on the three “square antiprism” shapes: they have identical
edge color signatures {B:6, R:12, Y:6} but different rotation+scale-
invariant pairwise-distance fingerprints — i.e. different aspect ratios
arising from projection along the three icosahedral-axis classes
(2-fold, 3-fold, 5-fold) that correspond to the 16-cell’s tetrahedral/D4
sub-symmetries.
These three shapes are the well-known D₄ triality of the 16-cell:
the three orbits of 2(1+i)·I (where I = binary icosahedral group) under
the D₄ outer automorphism, all giving 16-cell projections with identical
strut counts (8 connectors, 6 R1 + 6 R2 + 6 Y2 + 6 B2 = 24 struts) but
distinct geometries. This matches the construction described at
https://polytopologist.github.io/zome_pages/zometriality.htm, which
also notes that truncating any one of the three to its edge midpoints
yields the same 24-cell zometool model — independent confirmation that
these correspond to our 24cell_triality.vZome projection.
n = e_i): kernel is one axis vertex; image is a
centered octahedron (shape 2).n = (1,1,1,1)): kernel is the centroid of a tetrahedral
cell; image is a cube (shape 5).n = (1,1,0,0)): kernel is along an edge midpoint;
image is a flattened octahedron (shape 3).cd regular_4polytopes\lib
python run_search.py 16cell 3
➡️ Open this page on GitHub Pages to interact with the 3D models below (the embeds only render when this file is served via GitHub Pages, not in github.com’s markdown preview).