Result: exactly 1 distinct shape.
The 600-cell has 120 vertices and 720 edges, with H₄ symmetry. Under orthographic projection to 3D, all valid default-color projections give the same shape: a 75-vertex H₃-symmetric polytope with edge color signature {R: 288, B: 180, Y: 240} (12 of the 720 edges collapse).
This is the famous H₄→H₃ “icosahedral projection” of the 600-cell.
| rng | candidate directions | hits | distinct shapes |
|---|---|---|---|
| 2 | 1,819 | 30 | 1 |
| 3 | 20,474 | 60 | 1 |
| 4 | 135,750 | 100 | 1 |
Stable at 1 shape. All hits represent ℤ[φ]⁴ lattice points on a single orbit under the H₄ symmetry of the 600-cell.
Example kernel direction: n = (1, 0, 0, 0) (vertex-first projection)
or n = (1, 1, 1, 1) (cell-first); both give the same image up to 3D
rotation.
This polytope provoked a fingerprint-stability bug in the search infrastructure: numerically scaled-equivalent kernel directions can produce 3D images that differ by a 3D rotation arising from SVD basis ambiguity (degenerate singular values). The fix was to use a more generous tolerance in vertex deduplication (1e-3 instead of 1e-5) when computing the rotation+scale-invariant fingerprint.
After the fix, 600-cell saturates cleanly at 1 shape.
cd regular_4polytopes\lib
python run_search.py 600cell 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).