Zomeable 4-polytopes

Prismatic uniform 4-polytopes — zomeable orthographic projections

Scope

The non-prismatic convex uniform 4-polytopes (47 of them) are covered in ../README.md. The remaining convex uniform 4-polytopes — the prismatic ones — fall into three families, two of which are infinite. This document sweeps a bounded sample of each family and catalogues every zomeable orthographic projection found.

Sweep parameters: rng = 3, permute_dedup = False, agnostic search (no kernel-inheritance shortcut, no obstruction-based filtering). Each polytope is constructed from scratch and tested identically.

Polytopes covered: 204 (in scope)

Family Description In scope Hit ≥ 1 Total shapes
A Polyhedral prisms P × [0,1] 17 12 45
B Duoprisms {p}×{q} 170 6 8 + 2 inf families
C Antiprismatic prisms A_n × [0,1] 17 1 9
Total   204 19 62 + 2 inf families

Definitions

A zomeable projection is a rank-3 orthogonal linear map π : ℝ⁴ → ℝ³ whose image of the polytope’s vertex set lies in ℤ[φ]³, such that all edge directions can be aligned (by a global rotation) with canonical Zometool RGBY axes. Vertex multiplicity in the image is allowed (a kernel-aligned projection may collapse several 4D vertices to one 3D point).

Strict rank-3: the image must span a 3D subspace. A 4D → 2D projection (image planar, sitting in ℝ³ as z = 0) is not a zomeable projection under this definition. The sweep enforces this through _try_align (a rotation must exist that maps the projected edge directions onto Zome axes — planar images cannot satisfy this for non-collinear edge sets).

Methodology

Identical machinery to the 47-corpus sweep:

  1. Construct each polytope’s vertex set V ⊂ ℝ⁴ and edge list.
  2. Enumerate kernel directions n ∈ ℤ[φ]⁴ with |a|, |b| ≤ 3 (gen_dirs(rng=3, permute_dedup=False) — False because prismatic polytopes lack full S₄ axis-permutation symmetry).
  3. For each n: project, attempt RGBY-alignment of edge directions, group hits by 3D shape signature.
  4. For each unique shape: snap projected vertex coords to ℤ[φ]³ (multi-scale search) and emit a .vZome file.

Constructive follow-up (2026-05-29): the finite snap/signature probes for {4}×{6} and {4}×{10} missed inf families, because their fixed alignment/scale search was not a proof of boundedness. The exact emitter tools/emit_duoprism_inf_family.py now emits representative members directly from the arithmetic conditions below.

Post-audit note: the original generic emitter stopped after the first valid edge-axis alignment. Some symmetric projections have multiple valid alignments, where only a later alignment gives exact ℤ[φ]³ vertex coordinates. The emitter now tries all valid alignments before declaring snap_failed. Rechecking the feature-aligned snap failures recovered four additional models: tetrahedron_prism/face_first_square, truncated_octahedron_prism/cell_first_cube, duoprism_4_6/cell_first_cube, and duoprism_4_10/cell_first_cube.

Driver: tools/run_prismatic_sweep.py. Log: ongoing_work/prismatic_sweep_log.jsonl (one record per polytope).

Results

Family A — polyhedral prisms

17 in scope (the cube prism = tesseract is covered in the main corpus).

3D base polyhedron Distinct shapes
cuboctahedron (cuboctahedron_prism) 2
dodecahedron (dodecahedron_prism) 5
icosahedron (icosahedron_prism) 5
icosidodecahedron (icosidodecahedron_prism) 5
octahedron (octahedron_prism) 2
rhombicosidodecahedron (rhombicosidodecahedron_prism) 5
rhombicuboctahedron (rhombicuboctahedron_prism) 0
snub cube (snub_cube_prism) 0
snub dodecahedron (snub_dodecahedron_prism) 0
tetrahedron (tetrahedron_prism) 2
truncated cube (truncated_cube_prism) 0
truncated cuboctahedron (truncated_cuboctahedron_prism) 0
truncated dodecahedron (truncated_dodecahedron_prism) 5
truncated icosahedron (truncated_icosahedron_prism) 5
truncated icosidodecahedron (truncated_icosidodecahedron_prism) 5
truncated octahedron (truncated_octahedron_prism) 2
truncated tetrahedron (truncated_tetrahedron_prism) 2

Family B — duoprisms {p}×{q}

170 in scope (3 ≤ p ≤ q ≤ 20, skipping (4, 4) = tesseract).

6 duoprisms yielded ≥ 1 zomeable projection.

p q Distinct shapes
3 6 duoprism_3_6 → 1
4 6 duoprism_4_6 → 1 cell-first + inf family
4 10 duoprism_4_10 → 1 cell-first + inf family
5 10 duoprism_5_10 → 1
6 6 duoprism_6_6 → 2
10 10 duoprism_10_10 → 2

The remaining 164 duoprisms produced 0 zomeable projections. Most fail because at least one of the {p}-gon or {q}-gon circles lies in ℤ[√3] (p ∈ {3, 6, 12, …}) or ℤ[√2] (p ∈ {8, …}), which don’t embed in the icosahedral field ℤ[φ]. The icosahedral-compatible regular polygons in range are {4} (in ℤ), {5}, and {10} (both in ℤ[φ]).

Infinite families

For {4}×{q} with kernel n=(a,b,0,0), the {q}-gon plane is preserved and the square factor becomes four parallel polygon layers. The square edges are height differences a+b and a-b, not the individual projected square basis vectors; nevertheless these are still Q(phi) multiples of a single Zome axis.

Family C — antiprismatic prisms

17 in scope (n ∈ [4, 20]; n=3 = octahedral prism, in Family A).

1 antiprismatic prism yielded ≥ 1 zomeable projection.

n Distinct shapes
5 5_antiprismatic_prism → 9

The remaining 16 antiprismatic prisms (n ≠ 5) produced 0 zomeable projections: only the pentagonal antiprism embeds in ℤ[φ]³ via the icosahedron’s non-polar vertices.

Reproduction

# Family A (17 polyhedral prisms)
python tools/run_prismatic_sweep.py --family A --rng 3
# Family B (170 duoprisms)
python tools/run_prismatic_sweep.py --family B --rng 3
# Family C (17 antiprismatic prisms)
python tools/run_prismatic_sweep.py --family C --rng 3

# Regenerate per-polytope RESULTS.md files from the sweep log
python tools/build_prismatic_results.py

# Regenerate this doc + the manifest from the sweep log
python tools/build_prismatic_doc.py

# Emit constructive infinite-family examples for {4}x{6} and {4}x{10}
python tools/emit_duoprism_inf_family.py