Strict orthographic, zometool-buildable 3D projections of the A5 (5-simplex) uniform family. This note records the sweep method, the saturation result, and the per-shape symmetry/buildability audit.
The A5 symmetry group is the symmetric group S6 (order 720), acting on the
5-dimensional hyperplane H = { x in R^6 : sum x_i = 0 }. The uniform Wythoff
family has 19 members: the nonempty subsets of Coxeter nodes {1,2,3,4,5},
modulo diagram reversal. For a node subset S, the seed vertex has consecutive
coordinate differences equal to 1 exactly at the nodes in S; the full vertex
set is its S6 orbit.
| Example | Nodes | Seed vertex | Vertices |
|---|---|---|---|
| 5-simplex | {1} |
(1,0,0,0,0,0) |
6 |
| Rectified 5-simplex | {2} |
(1,1,0,0,0,0) |
15 |
| Birectified 5-simplex | {3} |
(1,1,1,0,0,0) |
20 |
| Omnitruncated 5-simplex | {1,2,3,4,5} |
(5,4,3,2,1,0) |
720 |
Every edge of every A5 uniform polytope is parallel to a root e_i - e_j, so a
projection sends all edges to zometool axes iff every difference of projected
coordinate-images is a zome axis (or zero).
Because S6 acts irreducibly on H, the centered vertex covariance of every
A5 polytope is proportional to the same projector Pi_H = I_6 - (1/6) J. Hence
a projection P : R^6 -> R^3 is strict-orthographic for one A5 polytope iff it
is for all of them; the exact integer condition is
6 * sum_i c_i c_i^T - s s^T = lambda I_3 , lambda != 0 , s = sum_i c_i
where c_0 ... c_5 are the columns of P. Translating the image only shifts
the model, so we gauge-fix c_0 = 0; the remaining columns are then zome-axis
vectors (within radius R) whose pairwise differences are also zome axes. This
makes the candidate set small (zome axes, not all of Z[phi]^3).
The sweep (tools/a5/zphi_a5_sweep.py) enumerates these candidate column tuples
with exact Z[phi] arithmetic, keeps the isotropic ones, and dedupes hits by a
scale-and-rotation-invariant 19-signature tuple (one signature per A5 source
polytope).
R=2 and R=3 return the same three projection geometries; R=1 finds one of them. Applied to all 19 A5 Wythoff polytopes, this gives 57 models:
| Shape | First three ball counts | Full-family ball-count range | Notes |
|---|---|---|---|
| A | 5, 11, 14 | 5-240 | buildable (Yellow+Green), scale phi^2/3 |
| B | 5, 11, 14 | 5-300 | direction-zomeable only, scale phi^2 |
| C | 6, 13, 14 | 6-304 | buildable (Green + 2x Blue), scale phi^2 |
Symmetry orders are full Euclidean point-cloud symmetries; order 48 is full octahedral symmetry.
A model is fully buildable when, at a single scale, every edge is a standard
vZome strut length phi^n (or exactly double one), matched independently within
each color orbit. 38 of the 57 models are fully buildable.
Shape C’s 5-simplex is the cleanest example: it projects onto a regular
octahedron, whose 12 edges are standard Green struts and whose 3 long
diagonals are double-length Blue struts. The octahedron edge-to-diagonal ratio
is sqrt(2); because the Green orbit’s base length already carries a sqrt(2)
factor relative to Blue, the doubling supplies the remaining factor of sqrt(2)
so both colors land on real parts at one scale.
Shape B mixes Blue/Yellow/Green edge lengths whose ratios are not reconcilable to standard strut lengths at any single scale, so it is recorded as direction-zomeable only (all edge directions are valid zome axes).
tools/a5/build_gallery_a5.py projects each shape, collapses coincident
vertices, centers the ball cloud on its centroid (exact golden-field), chooses a
(num/den) * phi^n scale preferring a pure power of phi, enlarges by phi^2,
deletes the auto-created origin ball when the centroid is not itself a ball, and
fits the vZome ViewModel to the model bounds.