Evan Angelone, Egon Schulte
6 min
Abstract
We present a construction of chiral and regular polyhedra from subgroups of the general affine group AGL(1,q) for odd prime powers q. In particular, we show that the full group AGL(1,q) occurs as the automorphism group of a chiral polyhedron of type {q-1, q-1} when q=1 mod 4, or types {q-1,(q-1)/2} or {(q-1)/2, q-1} when q=3 mod 4, and we compute the genus in each case. We also establish that subgroups of AGL(1,q) cannot serve as full automorphism groups of regular polytopes of rank 3 or higher, nor of chiral polytopes of rank 4 or higher, demonstrating that our construction captures all polytopes that can arise from this class of affine groups.
Alex: Huh—so σ₂ is the spinner, τ is a quick back-and-forth flip, and σ₁ combines them without overlap. But why does that capture the full AGL(1,q)?
Sam: Powers of σ₂ conjugate τ—meaning they apply σ₂, then τ, then undo σ₂—which produces all the pure shifts, or translations, across the field. Since a generates the multipliers and these conjugates fill the translation subgroup, the whole group generated is AGL(1,q).
Alex: So the spinning conjugates of the flip sweep out the entire translation layer, locking in the full affine group without mirrors.
Sam: Precisely. Crucially, there's no extra involutory automorphism that would swap σ₂ to its inverse while fixing others in a way that adds reflections—unlike regular cases. This keeps it chiral, yielding polyhedra of type {q-1, q-1} if q ≡ 1 mod 4, or {(q-1)/2, q-1} otherwise, on surfaces of specific genus.
Alex: Rotation subgroup meaning half the symmetries, without the flips? How do they tell which case it is?
Sam: Imagine taking that flip τ and twisting it with every power of the spinner σ₂—like grabbing a fixed reflection and rotating copies of it around. All those twisted copies generate a layer called the normal closure of τ, which covers the full set of flips and shifts in a subgroup. Since σ₂ doesn't fit inside that flip-shift layer—because its scaling isn't just a flip—the whole group becomes either the chiral group or the rotation half of a regular polyhedron's symmetries.
Alex: But what do these polyhedra actually look like in terms of their faces and connections?
Sam: The shapes have faces that are regular polygons with a certain number of sides, and at each corner exactly a certain number meet. Researchers label this as type {s,t}, where s is the number of sides per face and t is how many faces meet at each vertex. For the full group cases, it's {q-1, q-1} when q is 1 mod 4, or {(q-1)/2, q-1} otherwise. They live on orientable surfaces with a genus, which counts the number of holes through the surface, like a donut has genus 1.
Alex: Okay, familiar ones for small q like 5 and 7 on the torus. But are there catches—like when this makes regular instead of chiral shapes?
Sam: For certain parameters where the scaling factor's order divides p^{l/2} +1 with l even, it produces regular polyhedra instead, with the group as the rotation half, extended by an extra involution. But the criteria aren't fully general—there might be other ways subgroups yield regulars that aren't covered yet.
Alex: Huh. So multiple parameter choices might give the same shape up to relabeling.
Sam: Indeed, and for chiral polyhedra the face and vertex numbers must divide q-1. The paper proves no subgroup of AGL(1,q) works for regular polyhedra or higher-dimensional chiral polytopes, limiting this to rank-three cases. For odd prime powers q not 3, it yields an infinite family on orientable surfaces.
Alex: A solid, delimited advance—knowing exactly where these affine groups fit in the picture of chiral shapes. Thanks, Sam, that's clarified the scope nicely.
Sam: My pleasure, Alex. This work sharpens our understanding of how finite field symmetries shape combinatorial geometry.