Universal Kernel

quotient class

How are automorphisms named when no power is available?

The stabilizer class of the quotient of an incarnation by an automorphism; seams preserve it, and it names the three involutions of the object of size 84.

D8 = N(C2), the half-turn at its centreC40123456∞{0, ∞} {1, 6}{2, 3} {4, 5}the coxeter pairingV4a0123456∞{0, ∞} {4, 5}{1, 6} {2, 3}V4b0123456∞{0, ∞} {2, 3}{1, 6} {4, 5}
Plate 2.10The three involutions of the object of size 84, named by the three subgroups of order 4 of D8D_8 that contain its centre: C4C_4, V4aV_4^a and V4bV_4^b.
Definition

Let aa be an automorphism of an incarnation YY of an object. The orbits of ⟨a⟩\langle a\rangle on YY form an object Y/⟨a⟩Y/\langle a\rangle, and its stabilizer class is the quotient class of aa. If a(y)=nya(y)=ny with n∈NG(Gy)n\in N_G(G_y), the stabilizer of the orbit of yy is ⟨Gy,n⟩\langle G_y,n\rangle.

The name uses only the stabilizer class, applied to the quotient object.

Lemma(Quotient class)

The quotient class of aa depends only on aa up to conjugation in Aut⁡G(Y)\Aut_G(Y), and a seam s ⁣:Y→Y′s\colon Y\to Y' gives sas−1sas^{-1} the same quotient class.

Proof

Since aa commutes with GG, GG permutes the ⟨a⟩\langle a\rangle-orbits transitively. Each ak(y)=nkya^k(y)=n^ky has stabilizer GyG_y, so gg fixes the orbit of yy exactly when g∈nkGyg\in n^kG_y for some kk. A seam carries orbits of aa to orbits of sas−1sas^{-1} and preserves stabilizers.

Proposition(Natural involutions of the object of size 84) computed

The object G/C2G/C_2 has automorphism group NG(C2)/C2≅C2×C2N_G(C_2)/C_2\cong C_2\times C_2. The dihedral group NG(C2)N_G(C_2) of order 8 has exactly three subgroups of order 4 containing its centre, one in each of the classes C4C_4, V4aV_4^a, V4bV_4^b, so the three involutions have these three quotient classes, and the quotient class names the involution. The natural ones are:

(1) on Coxeter arcs, reversing the arc: C4C_4, the quotient being the set of edges;

(2) on centres with one of the four bitangents through them, re-pairing the four bitangents at the centre: the pairing into Coxeter edges has class C4C_4, the two other pairings V4aV_4^a and V4bV_4^b;

(3) on involutions tt with a Sylow 3-subgroup PP normalized by tt, replacing PP by the P′P' with ⟨P,P′⟩=G\langle P,P'\rangle=G: C4C_4; by the P′P' with ⟨P,P′⟩≅A4\langle P,P'\rangle\cong A_4 of class aa or bb: V4aV_4^a or V4bV_4^b;

(4) on quadrangles with an ordered pair of vertices, exchanging the pair: V4aV_4^a;

(5) on pairs of disjoint pairs of P1(F7)\Proj^1(\F_7) with cross-ratio {2,4}\{2,4\}, passing to the other such pairing of the same four points: C4C_4.

Sending a centre cc with a bitangent ℓ\ell to the Coxeter arc from the vertex of ℓ\ell to its neighbour fixed by the involution with centre cc is a seam, and it carries the Coxeter pairing of (2) to the reversal of (1).

Example

So for the object of size 84 every natural involution is named by a subgroup of order 4, and two natural involutions in different theories correspond under every seam exactly when their quotient classes agree. On the object of size 24 the invariant is coarser: both nontrivial automorphisms have quotient class 7:37{:}3, and only the power tells them apart.