Universal Kernel

marking

How are the symmetry groups of two theories compared?

An injective homomorphism from the reference group into the group a theory supplies; it makes the theory’s set a set acted on by the reference group.

0123456∞the antiflags (1, L){0, ∞} {4, 5} {2, 6} {1, 3}0123456∞the antiflags (p, 246){0, ∞} {2, 3} {1, 5} {4, 6}z ↦ 3z(1 3 2 6 4 5)not in psl(2,7)
Plate 1.3Two perfect matchings of P1(F7)\Proj^1(\F_7): the antiflags with the common point 1 (gold) and those with the common line 246 (blue). The map z↦3zz\mapsto3z, outside the group, carries one onto the other.
Definition(Marking)

Let a group Γ\Gamma act on a set YY. A marking of (Γ,Y)(\Gamma,Y) by GG is an injective homomorphism μ ⁣:G→Γ\mu\colon G\to\Gamma. The marked set YμY_\mu is the GG-set with underlying set YY and action g⋅y=μ(g)yg\cdot y=\mu(g)y.

Two theories rarely share a group on the nose: the automorphism group of the Klein quartic and the group of the Fano plane are different groups that happen to be isomorphic, and a seam between incarnations in the two exists only after both groups are marked. The word is borrowed from Teichmüller theory, where a marked surface carries an identification of its fundamental group with a fixed reference group. It is unrelated to Burnside’s marks.

Lemma(Change of marking)

Let μ ⁣:G→Γ\mu\colon G\to\Gamma be a marking and α∈Aut⁡(G)\alpha\in\Aut(G). (a) The stabilizer of yy in YμαY_{\mu\alpha} is α−1\alpha^{-1} of its stabilizer in YμY_\mu; so if YμY_\mu is transitive, st⁡(Yμα)=α−1(st⁡(Yμ))\st(Y_{\mu\alpha})=\alpha^{-1}(\st(Y_\mu)). (b) If α\alpha is inner, α(g)=hgh−1\alpha(g)=hgh^{-1}, then y↦μ(h)yy\mapsto\mu(h)y is a GG-isomorphism Yμ→YμαY_\mu\to Y_{\mu\alpha}.

So an inner change of marking does not change the isomorphism class of the marked set, and an outer one moves the stabilizer class by that automorphism, changing nothing when the class is fixed by Aut⁡(G)\Aut(G). The seams themselves do depend on the marking, even through inner automorphisms.

Proof

(a) μ(α(g))y=y\mu(\alpha(g))y=y if and only if α(g)\alpha(g) lies in the stabilizer of yy in YμY_\mu. (b) μ(h)μ(g)y=μ(hgh−1)μ(h)y\mu(h)\mu(g)y=\mu(hgh^{-1})\mu(h)y.

Theorem(Seams without markings)

Let Γ≤Sym⁡(Y)\Gamma\le\operatorname{Sym}(Y) and Γ′≤Sym⁡(Y′)\Gamma'\le\operatorname{Sym}(Y') be transitive permutation groups and HH a point stabilizer of Γ\Gamma. A permutation isomorphism is a bijection f ⁣:Y→Y′f\colon Y\to Y' with fΓf−1=Γ′f\Gamma f^{-1}=\Gamma'.

(a) A permutation isomorphism ff determines θf ⁣:γ↦fγf−1\theta_f\colon\gamma\mapsto f\gamma f^{-1}, and for every marking μ\mu of Γ\Gamma, ff is a seam Yμ→Yθfμ′Y_\mu\to Y'_{\theta_f\mu}. Conversely, if μ ⁣:G→Γ\mu\colon G\to\Gamma and μ′ ⁣:G→Γ′\mu'\colon G\to\Gamma' are isomorphisms, every seam Yμ→Yμ′′Y_\mu\to Y'_{\mu'} is a permutation isomorphism ff with θf=μ′μ−1\theta_f=\mu'\mu^{-1}.

(b) If there is a permutation isomorphism Y→Y′Y\to Y', the set of all of them is a torsor for the normalizer NSym⁡(Y)(Γ)N_{\operatorname{Sym}(Y)}(\Gamma), and there is an exact sequence

1⟶Aut⁡Γ(Y)⟶NSym⁡(Y)(Γ)⟶Aut⁡(Γ)[H]⟶1,1\longrightarrow\Aut_\Gamma(Y)\longrightarrow N_{\operatorname{Sym}(Y)}(\Gamma)\longrightarrow\Aut(\Gamma)_{[H]}\longrightarrow1,

where Aut⁡(Γ)[H]\Aut(\Gamma)_{[H]} consists of the automorphisms that preserve the class [H][H]. Hence ∣NSym⁡(Y)(Γ)∣=∣NΓ(H):H∣⋅∣Aut⁡(Γ)[H]∣|N_{\operatorname{Sym}(Y)}(\Gamma)|=|N_\Gamma(H):H|\cdot|\Aut(\Gamma)_{[H]}|.

Proof

(a) f(μ(g)y)=θf(μ(g))f(y)f(\mu(g)y)=\theta_f(\mu(g))f(y), and conversely a seam satisfies fμ(g)f−1=μ′(g)f\mu(g)f^{-1}=\mu'(g) for all gg.

(b) If f,f′f,f' are permutation isomorphisms then f−1f′f^{-1}f' normalizes Γ\Gamma, and fnfn is one whenever nn does. The kernel of n↦(γ↦nγn−1)n\mapsto(\gamma\mapsto n\gamma n^{-1}) is the centralizer of Γ\Gamma, which is Aut⁡Γ(Y)\Aut_\Gamma(Y), and the image preserves [H][H] because nΓyn−1=Γnyn\Gamma_yn^{-1}=\Gamma_{ny}. Conversely, if α\alpha preserves [H][H], then YY with the action γ∗y=α(γ)y\gamma\ast y=\alpha(\gamma)y has the same stabilizer class, so the stabilizer principle gives a Γ\Gamma-isomorphism n ⁣:Y→Yαn\colon Y\to Y^\alpha, which normalizes Γ\Gamma and induces α\alpha.

Corollary(Seams are markings)

Let XX be a rigid object of GG whose stabilizer class is fixed by Aut⁡(G)\Aut(G), and let Γ≤Sym⁡(Y)\Gamma\le\operatorname{Sym}(Y), Γ′≤Sym⁡(Y′)\Gamma'\le\operatorname{Sym}(Y') be permutation groups isomorphic to GG whose marked sets are incarnations of XX. Then f↦θff\mapsto\theta_f is a bijection from the permutation isomorphisms Y→Y′Y\to Y' onto the isomorphisms Γ→Γ′\Gamma\to\Gamma', and there are ∣Aut⁡(G)∣|\Aut(G)| of them.

In words: for such an object, once the two symmetry groups are identified the identification of the two sets is forced, and every identification of the groups occurs.

Example

Among the orbital graphs of the object of size 28 are two families of seven K4K_4’s: in the antiflags, those with a common point and those with a common line. On the projective line each K4K_4 is four disjoint pairs covering P1(F7)\Proj^1(\F_7). Which family corresponds to the points of the Fano plane depends on the marking: the map z↦3zz\mapsto3z of PGL⁡(2,7)\PGL(2,7), which lies outside GG and induces an outer automorphism, exchanges the two families.

The same marking fixes the labels aa and bb of the classes V4V_4, A4A_4 and S4S_4, and changing it by an outer automorphism exchanges them. The class of the object of size 28 is fixed by Aut⁡(G)\Aut(G), so the normalizer of GG in the symmetric group of its 28 points has order ∣NG(S3):S3∣⋅∣Aut⁡(G)∣=336|N_G(S_3):S_3|\cdot|\Aut(G)|=336; in the pairs it is PGL⁡(2,7)\PGL(2,7).

Example

Two theories asked to agree can single out a class of markings. On a Coxeter edge, read in the antiflag model, the point rule and the line rule trace directed 4-cycles that give mutually inverse elements of order 4, and the bracket rule on the projective line orients the edge’s harmonic pair of pairs by the square class of [a,c][c,b][b,a][a,c][c,b][b,a]. The bracket rule agrees with the point rule when the marking differs from μA\mu_A by an inner automorphism, and with the line rule when it differs by an outer one. So a marking carries the stabilizers of the bisection {0,1,2,5} ∣ {3,4,6,∞}\{0,1,2,5\}\,|\,\{3,4,6,\infty\} to the stabilizers of points of the Fano plane if and only if the bracket rule agrees with the point rule: the square classes of F7\F_7, which GG preserves and PGL⁡(2,7)\PGL(2,7) does not, tell the points of the Fano plane from its lines. The modular curve does the same for the inner class through its holomorphic structure.