§9

Commutators [A, B] — the soul of advanced solving

For two operations $A, B$ , their commutator is defined as:

[A, B] := A \cdot B \cdot A^{- 1} \cdot B^{- 1}

If $A$ and $B$ commute, $[A, B] = e$ . The commutator measures how far they fail to commute. In an Abelian group all commutators are trivial — but the cube group is decisively non-Abelian.

Why commutators are so powerful

When

A

and

B

nearly overlap — affecting mostly disjoint pieces but sharing one or two —

[A, B]

cycles those few pieces while leaving everything else untouched. This is the 3-cycle: the elementary atom of blindsolving and FMC.

For example,

[R U R^{'}, D]

is a clean edge 3-cycle. Extracting an operation that moves only 3 pieces out of a group of size

8! \cdot 12! \cdot 3^{8} \cdot 2^{12}

is the near-magical thing commutators do.

Interactive § Commutator [A, B] = A B A⁻¹ B⁻¹

The commutator measures how far A and B fail to commute. If they commute, [A, B] = e.

[R U R', D][U R U', L'][R, U][M, U]

expanded

R U R' D R U' R' D'

identity?

corner cycles

3-cycle

edge cycles

identity (no cycles)

9.1 Commutator atom library

Four typical 3-cycle commutators — each moves exactly 3 cubies. Blindsolvers treat them as an alphabet: any state can be reduced to a sequence of 3-cycles.

[R U R', D]

Edge 3-cycle (UD-axis)

R U R' D R U' R' D'

moves 3 edges, fixes the other 17 cubies

[R', D]

Corner 3-cycle

R' D R D'

moves 3 corners only

[M', U]

M-slice edge cycle

M' U M U'

slice-then-U commutator

[F R F', U]

F-slot edge swap

F R F' U F R' F' U'

a localized 3-cycle near the F2L slot

9.2 The commutator subgroup [G, G]

Definition 9.1

The commutator subgroup (also called derived subgroup) of G is:

[G, G] = ⟨ {[a, b] : a, b \in G} ⟩

The subgroup generated by all commutators. The quotient

G / [G, G]

is G's largest Abelian quotient — what remains after stripping out all non-commutativity.

For the cube group: $G / [G, G] ≅ Z /2$ . This means G's non-Abelian structure is almost everything — the only Abelian information is the parity bit. Equivalently, $[G, G]$ itself has order $∣ G ∣/2 \approx 2.16 \times 1 0^{19}$ .

Corollary 9.2

The set of all even-parity states (sgn = +1) equals [G, G]. So every parity-correct state can be written as a finite product of commutators. This is precisely why the commutator language is so central to blindsolving: it decomposes every reasonable state into elementary atoms.

9.3 Commutators and the centre

Definition 9.3 — centre

The centre of a group:

Z (G) = {z \in G : z g = g z \forall g \in G} .

The elements that commute with everything.

For Abelian groups: $Z (G) = G$ . For sharply non-Abelian groups: $Z (G) = {e}$ only.

Theorem 9.4

The cube group's centre is

Z (G) = {e, superflip}

, of order 2. Only the identity and superflip commute with every face turn.

This is a striking fact: among 4.3 × 10¹⁹ states, exactly two commute with all face turns — one being the celebrated superflip (also among the famous 20-step extremal positions).

Interactive § Centre check — does g commute with every face turn?

For each face turn X, check [g, X] = e. If all six pass, then g ∈ Z(G). Theory says Z(G) = {e, superflip} of order 2.

e (identity)superflipRR U R' U'U2 D2

✓

[g, U] = e

✓

[g, D] = e

✓

[g, L] = e

✓

[g, R] = e

✓

[g, F] = e

✓

[g, B] = e

✓ g ∈ Z(G) — commutes with every face turn

Why |Z(G)| = 2

Let g ∈ Z(G). Then g commutes with every face turn ⇔ g is invariant under conjugation by the symmetry group generated by face turns ⇔ g is invariant under all 48 outer cube symmetries (since face turns generate the full symmetry closure).

Conversely, any element of G fixed under all 48 outer symmetries must have a cycle type that is "self-symmetric" under all those rotations and reflections. This allows only two states:

cp = identity, co = 0, ep = identity, eo = 0 — the identity e
cp = identity, co = 0, ep = identity, eo = (1, 1, …, 1) — superflip

So Z(G) = {e, superflip} and |Z(G)| = 2.

∎

9.4 The Hall–Witt identity

Commutators satisfy a nontrivial algebraic relation analogous to the Jacobi identity for Lie algebras. Write $a^{b} := b^{- 1} ab$ for conjugation:

[[a, b^{- 1}], c]^{b} \cdot [[b, c^{- 1}], a]^{c} \cdot [[c, a^{- 1}], b]^{a} = e

Independently proven by Philip Hall and Ernst Witt in the 1930s. It says "three nested commutators, cycled a → b → c → a, multiply to the identity." For the cube, plug a = R, b = U, c = F: the identity holds automatically, giving an 18-token alg that necessarily equals e (though typically without a clean reduction).

9.5 Derived series — what's after [G, G]?

Define the derived series: $G^{(0)} = G$ , $G^{(k + 1)} = [G^{(k)}, G^{(k)}]$ , giving a descending chain $G \supseteq G^{'} \supseteq G^{''} \supseteq \dots$ . A group is solvable if this chain reaches ${e}$ in finitely many steps.

Term	Definition	Cube order
$G^{(0)} = G$	cube group	4.33 × 10¹⁹
$G^{(1)} = [G, G]$	even-parity states	\|G\|/2 ≈ 2.16 × 10¹⁹
$G^{(2)} = [G^{'}, G^{'}]$	even states with CO=EO=0 (A₈ × A₁₂ projection)	≈ 9.65 × 10¹⁵
$G^{(3)}$	commutator subgroups of A₈ and A₁₂ — both simple, so equal themselves	≈ 9.65 × 10¹⁵
$G^{(k)}, k \geq 3$	stabilises (no further descent)	≈ 9.65 × 10¹⁵

Since A₈ and A₁₂ are simple non-Abelian groups (Jordan 1875), $[A_{n}, A_{n}] = A_{n}$ for n ≥ 5. The derived series stabilises at $G^{(2)}$ , so the cube group is not solvable. This is significant: there is no "iteratively kill the commutator" path to a one-stage Abelian solver — one must invoke the §10 subgroup chain or the §22 global search.

9.6 Lower central series & nilpotency

The lower central series: $γ_{1} (G) = G$ , $γ_{k + 1} (G) = [G, γ_{k} (G)]$ . A group is nilpotent if this chain reaches ${e}$ in finitely many steps. Nilpotent ⇒ solvable, but not conversely.

The cube group is not nilpotent: by 9.5 it is not even solvable, let alone nilpotent. Intuitively, nilpotent groups have a "commutator tower" that flattens out; the cube's lower central series barely descends past $γ_{2} = [G, G]$ because the quotient sits in A₈ × A₁₂. The archetypal nilpotent groups are finite p-groups — and G, mixing ℤ/3, ℤ/2 blocks with a non-solvable A_n core, is the opposite of a p-group.

∎

cuberoot.me · Rubik's Cube as a Group · 2026