Before discussing the group structure of the skewb, an orientation must be constructed. First, orient the cube and label the sides by R,L,U,D,F,B. A 120 degree clockwise rotation of a corner is denoted by the letters of the three faces which meet at that corner. For example, UFR is the 120 degree clockwise rotation of the up-front-right corner. Such moves are the basic moves of the skewb and are called "generators" since the eight moves of this type generate the skewb group. Any generator of the skewb will permute three corners while leaving the opposite side of the cube alone. It should be noted that twisting about a corner and twisting about the antipodal opposite corner are equivalent moves up to a rotation of the entire cube.
Let
Let C denote the set of square center facets of the skewb and SC denote the symmetric group on C. Let V denote the set of vertices of the skewb and SV denote the symmetric group on V. Finally, let y(g) be an element of C38 denoting the twist on all eight corners.
Claim: There are homomorphisms
Claim: G is a subgroup of H.
Let GC be the group that acts only on the center facets of the skewb, and GV the group that acts only on the vertices. Now,
The group that acts on the vertices of the skewb is slightly more complicated. Unlike the Rubik's cube, there is no condition like conservation of twists which applies to the entire vertex set. Instead, we must split the vertices of the skewb into two 4-corner orbits. This idea is borrowed from Bandelow's notes on Mickey's Challenge (a puzzle similar to the Masterball). An orbit is constructed by starting with one corner and including the opposite corner of each face that meets at the first corner. Referring back to our original labeling of the skewb, the orbits are the odd corners {1,3,5,7}, and the even corners {2,4,6,8}. Let the orbit of odd corners be denoted by V(odd), and let V(even) denote the orbit of even corners. We now partition GV so that
Let h=(s,y(h),t,z(h)) be an element of GV.
First we will examine the permutations of the vertices of both orbits. Each generator produces a 3-cycle on the vertices, whether they are in the odd or even orbit. Therefore, we can say that the permutations of each orbit generate A4 by the same argument used for the center permutations. Now,
Claim 1: There exist h, such that s is an element of A44, y(h)= (0,0,0,0), t is an element of A4, and z(h)= (0,0,0,0). We know this is true because there are clean skewb moves which only permute 3 vertices.
Claim 2: There exist h, such that s=1, y(h)= any even permutation of (1,2,0,0), t=1, z(h)= any even permutation of (1,2,0,0). This is true because there are clean skewb moves which only twist vertices.
If we combine the moves of Claims 1 and 2, we should generate all of the possible moves of GV. The condition on each of the vertex 4-tuples will drop them in dimension to elements of C33. So we can conclude that GV is a subgroup of index 9 of the semi-direct product
Since GC= A6, and GV= (A4 x C33)x (A4 x C33), we can conclude that
In conclusion, it is interesting to note that if we let G' denote the illegal skewb group - where reassembly is permitted - then
Move Center Permutation Vertex Permutation UFR (1 5 2) (2 6 4) UFL (1 4 5) (1 7 3) DFR (1 2 6) (1 5 7) DFL (1 6 4) (4 6 8) BRU (2 5 3) (1 3 5) BLU (3 5 4) (2 4 8) DBR (2 3 6) (2 8 6) DBL (3 4 6) (3 7 5)
Move Vertex Orientation UFR (1 2 0 2 0 2 0 0) UFL (2 0 2 1 0 0 2 0) DFR (2 0 0 0 2 1 2 0) DFL (0 0 0 2 0 2 1 2) BRU (2 1 2 0 2 0 0 0) BLU (0 2 1 2 0 0 0 2) DBR (0 2 0 0 1 2 0 2) DBL (0 0 2 0 2 0 2 1) UFR*UFL (0 2 2 2 0 0 2 0) DFR*DFL (2 0 0 2 0 0 2 2)
Note: The orientations for the generator moves contain two repeated orbits - permutations of (1 0 0 0) and permutations of (2 2 2 0). Also, the combination moves are equivalent to 0 mod 3.