Orientation of Rubik's Cube Pieces
Often in ZZ tutorials, a definition for edge orientation is given as: give the colors
of U/D sides the highest priority, followed by colors of F/B sides and L/R sides. For
each edge, if the direction from high to low priority is the same between the two colors
of the edge itself and of the two center pieces adjacent to the edge, iff it's oriented
correctly. You would use F, F', B, or B' to flip the orientation of the four edges on
the side you rotated. With this definition, an x
move doesn't change the orientation of
any edge, while a y
move does. This shows us that the orientation of edges are
dependant only on an axis, and not on the whole orientation of the cube. (i.e. the center
pieces)
Orientating the edges of a cube also puts it in H1:=⟨U,D,L,R,F2,B2⟩.
(it's supposed to make LL faster, because ZBLL is less than 1LLL but "reasonable" and such)
It might be natural to ask, if we orient the edges along the F/B axis, then orient the edges
along the (say) L/R axis with moves in H1, will the cube enter
H2:=⟨U,D,L2,R2,F2,B2⟩?
Unfortunately no. The corners wouldn't be oriented correctly. For each corner piece,
we can define its orientation number as the number of clockwise rotations it takes to
make the face of the corner with the color of U/D actually on one of U/D faces. Then we
sum them up for all 8 corners modulo 3, and obtain 0. This can be proven by calculating
the number added to the sum for each quarter face turn. However, none of the moves in
H2 will change the orientation number of any single corner.
To prove edge orientation on a single axis implies a solution in H1, we use
a similar argument: [y' x': [F U2 F', R D2 R']]
is a move that changes the
orientation of the corners at ULB and DRF, so we can still reorient corners arbitrarily
using moves in H1. (ZZ itself can be seen as a constructive proof of this,
if you're lazy, but you need to generate OLL and PLL with moves in H1, so
beginner's last layer is better for your brain)
These results can be applied to calculating the number of different states a cube can
have: we can calculate numbers of states for the cube's permutation and orientation
respectively and multiply them up, because we can force a state to be reached by first
orientating the pieces correctly, then permute them to the correct place. 3-cycles with
setups (that's conjugating) generates A8 and A12 for the corner and
edge permutation respectively, but a quarter turn is an odd permutation in terms of both
corners and edges. That makes 8!12!/2 permutations. It's already established
that corner orientation has 38/3 states. There are 12 edges and you can put
different edges into the F side and flip their orientations at once, so there are
212/2 edge orientations. Multiplying them gives 43252003274489856000 total
states.
(Editor note: I got tired while writing)
You can also easily derive some facts about parity in blindfolded solving: exactly half of
the states require parity (not if you exclude some like WCA does though), their solutions
have an odd number of quarter turns.