[next] [prev] [up] Date: Sat, 03 Jan 81 02:48:00 -0500 (EST)
[next] [prev] [up] From: Chris C. Worrell <ZILCH@MIT-MC >
~~~ ~~~ [up] Subject: How to play with the corners of your cube (long message)

At this point I have found sufficient Algorithms such that given a
cube with everything correct except for possibly the four
corner cubies on one side, I can (with a little thought and
reference to my notes) solve the cube in 24qtws.


If all the cubes are in the right place, but possibly oriented
wrong, the following transforms are used to TWIST the corners to
the proper orientations:

T1:  F' (R' D' R D' F D F' D)^2  F        18qtws

T2:  L D (D L' D' L)^2 D' L' R' D' (D' R D R')^2 D R     24qtws

T3:  (D' L' D R D' L D R') (B' L D^2 L' B L B' D^2 B L')
    (FDL,RDF,BDR) => (DLF,DFR,DRB)                20qtws

T4:  (R D' L' D R' D' L D)  (L B' D^2 B L' B' L D^2 L' B)
    (FDL,RDF,BDR) => (LFD,FRD,RBD)                 20qtws

Note: T3 and T4 are inverses based on the same components, which
happen to commute. (see C1 and C2 below)

T5:  (L' U L F U F') D' (F U' F' L' U' L) D          14qtws
    (FDL,RDF) => (DLF,FRD)

T6:  L' F' D' L' D R D' L D F L F' R' F            14qtws
    (FDL,RDF) => (LFD,DRB)

Along the same line as T5 and T6, but not usefull in the
present discussion, shown to me in a private message from

T7:  F' R' D' R U R' D R F D F' U' F D'             14qtws
    (FDL,BUR) => (LFD,RBU)

If all of the corner cubies are not in the proper positions
it is more profitable to execute several corner moving transforms
rather then one corner moving one then one corner twisting one.

As presented here all transforms cycle the cubies in the same
manner (clockwise) , though twisting the cubies in all possible
ways. Their inverses (counter-clockwise) should also be kept
in mind.

C1:  D' L' D R D' L D R'                            8qtws
    (FDL,RDF,BDR) => (FRD,RBD,LFD)      (twist all clockwise)

C2:  L B' D^2 B L' B' L D^2 L' B                     12qtws
    (FDL,RDF,BDR) => (DFR,DRB,DLF)        (twist all clockwise)

Note: make reference in c1 (twist all counter-clockwise)

C3:  F L^2 D' R' D L' D' R D L' F'                   12qtws
    (FDL,RDF,BDR) => (RDF,BDR,FDL)       (don't twist at all)

C4:  F' R' B' R F R' B R                             8qtws

C5:  F L F' R F L' F' R'                              8qtws

Note: C4 and c5 have been adopted from those presented by DCP in
his message of 25 nov. 1308-EST

C6:  L F L' D^2 L F' L' F D^2 F'                   12qtws

C7:  R' D^2 R B' R' B D^2 B' R B                       12qtws

C8:  (C5)' (C1)'=                                       16qtws
     (R F L F' R' F L' F') (R D' L' D R' D' L D)

C9:  (C1)' (C4)'=                                       16qtws
     (R D' L' D R' D' L D) (R' B' R F' R' B R F)    (FDL,RDF,BDR) => (DFR,BDR,LFD)

These nine transforms are the only possible legal ones (along with
their inverses) which exchange three corners on a face (with the
possibility of twists0, though I can't guarentee minimum lengths for
any of them.

If all of the corners are not in their proper positions then there
are three possibilities:

1) One of the corners is in the right position and has the correct
to fix: do the appropriate transform, or its inverse from the
list given above.
max length=16qtws

2) None of the corners is in the proper position
     to fix: using C1,C4, or C5 (the shortest ones) move one of the
              corners to the proper position and orientation, then
             continue as in case 1.
              max length=8+16=24qtws

3) One of the corners is in the correct position, but is in the wrong
     to fix: Preferably using C1,C4, or C5 move theproperly
             positioned corner out of its spot and at the same time move
             another corner into its proper position and orientation, then
             procede as in case  1.
              If none of C1, C4, or C5 will do the proper thing then
             a combination of C2 and C3 must be used, C2 first, to orient
             the corners correctly (with respect to the bottom) then use C3 to
              position the corners correctly.
             max length=8+16=12+12=24qtws

These algorithms may be usefull to someone making a sides first,
corners second cube solving algorithm.

If anyone has any shorter algorithms for any of these transforms, please
send them to the list.

Unfortunatly I probably won't be able to answer any questions about this
method as I am going back to school (Caltech) tommorrow (today?)(Sat. 3rd)
and I don't have decent net access from there.

Chris Worrell (ZILCH@MIT-MC)

p.s.: sorry about the length of this message.

[next] [prev] [up] [top] [help]