Random Cubing Discussion

[jiggy]

Could someone please hook me up with the pure flip algs for these cases?

By pure flip, I mean only the two edges flip and nothing is permuted. Thanks a gazillion.


Added to list of things to do: Brush up on edge commutation.

 

[Forte]

There are a bunch here
(they're part of ELL)

 

[jiggy]

There are a bunch here
(they're part of ELL)

*Whacks head* ELL! Should'a thunk of that! Hero!

 

[Forte]

There are a bunch here
(they're part of ELL)

*Whacks head* ELL! Should'a thunk of that! Hero!

The one I use isn't explicitly listed there anyway
I use the inverse of the first one, but I'm really liking the second one done like M' U' M U2 M' U' M' U' M' U2 M U' M2

 

[StachuK1992]

Could someone please hook me up with the pure flip algs for these cases?

By pure flip, I mean only the two edges flip and nothing is permuted. Thanks a gazillion.


Added to list of things to do: Brush up on edge commutation.

(M' U)*4 (U M')*4

and L F [(M' U)*4 (U M')*4] F' L'
or r U R' U' r' U2 R U R U' R2 U2 R
or M U M' U2 M U M U M U2 M' U M2

edit: wow. ninja'd majorly.

 

[jiggy]

Haha, well thanks both of you anyway! I'm actually a huge fan of r U R' U' r' U2 R U R U' R2 U2 R. ELL seems to have some really nice algs...added to list of things to do? No, jiggy! Bad jiggy! Back to your thesis!

 

[Kirjava]

I'd like "Human Kociemba" to exist. Time to throw some ideas around;

F2 R U2 R U' L F U L' U2 F' L' B2 F' L' F' U' D F D' L B2 D R2 F'

1) y2x' DM'F'LU2 x' rUr'U2M'U2B'RBR'Ur'U'RUrB'R'BM2UMUMU
2) y2 R2U'F2UF2U'R2DL2U'L2uL2U

 

[jiggy]

You do some crazy stuff, at crazy o'clock in the morning, Kir!

This is pretty impressive, do you think you could break down your solution and explain your thought process a bit?

 

[Kirjava]

Sure. I've been playing around with this idea for a while now so I've had some practise getting good at it.

Step 1 is reduction to opposite colours on each side. It's sort of like reduction to the half turn group, but there is a parity that exists requiring non half turns to solve. I can do this consistantly sub10 for speed, or consistantly sub30 movecount if I spend enough time on it.

The actual solution I got was just a CN Roux solve split into these four steps;

y2x' DM'F'LU2 x'
rUr'U2M'U2B'RBR'
Ur'U'RUrB'R'B
M2UMUMU

Step 2 is the rest of the solution. Just like Kociemba, this should be solved with the <U,D,R2,F2,L2,B2> group. The general idea is solve some pieces -> solve the rest with an alg. I don't yet have a good system for this step or a list of useful algs yet - this is was I was hoping I'd get help with.

y2 R2U'F2UF2U'R2
DL2U'L2uL2U

 

[jiggy]

Step 2 is the rest of the solution. Just like Kociemba, this should be solved with the <U,D,R2,F2,L2,B2> group. The general idea is solve some pieces -> solve the rest with an alg. I don't yet have a good system for this step or a list of useful algs yet - this is was I was hoping I'd get help with.

Well, I don't know if this is what you're looking for, but you could break it down into

<U,D,R2,F2,L2,B2>
2a) Orientate U and D corners
2b) Permute U and D corners together (5 cases)
2c) Solve the remaining edges

Which is all very intuitive, and can be executed fairly rapidly. But it's probably adding quite significantly to your move count. This is quite similar to the end of Human Thistlethwaite, but I can't help feeling there's a more efficient way of doing things. I think I'm going to keep toying with your idea and see if I spot anything!

 

[Kirjava]

That obviously works, but the idea is to setup into a position you know how to solve from a collection of algorithms. I think this is the best approach for getting a low movecount. (Also, U and D corners will already be oriented by that point)

I've been generating a list of useful algorithms that range from 5 to 8 moves - I'm waiting until I can form a coherant method from them before I post them though. It's too hit and miss at the moment to be useful.

 

[jiggy]

Yeah, I thought you would likely be looking for something a bit more like this. I did have a reason for throwing in the orientate step (EDIT: By orientate, I mean for opposite colours, in a similar way to Phase 3 of H.T.), but I'd like to study the problem a bit more before I talk about it.

I'd be interested in hearing more about the kind of state you hope to set the cube up into before solving with an algorithm. What is your main motivation when you perform this step, or is this still an area you're working on?

 

[Kirjava]

I don't have a strict method for step 2 yet - it's more trial and error (which is what it will end up being with or without structure, anyway). I generally try and solve as many pieces as I can in the fewest moves possible and see what state it leaves me in.

I think with some more development and enough time given for a solve this step could be solved near optimally.

I think the worst cases are like 15 moves?

Sidenote; Step1 is essentially a 3 colour cube solve.

 

[Kirjava]

Just did a 1 hour FMC attempt with this method. Step 1 started off really well but I couldn't find a nice continuation. (I also cheated a bit and didn't quite bother to finish) I bet sub20 moves for step1 is easy with some work.

L F2 U2 B U R' F B U' D' R2 U F L2 U' D2 B R2 L B U' F' U2 D B'

xR'D2U - 1x2x3
xU'r'U'r2 - Triple xcross
yRUR'U'RU2R' - F2L (yuk)
FR'F'RURU'R' - LL
yB2R2U'M2U'R2D' - Trial and error win
yL2UR2UR2 - Super easy

35 moves

 

[Robert-Y]

Weird and kinda nice bad R perm for 2H:

LH: (U L' U' L U) (L F' L' U2) y x (L U' L U) (L2' B2' L)

RH: (U' R U R' U') (R' F R U2') y' x) (R' U R' U') (R2 B2 R')

 

[Robert-Y]

Wow this megaminx alg: R U2 R' U2' R' F R U2 R U2' R' F'

does nothing :fp

oh well

 

[hawkmp4]

R U R' U' R' F R U2 R U2' R' F' is an OLL, but solves the same case Sune does, so not very helpful.

EDIT: That IS a Sune. Exactly a Sune.

EDIT 1.5: By which I mean, the above algorithm has the exact same effect as R U R' U R U2' R'.

EDIT 2: Related only in that it's the same puzzle:
In case anyone was curious, I worked out that R BR BL L F is of order 756. I'm sick of counting cycles now. I've never done that before.

 

[Robert-Y]

If the current naming system for PLLs wasn't used, how would you name each PLL?

For the one commonly known as "R perm" I think maybe "7 perm" is more accurate

And also I think 2/4 G perms could be considered R perms. Example: do these moves: R2' u' R U' R U R' u R2 y' L U' L' U2. Wouldn't you agree that the incorrectly permuted pieces and the centre make an "R" shape?

 

[CubesOfTheWorld]

I wanted to bring this up:
The reason I think that skewb should always be an unofficial event because the times are so inconsistent, at least with my method. You can have a 40 second NL, and also have a 15 second NL. The averages would be so inaccurate. That is just my opinion.

 

[shelley]

That is not the correct context for the word "inaccurate". If the averages are calculated properly, they are accurate.