Random Cubing Discussion

[qqwref]

Might as well just do something like R U R' F' R U R' U' R' F R2 U' R2 U R' U' R' U' R' U R U R2 then...

 

[JonnyWhoopes]

I would like to laud whichever mod cleaned up this thread. It's not often I get to see you guys in action, it was cool to see it as it happened.

 

[Cheese11]

Spoiler

Mother of god. *takes off sunglasses*

 

[5BLD]

I just sat on my cube and did E moves to break it in. So many centre cap pops- i wonder if I'm putting too mch pressure on it...

 

[Czery]

I just sat on my cube and did E moves to break it in. So many centre cap pops- i wonder if I'm putting too mch pressure on it...

Wait - you SAT on your cube...
Doing that can effectively prevent caps from falling off.

At least it's not as bad as the MF8 Square 1 caps. Those caps fall of as if the caps were lubed themselves.

 

[Athefre]

I may be the only one that finds this interesting and am talking to myself, but here are some new and old uses for what I guess would be called Edge and Corner Transformations:

Spoiler: OLLCP

Setup: F' U' L' U L F L' U2 L U L' U L (From Robert Yau's document)

Usually the solution would be the reverse. But, another way is to turn it into a different case like this:

M' U' - F R U R' U' F' - U2 M

Spoiler: CLL+1

Setup: B L U' L' B' R' U2 L' U' L2 D F2 D' L' R
Solution: M' - R' U2 l U' R U L' U R' U r - U2M
Solves the green edge at UB.



Setup: L F U F' U' L2 B L' B' L2 F U2 F'
Solution: M2 - R' U2 l U' R U L' U R' U r - U2M2
Solves the green edge at UF.

Usually CLL+1 would be ~330 cases. But, using this simple M setup, the set is reduced to ~80 cases without a large increase in movecount and often gives the chance for a cancellation at the beginning of CLL+1 and a cancellation at the beginning of L3E.

CLL+1 can be reduced to 2, 3, 4, etc times the number of CLL cases.

Spoiler: ELL

Solution: M' U - M2 U M U2 M' U M'

The first two moves place an oriented edge on the D-layer and turn the case into an all oriented 3-cycle with the last M2 and setup undo (M) cancelling. Edge transformation isn't very useful for ELL because there are so few cases. The best thing is that it gives you the ability to easily create your own algs and understand how they work.

Spoiler: CLL

Setup: U L2 U' L U' R U R' F L' U' R'

Transformation: L2 - U2 L' U2 L' B L B' - U' R'

As with ELL, the biggest use here is being able to understand the relationship among cases. Full list and better description here.

This can be applied to almost anything. Its usefulness depends on what it's being applied to. Anything can be reduced to 0 cases. But there is a difference between a setup and an algorithm. The further you try to reduce, the more complex the setup is, but fewer cases. The less you try to reduce, the more cases, but maybe better algorithms. You have to find the right balance considering recall, move count, and comfort.

 

[Kirjava]

Case reduction is akin to 1look2alg systems for attacking large algsets. Both have their own advantages.

How about you try doing 1LLL case reduction?

EDIT:

02:56 <+Kirjava> also I like how all the kids are making stupid LS methods
02:56 <+Kirjava> while the real development is happening in LL

 

[Athefre]

Case reduction is akin to 1look2alg systems for attacking large algsets. Both have their own advantages.

How about you try doing 1LLL case reduction?

I've been working on this off and on but haven't finished all edge angles. Give me some time and I'll post. I'm not optimistic though because I'm unsure of how easy 1LLL recognition would be.

 

[JonnyWhoopes]

I still want CLL+1 to be a feasible thing.

 

[Kirjava]

It is. I'm releasing an alg list at some point in the near future.

 

[JonnyWhoopes]

It is. I'm releasing an alg list at some point in the near future.

I have never loved you more than I do right now.

 

[Specs112]

I have never loved you more than I do right now.

OTP OTP OTP

 

[Julian]

4x4: last F2L slot is solved corner, flipped edge. An even number of LL edges are oriented. Flip F2L edge with pure OLL parity to dodge regular OLL parity? I've been doing this when it (rarely) comes up.

 

[MostEd]

4x4: last F2L slot is solved corner, flipped edge. An even number of LL edges are oriented. Flip F2L edge with pure OLL parity to dodge regular OLL parity? I've been doing this when it (rarely) comes up.

If this happens to me i pure flip(inner slices) the f2l edge, try to avoid it tho, extra rotations suck

 

[Cool Frog]

I may be the only one that finds this interesting and am talking to myself, but here are some new and old uses for what I guess would be called Edge and Corner Transformations:

Spoiler: OLLCP

Setup: F' U' L' U L F L' U2 L U L' U L (From Robert Yau's document)

Usually the solution would be the reverse. But, another way is to turn it into a different case like this:

M' U' - F R U R' U' F' - U2 M

This can be applied to almost anything. Its usefulness depends on what it's being applied to. Anything can be reduced to 0 cases. But there is a difference between a setup and an algorithm. The further you try to reduce, the more complex the setup is, but fewer cases. The less you try to reduce, the more cases, but maybe better algorithms. You have to find the right balance considering recall, move count, and comfort.

That OLLCP (Since I know this one)
Roberts Alg is quite awkward for me and he just suggested to do the M' U' since it is faster (for me to perform)

There is another case


R2 D' R [[FRUR'U'F'] U] R' D R2

 

[zzomtceo]

New last slot method proposal

I would like to propose a method where you solve the last slot+corners or edges and then finish with ELL or CLL depending on which way you solved the last slot.

 

[JohnLaurain]

I would like to propose a method where you solve the last slot+corners or edges and then finish with ELL or CLL depending on which way you solved the last slot.

Algorithms?

 

[qqwref]

Improvement on OBLBL for big cubes: for the last floor[(N-2)/2] layers of centers (so 2 layers on 6x6/7x7, 3 layers on 8x8/9x9, etc), do one entire center at a time in layers. So for 7x7x7 if the three centers were green, orange, blue you would do:
- green layer 1, orange layer 1, blue layer 1
- green layer 2, orange layer 2, blue layer 2
- green middle layer, orange middle layer, blue middle layer
- green layer 4, green layer 5, orange layer 4, orange layer 5, blue layer 4, blue layer 5
This saves on commutators since you only have to use them for the last colored center (so, no more commutators than in reduction).

There are also two more center tricks: (1) if you have the last row on the right of the top face and the extra oblique on the back of the top face, you can save slice turns by doing something like x r' U' 3R' U r U' 3R U; (2) you can do y m' U2 m to move centers around without messing up the block, usually saving 1-2 comms.

Also, I find it really nice to pair up the final edges on M, rather than on E as I was trying before.

 

[cubernya]

I would like to propose a method where you solve the last slot+corners or edges and then finish with ELL or CLL depending on which way you solved the last slot.

I would say that corners would be better, because the ELL algs are better than L4C. However, could you figure out how many algs there would be for each set?

 

[zzomtceo]

I would say that corners would be better, because the ELL algs are better than L4C. However, could you figure out how many algs there would be for each set?

I'll try, but all I know is ZLS-E would use less. Right now I want to find the ZLS-C algs more because they are, like you said, better.