Random Cubing Discussion

[oll+phase+sync]

I've considered this before (many times), but thought it would be too difficult so I always threw the idea away.

I'm thinking I can cut down the number of cases for non-matching CLL to the normal 42 by presenting them in a different way. By combining each CLL's two cases together to make one image, recognition can be as easy as CLL and have the same number of cases. It would still be eight stickers to look at (just like CLL) and you would still always know where to look.

I expect I would still have a difficult time explaining how to use it and I don't have the website programming knowledge to make appealing pages.

I'm not sure I get it right looking at this http://www.athefre.110mb.com/NMCMLL.html and comparing the two E4-* cases (both in the first columne) they look so different! how would you combine them?

 

[Athefre]

For each L/R orientation, there are two usable patterns that work for this. Those patterns match up for each corner case (E4 for example). I would need to make a new page because, for this pattern merge, some of the patterns in my original image wouldn't work. Example:

Setup (With white/yellow on L/R): FU'L'UR2U'LUR2F'

- You see the white/yellow orientation at URF, UBR, LBU, and LUF.
- URF and UBR are opposite. LBU and LUF are opposite. This is your first pattern.

Setup 2 (for the "second" E4 case): U FU'L'UR2U'LUR2F'

- You see the white/yellow orientation at FUR, BRU, BUL, and FLU.
- FUR and BUL are opposite. BRU and FLU are opposite. This is the second pattern.

- Now look at the stickers at URF and UBR. Then look at the stickers at LBU and LUF. It's the same as the pattern in the first setup.

With this, recognizing the orientation of the L/R colors is the first priority. Then you look for the second pattern. You would need to know where to look for each L/R orientation. That's the only additional information needed compared to CLL, but it's still just like memorizing 42 cases. This may be something that someone would naturally develop if I had the right patterns shown in the NMCLL image.

 

[Kirjava]

this is awesome

 

[qqwref]

Does anyone remember my intuitive 2gen cycle alg: (U' R' U2 R U) (R U R2 U' R')?

I just realized that you can use it to create an uncommon but short 2gen U perm... watch this:
R2 U (U' R' U2 R U) (R U R2 U' R') U' R2
= R U2 R U R U R2 U' R' U' R2

Also, what the hell?: R U R' U' R' U' R' U R U
This is also a "two setup'd single moves" type alg: [R U: R'] [U' R': U].
And it's a cyclic shift of (URURU)(R'U'R'U'R'), which is a well-known F2L alg, and also one that I'm not sure how it works. It's got to be one of the shortest useful algs which can't be written as a fixed-centers commutator.

 

[irontwig]

Isn't obvious; you just do two 4e4c cycles ?

 

[qqwref]

Well, sure, but it's not obvious to me why it works so nicely

 

[oll+phase+sync]

- Now look at the stickers at URF and UBR. Then look at the stickers at LBU and LUF. It's the same as the pattern in the first setup.

That's cool, I never realized it so far: If I know one L/R-pattern a CLL-alg produces, but the pattern don't show up, then I just have to look for that pattern as a F/B - pattern.

But it don't works in NMCLL cases ? right?


Also you created 23 different LR patterns groups, (each with up to 4 subcases) to use LR pattern recognition, my idea would be to declare ~50% of these "pattern groups" to "not a pattern" wich means each time I see a L/R "not a pattern" I know that I have to look for an F/B pattern.

Am I still on track?

currently I see no systematic how Algs are distributed amog the pattern groups, so I'm not sure an optimal (42 case) "pattern" / "not a pattern" separation exists.

 

[Athefre]

But it don't works in NMCLL cases ? right?

It works for NMCLL and is the reason I posted about it. I'll admit, I haven't went through every case to test if this works perfectly. That's why I started with "I'm thinking" when I began describing this. I've only checked a couple of orientation sets. I have run across a couple of cases (H1 and H6) that behave a little differently, but it doesn't seem to matter because those orientations are so simple with fewer cases.

All you have to do to see how this works is apply the inverse of your sequence and find the L/R orientation. Then with solved corners again, do U or U' then apply the inverse of your sequence. Find the L/R orientation and you have your other pattern. It would take a little more time to memorize this compared to CLL because in addition to the 42 cases you would have to look over the 23 possible orientations and learn where to look for stickers (which wouldn't be the same as the image on my page).

 

[oll+phase+sync]

[cube]pzl=3&alg=zxFU'L'UR2U'LUR2F'[/cube][cube]pzl=3&alg=zxFU'L'UR2U'LUR2F'y2[/cube]

(normal case FU'L'UR2U'LUR2F') L/R is E4-2 pattern F/B is E4-1 - fine (expected)


[cube]pzl=3&alg=zxR2FU'L'UR2U'LUR2F'[/cube][cube]pzl=3&alg=zxR2FU'L'UR2U'LUR2F'y2[/cube]

(R2 case R2 FU'L'UR2U'LUR2F') L/R is E4-2 pattern U/D is E4-2 too - fine (surprising to me)

[cube]pzl=3&alg=zxR'FU'L'UR2U'LUR2F'[/cube][cube]pzl=3&alg=zxR'FU'L'UR2U'LUR2F'y2[/cube]


(R case R FU'L'UR2U'LUR2F') L/R is E4-2 pattern but I don't see another E4 pattern

Instead of E4-* pattern I should maybe say E4 pattern case.

 

[Athefre]

For your last example, the remaining E4 pattern is FUR/BUL (they are opposites) and BRU/FLU (they are opposites). Notice in your first example that you could say FUR/BUL are opposite and BRU/FUL are opposite. Notice that FUR/BUL are on the yellow corners and BRU/FUL are on the white corners.

 

[ben1996123]

 

[Robert-Y]

Not bad, I still prefer the one on anthony's website though: (R' U R') (U2 R U' R' U R U' R2)

 

[cuboy63]

Not bad, I still prefer the one on anthony's website though: (R' U R') (U2 R U' R' U R U' R2)

I prefer this too.
I think I can get 0.8x.
Wanna go on TTW and talk about 2x2 algs again, Rob?
Or we can talk about something else.

 

[Escher]

This is also good imo: (R' U R') U R' U2' R U' R U' R

 

[Robert-Y]

Bill: Sorry not today, I have to leave tomorrow for a competition

Rowan: yeah this one is also nice, I can't do eido U2's very well though :/

 

[riffz]

No one likes R2’ U' R U2 R' U2’ R U' R2 ?

 

[oll+phase+sync]

Notice that FUR/BUL are on the yellow corners and BRU/FUL are on the white corners.

I try to make a ruleset out of this (it seems to be more deterministic(easy) than I first believed):

1. identify the L/R case (white/yellow), there are 23 cases

2a. There are two corners with yellow stickers, these corners have 2 colors in common (one is yellow) -> ignore this
2b. The remainig two stickers on the yellow corners are always opposite colors. -> these are the first two stickers of the second pattern

3a+b. Just the same as step 2 to get the next two stickers for the second pattern.

4. the second patter is again one of the 23 cases and the combination tells me the alg.


my considerations (?):
a) the 23 cases are the same for both steps in theory, but in practise I memoryse pattern 1 as two set of stickers of the same color, while I have to identify pattern 2 as set of opposite colored stickers

b) there are only 42 pattern combos, but they can be combined in one of two orders.

 

[Athefre]

That's how I used to recognize NMCLL, but using the matching colors, which are actually the U colors, instead of the opposites. The problem is that it's not as fast as a set recognition pattern.

 

[Hershey]

Bump?

This L2L4 stuff is getting to my head.
First layer average solve times

Average of 5: 8.98
1. (6.98) R' L D' R2 F R2 F2 L' F2 L' U' D2 B' L2 U' D2 L2 F R' B2 D2 F R2 L D2
2. (19.31) L U' F' L R' U B' F R B D2 R' L2 F' D U B U L2 F2 U2 D L' B2 L
3. 8.31 F R' L2 D' L' F U' F2 D2 F L2 F U' L D2 U B R F' U2 D2 F' U F2 B
4. 8.37 L2 D2 L' D B2 U2 D2 L2 F B' U2 L' D' R' B' F U D' L2 D B2 U L R F
5. 10.27 R B2 R2 D B F2 R2 B R' L2 B2 R F L2 R2 F2 R2 U2 D2 B2 F2 U D2 L' F2

2nd "solve" was fail. I could solve the entire cube in less than 18 seconds.

 

[irontwig]

LBL trick from Josef Trajber's book: R2 U2 R' U2 R2. I was like :0