Random Cubing Discussion

[EVH]

Dan Knights Z perm: R B' R' B F R' F B' R' B R F2 U

Now reverse it: U F2 R B R' B' F R' F B R' B' R
WTF? It's a Z perm on D, plus the equivalent of an E move. I can't explain that.

That doesn't make any sense........

 

[flan]

I didn't want to create a new thread because its such a bad review but I just thought I'd share my thoughts of the DaYan 4x4.

So out of the box this cube is very bad but after lubing it gets much better. It feels sturdy and crunchy as you move. Not really clicky.

Overall I like the feel of this cube It has a pleasing feel to me which I like. One problem is this cube is very big. Larger than my v5. But thats ok because it looks like a normal 3x3 with extra layers getting me 'cube on steriods' comments at school which I never even got from my 7x7 because if the puny sized cubies.

The inner layers of this cube feel smooth and it isnt too bad solving this cube as a 2x2 except for the size.

Sorry for not commenting much on the speedcubing aspects of this cube, the reason I didnt was because I know nothing about 4x4s and have only just decided to get serious and I've only ever owned a rubiks. Just felt like giving my 2 cents (zimbabwe cents).

 

[KYLOL]

Here is a method I am working on. I average about 49 moves with it. The example solves below are some average solves.
I use Petrus normally. This method is nearly the same. I start with a 2x2x3 block. I then move on to finish F2L, but I build a 1x2x2 block on U layer and preserve it while I place my last CE pair into F2L. This will leave me with a ZBLL case containing a 1x2x2 block ( the number of cases is low, I've learned nearly half of them in just a few days). The trick is to get from your 2x2x3 block -> ZBLL case as efficiently as possible. The majority of the moves are wasted here, in preserving / moving around blocks. I've come up with really unique ways to fluently push around my blocks, almost ignoring them as I finish my F2L. I understand every few days someone starts a "Hey, check out my new awesome method that is 10 seconds slower than fridrich" thread, but this method I use is only a few seconds slower than Petrus ( I haven't practiced much with it yet...kylol still 2slow2Bpro ). ao12 petrus = 13.85, ao12 new method = 19ish.
WHERE IS MY PEMBRO? THOM-BOMB?

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


U' D' l D L D
U' F2 U R'B' R B
x
U' R' U R U2
R U' R U R2 U' R' U R' U' R U R2 U' R2 U R
U2
M' x' U2 L2 U' L2 U' L2 U' R U' L




L R B2 F' U2 R2 B' F L2 R B' R F2 L' F2 B' U' L2 B' U' R F2 R U2 B'

U2 l2 f
z2
U R U R2 U F' U' F U'
x
U2 R' U' R2 U R' U' R' U R' U' R U R2 U' R'
y
R U' R
y'
M'E M D
L F R' F' L' F R F'




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

D' R' U R U2 F D'
F' D F2
x' y'
R' U' R U' F R' F'
z' y2
R U' R' U2 L F L'
y'
U R U R U R' U2 R U2 R'
U'
R' F' L F R F' L' F

 

[qqwref]

holy **** it's kyle allaire

I haven't seen you in ages


EDIT: oh yeah and that method kinda resembles tripod which is: 2x2x2 block -> 2x2x3 block -> build another 2x2x1 block on one of the layers -> build yet another 2x2x1 block (on the other layer) -> add in a corner/edge pair -> last 5 pieces in one alg. I don't get how you're finishing the F2L with the block on U but it seems like doing that would be more efficient than the tripod way if you could manage it (and clearly you can, but I don't think I could).

 

[Kenneth]

Here is a method I am working on. I average about 49 moves with it. The example solves below are some average solves.
I use Petrus normally. This method is nearly the same. I start with a 2x2x3 block. I then move on to finish F2L, but I build a 1x2x2 block on U layer and preserve it while I place my last CE pair into F2L. This will leave me with a ZBLL case containing a 1x2x2 block...

I think Johannes Laire knows this method, including all needed ZB cases ([wiki]L3C[/wiki] makes half the group).

Just the other day I was thinking of this LL method but ignoring EO for the two unsolved edges, that would be twice the number of cases for the last step (108 including duplicates).

Edit: aha, tripod...

 

[jiggy]

Recently, I've been playing around with conjugated PLLs.

Set up: B2 L' D B D' B' L F2 L2 D L2 B2 U' F2 U F2 D' F2 D
(Set ups generated by a random online cube solver.)

As you can see, the simple example above requires just two edges to be swapped and two corners to be swapped. Of course, we could just solve this with OLL and PLL, but we'd be missing the opportunity of an easy 1LLL if we did!

In this situation, we can apply [F R S], as a set up move, followed by a J-perm before reversing the set up moves.


Of course, we can use this in other ways than a 1LLL; the other day, I came across this case:

Set up: L' F2 R U2 F2 R2 D2 B2 R D2 F2 R2 B2 R' U' L2 F2 R2 D' R

Which can easily be solved with (y') B [Y-perm] B'.


Mostly, I've been playing with these for fun. However, I think I could find them moderately useful for FMC...the biggest road block I've hit so far is if a piece is in the correct position, but with the incorrect orientation. That's a real pain!

Anyway, I don't for a second believe I'm the only person to have messed around with this! I'd be interested to hear any thoughts you guys might have about it. Have you found a good solution for my problem? What do you think about an FMC application? I'm worried that by chucking a PLL alg in there you would probably bump up your move count quite harshly.

 

[qqwref]

Yeah, I've tried this before for Tripod (I don't know all the algs yet lol). For the last 5 pieces you can often get a two corner and two edge swap, and setting that up to a PLL is relatively easy. Perhaps a faster way to do the first case is like this:
F R y x' (R2 U2 R' U' R U2 L' U R' U' M') y' R' F'

I think this technique could be useful in FMC but only if you are left with a 2-2 swap and you know some efficient 2-2 swap algs (such as the 10-move J perm, or the 11-move T).

 

[jiggy]

I wouldn't say it can only work with 2-2 swaps, I've also made this work for 3-3 swaps (G-perm) but the set up moves can get pretty hairy! (Although maybe I'm just not so good at spotting the optimum set ups yet!)

I think it is definitely a technique which can only be used in special situations. I wonder if a human could find a shorter solution with standard FMC methods, though...


EDIT:

F R y x' (R2 U2 R' U' R U2 L' U R' U' M') y' R' F'

This is an interesting alg. Just out of my own curiosity, does it belong to a particular method/subset or have you done something clever I haven't spotted?

 

[irontwig]

Well, it just looks like it's a bad case, it takes 10 moves to leave three edges and 11 to leave three corners.

 

[jiggy]

Just a quick example of a 3-3 swap I just met:

(This may be easier to see on the 3D cube below, or if you execute the set up)

Set up: F2 D2 R F R' B D B' D F2 U2 B R2 B R2 B2 U2 L2 F' L2


My solution, executed from the start point on the right, blue on front:

[D' L2 D2 B' M] (x2) G-perm (x2) [M' B D2 L2 D]

With the G-perm at 12 HTM, that works out at 6 + 12 + 6 = 24 moves for the final 6 pieces.

=/ Not great, to be honest.


For comparison, I'll give an M-turn heavy edge commutation + corner commutation a quick try.

[U S'] (M' U2 M) D (M' U2 M) D' [S U']
y2 x
[U' L] (F' U' F) D2 (F U F') D2 [L' U]

Possibly not the optimal edge/corner coms, but they'll do for now.

Edge: 3 + 5 + 1 + 5 + 1 + 3 = 18
Corner: 2 + 3 + 1 + 3 + 1 + 2 = 12
= 30 HTM

24 < 30, which makes the conjugated G-perm better than my attempt at coms. There may be better coms out there, however. Overall, I'm not sold, but I do think that it's a very easy thing to do, which is a plus for the lazy cuber! =)


Anyway, given that the whole cube should be (optimally) solvable in 20 moves, 24 for 6 pieces (even the final 6) is probably a bit expensive. What do you guys think?

 

[Escher]

Better edge comm: y' U' M2 B' R' B M2 B' R B U y (12 moves inc. U setup)

I'm sure theres a BH alg that's shorter too.

 

[irontwig]

Why don't you just use inserted three-cycles instead?

 

[jiggy]

Heh, I knew there would be better coms out there, I didn't really put too much time into it! =P Having said that, there may be a better way to set up for the G-perm too. Either way, I don't think conjugated PLLs hold much promise. But they are pretty fun, and at the end of the day, isn't that more important than an excellent solve?

What? No? Darn!

Why don't you just use inserted three-cycles instead?

Yeah, I'm sure if you were to have applied something like insertions at an earlier stage you would avoid this whole problem. I know I've been chatting about an FMC application, but the only application it ever really had for me was a bit of fun! ; )

 

[Kirjava]

y' U' M2 B' R' B M2 B' R B U y

U R' U' M' U R U' M

 

[number1failure]

currently teaching 2 people LBL. one is on the green side corners, other is still on cross

 

[qqwref]

This is an interesting alg. Just out of my own curiosity, does it belong to a particular method/subset or have you done something clever I haven't spotted?

It's just an R setup to a RU2R' etc J perm, which combines well in speedsolving and saves a move in FMC. It's equivalent to a J perm of the other type with flipped edges.

Here's an interesting alg I just found:
U' R' U' R U2 L' U R' U' L U' R
Reminiscent of a J perm. It effect is J perm + 2 corner twist (one of which is outside the J perm).

currently teaching 2 people LBL. one is on the green side corners, other is still on cross

Why green? You can teach LBL just as well on any color. But I think it's kind of mean to push people into being used to a slightly unusual coloring.

 

[JonnyWhoopes]

R' U' R U R' U' R U' R' U R U R' U R U2 R' U R U2

What would you call that type of alg? It's a two corner twist, but it doesn't affect any other pieces. Found by hand.

 

[qqwref]

I'd call it a pure 2-twist. That's a pretty neat one.

 

[waffle=ijm]

I decided to do some dots (E2 M' E2 M)

Cubes Solved: 100/100
Number of Pops: 0
Average: 0.56
Standard Deviation: 0.13
Best Time: 0.40
Worst Time: 1.21

I got tired around the 1-20 were bleh. I guess I wasn't warmed up enough. 21-49 - pretty average for me 50-60 I started to feel tired. The rest, I got a sudden boost. I noticed lots of .43-.45. Last 10 or so were mainly sub-0.50

Thom, you should try this with cubes like the Guhong so accurately turning doesn't count nearly as much and you don't have to follow through completely on the E2.

 

[riffz]

Better edge comm: y' U' M2 B' R' B M2 B' R B U y (12 moves inc. U setup)

I'm sure theres a BH alg that's shorter too.

U L2 U' M' U L2 U' M

edit: Didn't see Kirjava's post, but mine's better for that angle, anyway.