The New Method / Substep / Concept Idea Thread

[goodatthis]

I didn't say that the recognition was impossible, nor that it would take a long time to do. I meant that it takes significantly longer than just a glance at the sides to recognize PLL. What you have suggested is possible, but is exponentially more time consuming and requires much more thought than would be efficient in a speed solve. It works, it's just not efficient.

I completely understand your point, but I think that with some practice, differentiating an oriented PLL with a non oriented PLL can be relatively easy and fast. But only with practice, and I'm not sure that's something a lot of people will try .

 

[IRNjuggle28]

Do algorithms for CPEOLS exist? It might be useful for somebody who is both willing to learn a lot of algs and is so focused on OH that they don't need algorithms that are versatile.

To make sure there isn't a misunderstanding, the steps would be
Cross + 3 F2L pairs
create final F2L pair
orient edges and permute corners while inserting final pair
2GLL

 

[mcw0805]

Hello, everyone.

I think what I am about to explain is quite new, so if you are interested please continue reading. I don't know how efficient this approach is, but I sometimes use it and have gotten good times.

Recently, I have been thinking if there was a better way to look ahead for the 1st F2L pair and/or save some time on the cross.

Sometimes, don't you just have that one last cross edge to insert that just sucks? (probably applies to a lot of people who aren't color neutral) What I mean is that the last cross edge takes 3 or 4 moves to insert. Of course, people will solve the cross by trying to orient these bad edges, but still, the cross cases won't be pretty in every single solve.

Just a heads up, this approach will work the best if you solve the cross on the bottom.

SO.. my idea is that...

1. Instead of placing the last cross edge, you insert the edge of the opposite side of the cross in any orientation in one move. For instance, instead of placing the last white cross edge correctly on the bottom, you place a random yellow edge in place of that last white cross edge. This way, F2L can be solved normally, and the F2L edge will never be placed in the bottom. Here's the limitation because it won't be always possible to insert that edge in one move. And the reason why I say "one move" is because I think that placing the wrong edge in more than one move is not pragmatic at all.

2. Solve F2L as you normally would. ALL four pairs. Another limitation here because sometimes, you might get confused that the cross edge is the F2L edge because when it is stuck next to a corner, it may appear like an F2L pair. But I don't have any problems with that.

3. As you insert the last cross edge, you orient the top four edges, thus getting a cross on top. From there on, you can do COLL, ZBLL, or just solve as you normally would. Inserting the last cross edge will always require M slice moves (usually by using M' U M move). Sometimes, inserting the last cross edge and orienting the top four edges will be simple as M' U M or M' U' M or M' U2 M. But most algorithms to insert the last cross edge will require you to do M' U M (or M' U' M) moves twice with U/U'/U2 moves in between.

So a sample algorithm is (M’ U M) U2 (M’ U’ M). There are two sucky cases (three if you count the mirror), and one of them looks something like this: (M’ U’ M) U’ (M’ U M) U (M’ U2 M). The other sucky case looks similar to this but has a mirror case.

I think there are 12 cases excluding the mirror and 16 cases including the mirror. (yes, only four cases have mirrors)

Using only M and U layers never messes up the corner orientations, so it is ALWAYS possible to predict the COLL case as you solve the last cross edge. (This of course applies if you do use COLL). When all four corners are oriented, obviously, you have an OLL skip. And I noticed that I unusually get a lot of OLL skips using this approach.

Conclusion. I call this way of solving as an "approach", as opposed to a "method" because it's not viable in every single solve. The ultimate OBJECT of this approach is to get a better look at the first F2L pair (perhaps second pair) without worrying much about screwing up the cross or breaking up the first pair you see as you solve the cross. This is kind of useful if you are trying to make an x-cross by building a 2x2 block first, then adding the last two cross edges because those two edges will not always be in nice positions. This approach is probably not that practical, but it's just another weird way to solve the cube.

Please reply if you are interested, and if enough people are interested, I will post the algorithm sheet.

Until then, happy cubing!

 

[supercavitation]

Approach

Not really new. It's called making a pseudo cross, and it's been done before in a whole bunch of way, I believe including this one.

 

[goodatthis]

Mcw0805:

I actually had a similar idea, where you make 2 cross edges instead of 3, and orient edges while inserting the last cross edge. There's actually an alg set for it: L5EOP.

 

[Dane man]

Do algorithms for CPEOLS exist?

They most certainly exist. The question is if they are recorded somewhere, and if there are few enough to be used. We've got ZBLS which is already 302 algs. Adding corner permutation to that would be... well, impractical to say the least.

 

[supercavitation]

They most certainly exist. The question is if they are recorded somewhere, and if there are few enough to be used. We've got ZBLS which is already 302 algs. Adding corner permutation to that would be... well, impractical to say the least.

I assumed that he meant something like VHLS with CP included, rather than something like ZBLS with with CP included, but I could be mistaken.

 

[IRNjuggle28]

Do algorithms for CPEOLS exist? It might be useful for somebody who is both willing to learn a lot of algs and is so focused on OH that they don't need algorithms that are versatile.

To make sure there isn't a misunderstanding, the steps would be
Cross + 3 F2L pairs
create final F2L pair
orient edges and permute corners while inserting final pair
2GLL

They most certainly exist. The question is if they are recorded somewhere, and if there are few enough to be used. We've got ZBLS which is already 302 algs. Adding corner permutation to that would be... well, impractical to say the least.

I meant VH+CP, not ZBLS+CP. That's why I separated the two bolded steps.

 

[elrog]

If your doing OH, ZZ is the way to go because edges are preoriented. For a last slot, I suggest you place the last F2L corner and perform an algorithm that flips the corner, places the last F2L edge, and solves corner permutation. Robert Yau came up with this idea and I generated algs for it a while back. There are 26 algorithms and I like them all, but I am the one who made them . If I could find out how to make arrows with visual cube I'd post them.

 

[TDM]

If I could find out how to make arrows with visual cube I'd post them.

You need to end the URL with something like: &arw=U0U1,U2U3. Explanation of this:

&: You use & whenever you're adding something to the URL in visualcube.
you probably knew that already
arw: this means you're adding arrows
U0U1: the first letter U means the arrow starts on the U face.
0: the pieces on the U face are numbers from left to right (and then from top to bottom) with numbers from 0 to 8. So U0 means the top left corner of the U face, and U4 means the centre of the U face etc.
U1: the arrow ends in the second location you give, so here the arrow points from UBL to UB.
,U2U3: this just adds another arrow from UBR to UL, or locations U2 to U3.

Here's an example using faces other than U (&arw=U0R3,F2R6):

(... hopefully you'll understand that)

 

[goodatthis]

Well, EO+CP last slot (not combining other acronyms for clarity) isn't all that impractical. You have 32 algs for all mirrors and inverses (8 for just a regular pair, and you could do R U2 R' setup for split pair cases, then just mirror each) and 6 CP cases for each EO case, so 6*32= 192 algs for an 84 alg 1LLL. But again, if you wanted to mirror the algs yourself, it would just be 8*6= 48 algs, plus you would already know a few, so it would be actually pretty worthwhile to learn. The only trouble would be recognition, but in terms of speed, VHLS (let's just assume you're using 8 algs) + full ZBLL is probably slower than CPEOLS+2GLL.

If anyone would be willing to generate (elrog, I'm talking to you haha) CPEOLS algs for just the first 8 VHLS cases, that would be great. I think it's a decent thing to learn.

Either way, and I'm quoting kirjava here, there's lots of kids these days coming up with silly LS methods when the real development is in LL.

 

[elrog]

U0U1: the first letter U means the arrow starts on the U face.
0: the pieces on the U face are numbers from left to right (and then from top to bottom) with numbers from 0 to 8. So U0 means the top left corner of the U face, and U4 means the centre of the U face etc.
U1: the arrow ends in the second location you give, so here the arrow points from UBL to UB.

That's what I needed to know. Thanks!

@ goodatthis - It shall be done. Also, I wouldn't say that there isn't real development to be made in LS methods. For example (but not limited to): my previous post.

I will generate any algs that are requested as long as there is some reason to but please don't pressure me to get done with them very fast though. School is starting again, and I have been working on other (some cubing related) related projects as well.

 

[TheNextFeliks]

This is a little different. Can someone come up with approaches that:
1. solving a face first and then corners not waterman
2. A really nice corners first method

 

[teddyman1331]

Here is a method idea i have thats under development, and i have no idea if it would be faster or not:

Basically, its CFOP, but with a twist. The difference is that when you solve the cross, you switch two edges. For example, when solving white cross, you place blue edge next to green center and green edge next to blue center. The red and orange edges are placed correctly. (Same can be done other way).

Next you solve f2l and oll normally, which gives you a nonexistent pll case (because the two edges are incorrect on the bottom)

Now, by solving the bottom two edges with M2 U2 M2, you can influence what the pll will be. Depending on how the top layer is when you perform the alg decides what pll is created. This can be used to force easier cases.

Well, that's it. I know it's not much and is probably useless, but you never know.

 

[supercavitation]

Here is a method idea i have thats under development, and i have no idea if it would be faster or not:

See previous statements about pseudo cross. Daniel Sheppard uses something much like that if it's more convenient than a normal cross, but long story short, this has been proposed numerous times before.

 

[elrog]

Fun new method !!!

1 - Solve a 1x2x3 block on D (leaving the R layer free)
2 - Solve 2 E-layer edges using the set of U, u, and R (AuF until they are in the FL and BL positions)
3 - Using the above set, solve the remaining E-layer edges (their places can be switched) and place oriented pieces into the RFD, RBD, and RD positions
4 - Use OLL to orient the top layer (there might be a few shorter algs)
5 - Using R2 and U, solve the F2L
6 - PLL to solve

I didn't intend on making a speedsolving method when I came up with this, but its not bad. There are many different variation of this too. Enjoy!

 

[IRNjuggle28]

Fun new method !!!

1 - Solve a 1x2x3 block on D (leaving the R layer free)
2 - Solve 2 E-layer edges using the set of U, u, and R (AuF until they are in the FL and BL positions)
3 - Using the above set, solve the remaining E-layer edges (their places can be switched) and place oriented pieces into the RFD, RBD, and RD positions
4 - Use OLL to orient the top layer (there might be a few shorter algs)
5 - Using R2 and U, solve the F2L
6 - PLL to solve

I didn't intend on making a speedsolving method when I came up with this, but its not bad. There are many different variation of this too. Enjoy!

I like this--it seems like a speed optimized and advanced version of keyhole. I'm guessing that Petrus blocks/pseudoblocks will be more efficient than what you have for steps 1 and 2, but turning your method into Petrus sort of ruins the point of you making a method, lol. It looks fun, so gj for coming up with it.

 

[TDM]

1 - Solve a 1x2x3 block on D (leaving the R layer free)
2 - Solve 2 E-layer edges using the set of U, u, and R (AuF until they are in the FL and BL positions)

Why not solve a Rouxblock and then place DF+DB? Or are you just trying to make it as weird as possible?

 

[elrog]

I myself think the first 2 steps are very efficient. I find building a 1x2x3 block and expanding into a 2x2x3 is easier than a 2x2x2 into a 2x2x3 such as in petrus.

Also, I didn't choose to do a Roux block because you would either have to learn to recognize the LL with a misaligned M slice, or you would always have to solve DF and DB as the correct edges.

Maybe some example solves will clear up any confusion. I have one done. I'll do more later.

Scramble 1: x F2 L R' D2 R' U B2 R B' D L F L2 D' R2 L F2 U B2 R2 D2 U' R2 D' R' x'
1 - F U2 B R2 F2
2 - u R u R2' u
3 - u U2 R U R'
4 - R U' R2' F R F' U2 R U' R' u'
5 - U2 R2 U2 R2' U' R2 U R2' U R2
6 - U (l' R') D2 R U R' D2 R U' R x'
Red - u2'
Blue - R'
This turned out to be a good solve.

 

[Loupisar]

I watched Feliks' mini-video series and I was really interested in this case he showed: the 3 move OLL. For those who don't know it, it is a an LL case where you can "cheat" to orient the LL with an easy setup ( x' z L' U L ). Basically what you do is to rotate the cube (x' z), and then insert the pair (L' U L) to get an easy J-perm. Here is the picture:
Generator : B' F' L2 B' R' F U2 F' R B2 L' F
(x' z ) L' U L and then you get the J-perm
Watching this was really cool but one day in a solve, I met this exact case: ( Generator: L2 B' L U L U' L' B L2 F' L' F (could be a good alg!)
It looks very similar, but it's not. If you change your grip like for the previous case, you see that the pair you want to insert is not at the same place. I found this to solve this LL case: x' z L' U L U y' L' U2 R U' R' U2 L (+AUF). You just get the pair out of the slot, and then insert it with a Winter Variation case that skips the PLL.

In these 2 cases, there is a 2x2x1 block already setup, but you can find many other situations where this block is not fully setup, and the LL still looks awesome, like for example the usual OLL case people use most:
: Solved with r U2 R' U' R U r'
I found three more like this, if you like you can find a way to recognize them, and learn the algs.
: F R U' R' U R U2 R' U' F' (really cool alg, inverse could be used as well)
: F' L F L' U R U2 L' U L U2 R' (setup+ Winter Variation)
: F R U' R' F' U2 R U2 L' U L U2 R' (setup+ Winter Variation)