The New Method / Substep / Concept Idea Thread

[TDM]

I already posted this idea earlier and some have disagreed with me and I don't understand your explanation. http://www.speedsolving.com/forum/s...-New-Method-Substep-Concept-Idea-Thread/page9

You can also view it as 8 orientation cases, but then you only include each PLL once.

 

[rybaby]

I have thought of a new 4x4x4 method, which combines Yau, Waterman, K4, and Reduction.
1. Solve opposite centers.
2. Build a 1x3x4 block
3. Solve the last 4 centers without breaking up the two centers or the block.
4. Rotate so the block is on D, with the open side of the D face at F.
5. Pair up a dedge and place it in DF.
6. Use chain edge pairing to solve the rest of the dedges.
7. Solve two corners on D to make the layer complete aside from 1 dedge (this makes the next steps easier, but making a full layer could be faster in some cases).
8. Solve opposite corners with CLL.
9. Rotate so the almost completed layer is on D. Now fill in edges on R using the keyhole on the L face.

Last steps:
Method 1:
1. Solve the last R/L edge while fixing orientation parity (if necessary).
2. Permute the Middle edges while fixing permutation parity (if necessary).

Method 2 (simpler):
1. Solve the last edge on R/L.
2. Orient Middle edges.
3. Permute middle edges.
4. Fix any parity that remains.

The last few steps can be played around with and altered depending on the situation.

 

[cygorx]

FOPP Method?

Recently I've thought of an interesting way to solve the 3x3 cube. It's not spectacularly efficient, but I think it is worth sharing. It solves corners ("pillars") first, and then permutes the edges around the cube. I have named it the FOPP Method (tentatively) because the steps are of this order:

1. F2L pairs (first two layers) [side edges/corners only; no cross]
2. OLLC (orient last layer corners)
3. PLLC (permute last layer corners)
4. PE (permute edges)

The premise of the method was intuition; the ability to recognize the patterns very quickly, so that your execution time is your only substantial limit. Having the F2L in the beginning seriously advanced my look ahead speed and pattern recognition after a bit of pratice. I will try not to have a massive wall of text, but I will attempt to organize and explain each step. I will accept any criticisms, as this is a heavy work-in-progress.

Overview

Spoiler

The FOPP method contains four essential parts. Restating: F2L, OLLC, PLLC, and PE. It is a corners-first method, so there is no introductory cross, unlike CFOP. Most of the algorithms were not developed by me, and none of the algorithms were directly developed by me. I utilize 2-look Fridrich PLL algorithms, as well as Ortega OLL algorithms. Cube Explorer 5.01 aided me in the development of my edge flipping algorithms; however, they were optimized (just how optimal are they?). So, in the most basic form: You solve F2L pairs intuitively, use very short Ortega algorithms to orient the LL corners, use 2-look PLL algorithms to orient the corners, and use intuition, edge-flipping algs, and more 2-look PLL algorithms to permute the edges.

This is not a tutorial. This is a brief overview of the method.

Intuitive F2L

Spoiler

As in the Fridrich method, we pair up pieces and insert them. However, there is a massive benefit to solving this first. When we pick our scrambled cube up and begin inspecting it, we can start seeing into the first two or three or four F2L pairs. We do not have to worry amount messing up the bottom layer, so we can perform F moves to insert pairs without performing an F' afterward, for example. Other moves that are useful are found in the <l, r, d, L, R, D, M> group. I recommend you grab a cube, scramble it, and insert try inserting F2L pairs without the cross. If you practice the Fridrich method, you may find it confusing at first. After a bit of practice, you should begin to perceive the efficient ways of pairing and inserting.
An immediate weakness to this is that there are certain pair edges on the bottom layer; it can be difficult to look at the front edge pairs and the bottom layer for specific pieces at first. I suspect, with practice, the pairs will become easier and much quicker.

Algorithmic OLLC

Spoiler

I grabbed the Ortega 2x2 algorithms for OLL and implemented them in my method for orienting the corners. Basically, after solving the F2L pairs on bottom, we look at the top layer, select a case, and perform an algorithm with about 6 to 8 moves to make all the corners face in their solved orientation. You could, alternatively, use the Fridrich 2-look OLL algorithms for this, though they are less efficient. The only algorithm that I have changed is the H-Orientation algorithm, which cannot be performed with desired results on a 3x3. It mixes the middle layer. Perform the following algorithm first instead:

 Twizzle link 

F (R U R' U') (R U R' U') (R U R' U') F'

This step is also quite efficient.

Algorithmic PLLC

Spoiler

For PLLC, I simply used 2-look Fridrich PLL algs. Using the E-Permutation and CW Corner 3-cycle, I managed to create a somewhat efficient method of rotating the corners around. I suspect there is a faster way of getting them in the correct place, as, again, the edges need not be preserved.
If anyone can find such an algorithm, please tell me. I have been working on it for a while and have nothing worth sharing.
The pattern recognition is the exact same as in Fridrich: If you have two oriented corners in the correct position, put them in the back and perform a CW Corner 3-cycle. Otherwise, perform an E-Perm.

Here I will address two possible changes about the method.

1. I suspect that using 1-look PLL could benefit this, but for now 2-look will suffice as I have only recently began work on the method.
2. I think putting PLLC before OLLC could make this more efficient. That is, using a very basic cyclical algorithm or something similar to place the corners, then orient them using modified Ortega algorithms. Repeated algorithms, however, are filth. I will look into this.

Intutive + Algorithmic PE

Spoiler

This is the step in which we move the edges all around the cube. We first permute the bottom layer edges using <M, U> moves. For any correctly oriented edge, we can orient it vertically with its bottom layer slot and perform an M' U2 M to insert it. For any incorrectly oriented edge, perform a U M' U' M. This solves the bottom layer. We now use Fridrich PLL algorithms to move the edges around so that they are in the right position, though not necessarily the right orientation. When we have finished, we must flip certain edges as a "parity" problem. We can utilize one of three algorithms:
Z-Translation: F2 R' D' U F' L2 U2 L2 F D U' R F2 U2
XZ-Translation: F R' F' R' F2 L D R D' L' R' F2 R2
Orientation (special case, special name): L' R B2 L2 U2 L2 B' L2 U2 L2 B2 L R' U'
The Z-Translation flips two opposite edges. The XZ-Translation flips two adjacent edges. Orientation flips all edges on a face.

Note the discomfort of these algorithms. I doubt that I could ever obtain sub-1 on the Orientation algorithm. This is, I think, one of the primary weaknesses of the method.

Perceived Strengths and Weaknesses & Observations

Spoiler

Note that the method contains as many steps and exactly as many algorithms as the Fridrich method. Nevertheless, the move count is slightly less.

STRENGTHS

• Quick pattern recognition
• Most of the algorithms can be quickly performed
• Not many algorithms to memorize; mostly intuitive
• Optimized for look ahead
• Not very difficult to learn when properly presented
• F2L as a first step is effective

WEAKNESSES

• PE algorithms are uncomfortable, slow and annoying
• May take some time to get quick
• Last layer can be tricky
• Although the pattern recognition is easy, bad presentation of the method may lead to initial confusion
• Bottom layer edges can be hard to look for
• Some who are used to Fridrich F2L will have a difficult time transitioning

Thanks for reading. As previously stated, criticisms welcome. I really want to hear your opinions.

(Yes, this is a legitimate, non-troll thread, unlike my first thread when I may have come off as ignorant. I apologize.)

 

[mDiPalma]

Good thinking! Not very efficient, but lots of room for improvement.

Look into PCMS.

It's a more efficient version of the "same" method.

 

[Kirjava]

Step 2 and 3 should be one step. Step 4 isn't very good for speed.

 

[tx789]

step 2 and 3 can be combined into CLL it only has 42 cases.

 

[TheNextFeliks]

Sorry. Lots of people use this. It's called columns first or PCMS.

It's fun to do pairs, cross edges, oll, pll also.

 

[cygorx]

step 2 and 3 can be combined into CLL it only has 42 cases.

This was suggested to me by someone else as well. Again, the main goal was pattern recognition, so I tried limiting the amount of cases. Thanks!

Good thinking! Not very efficient, but lots of room for improvement.

Look into PCMS.

It's a more efficient version of the "same" method.

Wow, I had no idea someone had already thought of this. Thank you. It is very similar indeed. Do you think it is a good idea to try to continue, even if someone has already made something similar?

 

[qqwref]

Yeah, people have tried columns first before (my variant: http://mzrg.com/rubik/methods/col3/). There are definitely better pure edge flip algs than what you have (for instance, U (M'U)3 U (MU)3 for your "Z-Translation") but you should almost never need to do pure edge flips. PLLs for the last edges are also not recommended. Essentially, you need to take much more advantage of your freedom to move the M slice.

 

[cygorx]

Yeah, people have tried columns first before (my variant: http://mzrg.com/rubik/methods/col3/). There are definitely better pure edge flip algs than what you have (for instance, U (M'U)3 U (MU)3 for your "Z-Translation") but you should almost never need to do pure edge flips. PLLs for the last edges are also not recommended. Essentially, you need to take much more advantage of your freedom to move the M slice.

I had done some research but I didn't really find anything with columns first, but wow. Apparently it's pretty notorious. I see what you mean, I think I just thought an interesting way to do step 4 with slices, but it is similar to PCMS, a little.

 

[sameer26726]

I have invented a new method to solve Rubik's 3x3

I NEED HELP..... how should i patent it.. i mean i want my method to be included in speedsolving wiki can anyone help

 

[Stewy]

I NEED HELP..... how should i patent it.. i mean i want my method to be included in speedsolving wiki can anyone help

do you have an example?

 

[notfeliks]

A few months ago in this thread I posted an idea I had for a 1LLL subset; similar to ZBLL, only instead of edges oriented, it's corners oriented. I generated a few random cases and a majority of them had very usable algs.

Unfortunately, I do not have a good way to generate algs, so what I am requesting is someone capable of this. If the subset becomes popular, I will be happy to attach both our names to it. Thanks.

 

[IRNjuggle28]

I NEED HELP..... how should i patent it.. i mean i want my method to be included in speedsolving wiki can anyone help

Stop worrying about getting credit for your method. I already answered this question on facebook. If the method is any good, you will get the credit you deserve. Make an explanation of the method, and if people think it's good enough, it will get on speedsolving wiki. That said, the odds that you actually invented a new method that is good and has not already been thought of is very low. Do not worry about patents.

 

[supercavitation]

A few months ago in this thread I posted an idea I had for a 1LLL subset; similar to ZBLL, only instead of edges oriented, it's corners oriented. I generated a few random cases and a majority of them had very usable algs.

Unfortunately, I do not have a good way to generate algs, so what I am requesting is someone capable of this. If the subset becomes popular, I will be happy to attach both our names to it. Thanks.

I'd be happy to help out as long as the case number is under 500. Would you be able to calculate that number first?

 

[Renslay]

If the method is any good, you will get the credit you deserve. Make an explanation of the method, and if people think it's good enough, it will get on speedsolving wiki.

The Hoya method is a popular and fast method and was invented months/years ago, yet still not a single note about it on the wiki. Just saying.

 

[Clarkeeyyy]

The Hoya method is a popular and fast method and was invented months/years ago, yet still not a single note about it on the wiki. Just saying.

Although I do feel that Hoya deserves a mention on the wiki, I wouldn't say that it's a revolutionary method. It's just a slight variation of redux.

 

[TDM]

I'd be happy to help out as long as the case number is under 500. Would you be able to calculate that number first?

((2^3)*4!*4!)/(2*4) = 576
Reasons for each number:
2^3: EO
4!: EP
4!: CP
2: Parity
4: AUF

Although I do feel that Hoya deserves a mention on the wiki, I wouldn't say that it's a revolutionary method. It's just a slight variation of redux.

I wouldn't say that Yau is a revolutionary method. It's just a slight variation of redux.

 

[Clarkeeyyy]

I wouldn't say that Yau is a revolutionary method. It's just a slight variation of redux.

I agree.

 

[JKNK]

Hey guys, i thought of a 4x4 method . It's called Hoyau, a mix between hoya and yau. You create one center and put 3 edges in like in yau. Scince you have no top, it's really easy. After this you make the top center and finsih the cube like yau with 3-2-3 edge paring. My times reduced about 15 seconds.