The New Method / Substep / Concept Idea Thread

[Teoidus]

If dedge pairing takes more moves, then it'll pretty much be inferior to Yau/Meyer ZZ does not save that many moves compared to CFOP, and I think that still holds, even with EPLL+parity

 

[Shiv3r]

No, I saisd wrong. It's not too inefficient, It just is how you place edges, sometimes an extra move is needed to orient an edge, and there are some nice algs you can do to skip parity at the same time as pairing the last two edges. The algs are generally nicer than Yau, and it skips parity completely, so it is actually a little more efficient than Yau/Meyer

 

[Sion]

On my method for 4x4, I might just learn K4 LL, making my method BCFK (Block, Columns, Final two centers and FD Dedge, K4 Last Layer (CLL, Commutator ELL)

 

[AlphaSheep]

No, I saisd wrong. It's not too inefficient, It just is how you place edges, sometimes an extra move is needed to orient an edge, and there are some nice algs you can do to skip parity at the same time as pairing the last two edges. The algs are generally nicer than Yau, and it skips parity completely, so it is actually a little more efficient than Yau/Meyer

No, as I pointed out before, you don't skip parity. Try generating algs for all of the cases. Half of them will be nice short algs that can be reduced to some variation on slice, swap edges, slice back (and the algs are exactly the same as Yau), and the other half will be really nasty algs that are as bad as or worse than the standard OLL parity algs.

 

[xyzzy]

No, as I pointed out before, you don't skip parity. Try generating algs for all of the cases. Half of them will be nice short algs that can be reduced to some variation on slice, swap edges, slice back (and the algs are exactly the same as Yau), and the other half will be really nasty algs that are as bad as or worse than the standard OLL parity algs.

r U2 r U2 F2 r F2 l' U2 l U2 r2 is pretty nice actually. (Or its mirror, which seems to be more popular.) You can also use a conjugate [F': r U2 r U2 F2 r F2 l' U2 l U2 r2] to get an alg for another L2E+parity case.

It's a bit like whether you choose to do L2E algs on 5x5x5; people generally agree that L2E is good for 5x5x5, so provided that EO inspection is easy, this should be preferred over a separate OLL parity step. In Z4 and Z4-like methods most of the edge pairs are already oriented by the time you get to pairing the last two, so this shouldn't be too hard.

 

[Teoidus]

I feel like if you're going to go for this, it's much better to just keep track of which dedges end up flipped while chain pairing as normal, then doing EO the traditional ZZ way. Otherwise you can't chain pair and you have probability of getting a very long pairing case

Another idea would be to place all wings in their proper orbits at the beginning of the solve. idk how you'd do this but that'd be perrty damn cool

 

[Shiv3r]

The problem I have with sion's method is that its pretty much K4 but he does an OBLBL blockbuilding approach to get f3l.

 

[Sion]

The problem I have with sion's method is that its pretty much K4 but he does an OBLBL blockbuilding approach to get f3l.

I looked at OBLBL, It develops a 3x3 block from a 2x2 block, and then doing the final extension. Sion Does 1x3x3, 2x3x3, 3x3x3, 4x3x3.

As for last layer, it is still up in the air; I'm definitely thinking in terms of some form of a unique LNE (Last nine edges) of three steps, after solving COLL.

L5EO: do a general L5EO to flip most of the edges as if it were on a 3x3.

LNEA: Last Nine Edges Augment, flip remaining edges while inserting Last FD edge. this would also be the prime step to solve Parity cases.

EEPLL: extended EPLL, or just permute the already oriented edges, but unsolved edges in a single algorithm.

 

[Teoidus]

Movecount?

 

[Sion]

Movecount?

Not sure. I need to find software to help me gather algs. this method is still im very early development.

 

[Shiv3r]

I think I have come up with a nice 1look last layer with very few algs for ZZ-d. in a 2-gen solve, you can either do 2-gen OCLL and then EPLL, which isn't that nice, or full 2GLL, which has about 85 algorithms. I have come up with an algorithm set for solving the last layer in 1 look when the edges are oriented phased and the corners are permuted. The alg count? 28. It is just the ZZLL algorithms for the 2-gen COLL's. It may be similar to ZZ-orbit or things like that, but if this hasn't been come up with yet, Im calling it ZZ-dab, because Its ZZ-d with a ZZ-a/ZZ-b style 1LLL.

 

[TDM]

I think I have come up with a nice 1look last layer with very few algs for ZZ-d. in a 2-gen solve, you can either do 2-gen OCLL and then EPLL, which isn't that nice, or full 2GLL, which has about 85 algorithms. I have come up with an algorithm set for solving the last layer in 1 look when the edges are oriented phased and the corners are permuted. The alg count? 28. It is just the ZZLL algorithms for the 2-gen COLL's. It may be similar to ZZ-orbit or things like that, but if this hasn't been come up with yet, Im calling it ZZ-dab, because Its ZZ-d with a ZZ-a/ZZ-b style 1LLL.

This has been thought of lots of times before; the earliest I know of it being suggested was here. Most people who've discussed this have come to the conclusion that forcing the edges to be unphased would actually be better, since not only would LS be shorter, but the LL algs would also be better. Though obviously you'd want to not always for unphased edges, e.g. if there's a 3 move WV/SV (which could lead to a LL skip).

 

[Shiv3r]

This has been thought of lots of times before; the earliest I know of it being suggested was here. Most people who've discussed this have come to the conclusion that forcing the edges to be unphased would actually be better, since not only would LS be shorter, but the LL algs would also be better. Though obviously you'd want to not always for unphased edges, e.g. if there's a 3 move WV/SV (which could lead to a LL skip).

That would double the alg count, I believe. If that isn't nice, what about 2-gen Winter variation into EPLL? I have a 2gWV alg sheet if you want me to share it with you guys

 

[TDM]

That would double the alg count, I believe. If that isn't nice, what about 2-gen Winter variation into EPLL? I have a 2gWV alg sheet if you want me to share it with you guys

It would - there are 54 algs for LL - but it's still far lower than most other 1-look LL methods.

I'd be interested in seeing a 2-gen WV alg sheet - surely the algs aren't that good? I generated algs for 2-gen WV ignoring the edge a few days ago and some of the algs for that were already 11+ moves. But maybe the algs for normal WV aren't actually that much longer, in which case it might actually be good. I've never tried generating algs for this before so I honestly don't know much about this!

 

[Shiv3r]

It would - there are 54 algs for LL - but it's still far lower than most other 1-look LL methods.

I'd be interested in seeing a 2-gen WV alg sheet - surely the algs aren't that good? I generated algs for 2-gen WV ignoring the edge a few days ago and some of the algs for that were already 11+ moves. But maybe the algs for normal WV aren't actually that much longer, in which case it might actually be good. I've never tried generating algs for this before so I honestly don't know much about this!

here. Not all the algs are nice, but still... Its possibly the best way to solve a cube 2gen, I think...

 

[xyzzy]

Not all the algs are nice, but still... Its possibly the best way to solve a cube 2gen, I think...

Is it possible to get nicer non-2-gen algs that still preserve the 2-gen corner permutation? For example, F2 L' U' L U F2 and its inverse are 1-2 moves shorter than the 2-gen version (U' R U R' U2 R U' R') and still preserve CP, though in this case the last two pieces aren't already paired up and this isn't a lot nicer than the <R,U> one to begin with.

Or, just learn full 2GLL!!!!!111! Around half of the cases can be done with two Sunes (for most of these cases there aren't any better 2-gen algs), and some of the remaining cases are a combination of algs (Sune + Bruno = R U R' U R' U' R2 U' R2 U2 R) or a conjugate ([R U R' U: Sune U2]); in effect you really only need to learn about 15 algs and their mirrors/inverses.

 

[TDM]

here. Not all the algs are nice, but still... Its possibly the best way to solve a cube 2gen, I think...

Wow, nowhere near as long as I was expecting. The high chance of H/Z is still a problem if you want to stay fully 2-gen though (but isn't such a problem if you don't mind doing <M, U>).

 

[Teoidus]

I think the best way to solve the 2gen group wouldprobably either be something intuitive like LSE or some 2phase algorithm set

 

[Sue Doenim]

I think that it would be useful to have "partial Winter Variation" algs, more than partial corner control but less than WV, that would orient 2 corners, adjacently if if possible. This would be helpful for Roux, drastically reducing the number of OLLCP algs.

 

[shadowslice e]

I think that it would be useful to have "partial Winter Variation" algs, more than partial corner control but less than WV, that would orient 2 corners, adjacently if if possible. This would be helpful for Roux, drastically reducing the number of OLLCP algs.

This can be done intuitively by varying the final insert.

But Roux doesn't use OLLCP...