L2L4 - The Lost Methode ? Help Find it!

[riffz]

In fact with most of your thoughts about step sequence I agree, but: EO with 6 instead of 5 Edges ist just 3 algs more, CP on the other hand is an very easy step, but doing it with Oriented edges may opens the door for aditional shortcuts with simple recog.

I don't see how corner recog would be any easier if edges are oriented...

 

[deadalnix]

As far as I get it:
1) Build FirstLayer - Was the Center Alignment intendet? I dont See an Reason for it?
2) and 3) I would do U FR'F'R U2RU2R' y to CO/ CP for the last Layer -> 42 Algorithems similar to Roux - Middle Layer Edges are oriented - intention or luck ?
4) M'UM'UM'U2 MUMUM - this one was easy but this sounds like alot of algs ... only 18algs I believe
5) 8 edges 8!/2 (~6500) possibilities I think this is only possible in at least 2 substeps

Did I miss somthing important?
What does L2L4 mean by the way?

EDIT: WARNING I mixed in CLL wich is not present in L2L4, You may just read on.

This is known as ofapel method for years. You can have a look here for step 2 : http://www.francocube.com/ofapel/cell.php

 

[oll+phase+sync]

What is called CELL on francocube is in fact identical to the respectiv Watermann methode Step.

But L2L4 means (just repeating another post)
1) Solve first layer
2) Orient Corners Last-Layer + Place 1st Edge in Middle Layer
3) Permute Corners Last-Layer + Place 2nd Edge in Middle Layer
4) Orient all edges + Place 3rd Edge in Middle Layer
5) Position all edges ... done

Big Benefit of this is that you have few Steps and you are doing algorithmic (fast) solving in 4 Steps while using preinspection to do a fast First layer.

BIG QUESTION: What do do when I'm lucky?
Answer: Keep it

But seriously There is a ~20% chance to have a solved edge before Step 2 (CO). Plus ~17% that CORNERS are already Permuted correctly after Step2. But how to benefit from this? Any Ideas?

 

[Kirjava]

But how to benefit from this? Any Ideas?

Solve three other edges and do EPLL last instead of L5E?

 

[qqwref]

using preinspection to do a fast First layer.

Really? I use inspection for this.

But seriously There is a ~20% chance to have a solved edge before Step 2 (CO). Plus ~17% that Centers are already Permuted after Step2. But how to benefit from this? Any Ideas?

If you have a solved edge before CO, you can just use a normal CO alg and then proceed. If corners (that's what you mean, right?) are permuted before CP, you can insert another edge in any way which does not disturb the corner positions. There aren't a lot of true lucky cases with L2L4 - you'd have to skip an entire step to be able to avoid doing the alg. Cases with one part solved will have to be included in our alg count.

Solve three other edges and do EPLL last instead of L5E?

I don't recommend this, because the algs on the CO, CP, and EO steps are not guaranteed to preserve an extra edge (many of them will probably preserve only as many edges as are absolutely necessary). So your solved edge may well disappear, and then you have gained nothing.

 

[Kirjava]

I don't recommend this, because the algs on the CO, CP, and EO steps are not guaranteed to preserve an extra edge (many of them will probably preserve only as many edges as are absolutely necessary). So your solved edge may well disappear, and then you have gained nothing.

It depends on what type of algs you learn.

Anyway, this is a way to either gain something or lose nothing. Unless you can propose a way of /always/ gaining something I don't see why you wouldn't recommend this strategy.

 

[qqwref]

I mean that I recommend what I said above - treating it as a normal case. If you have a skipped edge before CO, you can do a pure CO alg, which is probably nicer than the algs which also insert an edge.

 

[oll+phase+sync]

Solve three other edges and do EPLL last instead of L5E?

I.
EPLL is nice if a) its a 3-cycle b) all edges areoriented c) no edge must be moved

{ c) is maybe just nice but not short} But the rest is pretty ugly and still pretty much, or am I wrong?


II.
Solving 3 middle Layer Edges - that are not yet in the Middle Layer Is ther an easy way?

 

[Kirjava]

All edges are oriented anyway in step 4.

 

[oll+phase+sync]

L5E for L2L4

I started investigating How L5E feels for Middle Layer (I only found solutions with on edge in the Bottom layer and 4 in Top Layer on the web)

I have choosen to fill FR Slot - is that good ?


3-Cycle UB->UR->FR->
R' U' R' U' R' U R U R
F' M' F U' F' M F

3-Cycle UF->UR->FR->
U2 F' M' F U F' M F
U2 R' U' R' U' R U R U R

3-Cycle UL->UR->FR->
U R' U' R' U' R2 U R U R

UR<->FR,UB<->UL
R U R' U' F2 L' U' L U F2

UR<->FR,UF<->UL
U' R U' R' F2 U' L' U L F2

UR<->FR,UF<->UB
U2 R' U2 B2 R2 B2 U2 L R U2 L'
U2 L' F2 U2 L2 U2 F2 L2 U2 L'
...

There should be 16 Cases at all if I have to place in FR or BR slot there are 12 more Mirrors?

There are no really short sequences though.

Regarding EO i believe there are 12 Orientation patterns:
-We must orient 4 Edges in Top and 2 Edges in middle layer.
-There can be 0, 1, 2 , 3 or 4 misoriented in Top.
-There can 0 , 1 ,2 misoriented in MiddleLayer.
- I Made the cases bold wich have 2 different patterns.
- I'm only allow to combine even with even and odd wit odd pattern.

Number of O-Cases =[1,3]*[1] + [0,2,4]*[0,2]
= 2*2 + 4*2 = 12

0-0
The "Skip" - Case with all Edges already oriented al leasat won't need any special algs.

0-2
But already the Case with 4 oriented on top and 2 bad in Middle has 3 cases:
1) "Target" Edge in Top Layer -> solvable by a 3 cycle similar to RUR'U'M'URU'r'
2) Target" Edge in its position -> something similar M'UM'UM'U2MUMUMU2
3) lucky the alg from 1) does it again.


1-1
One misoriented in each layer
1) Target Edge is misoriented and not in its Slot => any 3 cycle again
2) Target Edge is misoriented and in its Slot => again similar to 2) from above there might exist somthing better, but case seems to be rare
3) Target edge oriented in Middel Layer => 3cycle again
4) Targetedge oriented in Top Layer and Target Slot contains miss oriented => again 3 cyle (with one move conjugation)
5) If I would be able to do L5E with BR or FR Slot I could skip this an solf FR. but else I need algs:

UF~, FR~, UB -> BR-> ( ~ denotes orientation change)
R F R U R2 U' F' U' R' U2

UR->BR~ , FR~
R U F U R2 U' R' F' R'

UR~, FR~, UB->BR
R U L F R2 F' L' U' R'


... is that the way to go ?

 

[Zarxrax]

I suspect that this L2L4 method wont be faster than fridrich for speedsolving, due to recognition.

 

[jms_gears1]

I suspect that this L2L4 method wont be faster than fridrich for speedsolving, due to recognition.

I suspect recog should not be hard at all.

especially the first two steps considering you can see a good majority of the cube at any given time

Also i would position the first two edge i solved in B then you could see every edge youd need.

Also i would build the face on L like rouxs method but adding a 1x3x3 on top.

 

[Zarxrax]

But, you aren't really going to be able to look ahead, probably. You will have to stop, figure out the case, stop, figure out the next one, and so on. With fridrich f2l, you can look ahead through the whole f2l, so there are no pauses, and then you only have to stop to look for the last 2 steps.

 

[oll+phase+sync]

I don't see how corner recog would be any easier if edges are oriented...

In L2L4 Corner and Edge Lookup are combined - so having oriented edges earlier during the solve would improve lookup and reduce number of algs.

But my original idea to do EO after CO does even increase the number of cases (due to me only having five fingers and fiddling around with higher numbers )

I also tried to to do EO after building the first Layer, but I faild on finding something like Roux-iteration, the first layer preservation was unpleasent, and being forced to place some edges on the way it more felt like an petrus / heise variation . ZZ - User may feel differnt about that.

 

[oll+phase+sync]

But, you aren't really going to be able to look ahead, probably. You will have to stop, figure out the case, stop, figure out the next one, and so on. With fridrich f2l, you can look ahead through the whole f2l, so there are no pauses, and then you only have to stop to look for the last 2 steps.

Not saying fridrich is not fast - that would be very unrealistic - but that does not mean there are no other fast solution. And the ability to look ahaed is also based on training a lot.

For example ZB-LL or ZZ-LL are huge steps
and you benefit from these only after Mastering this Step. While in F2L and also in L2L4 also smaller Improvments in any step are (I hope) possible, and directly pay off.

For example I can predict then CP case before starting CO. (same as knowing/changing PLL-class before/througout OLL)

Also condider L2L4 will only reqire you to look up two thing at a time : an Egde and ... "something else" - reminds me of something (and "something else" is always at the same spot with no hiding spots )

 

[qqwref]

Yeah, the recognition wouldn't be too bad, about as bad as F2L I think. Solving the first layer is no problem; the CO and CP steps require finding an edge plus the easy corners recognition; EO is a little trickier but shouldn't be any worse than Petrus's step 3 because you only have 6 edges to look at (and the ones on E should be adjacent); EP recognition should be more or less similar to EPLL recognition, just with an additional piece.

 

[cincyaviation]

But, you aren't really going to be able to look ahead, probably. You will have to stop, figure out the case, stop, figure out the next one, and so on. With fridrich f2l, you can look ahead through the whole f2l, so there are no pauses, and then you only have to stop to look for the last 2 steps.

But for F2L turning speed is limited so that you have the ability to look ahead, this method nearly eliminates the need for lookahead, so that you can reach Max TPS for most of the solve.

 

[a small kitten]

But for F2L turning speed is limited so that you have the ability to look ahead, this method nearly eliminates the need for lookahead, so that you can reach Max TPS for most of the solve.

I don't understand what you are talking about. What?

 

[nickvu2]

Anyone up for an alg-genning session?

(Is there a better [short] verb than 'genning' for this?)

I might be up for some alg...manifestation.

 

[oll+phase+sync]

I might be up for some alg...manifestation.

Me too. I also did a pretty minor start at #30 in this threat.

Where/How should we collect our results?
Wich process do we use do identify the 'good' algs from all the possible ones?