new method

[doubleyou]

yesterday I came up with this idea for a method while falling asleep.
I was almost sleeping but when I got the idea I just had to get up again and turn the light on and try it out..

it did not work

but I think it might - with some work

ok, here is the outline: (this describtion is made in Fridrich perspective)

1. put 3 middle layer pieces in place. (so you got a broken "ring"/keyhole)

2. seperate pieces from 1st and 3rd layer. as you seperate the last piece you close the ring so you got a complete middle layer.

3. perform LL algs from Fridrich on both 1st and 3rd layer.

thats right - NO cross. NO f2l. I know this is some quick steps, but the idea for this came as most people is forever faster in the LL than in the f2l.

The thing is (as you propably worked out youself, when reading this) that you get more complicated/odd cases in the first of the two last layers you fix.

DO you think algorithms can be made up for this approach?

OR is this just another stupid ass newbish revolutionary"!!!" idea?

please bare with me,

-Sigurd

 

[joey]

It depends on how quick the first two steps could be done.
If you have a 4s LL, thats 8s for step 3. So the first two steps would need to be quite fast, and I think there would be a few wierd cases.

I just tried a solve with this method, and the only thing that was wierd was having 3 edges flipped when I came to step 3.

EDIT: 48s on my first try.

 

[pjk]

That sounds interesting, I will give it a shot later.

 

[deKeijzer]

The problem with this method is that you can have a parity error combined over the opposite layers.
As you know, there have to be at least 2 edges mis-oriented on the last layer. But with this method there could be one of those 2 edges on either layer.

 

[doubleyou]

deKeijzer, thats right!

but in theory, you could keep this parity problem in mind and avoid it in the end of 2.

I think this method would only be benefitial if you already had a 2Look

 

[doubleyou]

ok what deKeijzer said opened my eyes.
I didnt even realize the truth of that before after I replied.

IF you avoid this parity while doing step 2, then you get exact normal cases from there!

THEN my system works!!

EDIT!!

or what! I seem to be having problems with this.
more than two pieces can be oriented wrongly. this can be harsh to fix..


ok name:

what about "K S 1OP 3OP"

Keyhole - Seperation (+parity) - 1st layer Orientation, Permutatinon - 3rd layer Orientation, Permutation

or simply just "50/50"

 

[doubleyou]

btw. W00t!! a parity on 3x3x3 :wub:/

 

[joey]

The parity isn't hard to fix.
U'M'U'M'U'M'U'M' M'U'M'U'M'U'M'U'

EDIT:
I know theres a non slice version, Ill have a look for it.

EDIT2:
Here it is:
R U' R2 U2 R2 U R' U' R' U2 r U R U' r' R U R'

 

[doubleyou]

theory:

solve everything ignoring parity

then solve those last with commutators!

w00t!

edit:

THANKS joey for providing that alg!!!

 

[doubleyou]

btw. please say where the parity pieces should be before aplying the alg

I am newbish and this is not easily figured out on my own

 

[joey]

Just do the algs, and see what pieces are affected.

 

[pjk]

I tried this. It seems like it could be good. Separating the pieces can be done quickly with practice. You can have 1 incorrectly oriented corner in each layer, where you would have to fix the parity. But yeah, it looks good.

 

[joey]

Yeh, we need an alg for that case. I would open Cube Explorer, but I'm not on the right computer.

 

[Arakron]

The edge parity would be pretty easy to avoid, I think. You'd need to pay more attention to corners.
Also, I think you can combine steps 1 and 2, solve the ring while separating layers 1 and 3. This has been tried before, I think. There was a thread on twistypuzzles pretty recently about it, if I remember right.

 

[Cubinator]

I'd search for a parity fix with JACube, but I have no clue how you'd specify that the pieces are in the right layer but with an unknown destination in said layer.

A few notes:

This reminds me of solving a 4x4: you can have permutation and orientation parities for both the edges and the corners.

Edge permutation parity can be quickly fixed via M2 U2 M2 U2: that swaps UF and UB, and also DF and DB, while preserving everything else.

Getting the corners and edges to the right layers is irritating. The corners are a big problem. It'd be much easier to do edges first and worry about corners after that.

Because of the parities, LL is kindof tricky and will probably take a bit longer than a normal LL.

This really needs some special algs for it to work.

 

[pjk]

To fix the corner parity, I would just use a commutator that I use for CO of BLD. It is:
A=(R'D'RD)x2
To flip to corners and fix the parity, I would do:
U'AUA'

Or the inverse of that if the corners need to be rotated the other way. If you do a z' before performing this, it makes it execute much quicker.

 

[Johannes91]

I just found these:
R2 U2 R' U' R U' R' U2 R' for CO parity.
M' U M U' M' U2 M for EO parity.
M2 U2 M2 for permutation parity (this was already mentioned).

 

[doubleyou]

well I found that placing edges first is logical. for corners I just do the "edge with d face color on top, on top of a correctly placed edge piece" -move. RUR'U' 3 times. I ofc use it to exchange places of two incorrect pieces.

I know this is slow but it works.

I didnt know anyone else had done this before.. please lead me towards any describtions!

I still dont get your algorithms tho. But thank you so much for providing!

I know this isnt a very fast method, but I think in time it could be around or a bit slower than fridrich.

 

[Arakron]

Originally posted by doubleyou@May 2 2007, 01:51 PM
I know this isnt a very fast method, but I think in time it could be around or a bit slower than fridrich.

Fridrich is what most record-holders use, so that'd be quite fast even if it was a bit slower.

 

[doubleyou]

Originally posted by Arakron+May 2 2007, 08:16 PM--></div><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td>QUOTE (Arakron @ May 2 2007, 08:16 PM)</td></tr><tr><td id='QUOTE'> <!--QuoteBegin-doubleyou@May 2 2007, 01:51 PM
I know this isnt a very fast method, but I think in time it could be around or a bit slower than fridrich.

Fridrich is what most record-holders use, so that'd be quite fast even if it was a bit slower. [/b][/quote]
you know I know this ^^

I forgot to add the alg Jeyns use in his method in the last step where you orient corners in pairs.

you "customize" the alg as you see what direction the piece needs to be rotated.

UFL cw LD2L'F'D2F

then move the other corner piece that needs rotation in the UFL position and if ccw F'D2FLD2L'

this can be used in the CO parity