ZZ Cubers

[Escher]

damn you pochmann!

i didnt have a cube to hand.

and now i do.

and i am wrong.

EDIT

what IS correct, is if the bad edges are on the F and B slices, not faces, and you do a y rotation.

 

[Stefan]

what IS correct, is if the bad edges are on the F and B slices, not faces, and you do a y rotation.

You might want to double-check that... (and clarify what you mean with "on the F and B slices, not faces").

 

[Escher]

what...

okay. if 4 BAD EDGES ARE ON THE E SLICE, AND ON THE F AND B SLICES THEN DO A y.

i think i might just give up this whole posting lark.
... or just make sure what i say is exactly correct.
... or completely ignore stefan.

 

[Stefan]

if 4 BAD EDGES ARE ON THE E SLICE, AND ON THE F AND B SLICES THEN DO A y.

And that achieves what?

Consider the superflip. All 12 edges have wrong orientation. Do y rotation. Still all 12 edges have wrong orientation.

 

[Matthew]

what...

okay. if 4 BAD EDGES ARE ON THE E SLICE, AND ON THE F AND B SLICES THEN DO A y.

i think i might just give up this whole posting lark.
... or just make sure what i say is exactly correct.
... or completely ignore stefan.

If this edges are from U/D slice ok - move y/y' "orient" them.. but if this edges are from the E-slice.. it don't make a expected effect

Btw I have a 15.06 avg on the video - I will post this in the next couple of hours...

 

[JohnnyA]

Yellow/white edge stickers can be on U and D of course, but they can also be on F or B. Am I wrong here?

In terms of rotations, it's more complicated than we are thinking. If the edge has a U/D sticker and is in the middle layer, y/y' will change it's orientation. If it is in the top layer, no rotation will change it's orientation. However if it has no U/D sticker, y/y' will change orientation only if it is in the U/D layer.

So there is no "mathematical" way to determine how y/y' will affect orientation unless you count how many of each type (with U/D or without U/D) are in middle and U/D layers.

EDIT: On second thoughts, maybe it would be possible to do that - possibly by counting how many U/D stickers are in the middle layer, then you know how many are in U/D layer by subtraction, and you also know that the same number of middle layer edges are in the U/D layers. Would this be practical/workable?

 

[Matthew]

Matthew could you post up an average? I want to see really fast ZZ solves

My 15.06 avg of 12:

 

[bamman1108]

Old, but, if you look at both EOLine and the cross using half turn metric, wouldn't they be close to the same?

I'm not really one to talk though since I have no knowledge of edge orientation whatsoever, which is stopping me from learning ZZ as of now.

 

[yurivish]

Matthew—I noticed that you're using COLL for the sune case as well. Right now I use COLL for everything but sune and antisune. I wanted to learn those cases as well, but some of the algorithms were really slow.

Are the ones you use any good? I looked on a few sites and couldn't find any nice ones for some of the cases.

 

[JohnnyA]

Wow, those are some great solves Matthew. Hopefully I'll be able to get that fast with my ZZ when I work on it after French Open. For EOLine, are you colour neutral or do you always start with the same orientation of the cube?

Also, what do you do for blockbuilding? Do you fill in one half and then use petrus approach?

 

[Matthew]

Matthew—I noticed that you're using COLL for the sune case as well. Right now I use COLL for everything but sune and antisune. I wanted to learn those cases as well, but some of the algorithms were really slow.

Are the ones you use any good? I looked on a few sites and couldn't find any nice ones for some of the cases.

I collect algs for many sites - but you're right - some algs for Sune and a-sune coll's cases are horrible..

Wow, those are some great solves Matthew. Hopefully I'll be able to get that fast with my ZZ when I work on it after French Open. For EOLine, are you colour neutral or do you always start with the same orientation of the cube?

Yes - I always solve with blue on front and yellow on top.. (oc while f2l and LL - in EOLine i rotate the cube)

 

[yurivish]

Matthew—I noticed that you're using COLL for the sune case as well. Right now I use COLL for everything but sune and antisune. I wanted to learn those cases as well, but some of the algorithms were really slow.

Are the ones you use any good? I looked on a few sites and couldn't find any nice ones for some of the cases.

I collect algs for many sites - but you're right - some algs for Sune and a-sune coll's cases are horrible..

That's cool, you use the same color orientation as me. I know that ZZ himself uses red in front.

Have you noticed that doing Sune-COLL + ePLL is better than Sune + PLL? If a sune is faster than your average COLL and a PLL is slower than your average ePLL, then it should be roughly equal. Have you noticed any speed difference either way?

It seems that Sune + PLL would give some spike times, when you get Sune + easy PLL, and some worse times – when you get an N or an E.

 

[Matthew]

Have you noticed that doing Sune-COLL + ePLL is better than Sune + PLL? If a sune is faster than your average COLL and a PLL is slower than your average ePLL, then it should be roughly equal. Have you noticed any speed difference either way?

I don't know but.. using coll we have a higher possibility to EPLL-skip I don't say even about ZZLL

 

[blah]

if 4 BAD EDGES ARE ON THE E SLICE, AND ON THE F AND B SLICES THEN DO A y.

And that achieves what?

Consider the superflip. All 12 edges have wrong orientation. Do y rotation. Still all 12 edges have wrong orientation.

Exactly.

Keep this in mind:
After a y/y' rotation,
U/D edges on U/D are not flipped
U/D edges on E are flipped
E edges on U/D are flipped
E edges on E are not flipped

Yes I know that's extremely trivial, but I'm surprised by the number of people who don't realize this. Somehow people tend to think that doing a y/y' rotation is sort of a "black becomes white, white becomes black" move - it's actually a "gray" move, in which you have nonzero probabilities of getting more, fewer or the same number of bad edges.

I did a lot of calculations last night, probabilities and stuff. And this is what I found: If you have exactly 8 bad edges, the probability of you getting 6 or fewer bad edges after a y/y' rotation is 67.953%, which is somewhat considerably higher than 50%. The probability of you still getting 8 bad edges after that rotation would be 27.825%. Add those 2 figures up, and you can be *pretty* sure that after a y/y' rotation, you won't be worse off. But it's worth knowing this fact: P(10 or 12 bad edges after y/y' rotation | exactly 8 bad edges before y/y' rotation) = 4.222%; P(10 or 12 bad edges on a random scramble) = 3.271%.

I don't know about anyone else, but 68% just doesn't seem convincing enough to me. I think I'll stick to flipping those 8 edges when I get them. Should I get some numbers for cases when there are 10 bad edges as well? :rolleyes:

By the way, can someone do an independent calculation as well, just to confirm these figures? I may be completely wrong

P/S: Should I start this in a new thread? Anyone interested in discussing this?

 

[JohnnyA]

if 4 BAD EDGES ARE ON THE E SLICE, AND ON THE F AND B SLICES THEN DO A y.

And that achieves what?

Consider the superflip. All 12 edges have wrong orientation. Do y rotation. Still all 12 edges have wrong orientation.

Exactly.

Keep this in mind:
After a y/y' rotation,
U/D edges on U/D are not flipped
U/D edges on E are flipped
E edges on U/D are flipped
E edges on E are not flipped

Yes I know that's extremely trivial, but I'm surprised by the number of people who don't realize this. Somehow people tend to think that doing a y/y' rotation is sort of a "black becomes white, white becomes black" move - it's actually a "gray" move, in which you have nonzero probabilities of getting more, fewer or the same number of bad edges.

I did a lot of calculations last night, probabilities and stuff. And this is what I found: If you have exactly 8 bad edges, the probability of you getting 6 or fewer bad edges after a y/y' rotation is 67.953%, which is somewhat considerably higher than 50%. The probability of you still getting 8 bad edges after that rotation would be 27.825%. Add those 2 figures up, and you can be *pretty* sure that after a y/y' rotation, you won't be worse off. But it's worth knowing this fact: P(10 or 12 bad edges after y/y' rotation | exactly 8 bad edges before y/y' rotation) = 4.222%; P(10 or 12 bad edges on a random scramble) = 3.271%.

I don't know about anyone else, but 68% just doesn't seem convincing enough to me. I think I'll stick to flipping those 8 edges when I get them. Should I get some numbers for cases when there are 10 bad edges as well? :rolleyes:

By the way, can someone do an independent calculation as well, just to confirm these figures? I may be completely wrong

P/S: Should I start this in a new thread? Anyone interested in discussing this?

I figured you could work out the best orientation for your cube by counting the number of U/D pieces in the middle layer, which tells you how many of each type (U/D or Middle Slice) pieces are in each layer. But then I realised you need to know their current orientation in order to make any concrete decisions about whether to turn or not.

 

[Matthew]

I figured you could work out the best orientation for your cube by counting the number of U/D pieces in the middle layer, which tells you how many of each type (U/D or Middle Slice) pieces are in each layer. But then I realised you need to know their current orientation in order to make any concrete decisions about whether to turn or not.

But remember that we have only 15s of preinspection!

In 15s isn't easy to see a good solution to the full-EOLine - and you want to look even from two views - isn't better have in pinky the EOLine from one-view?

Apart from this - I don't like only cases when I have 10 or 12 flipped edges - but when I have 8 or 6 or low.. always is the way to see solution <=10 moves in 15s

 

[JohnnyA]

I figured you could work out the best orientation for your cube by counting the number of U/D pieces in the middle layer, which tells you how many of each type (U/D or Middle Slice) pieces are in each layer. But then I realised you need to know their current orientation in order to make any concrete decisions about whether to turn or not.

But remember that we have only 15s of preinspection!

In 15s isn't easy to see a good solution to the full-EOLine - and you want to look even from two views - isn't better have in pinky the EOLine from one-view?

Apart from this - I don't like only cases when I have 10 or 12 flipped edges - but when I have 8 or 6 or low.. always is the way to see solution <=10 moves in 15s

I personally find 10 and 12 edges easier than 6. 6 is the most difficult for me. When I get 6, I almost always rotate to get 8 or 4 or a different number ... it's because I don't like having just 2 edges left over to solve with some variation of F'UF. I don't think rotations are useful, unless you are like me and hate having a certain number of edges to orient.

Also, how about ignoring U edges during orientation and doing OLL+PLL if you don't know COLL yet?

 

[yurivish]

I figured you could work out the best orientation for your cube by counting the number of U/D pieces in the middle layer, which tells you how many of each type (U/D or Middle Slice) pieces are in each layer. But then I realised you need to know their current orientation in order to make any concrete decisions about whether to turn or not.

But remember that we have only 15s of preinspection!

In 15s isn't easy to see a good solution to the full-EOLine - and you want to look even from two views - isn't better have in pinky the EOLine from one-view?

Apart from this - I don't like only cases when I have 10 or 12 flipped edges - but when I have 8 or 6 or low.. always is the way to see solution <=10 moves in 15s

Also, how about ignoring U edges during orientation and doing OLL+PLL if you don't know COLL yet?

You can still gain from orienting the top-layer edges even if you don't know COLL, since you're always left with an easy cross OLL. And it increases the chance of an OLL skip to 1/27, which is actually noticeable.

 

[Matthew]

I personally find 10 and 12 edges easier than 6. 6 is the most difficult for me. When I get 6, I almost always rotate to get 8 or 4 or a different number ... it's because I don't like having just 2 edges left over to solve with some variation of F'UF. I don't think rotations are useful, unless you are like me and hate having a certain number of edges to orient.

Also, how about ignoring U edges during orientation and doing OLL+PLL if you don't know COLL yet?

Hmm.. problem is not to find a flipped edges.. a problem is to plan a comfort and quick and low-move count EOLine..

IMO finding flipped egdes should take You ~2-4s... rest of the time You should spend at planing your solution for EOLine..

And try solving EOLine blind - this should make your lookahead (findind first block pieces) better ;]

 

[JohnnyA]

I can do EO blind, but I can almost never get the red/orange edges (i.e. line) done without looking again ;[