# [Help Thread]4x4 Discussion and Help

#### xyzzy

##### Member
View attachment 20524
What's the best alg for this case? I currently do Na then Opposite Parity.
2R U 2R' U' r' F r2 U' r' U' r U r' F' R U R' U' r' D' r U2 r' D r
[algtwizzle=4x4x4]2R U 2R' U' r' F r2 U' r' U' r U r' F' R U R' U' r' D' r U2 r' D r[/algtwizzle]

(edit: hm, Twizzle integration seems to be broken currently)

#### Ekaufman

##### Member
Hello all
I’m still somewhat new to cubing. I can solve a 2x2 & 3x3 and I’m now working in solving a 4x4 using the beginners method. However I seem to keep running into the same problem and can’t get a yellow cross. Sometimes I do a recognizable pattern like the J or the bar and do those algorithms but it never brings me to the yellow cross. Just scattered yellow pierces. Do I need to do the last yellow edges before yellow cross and if so how? I’ve included pictures. These are brand new cubes that came to me solved and I scrambled myself. I did not have any pops or anything so there’s no chance a piece is flipped wrong. Please help I’m getting super frustrated

#### bulkocuber

##### Member
Hello all
I’m still somewhat new to cubing. I can solve a 2x2 & 3x3 and I’m now working in solving a 4x4 using the beginners method. However I seem to keep running into the same problem and can’t get a yellow cross. Sometimes I do a recognizable pattern like the J or the bar and do those algorithms but it never brings me to the yellow cross. Just scattered yellow pierces. Do I need to do the last yellow edges before yellow cross and if so how? I’ve included pictures. These are brand new cubes that came to me solved and I scrambled myself. I did not have any pops or anything so there’s no chance a piece is flipped wrong. Please help I’m getting super frustrated
I'm not sure if I understood what exactly your your cubes looks like because you didn't send include the images, but it's probably just OLL parity. Does your cube look like thisor like this?(an image I found or Reddit)

If it's the first case, it's just OLL parity, meaning that there's a single flipped edge that creates an impossible 3x3 case, so you just do the alg and you get one of the cases you can get on 3x3 (L shape, dot, line, or cross)
If your cube looks like the second image then you didn't finish pairing up the edges, so before getting to 3x3 stage remember to check if you finished edge pairing.
If your cube doesn't look like any of those then I failed and you could watch a tutorial on YouTube (or you could just send a photo of your cube)

#### LBr

##### Member
If you struggle to do the corners, then you may have a parity error commonly referred to as pll parity. It’s a simple fix tho

#### Timona

##### Member
Asides from computing power, is there any significant difference between 4x4 fast scrambles and regular WCA scrambles?

#### xyzzy

##### Member
Asides from computing power, is there any significant difference between 4x4 fast scrambles and regular WCA scrambles?
Yes. There's a non-negligible increase in chance of getting F2C skip on 40-random-move scrambles compared to proper scrambles.

Also random-state ("regular WCA") scrambles are easier to read because the first half of a random-state scramble is entirely outer-layer moves.

#### beniolenio

##### Member
I've just gotten back into 4x4 and I know last two edge edge pairing algorithms for every case but I've forgotten one of the algorithms that I really liked and was very easy. All I remember about it is that it has this move sequence in it: (Rw' F R F' r). Does anyone know this algorithm? I miss it. I believe it's supposed to pair opposite edges on the top layer.

Edit: I also can't easily find this alg on the Internet. The internet only really talks about 4 main algs for top layer edge pairing, none of which are the one I'm looking for.

So I know 2 OLL parity algs.

Rw U2 x Rw U2 Rw U2 Rw' U2 Lw U2 Rw' U2 Rw U2 Rw' U2 Rw' and Rw' U2 Rw' U2 Lw U2 Rw' U2 Rw U2 x U2 Rw2 U2 Rw' U2 Rw U2

The first one is the default parity alg that everybody uses, which (if you didn't know) affects PLL parity. The second one, which I found from Hashtag Cuber's video, doesn't affect PLL parity in anyway. If you did the second one and you had PLL parity during OLL, it would still give you PLL parity, but if you did the first one, you would avoid PLL parity after doing OLL.

So now my question is, is there a way to know if you have PLL parity during OLL Parity so that you can do the "right" OLL parity alg to be able to always avoid PLL parity?
You have it backwards. The first alg (that everyone knows Rw U2 x Rw U2...) doesn't affect PLL parity whatsoever. If you had PLL parity before doing the alg, you'll still have PLL parity after the alg. The alg does what is essentially an F perm to the cube while rotating two corners and flipping an edge.

The second algorithm you listed (Rw' U2 Rw' U2...) does affect PLL parity. If you had PLL parity before doing the alg, you won't after doing the alg. This alg does a single corner swap--an impossible PLL (as well as rotating two corners and flipping an edge).

Also, pretty much the only way to do what you're asking about is to either orient all the pieces in your head and think about if it's a valid PLL, or... know full ZBLL and recognize if you have a possible case (excluding flipped edges, i.e. imagining if all edges were oriented). Too much recognition time for that to ever be worth it for someone who doesn't know full ZBLL.

#### xyzzy

##### Member
All I remember about it is that it has this move sequence in it: (Rw' F R F' r). Does anyone know this algorithm? I miss it. I believe it's supposed to pair opposite edges on the top layer.

Rw' F R' F' R U' R U 2R is the closest-looking one from that thread, but you could modify that into Rw' F R F' R U' R' U 2R with essentially the same effect.

#### Timona

##### Member
You have it backwards. The first alg (that everyone knows Rw U2 x Rw U2...) doesn't affect PLL parity whatsoever. If you had PLL parity before doing the alg, you'll still have PLL parity after the alg. The alg does what is essentially an F perm to the cube while rotating two corners and flipping an edge.

The second algorithm you listed (Rw' U2 Rw' U2...) does affect PLL parity. If you had PLL parity before doing the alg, you won't after doing the alg. This alg does a single corner swap--an impossible PLL (as well as rotating two corners and flipping an edge).
Surely I am not tripping.

Take your 4x4, do PLL Parity, you should have the 2 dedges swapped at UF and UB.

Alg 1 - Rw U2 x Rw U2 Rw U2 Rw' U2 Lw U2 Rw' U2 Rw U2 Rw' U2 Rw'
Do the first alg and look at the permutation of the pieces, ignore the orientation. What you should have now is a conjugated T-Perm with an additional swap at UR and UL.

Alg 2 - Rw' U2 Rw' U2 Lw U2 Rw' U2 Rw U2 x U2 Rw2 U2 Rw' U2 Rw U2
Same steps, do PLL Parity then do alg 2. Take a look at the permutation. What you have now is a regular T-Perm, a non parity PLL.

In conlusion, if you have a PLL parity (based on impossible permutations) and OLL parity, and you perform the first parity alg, you wil end up with no PLL parity afterwards. If you do the second alg when you have PLL parity plus OLL parity, you will still have PLL parity after the alg.

Final notes, if piece orientation counts towards permutation on 4x4, this was all for nothing and i'm just dumb, I just typed all this on my phone so i'm malding.

#### beniolenio

##### Member

Rw' F R' F' R U' R U 2R is the closest-looking one from that thread, but you could modify that into Rw' F R F' R U' R' U 2R with essentially the same effect.
This is totally it (the modified version). Thank you so much!

#### beniolenio

##### Member
Surely I am not tripping.

Take your 4x4, do PLL Parity, you should have the 2 dedges swapped at UF and UB.

Alg 1 - Rw U2 x Rw U2 Rw U2 Rw' U2 Lw U2 Rw' U2 Rw U2 Rw' U2 Rw'
Do the first alg and look at the permutation of the pieces, ignore the orientation. What you should have now is a conjugated T-Perm with an additional swap at UR and UL.

Alg 2 - Rw' U2 Rw' U2 Lw U2 Rw' U2 Rw U2 x U2 Rw2 U2 Rw' U2 Rw U2
Same steps, do PLL Parity then do alg 2. Take a look at the permutation. What you have now is a regular T-Perm, a non parity PLL.

In conlusion, if you have a PLL parity (based on impossible permutations) and OLL parity, and you perform the first parity alg, you wil end up with no PLL parity afterwards. If you do the second alg when you have PLL parity plus OLL parity, you will still have PLL parity after the alg.

Final notes, if piece orientation counts towards permutation on 4x4, this was all for nothing and i'm just dumb, I just typed all this on my phone so i'm malding.
You must be tripping (lol). I don't know what else to say but you happen to be incorrect. This would be easier to see if you knew the alg that solves pure OLL parity.

r' U2 l F2 l' F2 r2 U2 r U2 r' U2 F2 r2 F2

Do the first OLL parity alg you listed. Now do the pure OLL parity alg. You will now see that the cube is solvable without doing a PLL parity alg.

Now do the second OLL parity alg you listed. And then do the pure OLL parity alg. You will now see that you need to do PLL parity in order to solve the cube.

#### Timona

##### Member
This would be easier to see if you knew the alg that solves pure OLL parity.
Thanks for the correction.

#### Timona

##### Member
Question for people who know and use ZBLL in their solves, how does PLL parity affect recognition and how do you know if you have PLL parity when recognising the ZBLL?

#### Thom S.

##### Member
Question for people who know and use ZBLL in their solves, how does PLL parity affect recognition and how do you know if you have PLL parity when recognising the ZBLL?
Not anymore, but I used to Know H and Pi.
My recognition was recognise the COLL from the same angle and look at UF and UR.
That does account for Parity not messing up recognition.

#### xyzzy

##### Member
Question for people who know and use ZBLL in their solves, how does PLL parity affect recognition and how do you know if you have PLL parity when recognising the ZBLL?
Stuff:
The BH recognition system has the problem that if the two adjacent edges you look at happen to be adjacent colours (relative to each other) and you have permutation parity, then you'll end up with the not-so-nice adjacent swap case. (Like this, where applying the pure corner twist will leave you with an adjacent swap. For this particular case it turns out that it's way better to do parity first then ZBLL, but in general you'd have to work out whether that's the case and that sounds like a lot of effort, plus you cannot use two-sided ZBLL recognition.)

Tran-style ZBLL recognition but using opposite colours instead of adjacent colours (i.e. you look for, say, the red and orange edges, as in hyperorientation) will force the opposite swap case if you do have parity, which is nice. And if you don't have parity then it's just ZBLL as normal, which is even nicer.
(click through to see the rest of the thread and some context)

If your ZBLL recog method is by looking at two edges, then:
(i) If those edges have opposite colours, it's safe to do the normal ZBLL. Either you solve it, or you end up with opposite parity.
(ii) Otherwise, if they have adjacent colours, it'll end up either solved or with adjacent parity. You can either take the 50-50 or check a third edge and decide from there.

Assuming you're good enough with ZBLL recog, you can sometimes notice "something wrong" about a parity ZBLL case and detect PLL parity that way too. Like if a case that's not supposed to have a bar does have a bar, or vice versa, or an impossible number of bars. That kind of stuff.

There are also a few cases where it might make sense to combine PLL parity with the ZBLL itself, like R U R' U R' r2 U2 r2 R u2 2R2 u2.

#### OreKehStrah

##### Member
Question for people who know and use ZBLL in their solves, how does PLL parity affect recognition and how do you know if you have PLL parity when recognising the ZBLL?
I rarely do even layer big cubes, but would just recog ZB like normal, do the alg, and deal with adj ep swap parity if it happens

#### Timona

##### Member
I rarely do even layer big cubes, but would just recog ZB like normal, do the alg, and deal with adj ep swap parity if it happens
You're comfortable with getting adj parity? What alg do you use?

#### GenTheSnail

##### Member
For 6x6 I'll generally just do COLL, parity if necessary, and then EPLL.
Sometimes if I see enough blocks or just happen to know what the entire zbll should look like I'll do that, but I generally try to avoid doing it cause it's more risky to do a longer zbll and still end up with bad parity.

You're comfortable with getting adj parity? What alg do you use?
For adj parity, I use R' U' R U (r2 U2 r2 Uw2 r2 Uw2) U R' U R

#### Bh13

##### Member
My apologies if this was asked earlier in this thread, but is there an algorithm for OLL parity that you can do during F2L? I have found that I can pretty consistently determine if I will have OLL parity during my third or fourth F2L pair, and I feel like there should be an alg for OLL parity that you can do during this step that is easier/fewer moves than the standard OLL parity algs, since you don't have to worry about preserving one (or two) F2L pairs. I'm not sure it would provide any benefit, but I would at least like to try it out and see how I like it if there is such an alg readily available.

LBr

#### xyzzy

##### Member
My apologies if this was asked earlier in this thread, but is there an algorithm for OLL parity that you can do during F2L? I have found that I can pretty consistently determine if I will have OLL parity during my third or fourth F2L pair, and I feel like there should be an alg for OLL parity that you can do during this step that is easier/fewer moves than the standard OLL parity algs, since you don't have to worry about preserving one (or two) F2L pairs. I'm not sure it would provide any benefit, but I would at least like to try it out and see how I like it if there is such an alg readily available.
I used to use these (from the wiki):
2R U' R U2 R' U 2R U2 2R U2 2R U R U2 R' U' 2R
or
3R U' R U2 R' U 3R U2 3R U2 3R U R U2 R' U' 3R
Essentially the same alg, just that you can choose whether you prefer doing the slices on the right side or on the left side.

There's also r U R' U' r2 R U2 r2 U r U2 r' U2 r U' r2 U2 r' (also from the wiki). Not that useful on its own, but can be used as part of a special-case OLL parity alg.

These aren't much better (arguably, not even better at all) than doing OLL parity normally. While there are shorter algs when you don't need to preserve all F2L pairs, it also seems like the algs we know of (i.e. the ones documented on the wiki) are rather unergonomic.

Other possibly useful stuff:
r U2 F2 r U' F R' F r2 U2 r' F2 r2 F' (almost looks good but good luck executing this without doing at least 3 regrips)

Last edited: