[Unofficial] 15.827 Ao12 using ZZ-CT: The LL-Skip Method.

[ottozing]

Happy to help! Keep pushing all the limits :tu

 

[AlphaSheep]

What are some good intermediate steps to use when you don't know all of the cases.

For TSLE, I can do an easy setup to WV in some cases or cancel into some form of sune quite often. When I speedsolve though, I tend to just insert the edge and any corner then do normal OCLL.

TTLL is a little harder. Whats a sensible 2 look TTLL thats not just triple sledge followed by PLL?

 

[ottozing]

You could try doing insert corner into F2L+CP into an EPLL for TTLL, not sure about TSLE

 

[4Chan]

What are some good intermediate steps to use when you don't know all of the cases.

For TSLE, I can do an easy setup to WV in some cases or cancel into some form of sune quite often. When I speedsolve though, I tend to just insert the edge and any corner then do normal OCLL.

TTLL is a little harder. Whats a sensible 2 look TTLL thats not just triple sledge followed by PLL?

For TSLE, there's two ways that I can see.

1. By intuitively knowing which corners are oriented by R U R', you can just intuitively orient down to 1-2 mioriented, so that removes approx half the cases from 3-4 misoriented. This is what I did.

2. By just ignoring everything and just using R U R' to insert the edge, and there are only like, 20(?) cases for edge already inserted, to orient everything.



For TTLL, if I got a case I didn't know, I'd use either an A perm or Y perm to reach the subset where every corner except UBL (the weird corner) was permuted correctly relative to one another. The corners then looked like they would in a U perm, H perm, or Z perm, since they are "pseudo-solved"

I did that approach because that subset really stands out because of the matching corner colours.

 

[Petro Leum]

Damn 4chan, back at it again with the evolution of cubing! This looks real nice.
I didn't look at it in detail, but i will later. How do you think this compares to ZZ with phasing + ZZLL, which i currently use?
also, reading the title, at first i thought you were going to solve LS and LL simultaneously 100% of the time (wait, i'm stupid, technically you do! it's just in 2 LS steps.
0.051% in ZZ(1 out of 1944 solves) - does this mean that with phasing i have a chance of skipping LL in 1 out of 648 solves? that's pretty neat and would fit my experience so far quite well.

 

[gyroninja]

What are some good intermediate steps to use when you don't know all of the cases.

For TSLE, I can do an easy setup to WV in some cases or cancel into some form of sune quite often. When I speedsolve though, I tend to just insert the edge and any corner then do normal OCLL.

TTLL is a little harder. Whats a sensible 2 look TTLL thats not just triple sledge followed by PLL?

Right now for tsle I've learned the 27 cases with the f2l piece in UF where you have 2 or 3 corners oriented (3 or 4 for one subset).

To setup to a tsle I know I just spam sexy move. There a few cycles where I had to learn a specific case for.

 

[FJT97]

The only big problem i see is that you have to limit yourself in f2l cause the fr slot has to be the last one... Thats kinda bad for efficiency, i guess.

 

[AlphaSheep]

The only big problem i see is that you have to limit yourself in f2l cause the fr slot has to be the last one... Thats kinda bad for efficiency, i guess.

I don't think it would be too difficult to mirror TSLE for BR, and after that it's just a D' setup for TTLL.

 

[4Chan]

Damn 4chan, back at it again with the evolution of cubing! This looks real nice.
I didn't look at it in detail, but i will later. How do you think this compares to ZZ with phasing + ZZLL, which i currently use?
also, reading the title, at first i thought you were going to solve LS and LL simultaneously 100% of the time (wait, i'm stupid, technically you do! it's just in 2 LS steps.
0.051% in ZZ(1 out of 1944 solves) - does this mean that with phasing i have a chance of skipping LL in 1 out of 648 solves? that's pretty neat and would fit my experience so far quite well.

Wow, thanks!!!

Mmmm, I think I'm not experienced enough to say. Phasing is a half step, so maybe a little longer, but I'm not 100% sure.

I think only time will tell!
I think it's down to personal preference at this point.
(Btw, I think you're cool for being the first sub10 official zz solver, that's my dream!)

I don't think it would be too difficult to mirror TSLE for BR, and after that it's just a D' setup for TTLL.

Omfg, the D layer adjustment works for every face, it can even cancel moves when the last move is D' of the alg!! It's so obvious, this is remarkable!!! This is really really cool, aND I had no idea previously!!WOW

 

[Petro Leum]

(Btw, I think you're cool for being the first sub10 official zz solver, that's my dream!)

It makes me really happy that some people actually noticed, not just my german friends whom i told. Feels like i really accomplished something. Of course, that's only the beginning for ZZ, but my focus is on OH for now.

Btw, i think you're cool for being the guy who keeps pushing forward the community in terms of alg sets and methods, and ZZ especially. You've inspired me multiple times to learn something new or keep going with something weird, and i'd say you're quite literally the reason why i pulled through with ZZLL

On topic: Is my assumption right? If the chance for LL skip with basic ZZ is 1/1944, it should be a third of that with phasing, right? That's actually pretty close to ZZ-CT dimensions.

Also, like Fin, i'm worried about limiting F2L blockbuilding to leave the FR slot unsolved, or even only FR/BR for that matter. That's why my gut feeling tells me that ZBLL-A will be "better" than ZZ-CT if you really invest the time to learn all of it - It's more intuitive and you have more possibilities to save moves in F2L. I know it's easy to keep ZZLL algs drilled and ZBLL is only triple the mass.

I really want this to be my next stepping stone on my way to ZZ-A though.... Where are you going to publish your algorithms? Do you have a homepage? I'd like to know if people are going to be able to submit alternative algs for OH and stuff, so we can really perfect this method.

 

[4Chan]

Thanks so much! I really really appreciate it!!! I'm glad that I could have an impact!

I'm not great on probabilities, but that number sounds very plausible to me.

Yes, I want to agree. At the very highest competitive levels, I think ZZ-A is superior.

My biggest issue with ZBLL was the time requirement.
People keep posting new algorithms, and I'm constantly refreshing my lists and I'm too old and too busy now to commit to it.
I wanted something a little bit lighter for me, and I think this is a nice compromise.


Next week, once I figure out better algs for about 14 cases, I will give it to Gyroninja, and he will post them on his personal site. But I think I can also post something on a google docs too!

 

[irontwig]

Oh, really? I though there would be 1/6 of the cases 2-gen in one orientation and 1/6 in the other.

Yeah... not sure what I was thinking, FRU (or RUD) can obviously permute corners. I also realize now that this is the domino redux thingy I mentioned in the ZZ-HW thread. I knew I had seen Chris describe this. I guess TSLE can be described like Roux EO; recursively?

 

[4Chan]

Just to clarify, although you did post about this, I didn't take your idea, which was posted on April 20th.

I was inspired by elrog on March 31st.

 

[SpeedCuber71]

Very Good! MGLS would be something similar to use for CFOP solvers who find this difficult (yeah it only skips OLL< but it is kinda similar)

 

[pinser]

I definitely see potential with this. It's much more manageable and only a tiny bit slower than ZZ-A, comprable to ZZLL, and a bit faster than COLL EPLL.
Keep working on this and I'll keep playing around with it!

 

[irontwig]

Just to clarify, although you did post about this, I didn't take your idea, which was posted on April 20th.

I was inspired by elrog on March 31st.

Just to clarify even further, this is what happened on my end:
1. I saw a blog/forum post or video from you describing this method
2. Got the idea in my post from messing around with domino redux, being inspired by Baldav and Basti
3. I saw your thread, didn't read it too thorough and posted.

 

[Isaac VM]

A couple of months ago I tried ZZ for the first time, I loved the fluidity during F2L (easy blocks), the higher chances to get oll and pll skip (using coll) and now this variant with higher LL skip it's just great. I am working so it can be my main method, I think this ZZ variant will be amazing. Keep it up!

 

[gyroninja]

I'm kind of stupid. I just realized that you can take the same 2gen algs I know for TTLL and conjugate them with a D (like [D: ttll]) to solve some more cases without learning new algs (still need to learn reco though). You can probably do the same with the others. It probably doesn't work as nicely as it does with 2gen though but I would cut the alg count in half? (39? non pll)

Would work with a mirrored version of the ttll (over S), but gl with mirroring them in your head.

 

[AlphaSheep]

TSLE inserts the last edge whilst orienting all corners. All states can be solved by a linear combination of at most three R U R', R U' R', or R U2 R' triggers. You just remember which trigger comes after which for all cases. ez pz

TSLE sounds like it should be pronounced "T-silly".

Anyway, recognition seems tricky, so I've been trying to come up with a reduction system like the way the EO step in Roux works, as @irontwig suggested.

Unless I've made a mistake, there are only 104 TSLE cases? There are 81 cases with the last F2L edge in the top layer, and I could only find 23 with the last F2L edge in place.

Spoiler

With one unique edge in U, there is no rotational symmetry, so all combinations of corner twist for the four corners in U are possible. The number of cases is 3^4 = 81.

When the F2L edge is solved, rotational symmetry comes into play, so the number of cases is reduced to 24, but one is the skip case. The cases are:

0. 0 0 0 0    (0.61 %)    Skip
1. 0 0 0 1    (2.42 %)    One CW
2. 0 0 0 2    (2.42 %)    One CCW
3. 0 0 1 1    (2.42 %)    Two adjacent CW
4. 0 0 1 2    (2.42 %)    T-OCLL
5. 0 0 2 1    (2.42 %)    U-OCLL
6. 0 0 2 2    (2.42 %)    Two adjacent CCW
7. 0 1 0 1    (1.82 %)    Two diagonal CW
8. 0 1 0 2    (2.42 %)    L-OCLL
9. 0 1 1 1    (2.42 %)    AntiSune OCLL
10. 0 1 1 2    (2.42 %)    CW-CW-CCW
11. 0 1 2 1    (2.42 %)    CW-CCW-CW
12. 0 1 2 2    (2.42 %)    CW-CCW-CCW
13. 0 2 0 2    (1.82 %)    Two diagonal CCW
14. 0 2 1 1    (2.42 %)    CCW-CW-CW
15. 0 2 1 2    (2.42 %)    CCW-CW-CCW
16. 0 2 2 1    (2.42 %)    CCW-CCW-CW
17. 0 2 2 2    (2.42 %)    Sune OCLL
18. 1 1 1 1    (0.61 %)    All CW
19. 1 1 1 2    (2.42 %)    3 CW, 1 CCW
20. 1 1 2 2    (2.42 %)    Pi-OCLL
21. 1 2 1 2    (1.82 %)    H-OCLL
22. 1 2 2 2    (2.42 %)    3 CCW 1 CW
23. 2 2 2 2    (0.61 %)    All CCW

I've managed to find combinations of three or fewer triggers (from [R U R'], [R U2 R'] and [R U' R']) for all but these four cases.

R' U2 R2 U R2 U R U2 R U R
R' U2 R' U R2 U' R U2 R2 U R'
R2 U2 R' U' R U' R2 U2 R U R'
R U R2 U' R U' R' U2 R2 U2 R'

It seems these cases, and only these cases need at least four? Although all four cases have 11HTM optimal solutions in <R,U>, so it may be worth it to just learn optimal algs for them.

EDIT: for anyone who wants TSLE algs, here are my raw unsorted algs for each case. You can probably copy and paste into Excel to get a more readable format.

 

[gyroninja]

R' U2 R2 U R2 U R U2 R U R
R' U2 R' U R2 U' R U2 R2 U R'
R2 U2 R' U' R U' R2 U2 R U R'
R U R2 U' R U' R' U2 R2 U2 R'

For those cases I use

U R U' R' U R U' R' U2 R U R'
U2 R U R' (Bet this makes you feel like an idiot)
R U R' U2 R U R' U R U' R'
R U2 R' U R U R' U R U2 R' U' R U' R'

Case wise I only count 100. You have the 27*3 (Think WV cases but one of the corners is twisted to use all three possible co on it) That's 81. I have 19 algs for when the f2l edge is in place.

Edit: Looks like I missed four cases.