• Welcome to the Speedsolving.com, home of the web's largest puzzle community!
    You are currently viewing our forum as a guest which gives you limited access to join discussions and access our other features.

    Registration is fast, simple and absolutely free so please, join our community of 40,000+ people from around the world today!

    If you are already a member, simply login to hide this message and begin participating in the community!

[Help Thread] CPLS and 2GLL discussion

mDiPalma

Member
Joined
Jul 12, 2011
Messages
1,534
I've recently been playing around with CPLS, and I find it very promising, especially from a Petrus perspective.

I've regenerated (what I consider to be) better algs for the cases with the corner in the U-layer:

Code:
2swap in back (these are just <RU> solutions):
CW: (U) R U R' U2 R U R' (7)
ACW: (U') R U' R' U2 R U' R' (7)
O: R2 U R2 U R2 U2 R2 (7)

2swap on left (these are just y' then <RU> solutions):
CW: (y' U) R' U R U2 R' U R (7)
ACW: (y' U') R' U' R U2 R' U' R (7)
O: (y') R2 U' R2 U' R2 U2 R2 (7)

UFL -> URF:
CW: [L D' L' : U] (7)
ACW: (y U') [L D L' : U] (7)
O: (y U2) R U' L2 U R' U' L2 (7)

UBR -> URF:
CW: (U) [R' D' R : U'] (7)
ACW: (y) [R' D R : U'] (7)
O: (U2) L' U R2 U' L U R2 (7)

ULB -> URF:
CW: (U2) [R' D' R : U2] (7)
ACW: (y U2) [L D L' : U2] (7)
O: (U2) u R2 u' F u R2 u' F' (8)
    or (R' F R F')*3 (12)
    or (U2) R' D' U2 R U R' U D R (9)

no motion:
CW: (U) R U R2 D' r U2 r' D R (9)
ACW: (U') R' D' r U2 r' D R2 U' R' (9)
O: R U R' U2 R U2 L U' R' U L' (11)

*note all the ACW single motion cases begin from a y rotation

In methods that solve EO before the LS, sliding in the last F2L edge only takes 2 htm on average. This number can drop if you pull in the F2L edge with a U-layer corner earlier in the solve.

Using the ergonomic algs posted above, as well as Baian's original algs for the cases with the corner in place (but r U2 R2' F R F' R U2 r' for the pure diagonal swap), gives you 8.22 htm for CPLS.

And I recently found that optimal <RU> 2GLLs are 13.15 htm.

Then, assuming 25 htm for EOF2L-1 (typical for ZZ, conservative for Petrus) and accounting for AUFs as executed, this entire method only requires ~50 htm. And if you use ZZ-Orbit for cases in which the F2L pair is already made, this number can also drop slightly. That's pretty good for a method with an ergonomic <RU> finish. And the efficiency actually beats out many LS/LL methods of a comparable alg count.


I also put together this CPLS Trainer to practice the CLL-esque recognition. You might need to zoom out slightly depending on your screen/resolution.
 
Last edited:
Joined
Aug 5, 2016
Messages
254
I really want to learn this but the recognition makes no sense i get the example that stacchu did, but when I try any other case it doesn't make sense. could you show how to recognize cpls with this setup: t perm triple sexy?
 

Shiv3r

Member
Joined
Mar 16, 2016
Messages
800
Location
San Diego or thereabouts
WCA
2016LEWI04
YouTube
Visit Channel
I really want to learn this but the recognition makes no sense i get the example that stacchu did, but when I try any other case it doesn't make sense. could you show how to recognize cpls with this setup: t perm triple sexy?
recog for CPLS is one of the reasons no one uses it.
If I were using it in ZZ (probably OH), I would suggest just learning the algs for ZZ-D(ZZ-Porky2, theres a doc somewhere, only 6 cases to recognize and the longest alg is like 3 moves long)
 

mDiPalma

Member
Joined
Jul 12, 2011
Messages
1,534
I really want to learn this but the recognition makes no sense i get the example that stacchu did, but when I try any other case it doesn't make sense. could you show how to recognize cpls with this setup: t perm triple sexy?

I think the recognition schemes proposed by Stachu and Cyragia are not ideal for speedsolving. Instead, I recommend either a CLL-type recognition (something like 160 cases) or intuitively identifying opposite/adjacent/not pairs of corners in the U-layer (the people I've taught to solve the cube use a Petrus-style LL and even they can figure this out in just a few moments).

For the latter, on your example, both the UFL and ULB corners have green stickers, but they wouldn't be on the same side if the corners were oriented. Therefore, they will have to swap places. The corresponding CPLS alg for that is [y'] R2 U' R2 U' R2 U2 R2.
 

Y2k1

Member
Joined
Jan 18, 2016
Messages
134
WCA
2017AHME03
sorry for not understanding this claearly, but is cpls inserting the edge and then inserting the f2l corner and cp, or is it doing cp and then doing the ls like normal?
 

CubingGenius

Member
Joined
Jul 3, 2016
Messages
238
Do you know what the average movecount would be if you solve the last pair and CP simultaneously? 120 unique cases if I'm right?
 

Arc

Member
Joined
Jan 27, 2017
Messages
63
Location
Washington
So reading this thread many times I see "it sounds really cool but the recognition seems really hard." I learned Teoidus' 2GR recently and realized that a similar (virtually identical) recognition scheme can be applied to CPLS. I don't take any credit for it (apparently it originates with Porkynator, and I wouldn't be surprised if people already did something similar intuitively), but I thought that it would be useful to have a good structured recognition approach posted on this thread.

As a prerequisite you have to know your LL corners by identity. I use numbers for their names for simplicity. I personally solve on blue/green (with black instead of yellow) and for me, the orange-black corner is 1, the black-red corner is 2, the red-white corner is 3, and the white-orange corner is 4. You can label your corners however you want but you need to be able to recognize them by name and know their relative positions.

0882b638db.png


Now, once you have set up to CPLS, follow the corners clockwise starting clockwise of the D layer corner (or starting in UFL if the D corner is in D). For example you could have 2 3 1. Now look at the relationships, as shown by the fancy diagram that I totally didn't make in paint. 3 is forward from 2, 1 is diagonally across from 3. Those two relationships (forward, jump) together correspond directly to which of the 6 CP states you have.

1255911174.png


This becomes very fast in an actual solve, once you practice it changes from identifying the corners and then the relationships to doing it first relationship then second relationship to simply looking at the three corners and immediately seeing their relationships.

RelationshipsSwap
Forward, ForwardNo Swap
Forward, JumpSwap on Right
Backward, BackwardDiagonal Swap
Backward, JumpSwap on Left
Jump, ForwardSwap in Front
Jump, BackwardSwap in Back
As for color neutrality, x2 y neutrality is trivially easy and generally what most people use. If you want to use different colors on top/bottom you'd have to learn new names for the corners. Since the only thing that matters is the relative positions of the corners, it doesn't matter what front you use, as long as the top is the same, you do not change a single thing about how you're recognizing. The corner names and relationships stay the same. When you flip the cube and use the opposite color for the up face, there's one minute difference. Instead of starting clockwise and reading clockwise, start anticlockwise of the D corner (or in UBR if it's in the D layer) and read the corners anticlockwise. The corner names and relationships remain the same.
P.S. Sorry for bump.
 

PapaSmurf

Member
Joined
Jan 4, 2018
Messages
1,103
WCA
2016TUDO02
YouTube
Visit Channel
I'll bump the thread. This is an idea I had over a year ago now, again inspired by 2GR recog, and I think that it has potential. It is based off a post somewhere on the forums, but made better. I'm not certain how brilliant or bad it is, but it's a thing that could work. Average movecount is 7+7+4+7+13+14.5=52.5, as the steps are eoline, block, cp, FL, right block, 2gll. The way it works is that you use the method described above to find the relationship between the corners for any position of the DFL corner. You then do an alg which solves cp and removes any pieces on L from R. LU to solve left pair, then RU for the rest of the solve. This is unfortunately something that I haven't worked on for a while nor do I plan to work on, as I believe that ZZ-A is far superior in terms of ergonomics, recog and movecount, but if anyone wants to fully develop it, that'd be great. Here's the incomplete doc: https://docs.google.com/document/d/1iJn6wzG9jYefRqOp0C3REtJCOAns1aAwYITV8UV0V5A/edit?usp=sharing
 

IsThatA4x4

Member
Joined
Jul 18, 2021
Messages
915
Location
UK
WCA
2022RITC01
Is it worth it to learn full or near full 2GLL to take advantage of it (like you would do with COLL) in a CFOP solve? The algs are generally short and fast and the only problem I can see is CP recognition. What do you think?
 

OreKehStrah

Member
Joined
May 24, 2019
Messages
1,435
YouTube
Visit Channel
Is it worth it to learn full or near full 2GLL to take advantage of it (like you would do with COLL) in a CFOP solve? The algs are generally short and fast and the only problem I can see is CP recognition. What do you think?
Yes! 2GLL would be a great subset of ZBLL to learn, especially for OH. Over time you'll get used to recognizing when you have solved CP, so that shouldn't be an issue either.
 
Top