[Help Thread] Roux Discussion and Help

[dominugget]

How to recognize L cases for CMLL?


Sent from my iPhone using Tapatalk

 

[muchacho]

I do it like this: (but I think other people do it from other orientation)


S = Same colors
O = Opposite colors

 

[TDM]

So I've started doing the daily CMLLs on reddit again. I should update the pastebin every day if you don't use reddit.

How to recognize L cases for CMLL?

On the daily CMLL list, I include my recognition method (with diagrams). Each case is on a different page unfortunately, so to make it easier I'd suggest learning recognition in pairs: L5 and L4, L1 and L6, L2 and L3. These are separated by the pattern of colours (opposite, adjacent, same) on the U layer.

 

[Teoidus]

I keep hearing that LSE should be one continuous step, where you look ahead to the next step while performing the previous, but I'm having some trouble wrapping my head around it. For example, I try and track UL/UR while orienting edges, but everything is moving around so quickly that it's hard to see/visualize where they'll end up--same for BU prediction. How do you guys approach this?

 

[GuRoux]

I keep hearing that LSE should be one continuous step, where you look ahead to the next step while performing the previous, but I'm having some trouble wrapping my head around it. For example, I try and track UL/UR while orienting edges, but everything is moving around so quickly that it's hard to see/visualize where they'll end up--same for BU prediction. How do you guys approach this?

first be able to do all you sub steps blindfolded. that way, while you solve eo, you don't have to think about solving eo, you can focus everything on UL/UR while muscle memory solves eo. then, it really just takes a lot of practice, getting yourself familiar with cases. try doing slow turning lse and see how much you can lookahead, and increase the speed when you think you've gotten better, etc.

 

[Teoidus]

Do you actually see the ULUR pieces coming up as you do EO, or are you just familiar enough with EO cases that you know where they'll end up?

 

[GuRoux]

Do you actually see the ULUR pieces coming up as you do EO, or are you just familiar enough with EO cases that you know where they'll end up?

usually when you start eo, you won't see both UL and UR. as you solve eo, you should be able to see both UL and UR. when you see them, you track them, by the time you finish eo, you know where UL and UR are. so it usually is not predicting but good lookahead.

 

[sub20oneday]

How do you recognize the M slice edge cycles quickly/while performing L/R reduction? They all seem to look similar to me. What I'm doing right now is looking at the U/D faces and if there's one square of opposite color on U, put it in it's proper position on the D layer (e.g. by performing M' U2 M U2), but that only covers some of the cases.

i dont just learn heise and petrus somehow merge them and use 2 gen last layer you will be better thank feliks in justa few days

 

[Yetiowin]

I use CFOP, but I like trying the other methods for fun. Are there any tips for doing the first block? I usually do a 2x2x1 block and then do a f2l pair.

 

[shadowslice e]

I use CFOP, but I like trying the other methods for fun. Are there any tips for doing the first block? I usually do a 2x2x1 block and then do a f2l pair.

Yeah, that's pretty standard.
Try watching pdf and guroux`s videos in the how to/guides section.

In essence though, for first block you try to do anything you possibly can to build a block and don't be afraid to try new things if you're not practising for speed (ie just messing around with a cube).

Sent from my M1005D using Tapatalk

 

[shadowslice e]

How about for CMLL we learn algs which separate UL and UR into U and D layers? I seem to get better solves with nicer/ easier LSE when that happens. Even before EO because it's easier to pair it up or even put it in place. Probably need 2 or 3 algs per standard case. I also reckon it is better than just using the OLLCP and forcing the EO skip because you don't have to do any extras set up moves and not much recognition.

 

[Dan Cochrane]

That's seems to me to be two separate alg sets or one very large set... Don't think that has any benefit over ul and ur in D + OLLCP

Sent from my ONE E1003 using Tapatalk

 

[shadowslice e]

That's seems to me to be two separate alg sets or one very large set... Don't think that has any benefit over ul and ur in D + OLLCP

Sent from my ONE E1003 using Tapatalk

You don't have to do any setup moves. Just standard SB and the movecount will be similar to the OLLCP I reckon. Plus the lookahead and skip forcing is just as nice.

 

[PenguinsDontFly]

I have now decided that standard FB SB CMLL LSE is the best. Even if I'm gonna get a 6 flip, I'll usually do my standard alg and just cancel into the 6 flip (Unless the flipping alg is just as fast). I basically stopped learning ollcp because there are so many other parts of my solve that can still be improved (most importantly: frickin CMLL recognition! Only pause in my solve). In the time it takes to recognize CMLL and pick an alg to skip EO or separate LR edges, I can probably recognize CMLL and be part way into the algorithm. Also, ollcp and sse's suggested algs are definitely slower than standard CMLLs so slower recog and slower exec for something that can be done in a few moves in the next step is pointless. LSE TPS is also much higher than CMLL TPS, so I would rather get CMLL over with faster so I can take advantage of the sexy 2 gen moo.

 

[shadowslice e]

I have now decided that standard FB SB CMLL LSE is the best. Even if I'm gonna get a 6 flip, I'll usually do my standard alg and just cancel into the 6 flip (Unless the flipping alg is just as fast). I basically stopped learning ollcp because there are so many other parts of my solve that can still be improved (most importantly: frickin CMLL recognition! Only pause in my solve). In the time it takes to recognize CMLL and pick an alg to skip EO or separate LR edges, I can probably recognize CMLL and be part way into the algorithm. Also, ollcp and sse's suggested algs are definitely slower than standard CMLLs so slower recog and slower exec for something that can be done in a few moves in the next step is pointless. LSE TPS is also much higher than CMLL TPS, so I would rather get CMLL over with faster so I can take advantage of the sexy 2 gen moo.

I agree that standard is better than OLLCP though the CMLL split is not much slower (maybe .1 to .2) and the LSE lookahead is faster and lookahead is nicer.

 

[sqAree]

So how you would you usually solve this. Setup: M' U M' U2 M2 U2.

The EO case is pretty easy, I'd go for M U M' to solve EO and UF/UB (instead of UL/UR), then just finish with U M2 U2 M2 U'.

The length of this solution is 8 moves, while the optimal solution has 6 moves at max.

The standard EO+ULUR+M approach seems to fail in this simple case. Would an experienced Roux solver be able to solve the LSE optimally in this case? And how am I supposed to approach a case like that?

Another question, how many moves does the worst LSE case take to solve?

 

[TDM]

So how you would you usually solve this. Setup: M' U M' U2 M2 U2.

The EO case is pretty easy, I'd go for M U M' to solve EO and UF/UB (instead of UL/UR), then just finish with U M2 U2 M2 U'.

The length of this solution is 8 moves, while the optimal solution has 6 moves at max.

The standard EO+ULUR+M approach seems to fail in this simple case. Would an experienced Roux solver be able to solve the LSE optimally in this case? And how am I supposed to approach a case like that?

Another question, how many moves does the worst LSE case take to solve?

Misoriented centres gets 7 if you see the cancellation between EO and ULUR/EP:

M' // EO
M' U2 M2 U2 M' // ULUR+EP
U' M // fix centres

or alternatively a normal misoriented centres approach of M' U2 M2 U E2 M E2 also gives 7 moves. I don't know how you would see the 6 move solution in a speedsolve though...

 

[sqAree]

Misoriented centres gets 7 if you see the cancellation between EO and ULUR/EP:

M' // EO
M' U2 M2 U2 M' // ULUR+EP
U' M // fix centres

or alternatively a normal misoriented centres approach of M' U2 M2 U E2 M E2 also gives 7 moves. I don't know how you would see the 6 move solution in a speedsolve though...

Thank you. But what exactly is the misoriented center approach then? Didn't find something like that in my Roux tutorial so far. It's a bit hard to follow what's going on and I couldn't find solutions like that for different cases.

 

[TDM]

Thank you. But what exactly is the misoriented center approach then? Didn't find something like that in my Roux tutorial so far. It's a bit hard to follow what's going on and I couldn't find solutions like that for different cases.

Misoriented centres is when you have the F/B centres on U/D. When you "orient" edges, you orient ULUR normally, but the other four edges need to be flipped. For example,

Setup: M' U M U2 M' U M
EO: M'

After this, you can continue normally:
U' M U2 M' U' // EO
M // EP skip

 

[skidmarkstan]

Does anyone feel like talking about colour neutrality ?

Right now I'm doing x2 y2, averaging about 29 seconds, but I want to decide if I should go for full y.