[Help Thread] Roux Discussion and Help

[shadowslice e]

Tips for learning CMLL.


Sent from my iPhone using Tapatalk

1) Find a good recog method
2) Learn in a specific order (I did it as Pi, H, T, U, L, Sune/antisune - you should already know the preoriented cases)
3) Do slow solves with them without timing yourself so you have time for recognition.
4) Get good algs (or at least ones that work for you) and get them from the beginning as it can be a pain to switch later.
5) Set yourself goals to help motivate you to learn the algs (for example, by this time next week I will know the pi and h-cases)- not so much a specific cmll tip but rather a good life rule (at least for me)
6) Patience- fast recog may take some time to come but it will come eventually so don't give up.
7) Practise, practise, practise!
8) And never forget, it will pay off in the end!

Hope this helps

 

[TDM]

Tips for learning CMLL.

Do a set at a time, one alg a day. If you think one a day is too easy then try two. Have a bit of a pause after each set, maybe a day or two, where you do lots of solves and try to practise recognition. Getting to be able to recognise and recall the alg quickly in a solve is the hardest part imo, and takes a bit of practice. You should be doing at least some solves every day to be practising new cases.

 

[GuRoux]

Do a set at a time, one alg a day. If you think one a day is too easy then try two. Have a bit of a pause after each set, maybe a day or two, where you do lots of solves and try to practise recognition. Getting to be able to recognise and recall the alg quickly in a solve is the hardest part imo, and takes a bit of practice. You should be doing at least some solves every day to be practising new cases.

personally i did a set a day, when i was comfortable with that set, i would move on to the next one; though, i did procrastinate halfway through for like a month not learning anything.

 

[TDM]

personally i did a set a day, when i was comfortable with that set, i would move on to the next one; though, i did procrastinate halfway through for like a month not learning anything.

I can't learn algs that quickly at all. One a day was the limit for me.

 

[theROUXbiksCube]

Practise recognizing on 2x2 if you know CLL is what helped me get faster recog after a while

 

[Knut]

I just watched some of the seminar videos from US Nationals a couple days ago, and Kevin Hays said something about not solving the cube (timed) until you finish memorizing a set and get the recog. down pretty well. On that note I learned the U set two days ago, reviwed it and practiced recog. yesterday, did a bunch of solves today, and didn't mess up a single U case.
It may not work for you, but it seems to be working for me so far.
Now for Pi.

 

[theROUXbiksCube]

I just watched some of the seminar videos from US Nationals a couple days ago, and Kevin Hays said something about not solving the cube (timed) until you finish memorizing a set and get the recog. down pretty well. On that note I learned the U set two days ago, reviwed it and practiced recog. yesterday, did a bunch of solves today, and didn't mess up a single U case.
It may not work for you, but it seems to be working for me so far.
Now for Pi.

Yes! I did this! After I learned a set I drilled it until I was comfortable with it

 

[Renslay]

Tips for learning CMLL.

1) Find a good recog method

Here is my recog method.
It is similar to Waffle's system.

 

[shadowslice e]

LSE "new"(?) idea

Leave one misoriented edge on the bottom of the cube after F2B as this will reduce the case count to 2 and those 2 cases are the best (discounting a skip) and still quite good (as well as not having the worst few cases).

Reasoning: doing this requires only a little edge control and even if you don't control the edges during F2B it is quite frequently only a 1 or 2 move fix compared to the higher movecount for the that standard LSE would have.

The only problem that I found with it is that you it is one more thing to think about but I don't think it is that much of a problem.

This is probably not a new idea but I haven't come across any information regarding it so I was wondering what people thought about it.

 

[TDM]

LSE "new"(?) idea

Leave one misoriented edge on the bottom of the cube after F2B as this will reduce the case count to 2 and those 2 cases are the best (discounting a skip) and still quite good (as well as not having the worst few cases).

What about 1 on D and one on U? Or two on D and none on U? Those aren't "the best cases". There are far more than 2 cases.

(E: unless I've completely misunderstood what you're saying?)

 

[JediJupiter]

What about 1 on D and one on U? Or two on D and none on U? Those aren't "the best cases". There are far more than 2 cases.

(E: unless I've completely misunderstood what you're saying?)

1 on D and 1 on U, and 1 on D and 3 on U are the two best cases. It sounds hard to me, but if you can do that stuff it sounds pretty helpful.

 

[shadowslice e]

What about 1 on D and one on U? Or two on D and none on U? Those aren't "the best cases". There are far more than 2 cases.

(E: unless I've completely misunderstood what you're saying?)

No, keep only 1 misoriented edge on D as this will leave only 2 possible cases- 1 on U and 3 on U and these are always in the same configuration (discounting rotations) and those 2 cases are the nicest (excluding skips) for the orientation step in LSE.

Normally, yes there are more than 2 cases but leaving 1 misoriented piece pics on U will reduce the possible number to 2 (as above)

Edit: see the post above this one. However, I don't think it would be too hard as all the cases van be converted in 4 moves- including the AUFs (most are 0-1- do M2 or skip) doing (setup move) M U2 M' or for the 6 edges misoriented the standard M U M'. If you factor in edge control, the average move count for this step could be 1 or lower
.
Edit 2: you could also learn 2 algs for each CMLL case (one that does not affect orientation- basically COLL- and one that misorientates 4 edges in the U layer as this will enable you to force the best case orientation (the Arrow) every time with only 84 algs and the recognition is only very slightly harder than normal CMLL and definately no harder than learning the multiple algs that some have already started to or done.

Edit 3: while doing the correcting that I detailed in edit 1, you also get a little bit longer to recgnose exactly what CMLL case you will need which can go some way to eliminating pauses.

 

[TDM]

1 on D and 1 on U, and 1 on D and 3 on U are the two best cases. It sounds hard to me, but if you can do that stuff it sounds pretty helpful.

No, keep only 1 misoriented edge on D as this will leave only 2 possible cases- 1 on U and 3 on U and these are always in the same configuration (discounting rotations) and those 2 cases are the nicest (excluding skips) for the orientation step in LSE.

Normally, yes there are more than 2 cases but leaving 1 misoriented piece pics on U will reduce the possible number to 2 (as above)

Right, I assumed you meant have at least one misoriented edge, not exactly one. Having exactly one would be a bit harder to force though.

Is 1/1 really a very good case? I don't like them as much. They're 7 moves, same as most of the cases, except the 3/5 move cases and 6-flip.

Edit: see the post above this one. However, I don't think it would be too hard as all the cases van be converted in 4 moves- including the AUFs (most are 0-1- do M2 or skip) doing (setup move) M U2 M' or for the 6 edges misoriented the standard M U M'. If you factor in edge control, the average move count for this step could be 1 or lower

I think a lot of people use r' F R U M' U' R' F' R or another 6-flip alg, rather than doing a long MU solution. I still don't see what advantage this has over normal EO though.

Edit 2: you could also learn 2 algs for each CMLL case (one that does not affect orientation- basically COLL- and one that misorientates 4 edges in the U layer as this will enable you to force the best case orientation (the Arrow) every time with only 84 algs and the recognition is only very slightly harder than normal CMLL and definately no harder than learning the multiple algs that some have already started to or done.

But these algs will be worse than most CMLL algs in most cases. The moves saved in LSE will be so few that I don't think it's worth it.

Edit 3: while doing the correcting that I detailed in edit 1, you also get a little bit longer to recgnose exactly what CMLL case you will need which can go some way to eliminating pauses.

But like I said, the difference will be minimal, and since the method excluding pauses is slower than normal Roux I don't think it's worth it.

 

[shadowslice e]

Right, I assumed you meant have at least one misoriented edge, not exactly one. Having exactly one would be a bit harder to force though.

Is 1/1 really a very good case? I don't like them as much. They're 7 moves, same as most of the cases, except the 3/5 move cases and 6-flip.

No, not a MU total 6-flip. Just do the M U M' , the 4-flip CMLL then you have the arrow case.

I agree that some of the algs will be worse than standard CMLL (hence why they are not included in the set) but they would probably be only a couple of moves longer than standard CMLL, there would be less cases than if you tried to learn all the CMLL cases to force good LSEs (and have much better recognition) so while this would probably not be much use for the slower people, it may have a small advntage for those who are sub-10 (and when you are that fast every little thing counts).

One final little point that I came up with (although this is probably not very useful) is to use VHLS or ZBLS to insert the final pair to force an edge orientation skip then using pure COLL. The recogntion would not be too difficult and it may have a slighly smaller average move count than standard (but it is probably moving more into a different method rather than a variant of roux)

 

[JediJupiter]

No, not a MU total 6-flip. Just do the M U M' , the 4-flip CMLL then you have the arrow case.

I agree that some of the algs will be worse than standard CMLL (hence why they are not included in the set) but they would probably be only a couple of moves longer than standard CMLL, there would be less cases than if you tried to learn all the CMLL cases to force good LSEs (and have much better recognition) so while this would probably not be much use for the slower people, it may have a small advntage for those who are sub-10 (and when you are that fast every little thing counts).

One final little point that I came up with (although this is probably not very useful) is to use VHLS or ZBLS to insert the final pair to force an edge orientation skip then using pure COLL. The recogntion would not be too difficult and it may have a slighly smaller average move count than standard (but it is probably moving more into a different method rather than a variant of roux)

That is the MU 6-flip.

 

[shadowslice e]

That is the MU 6-flip.

No, not M U M' U2 M U M U M' U M
M U M' (CMLL 4-flip) M U M'

 

[dominugget]

What are the most common CMLL case in you opinion?


Sent from my iPhone using Tapatalk

 

[theROUXbiksCube]

Niklas and Anti Niklas for some reason

 

[TDM]

[...]

I still don't think you're saving any time with this.

What are the most common CMLL case in you opinion?

The majority of the cases have equal probability. There are only a few (2-gen H, diag H, diag O, skip) that are less likely.

 

[shadowslice e]

I still don't think you're saving any time with this.

Meh, I might try it out for a bit. See what happens. As long as it doesn't end up slower (which it hasn't so far)