FMC Help

[Arcanist]

I have a competition in 3 weeks, and believe I can probably put in about 20 hours of work into it. I plan on mostly (maybe only) focusing on FMC. I have done a bit of FMC in the past and averaged about mid 40s with basic block building and NISS. I was wondering how I should practice and what techniques I should learn.

Currently: Mid 40 average.
Knows:
Basic Block Building
NISS
Pseudo Blocks
EO (not very well)

Goal: Global 6/37

 

[GloopyGum]

Learn insertions

 

[Fred]

I'd suggest practicing EO for sure. It's never wrong to start with EO, but if you see a nice blockbuilding start you can also do EO partway through the solve. It generally makes blockbuilding easier and increases your chances of finding a good finish. The next two techniques I'd suggest trying are commutators and insertions, which usually finish off solves much more efficiently than algs. I recommend SpeedCubeReview's FMC tutorial.

 

[xyzzy]

Other than insertions (as suggested by others above), analyse your attempts.

Use Cube Explorer or Nissy or csTimer to find out what you could have done better. E.g. if you did an 6-move 2×2×2, was there a 5-move block instead; what do those 5-move blocks look like; how can you incorporate that into your solutions moving forward?

---

If somehow you end up having F2L done (this shouldn't happen all the time; last layer is not a good FMC step), you shouldn't be doing last layer with standard 2LLL unless there happen to be a lot of cancellations.

If edges are oriented:
- If edges can be solved with AUF, do that, then fix the corners with insertions.
- If not, spam Sunes/Antisunes from every angle to reduce to L3C or a short 1LLL that you know.
- If you get opposite edge swap (like T perm or Z perm), it's usually crap.

If edges aren't oriented:
- Spam all sorts of EO algs like FsexyF', sexysledge, wide Sunes/Antisunes, etc. Again, hope for L3C or short 1LLL.
- If you have dot OLL: do the normal OLL alg; you don't get a PLL skip, just ignore this and find something better.

Instead of fully solving F2L, it's often better to solve just the last edge together with the LL edges, bringing you to L5C (last 5 corners).

---

Check more things. Aggressively throw out starts that don't look like sub-40 potential. As you improve, reduce this threshold gradually.

 

[Arcanist]

A few questions about an FMC attempt I am doing:
Solution so far:
Normal Scram: R' U' F R2 D F2 D' L2 F2 L2 D' F2 D' U2 R U L2 F2 R D' R F' L U' R' U' F
F2L -1 = U L2 U L U L' (6/24)
Top edges oriented: U2 L' U' L U2. (5/28)

Inverse Scram: F' U R U L' F R' D R' F2 L2 U' R' U2 D F2 D L2 F2 L2 D F2 D' R2 F' U R
EO: U' F' B' L' B' (5/5)
2x2x3: L D U R U' R2 U' R' U2 R B2 R' (12/17)

The solution so far solves the cube to the skeleton.
I put my NISS and normal solution together to get the alg for the skeleton:
U L2 U L U L' U2 L' U' L U2 R B2 R' U2 R U R2 U R' U' D' L' B L B F U.
However, the end result is different from when I do R B2 R' U2 R U R2 U R' U' D' L' B L B F U (the inverse on the inverse scram) + the scramble + U L2 U L U L' U2 L' U' L U2. Is anyone sure why?

I came up with 2 commutators to solve my skeleton. The first is [L, F R F'], the second is [R', F' L F]. If I add these to the end of my normal solution, I solve the cube.
U L2 U L U L' U2 L' U' L U2 *^ R B2 R' U2 R U R2 U R' U' D' L' B L B F U.
* = L F R F' L' F R' F'
^ = R' F' L F R F' L' F

However, if I try and do an insertion, the cube does not end up solved.
U L2 U L U L' * U2 L' U' L U2 R B2 R' U2 ^ R U R2 U R' U' D' L' B L B F U.

My last question is if you can put a commutator inside a commutator. For some reason I was under the impression that you could, but I am no longer sure if this is true.
For example:
If I have the first commutator:
L F R F' L' F R' F'
and insert the other one in it to cancel moves:
L F R ^ F' L' F R' F'
the commutator no longer solves my skeleton.

 

[Arcanist]

Sorry for the bump, but if anyone could help me that would be great.

 

[Fred]

I'm not sure why, but I've also noticed I need to make the skeleton before doing inserts, or else I end up swapping the wrong corners. But you can insert commutators inside inserted commutators just fine. You should not be inserting the same commutator elsewhere in your solution, but rather a different commutator the permutes the same three pieces in the same orientations. The idea is to go through your skeleton with stickers on the pieces you're aiming to permute with a commutator and look for points where they're in a good position to apply a commutator that doesn't require setup moves, and also cancels move(s) with your skeleton. You'll need to consider a new commutator for every such position. Also, sometimes one position can have multiple possible commutators that work. For example, L' U R U' L U R' U' and U' F' U B U' F U B' do the exact same thing to the cube, but will cancel different moves.

 

[xyzzy]

However, the end result is different from when I do R B2 R' U2 R U R2 U R' U' D' L' B L B F U (the inverse on the inverse scram) + the scramble + U L2 U L U L' U2 L' U' L U2. Is anyone sure why?

Why should they be the same?

[scramble] [moves on normal] [inverse of moves on inverse] is not the same thing as [inverse of moves on inverse] [scramble] [moves on normal]. These are conjugates of each other, so they'll have the same cycle structure (e.g. in your case, they're both 2c2c), but where the cycles are will generally be different between them.

Your linearised skeleton is [moves on normal] [inverse of moves on inverse]. Once you have that, you don't need to care about which moves were originally done on normal / which moves were originally done on inverse.

Same deal with multiple insertions. Do your first insertion to reduce to 3c (or whatever), then treat the new thing after cancellations as a new skeleton on its own and find insertions on that. Of course you can nest insertions; your new skeleton cannot know that it was originally made from a shorter skeleton and prevent you from doing that (it is, after all, just a bunch of moves and not sentient).

When writing out the explanation I just write the linearised skeleton once and have multiple insertion sigils (and if I'm feeling lazy I might even leave the skeleton in NISS form), but during the attempt I write out the full skeleton after every insert. Takes a while, but makes checking alternative insertions much cleaner and it's also easier to spot cancellations across different insertions.

 

[Arcanist]

I went to my comp awhile back and did alright.
DNF first solve. (wrote something else on the bottom of my solution sheet and writing wasn't super clear due to being extremely rushed on time. Would have been like a 45)
40 second solve
39 third solve.