[Help Thread] FMC Discussion

[Cubenovice]

In theory, any skeleton can be converted to an optimal solution for the scramble if you're willing to try long enough insertions.

True but those insertions would not be 3 cycles.
The insertion would actually be "a shorter solution + moves around it that cancel out your existing solution"

And by the way, I've contrived some tricky cases where I thought the existing Insertion Finder might miss the solution I expected due to needing a suboptimal 3-cycle (for instance), but it would still find it. So if you want to have it find solutions that the existing one misses, you probably have to consider a lot of suboptimal insertion algs.

Again I dare to ask: what do we consider optimal here?
An optimal edge cycle is only 6 moves but if there are no 6 moves insertions possible for your skeleteon then a 7 mover, then an 8 mover etc etc should be considered as the optimal solution for the specific cycle to be solved.

In FMC insertion finding I consider the shortest possible insertion "optimal".
For edges this is more often than not based on a 3 cycle longer than 6 moves


edit:
"suboptimal insertion alg" sounds like solving an 8 move case with more moves which does not make sense.
It would be something like using an A9 commutator to solve an 8 move case...
But in reality this would just cancel an additional move for the same net result.

You only need to include optimal algs for all posible cycles, where optimal for specific corners cases can be up to 12 moves ( Per special)

 

[irontwig]

True but those insertions would not be 3 cycles.

Sure it would:

Optimal generator: G
Skeleton that leaves three corners: S

The insertion at the end S' G' is surely a three cycle, right?

 

[CubeRoots]

I'm not sure how much more optimal an insertion finder should go than the one we have currently. The longer the insertions you consider are the more conofusing, non-intuitive output you are going to get, and an insertion finder checking extremely long insertions would inevitably end up becoming an optimal solver, which we already have plenty of.

I think that the function of a real insertion finder is to find the optimal solution that a human could feasibly find. The difficulty in creating an optimal insertion finder is judging when the line of "can be feasibly found by a human" is crossed.

 

[Jakube]

And after rotating these at all faces I came up with a list of 96 commutators. Are there any (true 8 move) edge commutator that I missed now?

There are algs like L F2 L' R U2 R'. This is not a true commutator, but it something, I would definitely use as an insertion.

 

[cubernya]

L F2 L' R U2 R'. This is not a true commutator

[L:F2][R:U2]

 

[Cubenovice]

There are algs like L F2 L' R U2 R'. This is not a true commutator, but it something, I would definitely use as an insertion.

note how the first move does not move the pieces of the cycle.
This means you can just "roll over" the first move and you get F2 L' R U2 R' L

and hey, the last piece in your alg asls does not move the pieces of the cycle so you could bring it to the front.

This is the good part of the 6 movers: many options to cancel moves.
Too bad the pieces never end up on the slice correctly when you need them to.

 

[Jakube]

just roll over the first move and you get F2 L' R U2 R' L

this is the good part of the 6 movers

I know this. I just wanted to add, that Marcel's InsertonFinder should look for such algs too.

 

[MarcelP]

There are algs like L F2 L' R U2 R'. This is not a true commutator, but it something, I would definitely use as an insertion.

Thanks! You are right.. I need these also..

 

[TeddyKGB]

If I do a solve that leaves me with 4 corners left and it is two sets of swapped corners I believe the best way to go about finishing this would be to do two insertions. My question is how to sticker these corners. What I did was put a sticker with a 1 on one corner then a sticker with a 2 where the 1 goes, then put a sticker with a 3 on the same corner piece as the 1 but where the 2 goes and did the same thing for the other set but with A, B and C. Not that this was particularly difficult it just got a little confusing when going through the skeleton looking for an insertion with all these stickers everywhere. I was just wondering if there was an easier way to do this.

Also, if I have the same situation but with a fifth corner twisted in place should I do three insertions or go back a few moves and try for a different ending?

 

[MaikeruKonare]

I want to learn how to do this. Why did you invert the scramble?

 

[Cubenovice]

I want to learn how to do this. Why did you invert the scramble?

Go here:
http://www.speedsolving.com/wiki/index.php/Fewest_Moves_techniques

Read the page and check out the links at the bottom of the page.

If I do a solve that leaves me with 4 corners left and it is two sets of swapped corners I believe the best way to go about finishing this would be to do two insertions. My question is how to sticker these corners. What I did was put a sticker with a 1 on one corner then a sticker with a 2 where the 1 goes, then put a sticker with a 3 on the same corner piece as the 1 but where the 2 goes and did the same thing for the other set but with A, B and C. ?

Whoa...
There is no need to sticker all faces of the corner piece.

All you need is 4 stickers, one for each corner piece involved.

First swap:
sticker a random sticker "1" on the corner piece
find where it goes and sticker it with another "1"

Second swap:
sticker a random sticker "A" on the corner piece
find where it goes and sticker it with another "A"

Insertions:
look for a cycle that brings one "1" to another "1" + involves another stickered corner piece.
Note that it does not have the be the sticker with the "A" on it. You just need to involve the "corner piece" with the "A" sticker.

Also, if I have the same situation but with a fifth corner twisted in place should I do three insertions or go back a few moves and try for a different ending?

5 corners with one twisted in place typically requires an alternative ending.
But if you have a really short skeleton you might go for it, I know Sebastien has done three insertions before and with good results.

 

[yoshinator]

So, I've been practicing a little bit of FMC before worlds, and I came up with this nice-ish beginning of a solution, and I'm looking for some help with the end.
Scramble: F2 D2 R2 U' F2 D' U2 B2 U' L2 R' U B' D' F2 D2 F' L' R'
Solution so far: U' R' F B L2 D' L2 F' U R U' F R' B' R2 D' R' D B L
Now, what should I do next? I can't seem to find a good way to solve the rest of the cube.

Thanks in advance for the help.

 

[mycube]

can you please explain your solution?
so it's easier to give some tipps or good endings

so far: easiest way would be a B at the end of the solution and insertions. there you would need 3 what is a lot..

edit:
change the F2L-1 to this:
U' R' F B L2 D' L2 F' U R U' F R' B' R2
L D' R' D L' B L B

there is a 5-cycle, what you can solve with two insertions

 

[cuBerBruce]

Whoa...
There is no need to sticker all faces of the corner piece.

That's not what he said. He only puts stickies on 6 facelets, not all 12.

But your method is assuming you have oriented 2-cycles. 2/3 of the time the 2-cycles will be misoriented. In the case of misoriented 2-cycles, you need to be careful which piece is which in the 2-cycle, or you may end up "solving" a piece with the orientation wrong. That's why TeddyKGB uses "1" and "2" instead of two "1"s.

One can ask if it's worth the time searching all possibilities for the 1st insertion. Obviously it takes more time to do so, and one might be better off doing a faster, more limited search.

 

[yoshinator]

can you please explain your solution?
so it's easier to give some tipps or good endings

so far: easiest way would be a B at the end of the solution and insertions. there you would need 3 what is a lot..

edit:
change the F2L-1 to this:
U' R' F B L2 D' L2 F' U R U' F R' B' R2
L D' R' D L' B L B

there is a 5-cycle, what you can solve with two insertions

U' R' F // Super obvious 2x2x2
B L2 D' L2 // pseudo-2x2x3 Then I decided that I'd try to do F2L minus one heise style
F' U R U' F R' B' R2 D' R' D B L // Rest of peuso F2L minus one plus edge plus OLL (OLL was lucky) [What I did was insert one edge then solve remaining F2L pair]

Just to confirm, an insertion is basically just something like R' D R U R' D' R? Is that right? Because that's what I did in the first place, which leaves a not so nice solution.

Edit: After messing around with your idea to create a 5-cycle, I'm just curious, how did you think of that?

 

[stevecho816]

I started doing FMC and I have a question, how do you find a good premove? Do you choose a random premove until you find a good start?

 

[mycube]

Edit: After messing around with your idea to create a 5-cycle, I'm just curious, how did you think of that?

It is normal to try some different options for an ending. so this was the obvious easy opportunity for an end with better corners
and normally at the end of my skeleton I use insertions for the corners because there are shorter solutions than an LL

if nobody did I will take a look at your solution tomorrow

 

[mycube]

played around with your start yoshinator

F2 D2 R2 U' F2 D' U2 B2 U' L2 R' U B' D' F2 D2 F' L' R'

Premove: L
U' R' F B L2 D' L2 - same start, but premove for a real 2x2x3
B' D B' L' D L - F2L-1
B' R' B' R - lucky pair, which I insert with a sledgehammer
the insertion finder found a 27 move solution with two insertions

I hope you can understand what I did.
maybe there are some good things on the inverse, I didn't check it.

I started doing FMC and I have a question, how do you find a good premove? Do you choose a random premove until you find a good start?

with this solution I can answer your question:
Look at the 2x2x3 without a premove. It's just a pseude-2x2x3. Now I do the premove L to have a 2x2x3. Without the premove I wouldn't have found the solution I found.

 

[Cubenovice]

That's not what he said. He only puts stickies on 6 facelets, not all 12.

But your method is assuming you have oriented 2-cycles. 2/3 of the time the 2-cycles will be misoriented. In the case of misoriented 2-cycles, you need to be careful which piece is which in the 2-cycle, or you may end up "solving" a piece with the orientation wrong. That's why TeddyKGB uses "1" and "2" instead of two "1"s.

One can ask if it's worth the time searching all possibilities for the 1st insertion. Obviously it takes more time to do so, and one might be better off doing a faster, more limited search.

Ah yes...
I see now that read the OP completely wrong. And the description implies misoriented swaps...

 

[PM 1729]

I had done a presentation at college about FMC. I hope someone finds it useful.
If you find any errors please let me know.

Link to all files: here

The zip file has everything, so you don't need to download anything else.

If you don't want to download the prezi, just download the examples. Here is a link to the prezi.