3x3x5 BLD?

[ninjabob7]

I've been thinking about this ever since I got my 3x3x5. I've seen a few solves on Youtube, but nothing else anywhere. Has anybody here done this?

I was thinking it could be done like this:
1. Memorize the 3 inner layers (like 3x3x3 memo). This is a bit tricky since u/d edges can be swapped and corners are hard to differentiate with only two stickers.
2. Memorize the outer U/D layers. The hard part would be determining where each piece will end up after solving the 3 inner layers.
3. Solve the inner 3 layers using a normal 3BLD method.
4. Solve U/D corners. I'm not sure of the best way to do this. 3-cycle might be better than 2-cycle since there's no "buffer area" to make setup awkward.
5. You might have a parity for U/D edges which you could fix by switching two identical u/d edges and the RF/RB "tredges" (actually u/d corners plus E edges).
6. Solve U/D edges. I would do this as 2-cycles where setup moves are {U, D, L2, R2} and the algorithms switch two U edges and the RF/RB buffer tredges.

The main problem with this method is the memo. It could be done, but it would be very easy to get confused especially with the identical inner edges. I can't think of any way to avoid it though. Maybe it would help to memorize the pieces by their positions instead of by cycles, but this would complicate the actual solving phase.

Can someone recommend a program to generate the step 4 and 5 algs?

 

[Lucas Garron]

Can someone recommend a program to generate the step 4 and 5 algs?

M2 and the obvious commutators should work pretty well.

 

[ninjabob7]

M2 and the obvious commutators should work pretty well.

I'm not sure what you mean. Are you talking about the M2 method or the M2 move? And are corner commutators even possible within the restrictions?

I didn't find a program that can solve a 3x3x3 with restrictions, but CubeExplorer does generate {U,D,F2,B2,L2,R2} for its first solution. Out of all the positions I tried, this J-perm is the best:
Swap UBL/UBR and UL/UB: R2 F2 U' F2 D R2 D' R2 U R2 U' (or the inverse)
This leaves {D,F2,R2} free for setup moves if UBL is the buffer and UBR is the target.

 

[irontwig]

ACube and Ron's solver can do restricted movesets.

 

[Lucas Garron]

Are you talking about the M2 method or the M2 move? And are corner commutators even possible within the restrictions?

Either. And yes, of course.

http://www.opticubes.com/cubing/pll/pll_j1.php

 

[ninjabob7]

I actually misinterpreted the parity problem. The U/D and u/d layers are independent, so each one must have even parity. This means there are three parity steps: one after inner corners, one after inner edges (because two indistinguishable edges can be swapped making it look like an odd parity), and one after outer corners.

Now my method looks like this:
1. Memorize inner corners and inner edges.
2. Memorize outer corners and outer edges.
3. Solve inner corners with Old Pochmann.
4. Parity 1 (same as in Old Pochmann 3x3x3).
5. Solve inner edges with Old Pochmann.
6. Parity 2 (T-perm, swap uL/uR, swap UL/UR while setting up buffer)
Algorithms: T-perm
R2 F2 u2 F2 R2 F2 u2 F2
U R2 U2 R2 U2 R2 U
7. Solve outer corners with a slight variation on Old Pochmann.
Setup moves: <R2, F2, D>
Algorithms: R2 F2 U' F2 D R2 D' R2 U R2 U' (target UBR)
F2 U' F2 D R2 D' R2 U R2 U' R2 (target DFR)
U' F2 D R2 D' R2 U R2 U' R2 F2 (target UFL)
8. Parity 3
Algorithm: U2 R2 U R2 U R2 U2 R2 U2 R2 U R2 U' R2 U2
9. Solve outer edges (fixed or floating buffer)
Algorithms: R2 U2 R2 U2 R2 U2 (UF-UB)
R2 U R2 U R2 U2 R2 U2 R2 U R2 U' R2 (UF-UR)

I did a test solve, writing down the cycles instead of memorizing, and it worked. I had parity 2, which was the one I was mainly worried about, but not parity 1 or 3. The inner corners are straightforward. The inner edges can be handled fairly easily: the first time you encounter (for instance) a green edge send it to white/green. The second time you encounter a green edge send it to yellow/green. The outer corners are not too difficult - you just have to get used to defining positions based on the scrambled inner corners rather than the centers. The outer edges are the hardest to memorize because of the single-sticker problem. For each position you need to find it in the inner edge memo, then go backwards to figure out where the position is on the scrambled cube. Because of this I would definitely use a journey method to memorize inner edges. A list/story could still work if it was linked forwards and backwards, but the journey seems more suitable.

I will try this out for real tomorrow. If anyone can suggest improvements (especially better parity algs) I would appreciate it.

 

[ninjabob7]

I just did this in 14:11.40 on my second try. I didn't look at the memo time but it was over 7 minutes. It was a fairly lucky solve - I didn't have any of the 3 parities and the outer edges took only 4 swaps to solve. I messed up the inner edges - two of them were flipped at the end. I just flipped them and hoped that was the only thing wrong (it was).

Spoiler

Scramble (from qqTimer): B2 F2 D' U B2 R2 B2 L2 D' U R2 U R2 B2 U F2 L2 U2 R2 U2 F2 B2 L2 B2 R2 / R2 L D2 F L' U L B2 U' L U2 F' R2 F' D B2 U2 F D' L2 B2 F D B' R

Last night it took 16 minutes to memorize and 4 minutes to DNF (I think I botched a setup move on the inner edges), so I made a few changes to my memo method which helped a lot today.

Taking off that blindfold and seeing the cube solved was easily the best moment of my cubing history.

I guess I should make a proper tutorial now.