ACMLL: A new algorithm set for Roux

[Athefre]

ACMLL

Introducing a new algorithm set for the Roux method. The user solves F2B with flipped pairs, swapped pairs, or even flipped/swapped individual pieces. Then F2B is corrected during the CMLL algorithm. This provides many advantages including a significant move-count reduction, an ability to plan more of F2B during inspection, and good algorithms. To provide an example of how this reduces move-count, there is ~50% chance that the final built pair for SB will be a flipped pair before being inserted. It will have to be moved over its correct slot and flipped into place. However, if this pair is instead inserted into the opposite side just using R and U moves, this provides an average reduction of 2 moves. You can then imagine this reduction applied to all of the pairs that are built for F2B. If the shortest insert is used for all pairs, or even pieces, this will greatly reduce the number of moves for F2B. It is also easy to do and doesn't require any extra thinking during the solve.

I actually started working on this around a year ago and have occasionally posted updates in the Roux Discord server and in the Roux discussion topic here on SS. It has gotten really good feedback. So far what has been developed are the single flipped pair blocks and the spiral blocks. These have been analyzed and determined to produce a great move-count reduction and to also have good algorithms. There is still a lot to be analyzed and many algorithm sets to be developed. It would likely be beneficial to run a computer analysis for which block types have the best results. If anyone wants to start that project, let me know.

 

[povlhp]

So CxLL on each block individually and handle the diagonal swap as its own case ? Can likely be optimized if any corner can be rotated.
inspecting CxLL for 8 corners sounds impossible.
Or do we assume correct oriented but permutated corners ?

 

[Athefre]

So CxLL on each block individually and handle the diagonal swap as its own case ? Can likely be optimized if any corner can be rotated.
inspecting CxLL for 8 corners sounds impossible.
Or do we assume correct oriented but permutated corners ?

The user recognizes only the CLL case of the four corners on the U layer. Then executes the algorithm that solves those corners while correcting F2B. This algorithm comes from the specific set to which the flipped/swapped block or blocks belongs. There is no need to look at F2B because you know what you planned and built during FB and you know what you built during SB.

 

[efattah]

This is kind of similar to what I suggested a few years ago, namely to insert the FR pair with a flipped edge and then correct it during L6E using Waterman L6E algorithms. There are many further variations of this, like putting the entirely wrong edge (in any orientation) in FR, DR, or BR then correcting during L6E with Waterman algorithms.

 

[abunickabhi]

This is an interesting algset. I have always wondered how Roux and Waterman could be brought together with method variants.
There is too many cases in this, but it is useful to know a few, for a beginner like me.

 

[trangium]

There is too many cases in this

Says the person who learned 126,720 5-style algorithms.

 

[Nir1213]

This is an interesting algset. I have always wondered how Roux and Waterman could be brought together with method variants.

I thought i saw in the wiki somewhere about a method called WaterRoux. Didn't really read it though just skimmed over it

 

[abunickabhi]

Says the person who learned 126,720 5-style algorithms.

I do not know all the algs currently, and neither are all the algs speed optimal according to the 3-style standards (40% faster than execution time of 2 3-style UF algs).
Also 5-style is useful as it can come in some LSE cases and that is useful.

 

[Cubing Forever]

This is sorta like EG-1 but preserving the E and S slices. Nice extension to Roux btw.

 

[Athefre]

Added another set. The tab in the document is called "L/R Flip Swap Front" for now. This one is when the front two pairs are swapped from left to right. Very similar to EG. Overall this set has good algs and I could see this kind of block type being used in non-linear blockbuilding situations.

This set was generated by @Rouxster. I then went through and generated additional algs to make the set even more ready for speedsolving use.

 

[Athefre]

The "Swap FL/BR" set has been generated by @OreKehStrah. This is when the dFL and dBR pairs are swapped. This should be a set that is of interest to learners early in deciding which sets to learn.