[Help Thread] Waterman Discussion

[efattah]

In Waterman you solve the first layer minus the DF edge. So Waterman CLL ignores the E-slice and the DF edge. Advanced Roux solvers use multiple CMLL algorithms to modify the edge orientation during corner solving to avoid 6-flips. Waterman himself near the end of his 'career' would inject M/M' and/or convert r to R during CLL, to solve a U-layer edge at the same time as solving the corners. So Waterman CLL is actually a lot deeper than just CELL.

 

[WarriorCatCuber]

In Waterman you solve the first layer minus the DF edge. So Waterman CLL ignores the E-slice and the DF edge. Advanced Roux solvers use multiple CMLL algorithms to modify the edge orientation during corner solving to avoid 6-flips. Waterman himself near the end of his 'career' would inject M/M' and/or convert r to R during CLL, to solve a U-layer edge at the same time as solving the corners. So Waterman CLL is actually a lot deeper than just CELL.

Only simplified waterman solves the first layer minus one edge. And officially, You don't solve an extra edge during CLL.

 

[efattah]

You have it backwards. The 1st Waterman (original) method solved the entire first face. The later more advanced method left out the DF edge since it would be destroyed often during the next phase anyway. And yes, the official Waterman document does explain the method of solving an extra edge during CLL.

 

[WarriorCatCuber]

You have it backwards. The 1st Waterman (original) method solved the entire first face. The later more advanced method left out the DF edge since it would be destroyed often during the next phase anyway. And yes, the official Waterman document does explain the method of solving an extra edge during CLL.

Um ok sure. What would the algs do then ? Cause I thought that those 380 algs would solve it like this :
1. 0 algs (First layer)
2. 42 algs (CLL)
3. dunno how many algs (solve 2 redges and place another r edge in the r layer)
4. dunno how many algs (solve the last 2 redges and orient midges)
5. roux 4c

 

[Chris_Cube]

I understood also Watermans Original algs and I think they are better to learn than LMCF algs. I understand them better. Recog is also really fast

 

[WarriorCatCuber]

This guy averages like 44 STM on simplified waterman with full CLL : https://www.speedsolving.com/threads/waterman-method-example-solves-by-ryan-przybocki.46824/

 

[efattah]

Um ok sure. What would the algs do then ? Cause I thought that those 380 algs would solve it like this :
1. 0 algs (First layer)
2. 42 algs (CLL)
3. dunno how many algs (solve 2 redges and place another r edge in the r layer)
4. dunno how many algs (solve the last 2 redges and orient midges)
5. roux 4c

CLL: 42
Solve one ledge and one redge, and/or solve 2 redges in the R layer, and/or solve ledge+2 redges: 110-120 algos
Solve last 2 redge while orienting the midges, which also includes the case of solve last ledge while orienting midges and solve last redge while orienting midges: 48*4 +32 = 224 algos
roux 4c: no algos, basically intuitive

 

[Chris_Cube]

So I found a very fast possibilty to build the First Layer minus one corner. The ZZ Users here will understand it.
First Layer fast:
1. Build a Line (DB + FB Edge)
2. Take a good Edge and pair it up with 2 Corners and insert it with L2 or R2 Moves
3. Insert the last 2 Corners. I usually orient them to insert them with L2 or R2. Very ergonomic

 

[fun at the joy]

So I found a very fast possibilty to build the First Layer minus one corner. The ZZ Users here will understand it.
First Layer fast:
FB Edge

you mean DF?

 

[Chris_Cube]

Yes excuse me. What do you think? Partitional Edge Orientation is something good for this step

 

[efattah]

The two corners that sandwich the missing DF edge can be placed in a swapped position. In this case, you must use EG-1 algorithms to solve the corners. Most EG-1 2x2 algorithms (which are the same as the LMCF ones), will preserve the 1x2x3 block at the lower back of the cube. Only a few algorithms would need to be regenerated. This would offer a much more flexible way of creating the first face. I do like Chris_Cube's idea of building the first face too, and adding to that the ability to insert the last 2 corners swapped would be ideal.

I would comment that it is still worth looking into rotationless waterman. In this case the first 'face' is solved on the L face, and you must regenerate the CLL+EG1 algorithms to solve the corners on the R-face. In this fashion, the first 'face', being on L, is actually a lot like Roux first block, except you must add the last two corners in. The solve now becomes rotationless.

 

[Chris_Cube]

Thanks for your Input. Waterman rotationless could be combined with my first layer idea and you could conjugate watermans cll algs so there is no need to learn more

 

[mukerflap]

So I found a very fast possibilty to build the First Layer minus one corner. The ZZ Users here will understand it.
First Layer fast:
1. Build a Line (DB + FB Edge)
2. Take a good Edge and pair it up with 2 Corners and insert it with L2 or R2 Moves
3. Insert the last 2 Corners. I usually orient them to insert them with L2 or R2. Very ergonomic

roux fb + corners is probably best approach usually

 

[Chris_Cube]

No not if you perform original Waterman. You then have to turn the Cube z' and after that you have to turn it backwards.... so there are more rotations than neccesarily

 

[Chris_Cube]

What I find fascinating about waterman (one of the reasons why i want it to use as my main method besides maybe roux) is, that it was so extremely developed already in the 1980s. Waterman and Kraamer made back in the days for this time a very large algorithm list. Did they used computers? Or was it Kraamer and Waterman who made the algs using commutators?

 

[Owen Morrison]

Does anyone have a document with all of the Waterman algs? I am really interested in this method and I want to see if the algs are good for fingertricks.

 

[Chris_Cube]

You find the Algs on rubikscube.info under the Introduction as a pdf file. All algs you need are in there incl. an explanation of the method in-depth.

I also have a question on the Algs in Table 5. Sometimes the algorithms don't work with the correct setup. It inserts the last redge correct but there are sometimes 2 misoriented midges. For example case 5a never works. And 2,3,4 work sometimes. Because of this (R) move. Can someone explain this to me?

 

[efattah]

I remember reading that Waterman and Kraamer generated all 350+ algorithms by hand, no computers. They mentioned how it is 'so much easier' today to generate algorithms now that we have computers.

 

[Chris_Cube]

But how do I execute them? Table 5 wont work. Sometimes 2 midges are then still misoriented

 

[mukerflap]

instead of doing last redge+ eo in an alg why not just do LSE?