Classic Pochmann on Megaminx

[Kenneth]

In a spin of from the Megaminx PLL project and because a any puzzle BLD competition is starting to get planned in Sweden ("Gangsta BLD day") I made a small guide for how to adapt the classic Pochmann method to Megaminx:

http://sites.google.com/site/permuteramera/special/old-p-bld

Please point out errors if you find any, I'm not a BLD-cuber in practice, just in theory

 

[Stefan]

Looks good. Though, the memo I think is unusual. I analyze the cycle structures during memorization and memorize along the cycles so that blindfolded I don't have to analyze anymore. If I memorized along a fixed path like you suggest, then I'd have to analyze while blindfolded and jump around in my memory. I thought it's common practice to memorize along the cycles, but now I'm not so sure anymore. Might be an interesting poll...

 

[Kenneth]

Ah, no, that is not what I suggest. To memo is like you do but my blocks are for helping to sort the diffrent positions. So when you got the next piece you first think of which block it belongs to, then it's position in the block. Rather than trying to memo the side it goes to. 12 sides in all diffrent angles are much more to keep apart than 5 blocks that goes from top to bottom, one for each side of LL.

 

[Stefan]

Ok, but when I for example come across the white/green piece while analyzing+memorizing, I memorize "white/green" (actually, a person corresponding to it) and next just look at the white/green place (i.e. the place between the white and green centers). I don't see the advantage of your way, but probably I just still don't understand it.

 

[Kenneth]

Ah, yes, that if you use colours to make names, red-green = tomato. But I always used realtive sides, R-side top-side oriented (A,O,E) makes a RAT. The advantage is that you can start from any orientation, RAT is still in RAT position but red-green is somewhere else than normally.

On the 3x3 I divided the puzzle the same way as here to make one block for each side and that is the first letter of my word R, B, L and F and then the layer is the last letter, in the mid I got the orientation letter. Here I can use about the same but I get one extra side for first letter and more layer letters (I did not work it out 100% because I will not describe it deeper and I will not try to BLD the mega =)

I got some post somewhere in the BLD memo method thread that descrcribes my approach. There must be more methods for memo that works about the same...

Here : http://www.speedsolving.com/forum/showthread.php?p=37779&highlight=Blocks#post37779

Better description after a bit of rework, but in Swedish : http://sites.google.com/site/kubbuth/minnesteknik

 

[Chuck]

Thank you, Kenneth.

I've watched Mike's and Ryosuke's videos, and have been wondering how they're doing it. So I guess your guide will help. I will try this in the future.

 

[Mike Hughey]

Mondo and I both used methods involving only 3-cycles, not the 2 2-cycle swaps that Kenneth is proposing here. The 3-cycles are commutators that are easy to understand; the problem with using 3-cycles is that you have to constantly alternate back and forth between 2 places being your buffer, which adds to the confusion when you're actually doing it. I never got very comfortable with it, which is why I quit practicing it. Perhaps the approach of using 2 2-cycles would be better.

Setups were always difficult, even when I switched to Mondo's and Stefan's algorithms, which were better than my original really bad ones for setup moves. I also had a lot of trouble memorizing; I would use first letter for face and second letter for piece within the face. I find Kenneth's suggestion of blocks for memorizing somewhat confusing, and I'm not sure I see how it work in practical use, but maybe I just need to try it a bit and it might make more sense. It certainly was much harder to memorize than a 4x4x4, both because I needed 2 letters to identify each piece, and because it was so hard to trace the pieces around on a megaminx. I suppose maybe I just have a very bad method.

Someday I guess I should try again - I'm very unhappy with how bad I was at this - it's bound to be possible to get much better than I was at it.

 

[trying-to-speedcube...]

Mondo and I both used methods involving only 3-cycles, not the 2 2-cycle swaps that Kenneth is proposing here. The 3-cycles are commutators that are easy to understand; the problem with using 3-cycles is that you have to constantly alternate back and forth between 2 places being your buffer, which adds to the confusion when you're actually doing it. I never got very comfortable with it, which is why I quit practicing it. Perhaps the approach of using 2 2-cycles would be better.

How so? I have 1 buffer for corners and 1 buffer for edges?!

 

[Mike Hughey]

Mondo and I both used methods involving only 3-cycles, not the 2 2-cycle swaps that Kenneth is proposing here. The 3-cycles are commutators that are easy to understand; the problem with using 3-cycles is that you have to constantly alternate back and forth between 2 places being your buffer, which adds to the confusion when you're actually doing it. I never got very comfortable with it, which is why I quit practicing it. Perhaps the approach of using 2 2-cycles would be better.

How so? I have 1 buffer for corners and 1 buffer for edges?!

I'm sorry - sloppiness on my part. I haven't done it in a while and I wasn't really thinking about what I was saying. With my 3-cycles, I would have 2 pieces that are constantly swapping back and forth, and a third location that is the piece I'm shooting to. I would shoot first from the first of the two pieces that are swapping back and forth, and then shoot from the second of those two pieces (by using the inverse algorithm).

 

[trying-to-speedcube...]

Oh, that makes a lot of sense! I never found a way to do 1 at a time for megaminx BLD, maybe I should use that. Keeping track of the setup moves is already hard for 1 piece at a time, let alone 2! Thanks, I'll have a look into that!

 

[Mike Hughey]

Oh, that makes a lot of sense! I never found a way to do 1 at a time for megaminx BLD, maybe I should use that. Keeping track of the setup moves is already hard for 1 piece at a time, let alone 2! Thanks, I'll have a look into that!

My first couple of tries, I tried doing 2 at a time, and yes, it was very difficult. Switching to one at a time suddenly made it very doable - hard, but doable.

 

[KJiptner]

wow. this guide sooo makes me wanna try this! thank you!

 

[Kenneth]

Mondo and I both used methods involving only 3-cycles, not the 2 2-cycle swaps that Kenneth is proposing here. The 3-cycles are commutators that are easy to understand; the problem with using 3-cycles is that you have to constantly alternate back and forth between 2 places being your buffer, which adds to the confusion when you're actually doing it. I never got very comfortable with it, which is why I quit practicing it. Perhaps the approach of using 2 2-cycles would be better.

How so? I have 1 buffer for corners and 1 buffer for edges?!

I'm sorry - sloppiness on my part. I haven't done it in a while and I wasn't really thinking about what I was saying. With my 3-cycles, I would have 2 pieces that are constantly swapping back and forth, and a third location that is the piece I'm shooting to. I would shoot first from the first of the two pieces that are swapping back and forth, and then shoot from the second of those two pieces (by using the inverse algorithm).

That is about how I solve the pieces that belongs to the slave swaps, I have to use the inverse 3-cycle from even swaps for uneven swaps.

I think my block memo method works pretty ok, it is not diffrent to remember the block a piece belongs to than from remeber the side, they are just lesser. The idea is to reduce to smaller numbers, here I get 5 instead of 12 main parts, on 3x3 it is 4 against 6.

You can give the blocks certain attributes that are diffrent for each one, say weather (sunny, cloudy, rainy, fog, snowfall) or whatever to distinguis them from each other.

But newer mind, it is my approach and as I wrote at the page, we all got diffrent ways to do memo...

Mike, Hardness, on 4x4 you have a clear definition and you are used to it, on mega you just tried it a few and probably did some experiments while going. With the same amount of practise and the same level of definition for the memo method it should be a bit easier than 4x4.

Kai, Chuck.. go for it

 

[KJiptner]

I would stick with the advanced TuRBo-like version of your tutorial, since it allowes you to always use the easiest setup move for a piece. Too bad, that this approach requires many algs for the 3cycle cases...

 

[Stefan]

If you have trouble with setup moves or memorizing, you could try to somewhat... memorize the setup moves.

With buffers in the U layer, use a d-turn (everything below U) as the first setup move for all 95 targets below the U layer. That reduces them to 19 targets nicely in the front, for example in a standard block like Kenneth's. So memorize a target as the setup d-turn combined with one of 19 letters. During memo, you can do y-rotations to simulate the d-turns for easier visualization. And you can use the remaining letters of the alphabet for targets in the U layer.

Or something like that.

 

[KJiptner]

hehe, this is excactly what I'm thinking of right now. I try to figure everything out at the moment. I don't understand why the algs target something in the back though. This makes setups a lot more annoying. Why not something in the front-right? I would like the algs more with excactly the opposite definition of slave-pieces (so shoot to the (F)R-edge). Same for corners. The case would be better with a y and opposite definitions again (shoot to the FR corner). Would require changing the "advanced algs" though.

 

[Stefan]

A bit more about what Mike mentioned, solving with alternating 3-cycles instead of double swaps (at least for corners):

Use buffers UFL and ULB.

1st, 3rd, 5th, ... target:
Set it up to LFD and do [L D L', U'] or
set it up to FDL and do [F' D' F, U'].

2nd, 4th, 6th, ... target:
Set it up to LFD and do [L D L', U']' or
set it up to FDL and do [F' D' F, U']'.

Again: Something like that. Make up your own algs for your liking.

 

[Kenneth]

Kai, I started to generate the edge 3-cycles even before you said you may need them, so now I'm adding them to the page. All corners cycles are already at a diffrent page at the site but I have not added the algs for them yet, it will soon be done.

Edit: while looking at it, is seems you will only need one edge cycle that flips the pieces, the one that moves from slave to slave must stay oriented (naturally) so it is only one case (the mirror is the same as the inverse)

 

[Henrik]

Here is an alg that flips two edges UF and UR it might be useful.

It also works on 3x3, but I found it for Megaminx when I looked into BLDing it.

F R U R2' U' F' R U R U' R' U' R' U R (15,16)

 

[Kenneth]

Now I got all 3-cycle corners of LL up at the page

http://sites.google.com/site/permuteramera/other-methods/3cll