Pyraminx BLD

[Kenneth]

What is the best? go for speed BLD or normal BLD?

To do speed you orient centres and tips, tracking the edges (max 4 turns) and then solve them in max 3 cycles if you do two at the time, newer any paritys.

To do it normally you got a little easier memo because there are no edges to track... but to orient a single centre is 8 turns giving max 32 turns for centres orientation = slow and then the edges are the same as for speed, 3 cycles.

Does anyone do Px-BLD? It must be somewhere in between 2x2x2 and 3x3x3 in difficulty. I'm thinking of trying, the speed variation

 

[ThePizzaGuy92]

i tired it speed-BLD style, but all three attempts ended in two+ flipped edges. :/

 

[Kenneth]

I just had the same, it was not a full solve, I scrambled and did corner orientation, just to test edges and got to two unoriented

Must learn some two-cycles, they seem to be pretty common.

 

[Swordsman Kirby]

I just had the same, it was not a full solve, I scrambled and did corner orientation, just to test edges and got to two unoriented

Must learn some two-cycles, they seem to be pretty common.

Heh, freestyle edges... algs that only affect tips...

I've only done it real BLD, but haven't tried speed BLD yet.

 

[Kenneth]

Looking for 2-cycles, here is the first I found, a two-gen that orients all four edges involved:

R' L' R L' R' L R L

I tried the normal BLD style for centres and tips (preserving edges), I used a timer and it looks that I can do that part in about 35 seconds (20 memo, 15 exec), Success rate is about 75%, still I'm not used to do BLD at all

A memo system I made up:

First find the orientation, pyraminx has got fixed orientation. Look for one colour that is the same on three centres (find one, find another, then the third). These three centres are always in one layer and that is your choosen face (I use yellow for U-face, orange R as my fixed position).

Memo the edges (I have not come so far yet, as now I'm trying visual, maybe I make a method later).

For centres I made a system. Decide a fixed order for the centres and then use A, E and O for the orientation needed, A = oriented, E = clock, O = counter . It makes a four letter string, example "AEOO" or "OEAO" that is easy to remember. (The letters I use, it's historical reasons, don't ask =)

The tips: after the rest is memorised you vivid the tips quickly and then down the blindfold, do the tips and then forget about them again (don't store to much rubbish in the memory =). This is a 5-10 second operation.

 

[Kenneth]

Today I made my first try for a full solve and it was successfull

Crazy, I did not expect to set the first ever try on full BLD solve I did on any puzzle exept 2x2x2, that I find pretty easy.

Time 8:10.16

Scrmble: b u L' R U' L R' U B' L' B R' L' U R' L B' U L' U' L U' L' U L

Wasted most of the time to look at the edges, until I understood that a pice was oriented and not unoriented, like half the time, then I saw the adjacent edge was the same = two solved edges, then I had it much easier

So, there is two edges solved in that scramble, a little lucky but not wery lucky, still had to do two cycles for the rest and before one I did setup too.

The first edge cycle I did visual but the second I memoed the stickers for and figured out the actual cycle to use (setup to one face) after I had solved the rest.

Centres was all unoriented so it was 32 turns, two tips and edges was done in 10 totaly, that makes 44 turns. Execution was around 1:20.

Did the centres 1-by-1 but I found a couple of algs (2-gen) to do them 2-by-2 last night, I post them here when I have written them down.

BTW: I skiped the idéa to do it speed style, you need energy that I do not possess to do that

 

[ooveehoo]

I've done it a few times. I just memorized the way the tips had to be twisted, the I visualized solvind the corners and thee edges (not that many moves), and if you know all the nine cases left (excluding mirrors) thats pretty easy. I have never timed it, but i think it's around 2-3 minutes memo and 15 seconds execution.

 

[Kenneth]

Ok Otto, that's the speed method. I guess if you practice it much and have a resonably easy solve you can get it to sub 60 or even better. I'm sure it is the fastest style for px BLD but the DNF frequency is probably pretty high.

BTW: I understand you got the nordic (Scandinavian) record for speedsolving the Pyraminx nowdays, I look forward to beat it at SOC a month from now

Dunno if I'm able but I can do around 10 for 3(5) as it is...

 

[ooveehoo]

Yeah, I DNF a lot. I tried your scramble, but DNFd it, I think I did a U instead of U', which is why it left me 4 wrong edges and a wong corner. It would require a lot of practice to even get times like 3-4 minutes consistantly with more than 50% success rate. I think it has been 10-20% for me.

 

[Kenneth]

OK, maybe I should try it sometime...

I found even better 2-gens for orienting two centres at the time.

5x(R L) ... does -R and -L ... OO in my system
5x(R' L') ... does R and L ... EE
5x(R L') ... does -R and L ... OE
5x(R' L) ... does R and -L ... EO

Note that the centres orient opposite to the turns used, R gives -R and R' gives R...

Can't be easier to remember and execute, really BLD friendly algs

Using this instead of only single twists makes this step max 20 turns, 12 better than before =)

 

[Kenneth]

Did a speedsolve

Well, it was not my plan but after I started the memo I found there was two edges solved relative to one of two unoriented centres so I started to track the pieces. And it was really easy so this came out to be rather lucky. To solve I had do do only a 4-turn cycle and then R and L and it was done, all tips were solved from start also

Was only practicing, no real attempt so I used a manual scramble so I can't reproduce it and I did not time it either

Per, FMC Pyraminx, this was FMC-BLD in 6 turns!

........

Bah! off by one centre

Still only practising, if I use a real scramble it is a serious attemt but if I do that I get nervous and go really slow, it is not good practising... But I'm learning, this one was around 3-4 minutes. Three centres to do, two tips, one edge solved, one 3-cycle in odd layers and two edges to orient = a pretty normal scramble I guess.

........

Now I got serious and did a real attempt, failed, off by two unoriented edges

Time 2:58.31 ... really good, but it was a easy solve

Scramble: r' U' B' R' L' R' U L' B' U' B U L' U' B R' L' B' R' U' L' B' L' U' R

Next seriuos try will be tomorrow...

 

[Kenneth]

I now learnt a couple of 2-cycles and that made this much easier, also practice did so when I'm practising now I got over 50% success rate and I can easily do 3-4 in a row before I get tired. But all my trys (4 so far) exept the first I ever made using a real scramble and timer are DNF's.

 

[Kenneth]

At last, a real attemt that I made!

Time 2:14.72 ... good, but it was fairly easy.

Scramble: l' r b' L' B L U' L U' L' R' B U' B U L U B' U R B L B U B

 

[Kenneth]

I know got a huge trick for Px-BLD...

I solve edges in 3-cycles, directly placing 2 and put one new into the buffer...

In some situations all tree pices to go are around the same centre. Then to solve just turn the centre and the cycle is in place, next orient if needed and then resolve the centre and the cycle is done.

Else these cycles needs 2 setup turns and the chance for doing errors in orientation is therefore pretty high, using the trick it's near zero.

More advanced you memo that you need to do the trick at some point..

I solve centres first but if this situation occures I do not solve that particular centre until I need to do the trick, in 1:3 of the cases the solution is the same as the trick turn, then I only need to do the centre turn and orient the edges (if it is needed).

The situation may be there in the scramble or it may come up after the first cycle is done, this makes the chance for it to happen pretty large, like in 1:15 solves or so.

I had a try today where I did the trick but restored the centre in the wrong direction, the only fault in the solve = one centre of and it was 2:13.xx, my record by a second if it was a success

 

[fanwuq]

http://www.speedsolving.com/forum/showthread.php?t=3523&highlight=advanced+pyraminx&page=3

Lots of stuff here. I still think my newest method idea is the way to go whether it is speedBLD or speed. It's only 2 looks usually, looking ahead a bit to make it one look shouldn't be difficult. I don't think an actual cycling method would work. Solving the centers would waste so many moves.
Pyraminx is simple enough to just do speedBLD. I've tried a few times before, but failed because I used a "4 look" solution.

Pyraminx: a usually speedsolve (ignoring tips) is about maybe 19 for me using old method. I haven't found the algs, but new method I expect sub-15 moves. I've done some sub-10 move lucky solutions of course.
http://users.skynet.be/gelatinbrain/Applets/Magic Polyhedra/ranking.htm
Some people have gotten 4 move solutions.

 

[Kenneth]

Thank's fanwuq. After practice I think you can average better with normal BLD than with speed. My average time is closing in to 2 minutes and I'm new to BLD, a pro must be able to do it in sub 60 every time.

 

[Kenneth]

Next huge trick is if it is a 5-cycle edges, it's the most common situation so this can be used often. When memo the edges I first quickly check how many pices it is from my buffer to the buffer pice, it may be in position, in a 2-cycle (bad case), 3-cycle (nice), 4-cycle+2-cycle (worst) or a 5-cycle (yeha!).

When I find a 5-cycle I do not do normal memo because I found the permutaion of the cycle can always be solved using 2-5 2-gen turns. So then I go hybrid and do permutation using speed turns and then orient edges/centres after that in a more normal BLD manner.

A 5-cycle means one edge is in position already, this edge connects to two centres, those two shall not be turned so it is the other two that solves the 5-cycle permutation. So the only things you need to track are the two centres you use to do the 2-gen.

This is fast

EDIT: was practising, had a 4-cycle+2-cycle, was thinking of how to do the double 2-cycle alg to make the best out of it when I understood I can do as for 5-cycles.

I needed 1 turn to go from 4-2 to 1-5 cycles and 4 turns to solve 5-cyle permutation, after that it was 2 edges to orient and 1 centre. Memo was really slow because of the new situation but execution was fast. when I knew what to do it was easy, 5 turn permut (PE), 8 turn orient edges (OE) and 8 turn orient centre (OC) + two tips, total 23 turns.

This hybride style turns out to be promising

BTW: I found a nice alg for superflip that is 9 turns long:

3x(R L U') .... easy to remember, three turns three times.

 

[Kenneth]

Had a try today, it was DNF (a missed setup) but the time was 1:27,xx

UWR: http://speedcubing.com/records/recs_bf_pyraminx.html

A litte more practising and I'm pretty sure I will get that, give me a week or so

 

[Kenneth]

Had a try today, it was DNF (a missed setup) but the time was 1:27,xx

UWR: http://speedcubing.com/records/recs_bf_pyraminx.html

A litte more practising and I'm pretty sure I will get that, give me a week or so

I wrote that 10 days ago:

43.40 seconds! ... Pyraminx BLD

l' r' u R' U' B R U' B L B L R' L R U B' U B R' L B R U' L'

It was a little lucky, I solved E-perm + C-orient in one go using B' U so after that it was only two edges and three tips to oritent.

But it is this hybrid method, normal BLD it is a 5-cycle edges + orientation parity (one is placed but fliped), two centres and three tips to orient = not at all lucky!

That I did E-perm in only 2 turns is not lucky, it's normal to have 2-5 turns for that and if it is more than 2 it is always repeated turns for the last 1-3 like R' L - R' L, U' B' L - B' L or R U - R U R so the permutation is always easy to both remember and solve (I simply remember the turns as it was an alg).

 

[Mike Hughey]

Wow, very nice! Congratulations, Kenneth!