Developing a better LL system

[MovingOnUp]

From what I understand, this looks freaking awesome, great work. Going to wait until "learn-able version" is organised though haha,

 

[Kirjava]

I think the mouseover is a great idea.

Since you divide these into sections by the first alg performed, I was thinking maybe:
-throw each picture under "sune" as class="sune"
-somehow use php to write javascript functions that deal with just adding "[sune]" to the end of the alg?

Not sure if that makes sense, but mouseover would be really nice. Throwing this on github would be nice for this, too.

The images are static and do not use visualcube. However, I worked out a way I can do this.

Easy image tag is assigned an id that corresponds to the orientation and permutation of the pieces. I can write code that applies the alg and works out the new position.

Going to wait until "learn-able version" is organised though haha,

This is the "learnable version". All there is to do from here is just improve it.

 

[bobthegiraffemonkey]

This is the "learnable version". All there is to do from here is just improve it.

Hmm, I'm not sure I would personally call it "learnable" quite yet since as it stands people would have to learn each case individually one section at a time (unless I'm missing something you've done already?), although it's certainly came a long way already. Do you have any suggested ideas about recognition?

 

[Kirjava]

Hmm, I'm not sure I would personally call it "learnable" quite yet since as it stands people would have to learn each case individually one section at a time (unless I'm missing something you've done already?), although it's certainly came a long way already. Do you have any suggested ideas about recognition?

Actually, I'd call it learnable - just not friendly yet. Learning cases is much easier than learning algs.

I don't think 1LLL recognition is very hard, it's the recall people have trouble with. Recognition will be improved and easier for some of the cases, but the majority will need 1LLL.

I can't seem someone learning this system without putting in some effort themselves to design methods of recognition and learning to aid them, so part of this will be left open to allow people to implement the system in a way they prefer.

 

[Speedmaster]

I really want to learn this kind of LL, but i can't imagine recognise the cases and learning all the combinations of the cases..., how many cases are there? and what's the best way to recognise them?? corners first? and then edges? and if it takes a long time (1s) it will be fast too, because you don't have to recognise 2 things, and the algrhitms, are fast too

 

[EMI]

Actually, I'd call it learnable - just not friendly yet. Learning cases is much easier than learning algs.

I don't think 1LLL recognition is very hard, it's the recall people have trouble with. Recognition will be improved and easier for some of the cases, but the majority will need 1LLL.

I can't seem someone learning this system without putting in some effort themselves to design methods of recognition and learning to aid them, so part of this will be left open to allow people to implement the system in a way they prefer.

hm, I think 1LLL is quite hard to recognize (recalling might be even harder). OLL and PLL are very easy to recognize and recall. I doubt this method will be as good as normal OLL/PLL.

 

[MovingOnUp]

I don't understand how your website is organised. You click on the OLL alg type i.e. "Sune" and then what are you supposed to do ? For every one of those cases you do Sune and then from there how does it work? I must be missing something

 

[aznanimedude]

i think the idea is that the last layer is solvable using a very small set of algorithms.

and the set of algorithms are ones that are really fast to execute.

so for the sune cases, you first do Sune, then you do another algorithm from the set up there.

recognition is going to be pretty beastly but man, if anyone got this down, i mildly envy the last layer they will have :|

 

[Muesli]

Thom, you are a beast. This is incredible stuff.

 

[Kirjava]

I really want to learn this kind of LL, but i can't imagine recognise the cases and learning all the combinations of the cases..., how many cases are there?

No idea yet. There are about 4000 at the moment but this number should be reduced by a lot eventually.

I learned a system 1/12th the size of this in about two months - and that was without any kind of organisation.

and what's the best way to recognise them?? corners first? and then edges? and if it takes a long time (1s) it will be fast too, because you don't have to recognise 2 things, and the algrhitms, are fast too

Dunno, however you like really. I don't personally find 1LLL hard.

I doubt this method will be as good as normal OLL/PLL.

There are many benefits to this system over OLL/PLL, but if you have trouble recognising fast enough or learning all the cases OLL/PLL might be a better system for you.

You click on the OLL alg type

lol

try reading the rest of the thread

 

[irontwig]

If I would be crazy enough to learn this I'd probably want it organized according to case rather than alg. And maybe for learning this kind of system you might want to use something like Anki or a similair flashcard program, since there's no real need to drill the algs into muscle memory, since you probably already know how to do e.g. a Sune, and also you would get the cases you've trouble with more often than the rest rather than the random distribution when just doing solves.

 

[Kirjava]

If I would be crazy enough to learn this I'd probably want it organized according to case rather than alg.

It already is. Click on full table.

 

[MalusDB]

Anybody else having problems loading the full page? Or is it just me and my Uni internet sucking hard..? Kir thats a serious amount of leg work you put in on this. Fair play.

 

[MaeLSTRoM]

Anybody else having problems loading the full page? Or is it just me and my Uni internet sucking hard..? Kir thats a serious amount of leg work you put in on this. Fair play.

Trying to load the full table is some serious load on a connection. Earlier in the thread there is a zip of the full table, if you download that and open it locally it should be OK.

 

[Speedmaster]

I don't really understand how this full table works..., if i look at the case above, there are just 100 different combinations of algrhitms under it, what do they stand for?
Do you think, this LL system can be used for speedsolves? or should i just learn on set of the algrhitms for speedsolves like Feliks learned some ZBlls for the U cases (I think)?

 

[MaeLSTRoM]

I don't really understand how this full table works..., if i look at the case above, there are just 100 different combinations of algrhitms under it, what do they stand for?
Do you think, this LL system can be used for speedsolves? or should i just learn on set of the algrhitms for speedsolves like Feliks learned some ZBlls for the U cases (I think)?

Basically the full table is the list of possible ways to solve that case using the 2 shorter algs. Certain combinations have been seected and then sorted into the other pages, but not all combinations are good. If you're using the full table then find a combination you like.
As for learning other sets, you can learn whatever you like. doing this will be a /lot/ of effort and time.

 

[Speedmaster]

So I only have to learn one algrhitm of the hundred ones which are under the picture??

 

[AHornbaker]

Sounds awesome, but im not sure i completely understand it. Full OLL/PLL suggests 1197 cases (57*21) for 1LLL. The most efficient way to break these up into two groups would be roughly 34 in each as 34*34=1156 which would be a total of 68 algs, a lot more than 46. Also, i looked at the tables and didnt understand most of it. Would love to learn this, can someone explain it to me in simpler terms?

 

[Noahaha]

Sounds awesome, but im not sure i completely understand it. Full OLL/PLL suggests 1197 cases (57*21) for 1LLL. The most efficient way to break these up into two groups would be roughly 34 in each as 34*34=1156 which would be a total of 68 algs, a lot more than 46. Also, i looked at the tables and didnt understand most of it. Would love to learn this, can someone explain it to me in simpler terms?

There are 1156 possible OLL/PLL combinations, but there are actually well over 10,000 possible LL positions. What Kirjava has done is taken every LL position and given one of 46 algs to execute that will reach a position solvable by one of those same 46 algs.

What you didn't take into account is that 46 algs can actually reach 46*4 = 184 possible positions due to being performed at different angles.

So 184*184>#of LL positions.

 

[AHornbaker]

By that logic, you could use at the least 34 algs, so 46 sounds achievable