Developing a better LL system

[Kirjava]

Maybe I missed something, but, by looking at the raw data table, you seem to have dozens of algorithms for each case. How are you (or anyone else) supposed to organize theses cases by algorithms if there is more than one alg listed for each case? I might be interested in helping, but I don't know where to start because of this.

What I've been doing is taking the list of all cases, and removing the cases that start with a specific algorithm. So first, I took all cases that start with Sune and seperated them into their own group. The rest of the combinations are ignored.

Sounds great. Are there any news on a potential recognition method?

Hm.

Not yet. I was going to look to find patterns after the data is organised in this format. However, I was talking to someone on IRC last night and after it's done this way will look into doing a flowcharting based system instead. I have a lot of ideas for this, and I can imagine by the time this is finished the data will be explorable in a plethora of different ways.

 

[Zarxrax]

What about applying this system to the 2x2 cube?
With only 40 cases for CLL, it could serve as a nice test of the idea before completing the whole 3x3 version. And I'm curious how nice the algs would be.

 

[Kirjava]

I've been able to do 'this system' on 2x2x2 for like 6 years. It's just niklas/sune/fruruf combinations.

 

[nickvu2]

I've been able to do 'this system' on 2x2x2 for like 6 years. It's just niklas/sune/fruruf combinations.

Do you have documentation for that!?

 

[Kirjava]

It doesn't really need to be documented. Most of the time you do CP then CO or CO then CP and it'll get it in two algs. You just have to learn the combos.

 

[irontwig]

Still working on this?

 

[Cubenovice]

Kir's brain is too busy

I prefer the "guide" to "brute forcing" optimal 2x2x2 and 2x2x3 anyway.

 

[Kirjava]

Still working on this?

Not as much as I was doing, as Nats is coming up and I've been doing a lot of practise.

However, I should get in some sort of releasable state soon.

Sorry for the delays. Life, etc. You're free to help out if you want to.

 

[Kirjava]

heh. thanks for the nudge.

sunk another hour or two into this today and have about 48% of cases complete.

by complete, I mean listed by alg. once this is fully done, I'll release it.

that'll be version 1! after that, work will be done on moving cases around, finding patterns, and removing isomorphic cases.

grindgrindgrindgrind

 

[StachuK1992]

Any updates?

 

[Kirjava]

I post updates when I have them

 

[dregerdoes]

When this is all together, it may turn out to be one of the most interesting ways to go about finishing it. The cases that could occur would be intriguing to learn and you would definitely have a new way that others wouldn't. If nothing else it is a way about finishing it with originality.

 

[Kirjava]

Update in this post

I'm getting tired of people assuming I'm not working on this and complaining that it is not finished, so I'm releasing my progress so far. This is a lot of work, stop complaining that it's taking so long. I have other things to do. I was going to hold off until I'd got more of it done, but cba to put up with people anymore.

What's done; Every case is listed under its initial alg. This was the bulk of the work and brings the system to the basic format for the 'learnable verson'.

What needs to be done; Further organisation of cases to reduce the number of sets, duplicate removal, smarter case grouping, case number and recognition reduction.

Def. the biggest project I've worked on so far - the grind combined with the nagging greatly increases apathy

TL;DR - still a work in progress, link is below

http://snk.digibase.ca/ll/

 

[miotatsu]

this is crazy, great job on all this!

 

[Brest]

Wow that's amazing. And daunting.

 

[MaeLSTRoM]

Wow...
Just wow...
Could you possibly put some images to show the second stage cases and the respective algs, so that it's easier to see if you've got the right angle for the first alg.

 

[Kirjava]

Wow...
Just wow...
Could you possibly put some images to show the second stage cases and the respective algs, so that it's easier to see if you've got the right angle for the first alg.

A mouseover would be cool but would be too much effort at this point.

Why would you have trouble getting the right angle?

 

[aznanimedude]

now that i actually see this, this is quite impressive yo~

 

[Kirjava]

While (if?) anyone is learning this and notices any improvements or things that can be removed or moved around, your input would be appreciated.

I was thinking of putting this thing in GitHub so we could collaborate? Progress would be a lot faster, but would anyone actually do this?

More ideas; ability to search for a specific case (I planned ahead and left info in the files in case I wanted to do this), AUF removal

 

[StachuK1992]

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.