OLS input

[obelisk477]

Recently I have been looking into OLS, as I think it is the best possibility to 'improve' CFOP out of all the other additional sub-steps out there. I have also recently become obsessed with CubeExplorer, and so I am going to begin a massive undertaking which will hopefully (years from now) result in the completion of all 42 OLS sets. Why? Because I wanna.

This first thread is to get some input. I will be naming all of the subsets OLS-1 through OLS-42. This will of course involve me re-naming some existing subsets (VLS, RLS, OLS-FE) to their respective F2L case numbers found on the wiki and algdb.net, as I can't imagine that their is a more efficient method for doing so. So for connected pairs, the set will be OLS-1, separated pairs it will be OLS-4, and OLS-FE would be OLS-37. You get the idea.

Another bit of input I would like is which other F2L cases people would like to see OLS generated for before others. I want to get out the more popular, useable ones first. The one I plan on doing after 1 and 4 is OLS-23, because you neither use a formed or split pair to solve it if you're using the 'non-intuitive' algorithm which seems to be standard for that case. I will probably work on other non-trivial F2L cases after that as well.

I realize that OLS-1 has already been generated, but the ones I have found for it seem to be focused more on how easy they are to remember instead of speed/fingertrickiness. So I am doing my own again with this latter category in mind.

Once I get 3 or 4 full sets generated, I will begin a new thread that will keep all of them in one place and will update it as more come out.

I look forward to hearing your suggestions and input.

 

[qqwref]

Good luck. How many algs is this?

 

[obelisk477]

216 per case * 42 cases = 9072. But the ones that I'm really interested in generating (the ones I think are most useful) is maybe for 7 F2L cases, so about 1500. I'll probably give up after those, but right now I'm determined

 

[Escher]

Cases 11-18 should be somewhat interesting given the multiple different approaches to solving them that are within a couple moves of each other. 25/26 will also be interesting because the way it can be broken down will probably lead to some neat tricks for other cases (an R U' to insert the corner turns 9 into '25').

You'll probably find a lot of free solutions for other sets as you generate them, it might be useful as you go along to try and crack a good way of categorising cases (like Kir was doing with 2-alg 1-look LL).

 

[StachuK1992]

I suggest you write a program to do this for you.

 

[obelisk477]

I suggest you write a program to do this for you.

Nahhh It's the thrill of the hunt, you know? Also, I was wondering if you were planning on adding OLS stuff to algdb at anypoint.

EDIT: Other than WV, of course.

 

[mark49152]

I'd like to see the cases that otherwise take most moves for a simple F2L solution. So cases like 23/24 and the ones with both pieces in the slot.

Actually the both in the slot cases might be good because you could AUF and so only have 58 (?) cases per set. You could even opportunistically but deliberately solve a slot wrong to save some time then do this instead of OLL to fix it. That might be a terrible idea, I'm bracing myself for the flames

To narrow it further, you could start with cases where both pieces are in the slot, the edge is good but the LL edges are all bad, meaning you'd get a dot OLL if you solved the F2L case 2-gen.

 

[qqwref]

216 per case * 42 cases = 9072.

LOL. Well, maybe you can get BindeDSA to learn it...

 

[obelisk477]

LOL. Well, maybe you can get BindeDSA to learn it...

His sig does say 'no set too large' or something to that effect

 

[Lucas Garron]

Nahhh It's the thrill of the hunt, you know?

With a data set that large, I can pretty much guarantee you will have mistakes all over the place (I think I had mistakes on my first MGLS sheet, and that was only ≈100 cases). If you're comfortable with writing computer programs, I highly recommend writing some code for this. At the very least, make it possible to check if all your algs are correct for the given cases.

Cmoputers are also much better at boring stuff like reformatting tables of things.

 

[StachuK1992]

obelisk477: I'll give you access to add OLS yourself if you want? (when the next version comes out)

 

[obelisk477]

obelisk477: I'll give you access to add OLS yourself if you want? (when the next version comes out)

I'm interested. It'll be no where near complete to begin with, and if it requires me to know any javascript or css or html or anything, don't count on me being very independent with the uploading

 

[StachuK1992]

You don't have to know any code to do it. I'll bug you once I'm ready.

 

[Egide]

l recently finished generating algorithms for the 502 cases needed for ZZ-c (a tiny subset OLS), so l'd be glad to upload them on algdb or sent them to someone who could.

 

[StachuK1992]

[/algdb talk, let's stop polluting. email [email protected]]
Edit: but thanks!

 

[obelisk477]

It turns out that its not 216*42 exactly (its less, thank goodness), because not every f2l case has 216 algs each. For example, the flipped edge case (F2L 37) should just be 8*4*2=64 cases after AUF, and even less after excluding mirrors.

Anyone care to calculate what the total is? I'm not so great with calculating permutations and orientations and excluding mirrors and whatnot, plus I have alot of alg generating that I'm working on.