CFOP Variation: CF

[ben1996123]

I'm currently trying to write a program to do this. So far I have an optimal (but slow) cross solver...

Edit:

U' L2 D F2 U' R2 U F2 R2 U F' R2 F' L' F' U' B L' F' R2 U'
20.704 seconds to find F R L2 D L

R' U2 R' D2 R B2 F2 U2 R F2 D' R2 D' B' L' U L2 U2 L2 R2
It's been 10 minutes and it hasn't found anything yet lolol Looks like a pretty bad cross though.

[evogler]

Hee hee, that was fun to read.

Several of you got it. It's just a program that checks all possible routes you could follow during F2L.

I tried to stick to what seemed "human". It's pretty clear that you'd use R U R', R U2 R', and R U' R'. I debated including sledgehammers. I stipulated that the program had to insert a pair if it had one it could insert in one "sequence", and no more than two "sequences" per pair were allowed.
Two points where it seems a little unhuman are some insertions where it chooses extra U2 instead of a Us, and also the crosses. I think for all of these I allowed crosses that were within 1 move of optimal. Obviously, the more crosses you have to start with, the more solves you can sift through, and the greater chances of finding lower move count results, but the less realistic it seems, and the more it feels like computer magic. (Btw, I just copied them in from Cube Explorer. It would be fun to go back and write my own cross solver, to have a "complete" cube program).

I might get around to posting some other results. I also have the program recording optimal OLL skips and just plain optimal F2Ls. E.g. for L F2 R' B2 L2 F2 R B2 F2 R F' L2 R2 B2 D B D2 L' U' R' B:
B U D F L U2 D' F2
R' U' R L U' L'
R U' R'
U2 R' U' R
L' U L
(Jperm)

I hope people had a good time with these solutions. I find them kind of uncanny. I'm not sure if I see much to learn from a human solving standpoint, but maybe.

ben1996123, if you want to trade notes, let me know.

[evogler]

I'm currently trying to write a program to do this. So far I have an optimal (but slow) cross solver...
U' L2 D F2 U' R2 U F2 R2 U F' R2 F' L' F' U' B L' F' R2 U'

U F D' L' U2 F2 D' R D R2
U' y' F' L F U' L'
U' R U2 R2 U R2 U' R'

(Depending on how much zeroing you can tolerate in your cross... )

[KingTim96]

lol aimbot. way to bring video gaming into this.

aimbot...

[qqwref]

Ooh, great stuff. Evogler, any chance of releasing the program? I'd be very interested to play around with it.

As for a cross solver: an optimal solver shouldn't be that hard, and if you generate solutions one by one with increasing length, you could just keep going until you find a good continuation.

[evogler]

It's just an ugly little python script, but I'd be happy to share it. I have a bit more tinkering to do first, then I'll put it up.

[evogler]

Okay, here's a script:
http://www.ericvogler.com/media/F2Lsearch.zip
I'm using Python 2.5. Hopefully it doesn't break in other versions.

[ben1996123]

So it's been a while, and I took a break from programming, but yesterday I started again and added some more stuff to my program. It's not finished yet (doesn't work if you need to AUF somewhere in the solve but it shouldn't be that hard to add, and some F2L cases don't work) but here's a solution that it found pretty much instantly (apart from the cross):

Scramble: F2 U2 R B2 F2 U L U L' R2 B L' R F U2 B' F2 R' D
Cross: F' R2 D2 F R'
F2L 1: F2 L F L2 U' L F
F2L 2: y R U' R' F' L' U2 L F
F2L 3: y R U' R' U R U' R'
F2L 4: y U F' U L' U2 L F

[evogler]

Good job!
So what's the collection of insertions you're using? Standard pairs of R(U/U'/U2)R' plus other you put in by hand? Or computationally discovered ones?

[ben1996123]

Good job!
So what's the collection of insertions you're using? Standard pairs of R(U/U'/U2)R' plus other you put in by hand? Or computationally discovered ones?

A giant alg table

Edit: oh and AUFs work now.