What 3x3 alg set do you want improved, optimized, or generated?

[pineapplepie]

I learnt 4lll on r, so it's faster for me. Also, I find that I can AUF faster when LL is on right.

 

[Sub1Hour]

I learnt 4lll on r, so it's faster for me. Also, I find that I can AUF faster when LL is on right.

LL on U is much faster for recog and the algorithms are better and more developed. LL on right is a bad habit that you should break ASAP and will only hinder your times.

 

[whaffle]

LL on U is much faster for recog and the algorithms are better and more developed. LL on right is a bad habit that you should break ASAP and will only hinder your times.

Around half of my PLL algorithms are on right, and from my experience, recognition is not really that slow. If you use 2-sided recognition, recognition time should be about the same. Also, if you look at Teoidus' PLL on right algs, they're sometimes shorter than the speed optimized PLL on U algs. Also, since the algs are mostly <R,U,D>, I find them easier to fingertrick than some of the normal PLLs.

 

[Devagio]

No such algset exists yet afaik but would probably a big thing in the future for OH: 1LLL where CP is already solved. All algs will be rRU gen, and there will be like 500 of them.

 

[Humble Cuber]

Over the years I have learned many algorithms and learned to use a couple generators, specifically for 3x3, so because I want to use that knowledge but ran out of ideas on what to do I'm asking you, what 3x3 algorithm set do you want improved, optimized, or even generated? Any method, any idea, however I am requesting it is not some alg set in the thousands like 1LLL or OLS.

Zblls because they have so much room to improve and become faster / more efficient

 

[AlphaCuber is awesome]

Idk how many algs this is but move optimal PBL would be cool.

 

[Sub1Hour]

Idk how many algs this is but move optimal PBL would be cool.

2x2, or square-1, or 3x3? Specify the puzzle, please!

 

[OreKehStrah]

Would you mind making WVCP?
[

No such algset exists yet afaik but would probably a big thing in the future for OH: 1LLL where CP is already solved. All algs will be rRU gen, and there will be like 500 of them.

IDK about this specifically but I know a similar set exists where all corners are oriented. Chris Tran mentioned some professor that knows all of this set and said he calls it anti-zbll

 

[ProStar]

Would you mind making WVCP?

A mod must've deleted my post, I guess they didn't look at what I linked. Anyway, here's a sheet that contains WVCP algs

 

[AlphaCuber is awesome]

2x2, or square-1, or 3x3? Specify the puzzle, please!

Or you could just read the thread title.

 

[Sub1Hour]

Or you could just read the thread title.

Well, looks like I goofed again. There should be around 400 algorithms not including "M2 U2 M2 Parity", forgive me for my transgressions for I have done a bad.

 

[kubesolver]

@AlphaCuber is awesome
I tried to generate move optimal PBL and after generating several thousands of them it feels pointless - there are so many cases it hurts. What is the purpose?

 

[AlphaCuber is awesome]

@AlphaCuber is awesome
I tried to generate move optimal PBL and after generating several thousands of them it feels pointless - there are so many cases it hurts. What is the purpose?

How many algs is it? I just thought it would be useful as an option for FMC with DR sometimes if you have a situation where you could cancel into it or something.

 

[kubesolver]

How many algs is it? I just thought it would be useful as an option for FMC with DR sometimes if you have a situation where you could cancel into it or something.

It's (72 PLLs on top x 4 AUF ) x (72 PLLs on bottom x 4 ADF ) x 2 (PLL parity)
72 * 4 * 72 * 4 * 2 = 165888

 

[Stryder73]

Hls, I can only find all or no edges oriented.

 

[Devagio]

It's (72 PLLs on top x 4 AUF ) x (72 PLLs on bottom x 4 ADF ) x 2 (PLL parity)
72 * 4 * 72 * 4 * 2 = 165888

Are you sure?
If you discount AUF and ADF, I believe there will only be 22x22x2-1=967 cases. If you’re allowed to do z2 rotation, then this is further cut down to 22x23-1=505 cases.

 

[kubesolver]

Is true you can cut it by two with z2.
But you can't discount for auf's especially in the fmc context.

 

[Devagio]

Is true you can cut it by two with z2.
But you can't discount for auf's especially in the fmc context.

Agreed. I wasn’t sure how you got 72 PLL cases though, is that a typo?
Edit: my mistake, I misunderstood what PBL means.

 

[Sub1Hour]

Regarding PBL, some PBL is useful on square-1 due to the nature of the puzzle but learning the whole set seems a little bit ridiculous, especially without CSP since it would be over 1700 cases. I'm not sure if anyone knows full PBL on square-1 as of right now but I don't think that it would benefit 3x3 at all since it's not that hard to make a complete first layer. Even from an FMC standpoint its not worth it since we have better methods than CFOP+PBL and it would be much easier to use something like NISS or comms rather than learn 800 algorithms.

 

[teboecubes]

Speaking of PLL, I still feel like PLL has a lot of room for improvement. Thirty years from now (or likely much less), I wouldn't be surprised if we've found superior replacements for more than 40-50% of the current "best" PLL algorithms (assuming we still use CFOP and that PLL is still necessary).

Maybe ones like the F, V, N perm, but the vast majority of them (A, H, Z, U, T, Y, J, etc.) I can't see being more optimized than they already are