The New Method / Substep / Concept Idea Thread

[Sue Doenim]

But Roux doesn't use OLLCP...

You can use OLLCP to fuse CMLL and EO in one step, but its not used very often due to the immense number of algs. I also just realized that it would be better to misorient all corners, like in ZZ-Blah, giving only Pi and H cases, instead of headlights, chameleon eyes, L, and solved. Only 60 algorithms would be needed, only 18 more than CMLL and 300 less than full OLLCP.

 

[shadowslice e]

You can use OLLCP to fuse CMLL and EO in one step, but its not used very often due to the immense number of algs. I also just realized that it would be better to misorient all corners, like in ZZ-Blah, giving only Pi and H cases, instead of headlights, chameleon eyes, L, and solved. Only 60 algorithms would be needed, only 18 more than CMLL and 300 less than full OLLCP.

I am well aware of Pinky pie and CMLLEO though I really don't think that either of them are better than straight CMLL with a tiny bit of edge control thrown in. Also, If you misorient all corners, you give an extra step to impede lookahead and you still have to implement 2 edges in the d-layer which means that you may really as well just build a full F2L

 

[Shiv3r]

Wow, nowhere near as long as I was expecting. The high chance of H/Z is still a problem if you want to stay fully 2-gen though (but isn't such a problem if you don't mind doing <M, U>).

Yeah, I'm learning ZZ-d(porky2 style) and also I'm learning these algs for OH. Its a really nice way to do 2-gen solving, because It's less algs than 2GLL(easier to recognize as well) and faster than 2gOCLL-EPLL.
One thing I noticed is that when I do this, I get a LL skip very often(about 1 in 4/5 solves or so). Isn't the chance of an EPLL skip more like 1/12 or something like that?

One of the inspirations for coming up with a faster thing than 2gen OCLL-EPLL was me trying to compete in the 2-gen category in the Reddit weekly competition and too lazy/Not liking recognition for 2gll algorithms.

Maybe 2-sided (E)PLL recognition may be a good Idea as well.

 

[Teoidus]

L/R centers
Dedge correction (place in proper orbits) + 2color separation on L4C
EO-COL+L4C (Like in Yau)
z'
Pair 4 edges at once with u2, (aka 4-4 pairing)
ZZF2L with EOCross

4x4 method?

 

[Teoidus]

Hm chainpairing is still a problem.

 

[Shiv3r]

Hm chainpairing is still a problem.

yeah... This is pretty much ZZ-Hoya/Z4-Hoya, which has been created already, and EOCross pairing is less efficient than normal EOline edgepairing.

 

[Teoidus]

yeah... This is pretty much ZZ-Hoya/Z4-Hoya, which has been created already, and EOCross pairing is less efficient than normal EOline edgepairing.

I don't know where you're getting these.. definitely not a hoya variant, I don't think there's a 4x4 method that performs 2-color reduction + orbit correction, and EOCross dedge pairing wont' be any less efficient than EOLine edgepairing. Unless you mean blockbuilding, in which case I think the inefficiency is compensated for by the fact that EOCross is QTM optimized (there is even data from gyroninja that says that EOCross is optimal QTM if you don't openslot)

My issue is that I dont' think there is a way to chain pair with u2s. One thing I can think to do is to place 2 "buffer" pairs and then pair 2 dedges at a time with u2s (so it'd be 2-2-2-2). I dont' think 4-4 is always possible.

 

[Shiv3r]

I don't know where you're getting these.. definitely not a hoya variant, I don't think there's a 4x4 method that performs 2-color reduction + orbit correction, and EOCross dedge pairing wont' be any less efficient than EOLine edgepairing. Unless you mean blockbuilding, in which case I think the inefficiency is compensated for by the fact that EOCross is QTM optimized (there is even data from gyroninja that says that EOCross is optimal QTM if you don't openslot)

My issue is that I dont' think there is a way to chain pair with u2s. One thing I can think to do is to place 2 "buffer" pairs and then pair 2 dedges at a time with u2s (so it'd be 2-2-2-2). I dont' think 4-4 is always possible.

I misread, I didn't see the orbit part. May be a nice Idea, but if Orbit means what I think it means it won't be pretty. How many algs for orienting edges? cuz it seems like Orienting dedges then pairing dedges seems a lot like doing dedgepairing twice... and yeah, the u2's would both be hard to chainpair and kill lookahead. EOpairing is probably a better Idea instead of almost edgepairing twice.

 

[TDM]

Yeah, I'm learning ZZ-d(porky2 style) and also I'm learning these algs for OH. Its a really nice way to do 2-gen solving, because It's less algs than 2GLL(easier to recognize as well) and faster than 2gOCLL-EPLL.
One thing I noticed is that when I do this, I get a LL skip very often(about 1 in 4/5 solves or so). Isn't the chance of an EPLL skip more like 1/12 or something like that?

One of the inspirations for coming up with a faster thing than 2gen OCLL-EPLL was me trying to compete in the 2-gen category in the Reddit weekly competition and too lazy/Not liking recognition for 2gll algorithms.

Maybe 2-sided (E)PLL recognition may be a good Idea as well.

Yep, it's 1/12. I'm sure you're not actually getting one in 1/4 or 1/5 solves - it's just a really noticeable thing to happen so you're more likely to remember it. Try leaving a comment after every solve with an EPLL skip and take a look at how many there are after 50-100 solves. You'll find it's not as many as you'd think.

I thought 2-sided PLL recognition was something you'd do anyway, it's easier when corners are already permuted

 

[mDiPalma]

@Shiv3r, it looks like you are doing 2 relatively low-cost, low-impact substeps during your solve (CP during first block, phasing during last insert). why not just combine them into a LS variant? the movecount will be comparable and it will give you more freedom during your blockbuilding.

sure, it's not true zz-d. but does that really matter?

 

[Y2k1]

@Shiv3r, it looks like you are doing 2 relatively low-cost, low-impact substeps during your solve (CP during first block, phasing during last insert). why not just combine them into a LS variant? the movecount will be comparable and it will give you more freedom during your blockbuilding.

sure, it's not true zz-d. but does that really matter?

So an extension to zz-orbit?

 

[mDiPalma]

So an extension to zz-orbit?

i heard u like cases so i gave u 3 cases for each case, so now you can solve cases while you solve cases

3*6=18, and many of them would just be the ZZ-orbit algs with sexy move somewhere inside

 

[Y2k1]

i heard u like cases so i gave u 3 cases for each case, so now you can solve cases while you solve cases

Im sorry to say i do not understand what you mean. Can you ex0lain please? Thanks

 

[mDiPalma]

Im sorry to say i do not understand what you mean. Can you ex0lain please? Thanks

i think you would only need 18 algs so phase while solving CP. 3 phasing cases per ZZ-orbit case (6 total).

Spoiler: meme

 

[Shiv3r]

i think you would only need 18 algs so phase while solving CP. 3 phasing cases per ZZ-orbit case (6 total).

Spoiler: meme

The actual method I am working on has been changed I am now working on a ZZ-D(recog is much better than CPLS recognition, if you use the porky2 method) LL variant where it is 2gWV+EPLL, which is less algs than 2gLL(31 vs, 85) and gives very nice OH cases due to 2gen-ness(the entire reason for me learning ZZ-D is because the OH potential)

 

[mDiPalma]

assuming last pair is made in U layer

if you use [RU] wv+epll
=.75+10.26+.75+12=23.76

if you just insert pair and use 2gll
=3+.75+13+.75=17.5

plus, setting up the pair in the U layer costs moves

I would say this approach is ~8 moves worse than just using 2gll

 

[Shiv3r]

assuming last pair is made in U layer

if you use [RU] wv+epll
=.75+10.26+.75+12=23.76

if you just insert pair and use 2gll
=3+.75+13+.75=17.5

plus, setting up the pair in the U layer costs moves

I would say this approach is ~8 moves worse than just using 2gll

I Think that 2gLL is hard to recognize(yes, I have tried). Same reason why people don't learn ZBLL(or even full ZZLL) because recognition is worse than COLL/EPLL or something similar. recog is worse enough, even, that you can count the people who know full ZBLL on your fingers. 2gWV+EPLL recog is comparable to a simpler OLL/PLL recog based on what you need to look at.

 

[Teoidus]

wut 2gll is not hard, you just recognize CO and look at 2 edges

 

[mDiPalma]

I Think that 2gLL is hard to recognize(yes, I have tried). Same reason why people don't learn ZBLL(or even full ZZLL) because recognition is worse than COLL/EPLL or something similar. recog is worse enough, even, that you can count the people who know full ZBLL on your fingers. 2gWV+EPLL recog is comparable to a simpler OLL/PLL recog based on what you need to look at.

yeah, i also dont think the main problem with 2gll is that it's hard to recognize. i think the main problem is that every alg is pretty much exactly the same thing (there is a huge learning curve, hard to remember which case is which). also, the 2gen state is usually deeper than the 3gen state, and you actively invest moves to get there.

 

[Y2k1]

i think you would only need 18 algs so phase while solving CP. 3 phasing cases per ZZ-orbit case (6 total).

Spoiler: meme

Lol thanks