OLS (VLS + HLS) Algorithms: by Mats Valk and Rowe Hessler

[rowehessler]

Over the next few weeks, Mats Valk and I will be uploading all of our OLS(OLL + Last Slot) algorithms onto our new youtube channel. Mats came up with this idea in 2009 and has been using OLS cases for years, but never published them. I eventually came up with the same idea, and published my algorithms on rowe.cubing.net. When Mats saw this, he contacted me, and we have been working together ever since.
Our OLS method is broken down into two parts:

1. VLS(Valk Last SLot): these are 216 algorithms, created by Mats Valk in 2009, that will force OLL skip with the last pair connected(solvable with R U' R'). There are 8 subsets of VLS, each containing 27 algorithms and based on edge orientation: Winter Variation(developed by Lucas Winter in 2005), UF, UL, UB, UFUL ULUB, UFUB, and all edges unoriented.

2. HLS(Hessler Last Slot): these are 216 algorithms, created by myself in 2013, that will force OLL skip with the last pair broken(solvable with R U R'). There are 8 subsets of HLS, each containing 27 algorithms and based on edge orientation: Summer Variation, UF, UL, UR, UFUR, UFUL, ULUR, and all edges unoriented. These algorithms will be released on our channel once we have finished uploading VLS.

Here is our introduction video:

Mats and I will be alternating with the uploading. So far I have uploaded Winter Variation and VLS(UF), and he will upload VLS(UL) and VLS(UB) within the next few weeks. I hope you enjoy the channel!

 

[Tim Major]

Sweet. Good videos :tu

I know some random skips, so I might add in some more when I feel like getting faster at 3x3. Practising is boring

First subscriber, I'll be keeping track of these videos.

 

[Bhargav777]

<3 Hopefully should be the first one to finish after the series is out
gj RoVa!

 

[Andreaillest]

Awesome! I'm going to keep a diligent eye on the videos for sure.
gj Mats and Rowe.

 

[stoic]

Woah...is this where CFOP is headed now?
OLL...PLL...then just another 432 algs to learn!
How many of these do you guys know now? have you learned all of your own sets? Each others?
Great job anyway

 

[rowehessler]

Woah...is this where CFOP is headed now?
OLL...PLL...then just another 432 algs to learn!
How many of these do you guys know now? have you learned all of your own sets? Each others?
Great job anyway

i know most of VLS, mats may know all of it, I'm not even sure. we only know a few easy cases for HLS.

 

[mark49152]

So this saves the RUR' or RU'R' and AUF - any other advantages? How is recognition? How does move count compare to OLL? It would be great to see some more analysis, and maybe a derivation of the 10/20/50 most worthwhile cases based on move count difference.

 

[rowehessler]

So this saves the RUR' or RU'R' and AUF - any other advantages? How is recognition? How does move count compare to OLL? It would be great to see some more analysis, and maybe a derivation of the 10/20/50 most worthwhile cases based on move count difference.

not sure what you mean by saving the R U R' and R U' R' and AUF. Advantages: you skip OLL a lot. These algs are for the UFR slot, so if your last f2l pair goes into UFR, with a few set up moves, you get either a VLS or HLS and you force an OLL skip every time. Nothing can be said about HLS right now, because I'm in the process of making it. VLS however, is incredibly fast. If you watch mats' solves closely, he using them quite a bit and gets fast singles, around 1 in 3 solves id say. The recognition is tough at first, but after doing it for a year now, its just as fast as seeing an OLL. one disadvantage is that there is no way to influence your pll, like COLL or OLLCP. another disadvantage is that its a crap ton of algs, and it'll probably be just as tough to get fast with OLS as it is to get fast with ZB.

The purpose of this is just to show the world what we've generated. I have no idea how fast this can be, but I've enjoyed generating them and they're fun to learn.

 

[scottishcuber]

Have you notice any improvements in your times?

I assume because you don't use it every solve you probably just get a few faster singles here and there, rather than whole averages.

 

[Bhargav777]

not sure what you mean by saving the R U R' and R U' R' and AUF. Advantages: you skip OLL a lot. These algs are for the UFR slot, so if your last f2l pair goes into UFR, with a few set up moves, you get either a VLS or HLS and you force an OLL skip every time. Nothing can be said about HLS right now, because I'm in the process of making it. VLS however, is incredibly fast. If you watch mats' solves closely, he using them quite a bit and gets fast singles, around 1 in 3 solves id say. The recognition is tough at first, but after doing it for a year now, its just as fast as seeing an OLL. one disadvantage is that there is no way to influence your pll, like COLL or OLLCP. another disadvantage is that its a crap ton of algs, and it'll probably be just as tough to get fast with OLS as it is to get fast with ZB.

The purpose of this is just to show the world what we've generated. I have no idea how fast this can be, but I've enjoyed generating them and they're fun to learn.

One more disadvantage
>Learns few cases
> Hopes to get it in an average
> Gets the mirror case of what was learnt.
> Rage quit.

 

[mark49152]

not sure what you mean by saving the R U R' and R U' R' and AUF.

I mean you execute n moves for the OLS alg, as opposed to 3-4 moves for the insert plus m moves for the OLL. So n has to be no more than 1-2 greater than m, to give a meaningful saving in terms of move count, allowing for the fact that with 432 algs it will be harder to reach the same tps as with 57 well-practiced OLLs. It would be really interesting to analyse OLS case-by-case and identify those cases with the smallest n-m, assuming the OLL reached by the simplest insert. Those would be the ones to learn first

For example, the reconstruction of Mats' 3x3 in his 2-4 relay, posted today. There he does a 10-move OLS including the AUF. The alternatives would be AUF + sledgehammer (+ AUF) + 9-move OCLL = 14-15 moves or RU'R' + 9-move lightning bolt = 12-13 moves. I'd guess that the OLS is probably only marginally better than the latter - although at Mats' level, marginal makes all the difference

Advantages: you skip OLL a lot.

It's probably just semantics, but it's not really an OLL skip if you still have to work for it. A simple 5-move alg instead of a 3-move insert could be called forcing an OLL skip, I guess, given that the extra couple of moves is more trivial than any OLL. But a 9-10 move alg? That would be more accurately described as combining OLL and LS, IMHO.

That's why I'm curious whether there are other advantages (other than move count which is easy to measure) for example in better lookahead or prediction of cases, or overall flow of the solve. Since it's developed by two of the fastest cubers in history it's got to be good, right? .

The purpose of this is just to show the world what we've generated. I have no idea how fast this can be, but I've enjoyed generating them and they're fun to learn.

Thanks for generating and posting these and for the discussion. LS shortcuts are great fun

Have you thought of generating a similar set but for hard LS cases, to save maybe 7-8 moves building the pair? For example where both LS pieces are in the slot but one or both is misoriented? Moves could also be saved by deliberately inserting an earlier pair wrong and fixing during OLS (like EJLS but extended to OLL rather than just corners) and that would also increase the probability of the algs being used. There would be fewer cases given that neither LS piece would be in the top layer (5*58, I think, roughly).

 

[Yellowsnow98]

Hmmmm...
Maybe I won't bother finishing OLL.

 

[KongShou]

CFP?

 

[Iggy]

Thanks for making this. I'm gonna start learning WV.

 

[TDM]

I already know the entire first video.

 

[fastcubesolver]

Awesome!

 

[Robert-Y]

Here are my algs for RUR' + OCLL:

000
U2 R' D' r U2 r' D R

001
R U R D R' U R D' R' U' R'
002
R U' L' U2 L U L' U L R' or SA
010
R U' R' F' R U2 R' U2 R' F R or SA
020
SA
100
SA
200
U2 R L U' R' U L'

012
R U' R' U' R U R' U' R U' R'
021
R U' R' U R' D' R U2 R' D R2 U' R'
022
R U2 R' U R U2 R' or L' R U R' U' L
101
R U' R' r' F R F' M' U' R U' R'
102
R U2 R' U' L' U R U' R' L
110
R U R D' R U' R' D R U R
120
R U' R D R' U2 R D' R2'
201
R U' R' U' R U' R'
202
L' U2 L U L' U L R U2 R'
210
U2 R U' R2 D' R U' R' D R or SA
220
U L' R U R' U R U' R' U2 L

111
R U' R' U2 L' U R U' R' L or SA
112
U' R U R D R' U R D' R2'
121
R U2 R D R' U' R D' R2'
122
SA
211
SA
212
R U' R' U' R U' R2 U' R U' R' U2 R
221
R U2 R' U R U' R' U R U2 R'
222
U2 R L' U R' U' L U2 R U R'

(SA means standard approach, so just do the pair normally, then OLL). The first number refers to the orientation of the UBL corner, 2nd number: UBR, 3rd number: UFL.

I'm not really sure if it's worth learning HLS because I don't think you save that many moves. The reason why VLS is so useful in my opinion is because most decent OLL algs require you to take out a pair or two and play around with them, then return them to their rightful slots. You rarely use OLL algs which start like R U' R'. In fact, many of the algorithms I've listed require you to join the pair up first. Only roughly a third of my algs start with something else other than R.

 

[Mvcuber12]

Woah...is this where CFOP is headed now?
OLL...PLL...then just another 432 algs to learn!
How many of these do you guys know now? have you learned all of your own sets? Each others?
Great job anyway

I don't exactly know how many I know, but quite a lot, I think approximately 150 of the VLS set. I didn't learn them in any specific order but just when I thought an algorithm was useful to know or when I found on accident while I was playing with the cube.

So this saves the RUR' or RU'R' and AUF - any other advantages? How is recognition? How does move count compare to OLL? It would be great to see some more analysis, and maybe a derivation of the 10/20/50 most worthwhile cases based on move count difference.

The maximum number of moves you can save with this is in fact 7. RU'R' instert + RUR' + U + OLS = OLL, twice a 3 moves instertion plus an U turn makes 7 moves. However, I'm to lazy to calculate the actual average amount of moves that is saved, but I guess it's around 3~4. It doesn't look like much, but when you're turning on a speed of, let's say, 8 TPS, that's still around 0.4~0.5 seconds. Which is in my opinion definitely worth it if you want to be a good speedcuber.

About the recognizion, at first it feels really weird, but after some practice it's just like recognizing an OLL. I've been over the point for quite some time now where I can predict the OLS case while making my last pair, sometimes even while making my third pair.

Have you notice any improvements in your times?
I assume because you don't use it every solve you probably just get a few faster singles here and there, rather than whole averages.

I haven't seen an significant improvement on my average time. However, I've had so many times where I was able to turn a bad solve into a normal solve or a normal solve into a good solve just because I was able to skip the OLL with OLS. This makes me quite sure it really helped me improving my average time.

One more disadvantage
> Learns few cases
> Hopes to get it in an average
> Gets the mirror case of what was learnt.
> Rage quit.

Lol, a few things.
- Most of the times you can do cases intuit ve, so it's not really learning a new algorithm but just knowing how to do it.
- Even tough you won't get it every solve, it'll definitely help on the solves you can do it.
- A good cuber is able to mirror an algorithm without many problems.


To everyone, I might make a video soon with all the easy cases and therefore the ones I'd recommend to start learning with.

If you've found an algorithm that you think is faster or at least equal in speed than one we've used in a video or that is on the website, feel free to post and discuss about them here so we can have the best algorithms possible at one place.

 

[Kirjava]

is this some kind of new beginner method

 

[Bhargav777]

Lol, a few things.
- Most of the times you can do cases intuit ve, so it's not really learning a new algorithm but just knowing how to do it.
- Even tough you won't get it every solve, it'll definitely help on the solves you can do it.
- A good cuber is able to mirror an algorithm without many problems.


Also, I might make a video soon with all the easy cases and therefore the ones I'd recommend to start learning with.

If you've found an algorithm that you think is faster or at least equal in speed than one we've used in a video or that is on the website, feel free to post and discuss about them here so we can have the best algorithms possible at one place.

You're right. But that was OH I learnt a UF case that was already hard as it had F' since I got the left case, I had to rage quit. VLS seems epic!