• Welcome to the Speedsolving.com, home of the web's largest puzzle community!
    You are currently viewing our forum as a guest which gives you limited access to join discussions and access our other features.

    Registration is fast, simple and absolutely free so please, join our community of 40,000+ people from around the world today!

    If you are already a member, simply login to hide this message and begin participating in the community!

Roux-breaker? The YruRU method

Cuberstache

Member
Joined
May 7, 2018
Messages
1,042
Location
Washington State, USA
WCA
2016DAVI02
YouTube
Visit Channel
What is the best way to go about learning all of the circlets? Is there anywhere I can see all of them or do I need to figure out myself?
Swapping 5 and 6 simply reverses the circlet (this was dismissed before but with more testing, it is in fact true). So there are only a few circlets you need to learn: if 5 and 6 are in 1-3, 2-4, 1-5, 2-5, 3-5, 4-5, 1-6, 2-6, 3-6, and 4-6. Look for some kind of visual thing to help you remember it. Like just how a line connecting the pieces in the right order would look. I'll film part three of my tutorial soon, in which I'll cover this sort of tracing thing as well as pEO.
 

Owen Morrison

Member
Joined
Aug 16, 2019
Messages
1,953
Location
Tennessee
WCA
2017MORR06
YouTube
Visit Channel
Swapping 5 and 6 simply reverses the circlet (this was dismissed before but with more testing, it is in fact true). So there are only a few circlets you need to learn: if 5 and 6 are in 1-3, 2-4, 1-5, 2-5, 3-5, 4-5, 1-6, 2-6, 3-6, and 4-6. Look for some kind of visual thing to help you remember it. Like just how a line connecting the pieces in the right order would look. I'll film part three of my tutorial soon, in which I'll cover this sort of tracing thing as well as pEO.
Thanks!
 

Devagio

Member
Joined
Apr 21, 2020
Messages
266
While we’re on this topic, here’s another trick that I guess I haven’t mentioned on this forum:

When DL corners are solved, we try to set up a pair of corners to
UFR-UFL: F’ U’ F
UFR-UBR: F’ U F or F R’ F’
UBR-UBL: F R F’

But here’s the entire list of where you can set up the pair of corners before doing the triggers:
F’ U’ F: UFR-UFL or UBR-DBR or UBL-DFR
F R F’: UBL-UBR or UFR-DFR or UFL-DBR
F’ U F: UBR-UFR or UBL-DBR or UFL-DFR
F R F’: UBR-UFR or UBL-DFR or UFL-DBR

I find this really fascinating, it really opens up so many possibilities! It reduces the average movecount of CP-line by one whole move (need to check that figure) because 80% of the cases will now have a 3 move CP solution. I haven’t used these in speedsolves yet, but I intend to very soon.

Another thing, I’ve sort of figured out better ways to handle that annoying case where DL corners are unsolved but CP is solved. It was obvious we could do better than the two algorithms we have used so far.

Set the DFL corner to be an F2 away from being solved and do F U F or F’ R’ F’.
There are a few more ways, I’ll edit this post and add the exhaustive list in a while.

I guess these are major developmental leaps right here!
 
Last edited:

Devagio

Member
Joined
Apr 21, 2020
Messages
266
Updates:

1. The EOBF section of the website is ready; that makes 4/7 sections. You can visit the home page here.
I hope to finish the F2L section in a day or two since there's nothing much to be done there. A lot of academic load got dumped on me all of a sudden, so the CP line and extension sections will have to wait.

2. I found myself improving fairly well in all parts of the solve, except the pEO extension. It would be wrong to say there wasn't any improvement, but I could feel it was sub-par. Rather than waiting more and trying different ways to improve at it, I'm trying to see how the step can be done differently, even considering freestyle pEO or something. I am extremely pleased with how EOBF has turned out simply by eliminating the DB blindspot, so the aim of pEO-extension will remain somewhat the same [LEOR OH people can make good use of this information as well, instead of planning FB + some EO, you could plan FB + EO of DB edge and see how that goes; in case that hasn't been tried seriously].
 

Devagio

Member
Joined
Apr 21, 2020
Messages
266
Apologies for multiposting, but I'm not sure how to go about it when this post is new information building on the previous post.
The F2L section is ready. I ended up adding quite a bit more than what I initially intended, but I think its for the best. You can visit the home page here.
I'll be taking a break from adding more to the website for now.
There's some awesome pEO stuff that I'll put up here soon enough; though first I'd like to know the thoughts of people who even briefly tried this method out with YruRU style CP; basically why you stuck to it or quit it. Simply put, feedback to see what worked for others and what didn't, and see if we can address those things.
 

Cuberstache

Member
Joined
May 7, 2018
Messages
1,042
Location
Washington State, USA
WCA
2016DAVI02
YouTube
Visit Channel
There's some awesome pEO stuff that I'll put up here soon enough; though first I'd like to know the thoughts of people who even briefly tried this method out with YruRU style CP; basically why you stuck to it or quit it. Simply put, feedback to see what worked for others and what didn't, and see if we can address those things.
I think the ergonomics were really intriguing and I wanted to try the method out. Even at first, before things were developed, I thought it could be the best one-handed method. As you already know, the extension is the hardest part of the solve, but apparently you've figured that out already (!!). My inspection is sub-15 now so YruRU CP seems to be the way to go.
 

Devagio

Member
Joined
Apr 21, 2020
Messages
266
I think the ergonomics were really intriguing and I wanted to try the method out. Even at first, before things were developed, I thought it could be the best one-handed method. As you already know, the extension is the hardest part of the solve, but apparently you've figured that out already (!!). My inspection is sub-15 now so YruRU CP seems to be the way to go.
Thanks! Knowing that you’ve been using the method for over a month now, a more in depth critique of what good and bad things you think of each step/transition would be really helpful. Personal opinions totally welcome.

but apparently you've figured that out already
Haha, wouldn’t say figured out, more like an experimental trick for now. Though glad to know that we’re on the same page that the extension step requires the most attention.
 

Cuberstache

Member
Joined
May 7, 2018
Messages
1,042
Location
Washington State, USA
WCA
2016DAVI02
YouTube
Visit Channel
Thanks! Knowing that you’ve been using the method for over a month now, a more in depth critique of what good and bad things you think of each step/transition would be really helpful. Personal opinions totally welcome.
CPLine: Inspection is good with practice. I can sub-15 basically every time now and look ahead to extension sometimes. It can be awkward with F moves and I have to do rotations sometimes, but obviously we have to start somewhere. Bad ergonomics for this short step make for really good ergonomics for the longer steps. The several options for tracing and solving CP make efficiency really good for almost any scramble. The F U F trick for 2GR solved CP was basically the last thing to make any case good. I guess now the only thing missing is when all corner pairs are twisted ;)

Extension/pEO: Finding the edges can be hard if they're in blindspots. I personally don't mind the u moves like others do. This step seems inefficient sometimes, compared to a movecount for a standard Roux FB. I do pEO completely intuitively, which is really easy, though I have gotten 8 bad edges a few times doing it intuitively. I don't have too much to say here; I'll wait to hear your new information.

EO: I still usually pause briefly to recognize the EO case. That should go away with more practice. There are a few cases I struggle to do optimally, but that's just me being lazy and not studying the EO algs posted earlier in the thread. Not much to say here, this is where ergonomics begin to be really nice, so high TPS is possible.

BF: Lookahead to this step can be hard sometimes, but I've noticed it getting better. Same with deciding how to do the last two moves of EO to influence this step. Usually I just do U' because it's easier, but I've had a noticeable improvement in doing a U if it's better or even placing the centers solved instead of D on U when there's one edge done already.

F2L: Not much to say here, the only real thing is that I rarely blockbuild, which is something I might have to change soon.

2GLL: Again not much to say. Here we get rewarded for the work we did at the beginning of the solve. I know almost full 2GLL now and I'll have it done by next week. I've been using Roman's ZBLL trainer to practice the algs, and that's been absolutely crucial to learning these algs well. Recognition is really fast for most cases and will get faster as I get more used to the algs.

I love this method :D
 

Owen Morrison

Member
Joined
Aug 16, 2019
Messages
1,953
Location
Tennessee
WCA
2017MORR06
YouTube
Visit Channel
CPLine: Inspection is good with practice. I can sub-15 basically every time now and look ahead to extension sometimes. It can be awkward with F moves and I have to do rotations sometimes, but obviously we have to start somewhere. Bad ergonomics for this short step make for really good ergonomics for the longer steps. The several options for tracing and solving CP make efficiency really good for almost any scramble. The F U F trick for 2GR solved CP was basically the last thing to make any case good. I guess now the only thing missing is when all corner pairs are twisted ;)

Extension/pEO: Finding the edges can be hard if they're in blindspots. I personally don't mind the u moves like others do. This step seems inefficient sometimes, compared to a movecount for a standard Roux FB. I do pEO completely intuitively, which is really easy, though I have gotten 8 bad edges a few times doing it intuitively. I don't have too much to say here; I'll wait to hear your new information.

EO: I still usually pause briefly to recognize the EO case. That should go away with more practice. There are a few cases I struggle to do optimally, but that's just me being lazy and not studying the EO algs posted earlier in the thread. Not much to say here, this is where ergonomics begin to be really nice, so high TPS is possible.

BF: Lookahead to this step can be hard sometimes, but I've noticed it getting better. Same with deciding how to do the last two moves of EO to influence this step. Usually I just do U' because it's easier, but I've had a noticeable improvement in doing a U if it's better or even placing the centers solved instead of D on U when there's one edge done already.

F2L: Not much to say here, the only real thing is that I rarely blockbuild, which is something I might have to change soon.

2GLL: Again not much to say. Here we get rewarded for the work we did at the beginning of the solve. I know almost full 2GLL now and I'll have it done by next week. I've been using Roman's ZBLL trainer to practice the algs, and that's been absolutely crucial to learning these algs well. Recognition is really fast for most cases and will get faster as I get more used to the algs.

I love this method :D
One thing that I don't understand is the pEO step, I thought you were just supposed to expand the 1x1x3 into a 1x2x3 but it seems like you are supposed to orient some edges while doing that. How do I do that?
 

Cuberstache

Member
Joined
May 7, 2018
Messages
1,042
Location
Washington State, USA
WCA
2016DAVI02
YouTube
Visit Channel
One thing that I don't understand is the pEO step, I thought you were just supposed to expand the 1x1x3 into a 1x2x3 but it seems like you are supposed to orient some edges while doing that. How do I do that?
Make sure white/yellow isn't on top by making some moves wide moves, and just put a good edge into DB after the extension with (U/U'/U2/nothing) r. Check this out, and read the rest of the page too.
 

Devagio

Member
Joined
Apr 21, 2020
Messages
266
I guess now the only thing missing is when all corner pairs are twisted
Forgot to mention this one :p I haven’t updated that post that I said I will “in a while”.
You could do normal tracing, put a swap pair in UFL and DFR (this will take 0 or 1 move 94% of the time) then do one of the following 2gen triggers based on the orientation of the twist: F2 R’ F R’ F R’ F or F2 U F’ U F’ U F. Will give you an 8 move 2gen CP-line; probably bearable.
This step seems inefficient
Yeah, working on pEO-extension. I’m getting a feeling it’s not that difficult an issue to resolve, but haven’t really found a great way to do it. I’ll put up the little trick I’ve been using; though my plan is to write a code to generate say 100 scrambles and solve pEO optimally; and see what patterns I can find.
I still usually pause briefly to recognize the EO case.
I’m not sure if that’s something that’ll go away completely; at least not with the current pEO approach. I feel it’s something like OLL recognition, like how after F2L you somewhat know which OLL is coming up, but usually sort of need to pause just a little to exactly know which one.
Also, use the algs on the website. They’re organised in a much better way, plus I think the visual representation helps.
I love this method :D
Yay!
Stuff I just realised: there’s 3 past posts that I said I will update but haven’t; the inspection post (though I don’t need to anymore, the rest of the stuff I just put up here), the 500 alg method post, and the of course the pEO trick post. Let me just make those right here.

pEO Trick:
If done right, you usually will not have to use Rw moves during the pEO extension step. You may have to use it “before” the step to set up the centres, and “after” the step to put in DB; but during the step, due to the lack of Rw moves, the DB and DF edges do not move. So during the step (or maybe even during inspection or CP line) check out the orientation of the DF edge. Half the time, you will get off simply by doing and Rw’ after extension. If you want to do more pEO, you can put a good edge in UL before doing the Rw’. Just an awesome idea to play with till further development; basically gives hope that there will probably be such a simple General way to do this step.

~500 alg method:
Not sure whether this method is worth it; though it certainly is much more efficient than YruRU can ever hope to be.
The general idea is pretty obvious:
1. Make CP line
2. Somehow solve F2L
3. Solve LL in 1 alg
Since CP is already done, the LL cases will be reduced, much like ZBLL; but it’s better here because all the algorithms will be [r, R, U]-gen which is a great moveset for OH, the alg length will be about the same, and the recognition is better than ZBLL (ZBLL has 1 orientation and 2 permutations to recognise, this thing here has 2 orientations and 1 permutation to recognise, orientations are easier to recognise than permutations, for eg OLLCP is easier to recognise than ZBLL).
Once we figure out how to do CP (YruRU CP takes care of that, and CP-line here will be more efficient than YruRU since we don’t care about EO at any point during the solve, so x2 y neutrality will be the way to go) the only non-trivial part is to figure out the best way to solve F2L (Unlike ZZ where it’s pretty obvious we gotta do LB and RB).
I have a couple of candidates on how to do F2L, none of them particularly great; though I can totally see this method averaging around 45 movecount in brain dead speedsolves; the last 15 moves of which will be an ergonomic algorithm, the first 5 moves will be planned, and the middle 25 moves will have [r, u, R, U] moveset.
The reason I do not expect much out of this method is mainly the fact that I do not like high alg count methods, and that I feel methods with asymmetries are just bad (the F2L part will most likely have some asymmetrical component). Though from an objective point of view, it’s probably worth a try.
If someone is interested in taking up developing this, I’ll be happy to assist if required.
 

Devagio

Member
Joined
Apr 21, 2020
Messages
266
Analysed over hundred HTM and QTM optimal solutions of pEO extension after the line is made. Just to be clear, by pEO extension, what I mean here is extending the 1x1x3 line on DL to a 1x2x3 block on left, having either white or yellow centre and top, while placing a visible edge in DB with correct orientation. Here, a visible edge is some edge that was in in the U layer or in DF at some point during the step (which means even if extension is done, centres are proper, and DB edge is oriented, we may not know that the DB edge is oriented and so would have to add something like say U r2; this has been incorporated in finding the optimal solution).

First things first, the setup > extension > pEO, while being a great stepping stone, is not all that efficient. The algorithmic approach and easy recognition may appeal to some; plus ensuring 8 bad edges never comes up and 6 bad edges comes up very infrequently is a big positive. However, with DB edge oriented, 6 edge cases aren’t really worse than 2 or 4 edge cases recognition or execution wise, and there is only one possible 8 bad edges case, which is still really rare and avoidable.

That said, after a fairly long consideration I believe freestyle pEO extension is the way to go (Simply doing extension is not a good idea because EO is horrible). There are some neat tricks that become apparent with practice, I’ll try to be active on the example solves thread and try to cover as many of them as I can; there aren’t very many. I tried to categorise them in some sense but I guess it’ll have to be like F2L tricks, you see it and you do it on a case by case basis.

Here are a couple of thumb rules (in order of priority) to get you started:
1. If the centres aren’t proper and DF is “misoriented”, then most often an optimal or almost optimal solution will include exactly one Rw’ while doing the extension, and no other wide R move.
2. Often, extension followed by an Rw or Rw2 can optimal / almost optimal; so if you cannot find a good way to put an oriented edge in DB or make centres proper during extension, simply track candidate edges for DB during the execution of extension, and follow up with one of the said moves.
3. U R u2 inserts are preferable over u R u inserts due to the freedom between R and r in the former type of inserts. Another kind of insert which is useful when the left centre is in U or D face is a “u r u” type insert, where one edge is connected to the centre and another is in the E slice.
Optional: If you wish to preserve the orientation of some edge (which will typically be on the U layer since you saw it), bring it to UL; it will likely stay oriented for the rest of the step. This will ensure 8 bad edge cases do not pop up. Also, it will reduce the likelihood of 6 bad edge cases, though with DB oriented, this isn’t annoying anymore.
Optional: if you have a choice, you can try to have DF misoriented at the end of the step, since EO algs are much better this way.

The best way to learn this will definitely be through example solves and practice, just like once down with basic F2L, advanced F2L is best learned through examples and not algsheets.

Side note: I tried considering whether CPFB is at all viable, since that’ll open up some good possibilities; while you will have a bigger pause for EO with possibly worse efficiency, making it not suitable for rRU>RU reduction, it may be possible to use it for rRU>MU reduction effectively. Checked out some ways I found proposed on a few threads, tried to come up with my own, etc. however, there is just too much to do in inspection that it seems ridiculously hard to achieve, let alone do it consistently. It may be my lack of foresight, however, I do not see myself being able to do it with any conceivable method that may even pop up in the future; since it will at the very least involve planning FB, identifying CP, and calculating what moves to add to solve CP without destroying FB; it’ll at best be a gamble to try and fit it in.

PS if someone wants to play around with the code (.py) to find the optimals they can DM me their e-mail and I’ll send it over, though honestly it’s a beginners’ exercise. Just brute force it, since the only possible moves are r, R, u and U; and most solutions are like 6 moves long.
 

Devagio

Member
Joined
Apr 21, 2020
Messages
266
At long last! Done with the website!!
Here it is.

Whatever your thoughts about the method, just check out the website once. Took really a lot of effort to make that, though learned a lot about version control and front end too.
There are most likely many errors in there which I will resolve over time as when they are pointed out; also advanced/intuitive pEO extension page hasn't been put up since I myself need more experience with it before putting it there.
 
Last edited:

Owen Morrison

Member
Joined
Aug 16, 2019
Messages
1,953
Location
Tennessee
WCA
2017MORR06
YouTube
Visit Channel
So I have been trying to switch to the advanced way of doing CP, and as I was memorizing the circlets there is one that never solves CP for me.


Screenshot 2020-06-08 at 3.49.48 PM.png

If 5 is in UFL, I have been reading the 'N' from the UBL corner to the DFR corner, and if 6 is in UFL then I read it from DFR to UBL. Am I making a mistake or is this circlet incorrect?

Oh and btw your website looks awesome!
 

Devagio

Member
Joined
Apr 21, 2020
Messages
266
So I have been trying to switch to the advanced way of doing CP, and as I was memorizing the circlets there is one that never solves CP for me.


View attachment 12489

If 5 is in UFL, I have been reading the 'N' from the UBL corner to the DFR corner, and if 6 is in UFL then I read it from DFR to UBL. Am I making a mistake or is this circlet incorrect?

Oh and btw your website looks awesome!
Thanks!
Well, you are doing it correctly, and this is what I do too; probably making a mistake elsewhere.
 

Owen Morrison

Member
Joined
Aug 16, 2019
Messages
1,953
Location
Tennessee
WCA
2017MORR06
YouTube
Visit Channel
hmmmmm I am trying it over and over again and It never works:confused: :? I can't find anything that could be making it not work. I will keep trying and update this post if I figure it out.
Okay so I found out that if I have 5 in UFL and 6 in UBR if I have the circlet UBL>DFR>UFR>DBR it works and solves CP. Not really sure what is going on.
 
Last edited:
Top