PCMS Discussion Thread

[Sion]

Hello!

I decided to use PCMS as my primary method for speedsolving (especially after re-discovering a more amateur form of the method), and I need some help with L5EOP.

I understand the concept that you orient all the edges and permute the FD edge to get an ensured EPLL case, however, when I use such algorithms, I usually end up with an [annoying] OLL case.

I'm trying the algorithms on this page: https://www.speedsolving.com/wiki/index.php/L5EOP , yet when I attempt, it somehow never works.

I also considered L5EP, but I can't seem to decipher the diagrams of the cube: https://www.speedsolving.com/wiki/index.php/L5E .

I deeply appreciate your help. Thank you!

 

[Tao Yu]

Have you tried clicking on the links on the algs to see the animations? If you do the moves exactly as shown by the animation it should work.

Also, just as a quick check, are you doing M/M' moves right? M' goes up (it follows R) and M goes down (follows R'). That's a common mistake.

To be honest, I recommend learning Roux LSE instead. So just solve RD and LD then use LSE from there (don't place the BD edge). It's a much more optimised method at this point, and it's completely intuitive and easy to learn (loads of resources available). See waffo's page or Roux's page for more.

I gotta say though, I think Roux is a much better method than PCMS. I've written my thoughts about PCMS vs Roux here.

 

[Sion]

Thanks Tao!

Would It make sense to put down the RD and LD edges then solve LSE for a PCMS-Roux hybrid? And, can you do an average between that and normal Roux so I can see?

 

[Tao Yu]

Thanks Tao!

Would It make sense to put down the RD and LD edges then solve LSE for a PCMS-Roux hybrid? And, can you do an average between that and normal Roux so I can see?

Yeah, that's actually exactly what I mean. I think it would be faster than normal PCMS. I also think it would be faster to solve the corners after placing RD and LD, because the CMLL case stays the same after you solve the 4 CE pairs, and that allows you to recognise the case while simultaneously placing the two edges, giving you zero recognition time for the CMLL case. There isn't much benefit in having the S slice free when solving the top corners, the algs are going to be close to the exact same because very few algs use S/S' or f/f' moves anyway.

If I were to do an average with both methods, I'd be way faster with Roux for sure. I don't think it would really prove anything other than the fact that I've had a lot more practise with Roux. As I've said in my reddit post though, I think Roux will always be faster than PCMS/this hybrid method.

 

[Deleted member 51471]

Would this be viable as a PCMS variant?
1. Columns, same as in PCMS
2. EO
3. Two opposite edges on bottom
4. Two opposite edges on top
5. Kind of like roux-4c, but the layers at the right and the left might be like pseudo blocks. Solves pretty much like roux 4c.

Of all the PCMS variants, I am fastest with this.

 

[Cubinwitdapizza]

Would this be viable as a PCMS variant?
1. Columns, same as in PCMS
2. EO
3. Two opposite edges on bottom
4. Two opposite edges on top
5. Kind of like roux-4c, but the layers at the right and the left might be like pseudo blocks. Solves pretty much like roux 4c.

Of all the PCMS variants, I am fastest with this.

Lol nice 4 year bump. I don’t think this was supposed to be a pcms discussion thread just a quick question.

 

[Deleted member 51471]

Lol nice 4 year bump. I don’t think this was supposed to be a pcms discussion thread just a quick question.

It's the best thread I could find for this.

 

[Cubinwitdapizza]

It's the best thread I could find for this.

You should probably propose that in the method thread.
Also, if you want pcms help, you might as well make a pcms discussion thread.

 

[Sion]

You should probably propose that in the method thread.
Also, if you want pcms help, you might as well make a pcms discussion thread.

Updated title.

My main worry is how to limit regrips and cube rotations. if I could control that, I might consider practicing pcms more.

But for now, I'm sticking with learning CMLL for roux. My blockbuilding has been getting better, but I miss when I could inspect 3 columns :/

 

[Deleted member 51471]

Updated title.

My main worry is how to limit regrips and cube rotations. if I could control that, I might consider practicing pcms more.

But for now, I'm sticking with learning CMLL for roux. My blockbuilding has been getting better, but I miss when I could inspect 3 columns :/

When you used PCMS would you use the Roux LSE variant?

 

[Sion]

When you used PCMS would you use the Roux LSE variant?

I still use the L5EOP variant, still a bit of a habit for me.

however, I might start experimenting with LSE. The main worry would be making sure the cubies to make the blocks aren't in awkward positions.

 

[ProStar]

I still use the L5EOP variant, still a bit of a habit for me.

however, I might start experimenting with LSE. The main worry would be making sure the cubies to make the blocks aren't in awkward positions.

Me and WCC were working on the LSE variant a couple months ago, and both being pretty bad at Roux & PCMS we averaged around 55 STM. Also I've gotten some mid 40 solutions with it without much practice at all

 

[ProStar]

Whenever I'm going for low movecount PCMS, I almost always end up with the case where two U layer edges are swapped adjacent and two D layer edges are swapped adjacent. What's the move optimal way to solve this(don't care about fingertricks, STM)?

 

[xyzzy]

Whenever I'm going for low movecount PCMS, I almost always end up with the case where two U layer edges are swapped adjacent and two D layer edges are swapped adjacent. What's the move optimal way to solve this(don't care about fingertricks, STM)?

S' M' U M U2 S? There's also F2 U' (R2 F2)3 U F2 which is optimal in FTM but not STM.

 

[kubnintadni]

What options are there for finishing a PCMS solve? Here are the ones I've been able to find either on the wiki, here, or just thinking (alg counts will likely be in error for many since I was unable to find documentation for some of these):

There are probably worse methods section:

On the Wiki:
1. Solve 3 CeE pairs on D.
2. L5EO (5 algs)
3. L5EP (16 algs)

Why not?:
1. Solve 3 CeE pairs on D.
2. L5E (16 perms * (1 no bad + 10 two bad + 5 four bad) = 256 algs)

Probably best:
1. Solve an LMCF E2L pair while or after solving the L and R centers.
2. LMCF L6E
Note that finishing the solve as in Roux is a subset of this.

Probably silly:
1. Solve the DL-L CeE pair.
2. L7EO (7 two bad + 9 four bad + 4 six bad = 20 algs)
3. L7EP ((4 * 3 no misseparated) + (4 * 3 * 3 one misseparated) + (4 * 3 * 2 * 3 two misseparated) + (4 * 3 three misseparated) = 132)
Seems counterintuitively like step 1 of this is too restrictive, not to mention it doesn't completely solve the recog problem. Seems like this is just plain worse than the previous and next approaches.

Cool but probably no good:
1. L8EO (5 two bad + 6 four bad + 5 six bad + 1 eight bad = 17 algs)
2. Separate L8E (1 one misseparated + 4 two misseparated + 1 three misseparated + 1 four misseparated = 7 algs)
3. PEBL (4 * 4 = 16 algs)
(or combine two steps, but as far as I can tell that would change the poor recog of this to atrocious recog)

Bella Cullen could devise a better method section:

HKPCMS:
1. Skip CMSLL and solve the 3 D-layer CeE pairs except DR-R directly instead.
2. HKOLL
3. HKPLL

Roux PCMS:
1. Switch to Roux since PCMS is essentially just a worse Roux.
2. ???
3. Profit

LMCF PCMS:
1. Switch to LMCF since PCMS is essentially just a worse LMCF.
2. ???
3. Profit

Have I missed any obvious ones? Which are considered the least bad?

 

[kubnintadni]

I underestimated alg counts for PEBL, but it still feels of a reasonable size. Has anyone generated algs for it? So far my only success has been in setting up to EPLL. I assume that setting up to Roux L6E will be better, but so far I have been unable to successfully do it.

I know L8EO algs have been generated, but I can't find them anywhere; does anyone have them? They're pretty intuitive, though, it seems. Same with SL8E.

 

[Silky]

On the Wiki:
1. Solve 3 CeE pairs on D.
2. L5EO (5 algs)
3. L5EP (16 algs)

Wrong, it's 4 CE pairs.

Roux PCMS:
1. Switch to Roux since PCMS is essentially just a worse Roux.
2. ???
3. Profit

LMCF PCMS:
1. Switch to LMCF since PCMS is essentially just a worse LMCF.
2. ???
3. Profit

PCMS is not a worse version of LMCF or Roux, ridiculous. Roux is significantly different. One of the main downsides of Roux is that CMLL is a pretty awkward step. PCMS is also pretty different form LMCF specifically ergonomics wise, corner wise, lookahead wise, and alg wise. L6E is definitely better than L5E, no real argument here. PCMS advantage is that the 4 CE pairs is a very open step and benefits from inspection. Salvia-esque approach could be very good since it would make make CMSLL prediction very easy. LMCF tricks are definitely an advance since it can be incorporated, although not a necessity. Also you an avoid S moves by preforming a D/D' and solving DL/DR after Center perm of course. If you're going to evaluate a method consider judging it by it's own merits than just comparing it haphazardly to other methods. By the same logic ZZ is just a bad CFOP. LMCF is a bad Roux. Petrus is a bad FreeFOP. Waterman is just a bad LMCF and Roux.

Edit: it's also so frustrating to see the narrative that a method is just bad/worse version of X. The only way to really tell if a method has potential is but investing and pursuing that method. Saying a method is bad means that people aren't going to pursue this method which just reaffirms your bias assumption. Also @Aerma has achieved sub-12 averages using this method.. Also maybe move this to Method Debate Thread ?

Hope @Cubing Forever proves you wrong

 

[kubnintadni]

Wrong, it's 4 CE pairs.

I'm pretty sure it's not... It's 3 CeE pairs which then leaves 8 edges to be oriented and permuted. The wiki suggests DL, DB, and DR.

PCMS is not a worse version of LMCF or Roux, ridiculous.

I was mostly joking with the last few versions, but thank you for the reasoning as to why it's not; I'm always glad to see eclectic methods be, if not the best, at least viable.

 

[Silky]

I'm pretty sure it's not... It's 3 CeE pairs which then leaves 8 edges to be oriented and permuted. The wiki suggests DL, DB, and DR.



I was mostly joking with the last few versions, but thank you for the reasoning as to why it's not; I'm always glad to see eclectic methods be, if not the best, at least viable.

Sorry got confused. Though you meant that first step was 3 pairs instead of 4 pairs lol. Sorry if I got a bit heated, just am very passionate for nonstandard methods ( hence the Uncommon Method Competition ) and get a big frustrated..appreciate the understanding though !