• 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!

New Method?

IAmAPerson

Member
Joined
Mar 21, 2014
Messages
180
Location
US
WCA
2015LANE01
YouTube
Visit Channel
Hello everyone! I am in the process of creating a new method for a 3x3 solve. It has 4 basic steps:

Double cross (cross on white/yellow or any two opposite sides for colour-neutral solvers)
Corner layer set (putting the corners on their correct layers)
Corner orientation and permutation (self explanatory)
Finishing the M/E slice (M or E depending on how you hold the cube)

I am still in the early stages of making this method, but I have done many solves with it, trying new algorithms, techniques, etc. The method is very intuitive and I can easily get sub-2 minute times while reading the algorithms off of the screen. Thus, I'm expecting sub-30 second times once I learn all of the algorithms by muscle memory. In total there are 28 algorithms to learn (this is only counting the cases I've found and have algorithms for). The algorithms I have are not optimal as I've used Cube Explorer for most all of them (as in all except 2 or 3). I've done it this way to save time (I can always optimize it later).

I have run into one small problem. I cannot get a quick way to orient all of the corners. To orient them all in one algorithm requires memorization of 2186 algorithms. I can't do the basic algorithms provided with CFOP and other methods because all of the algorithms don't work under my conditions. All of the algorithms I've found on the internet require at least one layer to be completely oriented. So if someone can help me with this, that'd be AWESOME.

I've provided a .zip file that contains all cases + their algorithms that I've encountered so far if you want to do some solves yourself. If you find any that I haven't written down, please inform me. I have also linked a video that has a couple of walkthrough solves (just explaining how to use the algorithms to solve the cube). ANY input is appreciated. Also, before you ask, I indeed have looked at a good bit of other methods. This one is inspired by: Beginners & CFOP (the cross idea), Roux (the M-slice turning), and Adams aka Corners First (solving corners separately from edges). I've also looked at methods like ZZ, Petrus, and Triangular Francisco. So...yea. I'd love feedback and I welcome you to do a few solves! Thank you so much for you guys' time!

Algorithms Download - http://www.mediafire.com/download/bu5a1rjtd1j64gr/PhillipsMethod.zip (SAFE! I promise! It's only some .txt and .png files)
Method Explanation - https://www.youtube.com/watch?v=TISqcqGnEwY
 

MrMan

Member
Joined
Nov 26, 2014
Messages
101
Location
France
In the video did you used an algorithm to make the second cross ? If yes there is no need for any algorithm to make the second cross.
 

IAmAPerson

Member
Joined
Mar 21, 2014
Messages
180
Location
US
WCA
2015LANE01
YouTube
Visit Channel
This, for example.

I don't believe I've ever come across this. If I had to solve from this, I'd do an F2 U2 F2 U2 F2 (green on front and white on top) and continue to solve like the method states.

In the video did you used an algorithm to make the second cross ? If yes there is no need for any algorithm to make the second cross.

I used an algorithm for the yellow cross, but just because it was easier. I'm aware that it isn't necessary (CFOP proves this point).
 

mDiPalma

Member
Joined
Jul 12, 2011
Messages
1,534
good idea! but why don't you just:

1) solve D layer (roux block+3 pieces, intuitive)
2) solve U layer corners (42 algs)
3) ELL (29 algs)
4) E-slice (12 algs, I think, provided you permit z2 and x rotations)

or

1) solve D layer except for 1 corner (intuitive)
2) solve last D layer corner while orienting and permuting edges of LL (mostly intuitive, or can use LPELL, 6 algs)
3) COLL (42 algs)
4) E slice (12 algs, I think)


my opinions (on the variants i described above):

better than CFOP? yes.

better than Petrus? no.

better than Roux? significantly.

better than ZZ? no comment.
 

MrMan

Member
Joined
Nov 26, 2014
Messages
101
Location
France
It made me thought of something, ZZ EO Solve all the edges. Put Corner in the good layer. And the just use two COLLs ! -- Doesn't work eh.
 
Last edited:

deadmanlsh

Member
Joined
May 15, 2014
Messages
37
WCA
2014HALE02
I downloaded the zip file and looked through your algorithms. When you are generating algorithms, try to make them good or at least easy to remember. What you currently have fulfills neither of those conditions.

I tried a couple of solves with your method. All the steps past the first one is honestly quite bad. Here's a breakdown of why I think this method is generally inferior, arranged according to the steps of your method.

Step 1: It certainly is easier than 2x2 to 1-look and is not too terrible a step if other steps were better.

Step 2: Since you only need to look at the U/D or L/R stickers, it's not that bad, but it still requires you to see stickers around the whole cube. Also, There's very little flexibility already since you have to maintain the "Double Cross"

Step 3: Corner orientation is going to be a big problem. With the double cross, I doubt the algorithms can get very good even if they are properly generated (Of course, this may be refutable). Again, the problem of recognising the case comes into play. The number of algorithms makes one-looking unfeasible, so it's not like it can be done very fast even with superior recognition. Corner permutation is a pain as well since the edges are fixed. Currently your method for corner permutation is two-look, which really adds up the number of looks this method requires.

Step 4: M/E slice edge recognition is very difficult. This is already a slight problem in Roux, but those who are fast at it can generally develop systems to look ahead and figure out the case (by looking at the BU sticker, keeping track of the DB edge etc.) Note that edges are already oriented in Roux. This means that the task will become even more difficult. Again, your current method is two-look, so it's probably going to be slow both execution- and recognition-wise.

Assuming a one-look step 1, you have a total of 8 looks (correct me if I am wrong here). This isn't that bad, but it's still a little too many for it to be a viable method. Since you method is very algorithmic, you either need a very advanced prediction method or a really good look ahead to reduce the pauses between the steps. Recognition is an atrocity for the most part, since you constantly need to come back to all corner cubies. Sure, I think it can be pretty fast with an extremely advanced recognition and prediction method, but then you might as well learn to make more efficient blocks or learn F2L tricks.

Those are just my thoughts on your method. I hope I didn't come off as demeaning, because I didn't intend to. Hope to see more interesting stuff from you in the future.
 
Last edited:

IAmAPerson

Member
Joined
Mar 21, 2014
Messages
180
Location
US
WCA
2015LANE01
YouTube
Visit Channel
I downloaded the zip file and looked through your algorithms. [...] Those are just my thoughts on your method. I hope I didn't come off as demeaning, because I didn't intend to. Hope to see more interesting stuff from you in the future.

Thank you for your feedback. I did notice that corner recognition is a pain, and the algorithms are quite bad (Cube Explorer algorithms were really only temporary). I'll try to work with the corners to make it easier to work with.
 

TDM

Member
Joined
Mar 7, 2013
Messages
7,006
Location
Oxfordshire, UK
WCA
2013MEND03
YouTube
Visit Channel
I have run into one small problem. I cannot get a quick way to orient all of the corners. To orient them all in one algorithm requires memorization of 2186 algorithms. I can't do the basic algorithms provided with CFOP and other methods because all of the algorithms don't work under my conditions. All of the algorithms I've found on the internet require at least one layer to be completely oriented. So if someone can help me with this, that'd be AWESOME.
Orient one layer, then orient the other layer. You can orient two corners with combinations of sunes, e.g.
U twist: Right Antisune + left Antisune
T twist: Right Sune + left Sune
Other twists could be done as OLL + PLL, or CLL+ELL:
S twist: Right Sune U' Ua perm U'
A twist: Right Antisune U' Ub perm U'
P twist: F double sexy F' [R U R' U', M']

You might get parity, but then you can just twist one of the corners on each layer to make both layers solvable.
 
Top