Full LMCF 3x3 method now available

Discussion in 'General Speedcubing Discussion' started by efattah, Feb 26, 2017.

Welcome to the Speedsolving.com. You are currently viewing our boards as a guest which gives you limited access to join discussions and access our other features. By joining our free community of over 30,000 people, you will have access to post topics, communicate privately with other members (PM), respond to polls, upload content and access many other special features. Registration is fast, simple and absolutely free so please, join our community today!

If you have any problems with the registration process or your account login, please contact us and we'll help you get started. We look forward to seeing you on the forums!

Already a member? Login to stop seeing this message.
  1. That's an awful idea considering FMC is counted in HTM and not STM...

    Also this seems like an extremely interesting method, would you mind updating the original post with the revised PDF, for convenience?

    Awesome stuff :)
     
  2. Attila

    Attila Member

    271
    17
    Nov 7, 2010
    Halásztelek, Hungary
    WCA:
    2012HORV01
    I don't agree. The currently known best official result (with CF method) is 22 single, and 27 mean of 3. Unfortunatelly, a few cubers use this.
     
  3. Yes but if you use LMCF as described here then you will use many slice moves as you just follow a load of algs.
     
    Octavian-360 likes this.
  4. efattah

    efattah Member

    235
    166
    Feb 14, 2016
    Some people who have read the method express concern over the ergonomics of the E2L phase. This is a valid question but there are many advanced techniques used mid-solve to increase the ergonomics. The 2-gen pair algorithms are the fastest and there are ways of deliberately modifying the solve to increase the appearance of 2-gen pairs (vs. 3-gen, 4-gen pairs). Some of this stuff is intricate and isn't in the document; I will post a tutorial video where I explain some of these tricks. Furthermore it is true that some of the pair algorithms are not that ergonomic depending on which side of the cube they occur on (pair algs which have B moves will have F moves instead if the setup happens from the back of the cube). Anyway in many cases I will skip a pair if the alg is not ergonomic; the main time I will perform a non-ergonomic pair is if it is the last pair/triplet that sets up a favorable L6E case.

    There are many more tricks near the end of E2L; by modifying the way you solve the last pair/triplet, you can force an L6E case that you already know. Since LMCF has so many algorithms and the average user would not know all of them, it is very important to finish the solve with an L6E case that you know and this is easily accomplished if you adjust the way you solve your last pair/triplet.

    I am also compiling statistics of the appearance of the different L6E cases on normal solves with no control over the case. This gives a much better guideline over which algorithm sets are really useful. As with ZBLL, there is less value in learning sets that only occur in 1% of solves.

    The last thing which really needs clarification is the transition phase. During the transition phase, you have solved the corners, and depending on your color neutrality, it is often useful to solve edges of the top and bottom layers during the transition into E2L. This step is extremely important; the edge solves during the transition phase allow U/D moves while solving the edges and these U/D moves allow you to *see* every edge piece on the cube without rotating the cube, so when you start the E2L phase you have already seen your first pair or triplet. If you don't perform this transition properly you will enter the E2L phase blind and there will be a pause in your solve as you experience 'lookahead failure.'

    On a normal solve you solve 2 edges piece in the transition phase (U & D layers), and in the process you 'see' the entire cube and immediately roll into your first E2L pair; during the first pair algorithm you can see the next pair; after the second pair you have now solved all 6 required edges of the E2L phase and hopefully during the last E2L pair you set up an L6E case that you know and can roll into that directly.
     
    Octavian-360 likes this.
  5. efattah

    efattah Member

    235
    166
    Feb 14, 2016
    muchacho likes this.
  6. muchacho

    muchacho Member

    1,237
    859
    May 27, 2015
    Spain
    WCA:
    2016ALAR01
    YouTube:
    davidalejosalarcia
     
  7. efattah

    efattah Member

    235
    166
    Feb 14, 2016
    Fixed to public!
     
    muchacho likes this.
  8. efattah

    efattah Member

    235
    166
    Feb 14, 2016
    I reconstructed my 28-move speed solve from several weeks ago:


    Scramble: F2 L2 B2 U2 R' B2 L' U2 R2 F2 D2 B' L2 D R' U F' L2 B' D2 U2
    z2 U R2 U l U' R' // green face and CLL skip
    U M U2 M' // solve blue-red edge on U face
    z F U' M U M' F' // E2L triplet
    L2 R2 M F U' M2 U F' // E2L triplet
    U2 M U2 M' // permute midges
    // Total 28 STM
     
  9. Spencer131

    Spencer131 Member

    10
    1
    Feb 16, 2016
    Here's an idea for my variation of the method, I will still work on this more.

    Solve a 1x2x2 square before solving the corners. Solve corners like normal using wide turns when necessary to preserve the 1x2x2. Then you have 2 edges done for the e2l step
     
  10. efattah

    efattah Member

    235
    166
    Feb 14, 2016
    Yes, this is a great idea, I have experimented with this a little already and it shows great promise, to keep the 'blind' part of the cube solved. Of the 80 EG1+CLL algorithms I selected in my LMCF document, only a few 'break' any of those 2x2 block edges. So only a couple of those algs would need to be swapped with CMLL or other EG1 variants.

    This variant also results in a rotationless solution. Kind of a Roux-LMCF hybrid.
     
  11. bren077s

    bren077s Member

    17
    0
    Dec 17, 2015
    This looks like a really cool method. I think I am going to learn it. That being said, despite not having a high alg count for the beginner version, I feel that it will be harder to get use to than the big 4 methods. Currently, I average about 35 with ZZ, and I suck at look ahead in general. So, we will see how this goes. Any general advice for learning and getting good at this method/ what to focus on? Also, how do you think this compares to waterroux?
     
  12. efattah

    efattah Member

    235
    166
    Feb 14, 2016
    Block building is hard, and LMCF doesn't have any block building, and in that sense it is actually pretty easy to learn. I suggest using the multi-phase option in csTimer (or other timer) so you hit the space bar and each milestone in the solve; hit the spacebar after solving the corners, then hit the spacebar again before you perform L5E, then again at the end. For starters if you are using Ortega to solve the corners you can aim for 9 seconds on the corners, 13 seconds on the E2L phase, and 7 seconds on L5E. That would give you a 29 second average which is already a lot better than you have, and that should be achievable very fast.

    Make sure you have good finger tricks for the M-Slice. Practice U M' U' x 100, U' M' U x 100, U M U' x 100, U' M U x 100 until they are so blazing fast as to be invisible.

    For the E2L edge phase, do a lot of slow untimed solves to learn to follow the pieces and get an intuitive understanding of this phase. Slow untimed solves will improve this phase a lot more than timed solves. Understand that with basic keyhole techniques (L, R, U M' U', U' M' U, U M U', U' M U), you can break any edge on the L/R faces and put it in any other slot, even without any E2L algorithms. Once you learn that basic style, you can add the two E2L pair algorithms that I suggest.
     
  13. bren077s

    bren077s Member

    17
    0
    Dec 17, 2015
    That is nice to know. As a beginner, I am not nervous about corners(I average 15 with ortega, but I have only done like 20 timed solves. Dropping it to 9 should be easy with very little practice), but edges look like they will take a lot of thinking as a beginner. I have not tried solves yet, so this is just based on my assumptions after watching a few of your walkthrough solves in your tutorial.

    I will probably post a more educated opinion once I have done enough solves to see how fast I am improving. I have only been seriously speedsolving since about january. With ZZ, I got my first sub 40 Ao100 about 10 days ago and sub 35 Ao100 yesterday. So I am improving fast. This method I feel has a lot of potential, so I will probably end up switching.
     
  14. bren077s

    bren077s Member

    17
    0
    Dec 17, 2015
    Hey efattah,

    Do you think that at some point you could upload an average of 12? You have a lot of nice singles on your youtube channel but I am just wondering how this method performs on average.

    Also, do you always solve with either the blue or green center on top. If so, does insuring the blue or green center is on top make your 2x2 phase slower/more restricted?
     
  15. efattah

    efattah Member

    235
    166
    Feb 14, 2016
    OK I uploaded a random practice session:


    As you'll see in the video I don't always solve on blue/green when doing the corners, but I always hold blue/green on L and R when doing E2L. Lookahead on E2L is reduced if I solve other colors on L/R.

    Watching the split times is quite interesting. The first solve in the series is 14.56 of which I spend 6.98 solving the corners and 7.58 solving the entire rest of the cube. Given that an expert can solve the corners in 2 seconds and turn twice as fast in general, you see the potential.
     
    Teoidus and muchacho like this.
  16. bren077s

    bren077s Member

    17
    0
    Dec 17, 2015
    Thanks for the video. It is nice to see what an average looks like and how much room for improvement there is.

    I did my first couple of solves and then did some timed solves, plus some slow solves to see my movecount. Currently I am utterly terrible at e2l recognition and execution. I did 5 timed solves and got this average(corners/e2l/l5e/full solve): 12.45/1:00.91/25.24/1:38.70 Clearly I have a ton of room for improvement(like memorizing algs so I don't have to look at a sheet of paper all the time for starters), but this is day one. I did 5 untimed solves to look at my potential movecount and got this average mc(corners/e2l/l5e/full solve): 24/27.8/13.6/65.4 So not terrible but again room for improvement.

    I have a question about the beginner variant e2l. When trying to solve e2l pairs, I often got cases where I had a ledge to solve and in the place where it goes is another ledge or midge. If i didn't have a redge to flip, then I only saw a way to put in a single edge without the e2l Pairs Set algs(page 8). For example, if you had the case U M' U' l' L' U M' U' L2 with red front and yellow top, is there a good way to slot in the blue white edge with another edge to make a pair only using the beginner variant algs?

    I hope I explained that issue well enough. If not, I can try and clarify.
     
  17. efattah

    efattah Member

    235
    166
    Feb 14, 2016
    To answer Bren's question:

    Holding the cube with red on front, yellow on top, and applying scramble U M' U' l' L' U M' U' L2.

    1. If you know the full E2L set the solution would be:
    L2 M D M' U M [U'+D' same time], then M, and then you are set up for L5E
    [9 moves]

    2. If you don't know the E2L set and only know the 3 beginner algorithms the solution would be:
    U M' U' (solve blue yellow edge)
    L2 M U M U' (solve blue white edge), then M2, and you are set up for L5E
    [9 moves]

    So in this case there is no move savings from executing a pair algorithm; any time you have a blue edge on the green face (but disoriented) it is easily solved by a U/M/U move and similarly any green edge on the blue face (disoriented) is also easily solved with a U/M/U style move.
     
  18. bren077s

    bren077s Member

    17
    0
    Dec 17, 2015
    Ok, Thanks for the comment. I am slowly getting better. I simply need to do a ton of slow solves so I can get down recognition and becomes more move efficient. I also need to memorize all of the algs. Definitely need to work on putting in pairs instead of single edge pieces. Any general advice on E2L?
     
  19. efattah

    efattah Member

    235
    166
    Feb 14, 2016
    Many of the E2L pair algorithms are not ideal in a speedsolve due to ergonomics, unless they result in triplet solves where the benefit of a multi-edge solve is so huge that it is okay to suffer a delay from setup and a less than perfectly ergonomic sequence. The key to E2L is to learn intuitive pair solving, sometimes called cycling.

    Here are some examples. Hold the cube with orange on the front and white on the top:

    Scramble 1: L U M U' L' M U M' U'
    We want to solve the green white edge (at DF) into the UR slot. We see that the blue-red edge is in the UL slot. We want to solve both the green-white edge and the blue-red edge at the same time intuitively.
    Step 1: Prime the blue red edge with U M U'. Now the UR slot has been 'primed' and holds the blue-red edge, ready for an easy pair. The green-white edge has been displaced to the BD position.
    Step 2: Bring the green-white edge back to the DF position with an M' move.
    Step 3: We can now solve the green-white edge and the blue-red edge at the same time. Do an L move to put the blue-red slot in the UL position, then solve the pair with U M' U'.
    The above will work any time the UL slot contains a blue edge piece on the blue layer that is oriented; using this intuitive method you have just covered 3 possible E2L situations and this intuitive pair method is extremely fast and ergonomic (no F or B or D moves, just L/R/U/M).

    Scramble 2: Hold the cube with orange on front and white on top, U M U' l U M U' M'
    We see the blue-red edge is in the UR slot in an ideal position. We also see the blue white edge at the UF position. We can cycle-solve both the blue-white and blue-red edges at the same time. First do an M move to put the blue white edge at the DF position. Now solve the blue-red into UL with U M' U', and this simultaneously places the blue-white edge into the UR slot. Now we solve that one next; L' U M' U

    Scramble 3: (white on top, orange on front) U' M U R2 U' M' U
    We want to solve the green-yellow edge (at BD) into the UR slot. We see there is already a green-white edge there. We can cycle solve them both with no algorithm. First do U' M U, and this solves the UR slot with green-yellow; it also primes the UL slot with the green-white edge. Now quickly do an r2, then U' M' U to solve the green-white edge piece (I deliberately didn't solve the blue-white edge at the same time, this was possible but unlikely to occur so nicely in a real solve)

    Scramble 4: (white on top, orange on front) U2 R' U' M' U R U2 R'
    The blue face is solved. We want to solve the green-red edge piece into the UR slot. Green-red is at BD. However there is a green-white edge piece already in the UR slot and we can solve it intuitively. This situation is defined as trying to solve an edge piece where there is another edge piece of the same side already in the slot but disoriented. In this case it could be green-white (as it is), or green-orange too for example. It is all the same.
    Solution: We set this up as a basic U/M/U style pair and with all intuitive solving it is about 'priming' edge pieces for an easy pair. First do a U2. Now you have sort of inverted the blue/green faces. The green-white edge piece is now at UL and in a very favorable primed position for a pair. The green-red edge piece is still at BD and now it needs to go into UL. The green-white edge piece will be displaced to UR, so we need to put the green white slot at UR with an R move. Now solve the pair with U M U'. The edges are now both solved and we need to 'restore' the cube. We restore it with R' U2. Once you understand this 'style' of flipping the top slice with U2 to 'prime' a pair that exists on the same side, you can use it to solve any similar situation.

    Scramble 5: (white on top, orange on front) F U' M2 U F' M'
    We want to solve the blue-white edge (at DF) into UL. We see that UR (green-orange) and FR (green-white) are both disoriented and swapped. In a real situation this might happen, but more than often only one of UR and FR will hold such an edge piece. This is more of an algorithm but it is quite intuitive. We do M F U' M2 U F' and it solves all three pieces. This can be used in many situations such as this:
    (white on top, orange on front) F U' M2 U F' U' M' U R
    We want to solve the blue white edge at DF, into UL. We seed the green-white edge at UR would need to go to BR.
    First do an R'. Now we are in our familiar situation. By doing our mini-algorithm we know the UR and FR will be swapped and oriented. We see green-white at FR, and this will swap into UR and its orientation will flip. So we execute our mini-algorithm M F U' M2 U F' and both edges are solved. This works any time either of UR/FR need to be swapped.
     
  20. bren077s

    bren077s Member

    17
    0
    Dec 17, 2015
    Thank you. This is extremely helpful, especially scrambles 4 and 5.
     

Share This Page