ZZ F2L Move Count

Discussion in 'Puzzle Theory' started by Cride5, Oct 7, 2009.

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. Cride5

    Cride5 Premium Member

    1,225
    18
    Jan 27, 2009
    Scotland
    WCA:
    2009RIDE01
    Does anyone know the average and maximum move count for completing ZZ F2L (from EOLine) optimally, using <R,U,L> moves?

    I found some statistics for completing F2L from EOCross here, but would be interested to know how many moves it takes from EOLine.

    If possible I'd like the averages for F2L broken down in the following ways:

    (1) L & R 1x2x3 blocks completed simultaneously
    (2) Entire 1x2x3 L block, followed by entire 1x2x3 R block
    (3) BL 1x2x2 followed by FL 1x1x2, then BR 1x2x2 followed by FR 1x1x2
    (4) BL 1x2x2 followed by BR 1x2x2, then FL 1x1x2 followed by FR 1x1x2

    Cheers.

    EDIT: Also, how does it compare with the optimal move count using <L,R,U,D,F2,B2>?
     
    Last edited: Oct 7, 2009
  2. rjohnson_8ball

    rjohnson_8ball Member

    735
    1
    Apr 17, 2008
    Massachusetts, USA
    WCA:
    2009JOHN07
    YouTube:
    rjohnson8ball
    You cannot flip an edge using only <R,U,L> moves. Wouldn't you need to do at least one cube rotation or an r,u,l,F,B,M,E, or S? Sorry I cannot help with your primary questions.

    Edit: Oops, sorry, I didn't know what EOLine meant before I posted.
     
    Last edited: Oct 7, 2009
  3. Cride5

    Cride5 Premium Member

    1,225
    18
    Jan 27, 2009
    Scotland
    WCA:
    2009RIDE01
    I'm looking for the statistics to complete F2L from EOLine (edges already oriented and DF and DB placed). EOLine takes ~6.127 moves avg and 9 moves max (FTM) to do optimally.
     
  4. waffle=ijm

    waffle=ijm Waffo

    3,511
    6
    Jun 9, 2008
    Waffletopia
    WCA:
    2008MANA02
    YouTube:
    cubes4lyfe
    Perhaps PM Johannes91
    I did and got lots of data for Roux move count.
     
  5. Cride5

    Cride5 Premium Member

    1,225
    18
    Jan 27, 2009
    Scotland
    WCA:
    2009RIDE01
    Yup, already have, cheers :)

    ...looks like another coding job, but I'll have to put it on the back burner for now, too many other things to do! If anyone can find these stats in the mean time I'd be very grateful, along with a few other ZZ'ers ;)

    EDIT: Just for some quick info on the task involved, these are the numbers of cases which need to be solved:
    (1) L + R blocks: 3^4 * (8!/4!) * (10!/4!) = 20,575,296,000 <-- a biggie :eek:
    (2) L: 8*7 * 10*9*8 * 3^2 = 362,880 R: 6*5 * 7*6*5 * 3^2 = 56,700 (419,580 total)
    (3) BL 1x2x2: 8 * 10*9 * 3 = 2160, FL 1x1x2: 7 * 8 * 3 = 168, BR 1x2x2: 6 * 7*6 * 3 = 756, FR 1x1x2: 5 * 5 * 3 = 75 (3159 total)
    (4) BL 1x2x2: 8 * 10*9 * 3 = 2160, BR 1x2x2: 7 * 8*7 * 3 = 1176, FL 1x1x2: 6 * 6 * 3 = 108, FR 1x1x2: 5 * 5 * 3 = 75 (3519 total)
     
    Last edited: Oct 8, 2009
  6. Lt-UnReaL

    Lt-UnReaL Member

    1,142
    0
    Dec 6, 2006
    Rochester, NY
    WCA:
    2008CLAY01
    Coincidence?
     
  7. Cride5

    Cride5 Premium Member

    1,225
    18
    Jan 27, 2009
    Scotland
    WCA:
    2009RIDE01
    Yup. There is a reason the total number of cases for (3) and (4) are different. The number of cases depends on (a) the number of pieces involved and (b) the number of possible orientations/permutations of these pieces.

    Ignoring the first and last cases which have equal case counts: In (4) doing the 1x2x2 first means that (a) and (b) are both maximal and then both minimal when multiplied together, where as in (3) when (a) is maximal (b) is not and vice versa, yielding a lower overall total.
     
  8. Cride5

    Cride5 Premium Member

    1,225
    18
    Jan 27, 2009
    Scotland
    WCA:
    2009RIDE01
    That's great, cheers ... edging closer to the goal :)

    So based on 100 random scrambles the statistics for strategy (4) is ~24.64 moves or ~30.77 including EOLine.

    You wouldn't happen to have kept the averages for each sub-step would you?
     
  9. Stefan

    Stefan Member

    7,287
    12
    May 7, 2006
    WCA:
    2003POCH01
    YouTube:
    StefanPochmann
    No. Your (4) solves the four 2x2x1 blocks in a specific order. I didn't. Always solved the shortest one first.

    Looks like I didn't keep them. Can compute it again though, but after the world champs.
     
  10. miniGOINGS

    miniGOINGS Member

    3,052
    3
    Feb 27, 2009
    I might do this with 100 different scrambles and post my average movecount for (1), (2), (3), and (4).
     
  11. Cride5

    Cride5 Premium Member

    1,225
    18
    Jan 27, 2009
    Scotland
    WCA:
    2009RIDE01
    Hi that would be a great help, thanks. Could you also use this or something like it for your scrambles? Most online scramblers don't actually create statistically random cube states. See this thread for details..

    Cheers :)
     
  12. miniGOINGS

    miniGOINGS Member

    3,052
    3
    Feb 27, 2009
    Although I have a few questions:

    1. Would this be without EOline? (I probably would switch to ZZ, at least for OH, if I could figure out a fast way of doing EOline) Yes.

    2. Should I use fixed colours to reduce the chance of luck?

    3. Should I include the solutions?

    EDIT: I answered my first question by reading your post. My move count for (1) is probably going to be really bad just because I don't have any practice with it.

    EDIT2: So I just realized that when doing ZZ F2L, I pretty much always do either (3) or (4) and even when I try to do (1) or (2) it turns out to be (3) or (4).
     
    Last edited: Oct 9, 2009
  13. Cride5

    Cride5 Premium Member

    1,225
    18
    Jan 27, 2009
    Scotland
    WCA:
    2009RIDE01
    Sorry I misunderstood you, I didn't realise you were planning on doing this manually! :eek: We're looking for optimal move counts. Doing this manually for (3) and (4) is possible (if you really know what you're doing), but for (1) and (2) you really need to be using a computer solver which guarantees optimality. The only ones I know of are:
    Graphical: Cube Explorer in huge solver mode (windoze only)
    Command line: ACube, this, this and this.

    Generally I wouldn't recommend tackling this kind of problem manually.. even if you find a good solution its very difficult to prove with absolute certainty that its an optimal one. Because its possible to do an exhaustive search with a computer solver, we can guarantee that solutions are optimal..

    The added benefit of a computer is that its a **** load faster then even the best human solvers we know ;)
     
    Last edited: Oct 10, 2009
  14. miniGOINGS

    miniGOINGS Member

    3,052
    3
    Feb 27, 2009
    Yea, I planned on doing it manually. Nevermind then I guess, it would take me too long to use program, sorry.
     
  15. Cride5

    Cride5 Premium Member

    1,225
    18
    Jan 27, 2009
    Scotland
    WCA:
    2009RIDE01
    In order to clear up some recent discussions on ZZF2L move count, I've generated some quick stats for strategy (2) - solving an entire 1x2x3 block at a time. It was done using Johannes's web solver.

    I took 100 samples based on solving the easiest 1x2x3 block first. The average optimal move count for both blocks (excluding EOLine) was: 18.72 HTM
    Adding the average of ~ 6.13 for EOLine gives a total of 24.85 for all of F2L.

    Actual data:
    16, 16, 17, 19, 18, 21, 17, 19, 21, 20, 19, 18, 21, 17, 19, 16, 19, 14, 20, 19, 20, 18, 17, 18, 21, 19, 19, 19, 18, 19, 20, 15, 21, 17, 19, 20, 19, 18, 18, 13, 17, 19, 17, 18, 19, 19, 19, 18, 18, 19, 20, 20, 19, 18, 19, 20, 20, 20, 17, 20, 20, 18, 19, 19, 17, 19, 20, 21, 22, 19, 18, 20, 19, 21, 17, 20, 17, 19, 19, 19, 18, 20, 17, 20, 20, 18, 21, 19, 19, 21, 18, 16, 20, 20, 18, 18, 18, 20, 20, 18
    This provides an estimate for the absolute lower bound for ZZ F2L. Data from Stephan shows that ZZ F2L takes around ~30.77 optimally if each side is broken down into 1x1x2 and 1x2x2 blocks and solved in easiest-first order.

    In reality a human solver is rarely going to be able to solve an entire side optimally, but with experience a solver may sometimes solve an entire side or employ other tricks to achieve a lower move count than by breaking F2L down into four blocks.

    So to compare with Fridrich:
    Cross + 4xSlots: 5.81 + 23.03 = 28.84
    CN Cross + 4xSlots: 4.81 + 23.03 = 27.84
    EOCross + 4xSLots <R,U,L>: 7.53 + 28.1 = 35.63
    EOLine + 4x F2L Blocks: 6.13 + 24.64 = 30.77
    EOLine + 2x F2L Blocks: 6.13 + 18.72 = 24.85

    EDIT: Thanks to cuBerBruce stats I've added EOCross..
     
    Last edited: Mar 16, 2010
  16. a small kitten

    a small kitten Member

    1,627
    11
    Jul 21, 2008
    WCA:
    2010YUPH01
    YouTube:
    asmallkitten
    Does that mean you are saying that the EO line takes as many moves as the EO cross?
     
  17. Cride5

    Cride5 Premium Member

    1,225
    18
    Jan 27, 2009
    Scotland
    WCA:
    2009RIDE01
    No, not at all! I didn't post any figures for EOCross.

    Just to clarify what I meant:

    * Cross + 4xSlots = Just regular Fridrich - cross followed by four slots.

    * CN Cross + 4xSlots = Just regular Fridrich but colour neutral.

    * EOLine + 4x F2L Blocks = EOLine, then in easiest first order: RH 1x2x2, LH 1x2x2, RH 1x1x2, LH 1x1x2 (ensuring the 1x2x2 comes before the 1x1x2 on each side)

    * EOLine + 2x F2L Blocks = EOLine, then in easiest first order: entire RH 1x2x3, entire LH 1x2x3

    The optimal move count for F2L based on EOCross is pretty bad. From this thread, excluding EO its an average of 32.2 moves. I don't know the figures for optimal EOCross, but assuming it is something like 7-8 moves that would make EOCross-based F2L over 35 moves avg.
     
    Last edited: Mar 15, 2010
  18. cuBerBruce

    cuBerBruce Member

    912
    4
    Oct 8, 2006
    Malden, MA, USA
    WCA:
    2006NORS01
    YouTube:
    cuBerBruce
    For what it's worth, I've computed optimal distributions for adding left side 1x2x3 from EOLine in both <R,U,L> and <U,D,R,L,F2,B2> (face-turn metric). I note that this if for a specific 1x2x3 block, and not for the easier of the left and right block. The average distance for <R,U,L> is about 9.6922, while the average distance for <U,D,R,L,F2,B2> is about 8.4048.

    Code:
    distance  <R,U,L>   <U,D,R,L>  <U,D,R,L,F2,B2>
        0           1           1           1
        1           3           3           3
        2           9           9           9
        3          48          53          67
        4         182         236         363
        5         769        1122        1842
        6        2938        4862        9169
        7       10783       19743       39784
        8       35330       66495      122543
        9       89070      145949      165797
       10      136195      113465       23277
       11       77681       10907          25
       12        9769          35
       13         102
    
    I adapted my optimal 2x2x3 block program to generate these results. For <U,D,R,L,F2,B2>, I did a breadth-first search to generate all 47900160 reachable positions of the 7 cubies, and then filtered the result to only those corresponding to EOLine positions.

    EDIT: I've added <U,D,R,L> data (requested by Stefan) to the above table. The average move count for <U,D,R,L> came out to about 9.0241.
     
    Last edited: Mar 15, 2010
  19. Cride5

    Cride5 Premium Member

    1,225
    18
    Jan 27, 2009
    Scotland
    WCA:
    2009RIDE01
    Wow, excellent work! Thanks :)

    So in this case, using <U,D,R,L,F2,B2> rather than <R,U,L> improves efficiency by ~13%

    Would it be possible to adapt the program to calculate the move count for the other side as well, to give a total for left + right (solved separately)?

    Also, if you have the time it would be interesting to know the average optimal move-count for solving both sides simultaneously, as well as solving all four Fridrich F2L slots simultaneously. The information wouldn't be very meaningful from a speedsolving perspective, but would certainly be of interest to FMC'ers.
     

Share This Page