EG(D?) 2x2 method- 12.35 move avg.

[Stefan]

Bump, got new data (and some comments).

basically you solve one side of the cube, not necessarily permuted (like Ortega first layer.) i believe this step averages 3 moves.

Has this been more thoroughly analyzed in the meantime?

there are 143 algorithms for this

The wiki now says 120 algorithms in the box on the right and 120 non-PBL algorithms in the text. I think all of these numbers are wrong, but I'm not sure enough to "correct" them. I think there are 128 algorithms, 42 for EG 2 (=CLL) and 43 for EG 1 and EG 0. Or 123 if not counting the five PBLs (though I don't see why one would ignore them). Maybe fewer algs are needed by clever reuse, but that's not obvious to me and should be explained. Thoughts?

Case 1: first layer is correctly permuted. this is just COLL optimized for 2x2. there are 47 algs for this case, averaging 8.79 moves.

Case 2: there is a "bar" in the first layer, where only two pieces need to be switched. you place the "bar" in the BD position and the solve the cube. there are 48 algs for this, averaging 8.43 moves.

Case 3: this is the one i designed. it is where the first layer is entirely mispermuted, where a diagonal swap is needed. the FL does not need to be in any specific position to solve the cube, maybe an AUF at the end. there are 48 algs for this case, averaging 8.58 moves.

This counts algs multiple times, probably because Erik arranged them in a 6x8 grid and didn't want to leave cells empty and thus duplicated algs (see first row here). Also, it would be nice to know whether the averages were weighted.

I also tried it myself now (computing my own optimal-length alg set) and came up with these statistics:

            CLL=
            EG 2    EG 1    EG 0

 0 moves      1
 1 moves
 2 moves
 3 moves                      1
 4 moves
 5 moves      1
 6 moves      1       2       1
 7 moves      9       5       3
 8 moves      4      25      14
 9 moves     20      11      17
10 moves      6               6
11 moves      1               1

average     8.30    8.05    8.51
weighted    8.46    8.04    8.62

David       8.79    8.43    8.58

The main table shows how many cases need how many moves, then there's the average and weighted average (taking into account how often each cases occurs) and David's numbers for comparison. David, do you still have your statistics, and can you compare them to mine? (not counting duplicate algs, i.e. 42/43/43 instead of 47/48/48). I'll polish my statistics as well and post my alg set later...

this gives step two a total average of about 8.6 moves.

With my new computations, and assuming to really only focus on solving a face in the first step, I now get:
(8.46 + 4*8.04 + 8.62) / 6 = 8.21

 

[deadalnix]

The wiki now says 120 algorithms in the box on the right and 120 non-PBL algorithms in the text. I think all of these numbers are wrong, but I'm not sure enough to "correct" them. I think there are 128 algorithms, 42 for EG 2 (=CLL) and 43 for EG 1 and EG 0. Or 123 if not counting the five PBLs (though I don't see why one would ignore them). Maybe fewer algs are needed by clever reuse, but that's not obvious to me and should be explained. Thoughts?

You are just right. I know EG and this is it.

Anyway, case EG 2 and EG 1 can be merged easily with several techniques.

 

[Kenneth]

The wiki now says 120 algorithms in the box on the right and 120 non-PBL algorithms in the text. I think all of these numbers are wrong, but I'm not sure enough to "correct" them. I think there are 128 algorithms, 42 for EG 2 (=CLL) and 43 for EG 1 and EG 0. Or 123 if not counting the five PBLs (though I don't see why one would ignore them). Maybe fewer algs are needed by clever reuse, but that's not obvious to me and should be explained. Thoughts?

A number of the EG 1 cases can be solved as CLL if you have one half of LL solved and put that part on the same side as the pair in FL.


Not important but related and a bit fun : R U R' L' U' L R U R' for setup, that makes a self simulating case, re orient so the B-side becomes D

 

[qqwref]

Not important?! R U' R2' F R2 U' R' (an alg for that case) is very useful for Ortega.

 

[Kenneth]

Did not say the alg was unimportant but the fact that the case is self simulating. Who needs to know that?

And yes, I know your alg but I use mine, it is shorter than it looks because you execute it doing R and L simultainously, that way it is only 7 turns.

Edit: (to save a short post) David, yes, same... but diffrent =)

 

[DavidWoner]

They are the same alg.

Edit: I do not have any statistics. If anything was saved, it is on a computer I no longer have access to. However I feel whatever slight differences in movecount that arise are wholly inconsequential. The ability to execute quickly is not determined solely by movecount, nor the value of a method. I understand the interest to evaluate it from a statistical point of view, but I am no loner concerned with it.

 

[Killermanp]

Hi guys, I have recently begun to learn EG, I had all the algs saved on my flash drive, but my flash drive ghot zapped. Does anyone know the site?

 

[That70sShowDude]

Did you learn from here?

 

[riffz]

http://dtwoner.110mb.com/index.php?p=1_42_EG1

http://www.speedcubing101.com/eg-1.html

In the future, though, post a question like this in the One Answer Question Thread.

 

[Killermanp]

Thank u riffz!!! the one on speedcubing.com is the one I was looking for.

 

[Anthony]

Thank u riffz!!! the one on speedcubing.com is the one I was looking for.

I'm assuming you mean speedcubing101.com..
Anyway, it's funny that you bumped this thread today. I just added EG-2 to my site.