Difference between revisions of "Hex9 Documents"

Revision as of 05:11, 18 March 2011

Andrey:

With 9 colors first stage was easy, but edges are too much connected. The best thing that I could develop was 22-twist commutators for 3-cycle of edges and 10-twist for re-orientation of the couple of edges. And I had to remember them in normal and mirrored forms in all orientations of the plane. Terrible...

Nan:

I just finished my {6,3} 9 colors, 3 layer factor = 1.29903810567 (the sweet spot Roice provided). For the edges, I used a 10-move commutator for 3-cycle, which is not too bad. It's actually easier than I expected yesterday. I notice you said your 3-cycle is 22-twist but re-orientation is 10-twist. That's a little weird. For me re-orientation is usually two 3-cycles.

Andrey:

Yes, it's strange for me too.

Let's mark centers as

..G.H.I

.D.E.F.

A.B.C.

I've started with commutator [A,E']. Among other things it moves three edges around center D and changes orientation of two of them. So when I made [[A,E'],D^2] (10 twists), it was pure reversing of edges DF and DG. But I failed to make pure 3-cycle from it. Then I tried [A,E]. Again, it moves 9 edges, and the best I made from it was 5-cycle [[A,E],C^2]. Third commutator converted it to 3-cycle: [[[A,E],C^2],D^2]. Not very good, but enough for solving.

Nan:

Your reorientation algorithm is interesting. Using your notation, my 3-cycle is [[C,E],D] = (C,E,C',E'),D,(E,C,E',C'),D'. D can be replaced by D2, D3, D', etc to get some variations.