Message #364

From: Melinda Green <melinda@superliminal.com>
Subject: Re: [MC4D] There should be some prizes for contests :P
Date: Wed, 03 Jan 2007 14:22:25 -0800

Remigiusz Durka wrote:
> Melinda wrote:
>
> And now for the new challenge:
> Below is a state-only log file for what I suspect may be the 4D
> equivalent of the "superflip" pattern. I call it the "superduperflip" . :-)
> The challenge is to come up with the shortest solution for this one.
> There’s lots of regularity–in fact I think that it has the same
> symmetry as the solved state–so it should be pretty easy to find a
> short solution, right? Well, grab the latest version of MC4D and give it
> a try. Be sure to also copy the period at the end when you go to paste
> it into a log file.
>
> Good luck!
> -Melinda
>
> ___________________________
>
> MagicCube4D 2 0 0 3
> 000010000020304050000060000
> 111171111121314151111101111
> 222212222272324202222262222
> 333313333323730353333363333
> 444414444424047454444464444
> 555515555505354575555565555
> 666606666626364656666676666
> 777767777727374757777717777
> .
> —————————-
>
>
>
>
> Hello!
>
> Finally I found time to try this.
>
> Result: 68 moves
>
> MagicCube4D 2 0 68 3
> 666666666666666666666666666
> 777777777777777777777777777
> 555555555555555555555555555
> 333333333333333333333333333
> 444444444444444444444444444
> 222222222222222222222222222
> 000000000000000000000000000
> 111111111111111111111111111
> 122:2 412 122:2 412 122:2 412 122:2 412 016:2 122:2
> 412 122:2 412 122:2 412 122:2 412 010:2 416:-1 614:-1
> 218 016:2 414 14:2 414 14:2 414 14:2 414 14:2
> 010:2 414 14:2 414 14:2 414 14:2 414 14:2 28
> 614:-1 012 26 614:-1 310 616 316 610 612:-1 314:2
> 216 314:2 216 314:2 216 314:2 216 04:2 314:2 216
> 314:2 216 314:2 216 314:2 216 022:2 614:-1 616 310
> 610 316 612:-1 220 014.
>
> It was probably the same effort like 2^4 shortest contest - okay, maybe bigger.
>
68 twists is much less than a typical short full solution so you clearly
must have been able to exploit the symmetry of the state. Good job Remi!
> BTW:
>
> On second day christmas (using java version with macros) I solved 3^4 in 18 minutes 29.24 sec
>
Congratulations!! So still not enough contestants to hold the first
hyper speed solving competition? I suspect that you have scared everyone
away.

> Jesse wrote:
>
> So here is an interesting question.
> On a 3d cube, you can generate a "checkerboard" pattern by rotating all 3 middle slices 180°, looking like Fig 1.
> Can anyone else guess how to do this on a 4d cube? See Fig 2.
> Furthermore, is there an analog to the Superflip in 4d?
>
>
>
> Fig2. Hmmm, I did this in 40 moves -> http://www.superliminal.com/cube/3checkerboard-40.log
> -?there is whole category with this in Hall of Fame
>
> Fig.1 (in 3D) we change 2C pieces.
> Fig.2 (in 4D) we change 3C pieces.
>
> 2C - two colored pieces
> 3C -three colored pieces
>
> I cannot think about 3D analog of Melinda’s superflip… Flip of 1C? :) or just flip of 2C on 3x3x3 ?
>
As I mentioned in a previous msg today, my superduperflip is my guess at
the 4D equivalent of the 2C superflip defined in 3D.
> OH, MELINDA -> we found with Kirfo reason of my problems with macros on 2^4! It was reference sticers. I choose 3 different pieces on one "face"(Actually on the one surface of the face) but program took them as the one piece!
Yes, we figured this out too and now understand the problem. It turned
out not to be a coding bug but a misunderstanding between Don and my
ideas of what these reference things are. Our different ideas collided
where our code implementations meet each other. In other words, our code
had a similar misunderstanding. Don had always been thinking about
macros in terms of reference "stickers" whereas I had always been
thinking in terms of reference "grips". A grip can be thought of as one
of the 26 rays that run from the center of each hyperface out through
each of the outer 26 stickers of each face on the 3^4. Notice that there
are 26 grip on each face in all of the different sized puzzles even
though the larger puzzles have many more stickers per face. Much of our
code operates in terms of these grips. My misunderstanding was to assume
that macro definitions would also be based on grips because that would
mean that macros defined on one puzzle could still be used on the
others. This is what the code currently does but as you noticed, this
breaks down with the 2^4. None of the possible solutions seem perfect
and so some compromise will likely need to be made. We therefore plan to
use the Yahoo group polling functionality to get your feedback on the
possible solutions that we are considering, so stay tuned. Just know
that we haven’t forgotten your bug report; It just ended up going much
deeper than we expected.

-Melinda