Beruflich Dokumente
Kultur Dokumente
I wonder if anyone knows very much about the nature of that metric
anyway? I understand that it is known that no two points are more
than 94 (or is it 93?) twists apart (disregarding the extended
problem). I don't know if that number is actually attained, or if it
is only the currently known upper bound based on the best algorithm.
(Or perhaps there isn't an algorithm that good yet, just a proof of
the fact.)
I believe that you and ACW and I once did the math to show that
<Greenberg at MIT-Multics>
<Greenberg at MIT-Multics>
x y x
x y x
x y x
Only the top or bottom look like this; this is what you have to
remember to look for after aligning centers to taste.
We're gonna rotate the y y y band into the horiz position.
Do this exactly as for the CC above, producing
(x x x/ y y y/x x x)
Next goal is again to complete the solid band of the floor-parallel
centerslice by doubleswapping front/back so that
the x x edgecubies,w hich would complete that band, go to
the back. Of course, we must temporarily rotate the
back during this doubleswap, so that they go to the side
positions ofthe back when swapped. Do so, completing the
solid color-band of the floor-parallel slice.
Now consider the top and bottom. You note that exactly one appropriate
doubleswap between top and bottom would give us solid
crosses on both. Do it.
Take what had been the top just now, and call that the front.
Note that there are solid crosses on front and back, and the
body-parallel plane is correct and complete.
Think about the front: it looks like
a b a
b b b
a b a
Although it looks right fromt the front, the two vertical b-edge
cubies want to be the two horizontal b-edge cubies, as a cursory
inspectionof the top bottom and sides of the cube will show.
This is true of the back, as well.
Tofix up the FRONT do this:
1. Doubleswap front/back
2. Rotate the FRONT (temporarily) 90 degrees sothat the two
vertical b-edge cubies are gonna come to the right place,
3. And doubleswap front/back
4. Undo 2.
5. Doubleswap front/back.
Now you see all is right save the back. It wants the same
thing done to it. Do it for it; Do this same thing
just doNe in the last 5 steps for the back (viewing it as the
temporary front).
It is done. Consider it.
An exquisite variant ont he CP is obtained by taking on of the
trebly-bound sides and rotating the centers via the well-known
center-cubie rotating algorithm. As the centers are rotated
left or right, either a sextuple checkerboard or a stunning
triply-rotated canon of centers , edges, and corners appears.
The checkerboard is amusing insofar as it appears to a
novice cubist to be the Pons Asinorum 6tuple checkerboard
made by 6 twists (described earlier today), but cannot be
fixed (solved, or produced) without the consummate hair
<Greenberg at MIT-Multics>
...
The Higher Crosses are fascinating insofar as they appear to be
very simple edge-cube hacks, but are in fact quite "far"
from home; the CP being exactly twice as "hairy" (far)
as the CC (discovered by ALAN) is in itself a source of
wonderment.
I wish I could really say something was "exactly twice as far" from
home as something else. Unfortunately, as I complained before, the
metric by which one measures the distance of one configuration from
another is not well enough understood to be able to make claims like
this. It might well be that the CP is less than twice as far as the
CC, all we can really be sure is that it is not any further than that.
Date: 17 July 1980 22:52 edt
From: Greenberg.Multics at MIT-Multics
Subject: Postscript to above
To: cube-hackers at MIT-MC
I should have noted in the above flamage
(btw, will all of those of us who are paid by our respective employers
to hack this lunacy please let me know at once)
that there is room for opitimization and lookahead in the final
doubleswaps in correcting the top/bottom, and the
doubleswap preceding it, but I do not do this, so that
I might better keep track of what I am doing.
Date: 17 July 1980 22:54 edt
From: Greenberg.Multics at MIT-Multics
Subject: Bug in above
To: cube-hackers at MIT-MC
In the terminOlogy section, note that the
body-slcing and FLOOR-PARALLEL centerslices meet in the front face,
not the body-parallel and body-slicing as given.
ED@MIT-AI 07/18/80 00:12:53 Re: Patterns, designs &c.
The file which contains all cube-lovers mail is ALAN;CUBE MAIL on MC.
"Games People Play" on Mass Ave, near the fork where Mt. Auburn
St. branches off, carred them the last I knew. According to
rumor, the domestic product ("Rubik's Cube"), selling for
about $7, is mechanically inferior to the Hungarian import,
costing $15. I don't know how everyone feels about money,
but to me, not having to fight with the thing would be worth
the extra $8.
Bernie's explanation of how to achieve the Plummer and Christman
Crosses is a prime example of why we need a cube language.
Since no one has proposed anything, I will jump into the fray.
Center the cube at the origin of a 3-d coordinate system. The
axes of the coordinate system protrude from the centers of the faces.
Make a hitch-hiker's gesture with your right hand and point the thumb
along the X axis. Imagine rotating the WHOLE CUBE one quarter turn
in the direction your curled fingers are pointing. I call this operation
"I". (The X axis is the horizontal axis that does not skewer you.)
If the cube was lying on a table, "I" would roll it toward you.
Now point up, along the Y axis, with your thumb. A quarter-rotation
in the direction your curled fingers point is the operation I call "J".
The Z axis goes right through your belly. A quarter turn around it
I call "K". Actually, K=IIIJI.
To simplify things a little, we define I'=III, J'=JJJ, and K'=KKK.
In general, M' is M done backwards. If we call the do-nothing
manipulation "1", then MM'=M'M=1.
For my own nefarious reasons I define "H" as the operation (unachievable
in real life) of REFLECTING the cube right-to-left through the YZ plane.
We note H'=H.
Twisting the front (Z=1) face 90 degrees counter-clockwise I call
"T".
One more piece of notation: For any manipulations M and N, I write
M'NM as N[M], reading this as if N were a function: "N of M".
Note M[1]=M.
Examples:
T[I] means "Twist the top face"
T[II]=T[JJ] means "Twist the back face"
T[I'] means "Twist the bottom face"
T'[J] means "Twist the left face CLOCKWISE"
T[I] T'[I'] J' means "Rotate the floor-parallel center-slice a
quarter turn counter-clockwise as seen from above."
Note that (MN)'=N'M', and (N[M])'=N'[M].
that the maximum order of any element is 1260 assumes that these
center faces remain fixed. There is nothing wrong with that
assumption, it simply relects the intuition that picking a cube up and
putting it down again on a different face doesn't change the
configuration at all. If we decide that the orentation of the cube
DOES matter, then we get transformations like ED's here. The group
we are dealing with is also 24 times larger than before. And my proof
now shows that no element has an order greater than 2520 (twice as big).
Could it be that ED's transformation is not actually a maximal one?
Could there be one with higher order? Or can my proof be tightened up
some, so that even in the larger group the maximum order remains the
same?
To answer these questions (hopefully) I present a transformation
that I claim has order 2520:
D2 AB A0
DC AX AE
D5 AD A1
C2 CD C5 F5 DA D1 E1 EA E0 B0 BA A2
CA CX CF FC DX DE ED EX EB BE BX AC
C3 CB C4 F4 DF D6 E6 EF E7 B7 BF A3
B4 FD F6
BC FX FE
B3 FB F7
(This transformation, like ED's, is not a member of the usual group,
but the larger one where the center faces are allowed to move.)
It is easy to check that this permutation has order 2520, the real
question is whether you can get to this configuration from a solved
cube. I haven't actually tried to do that, but I am fairly certain
that it is possible. If anyone would like to try it, and report to me
their success or failure, I would be very grateful.
Date: 19 JUL 1980 0346-EDT
From: ALAN at MIT-MC (Alan Bawden)
Subject: OOPS
To: ACW at MIT-AI
CC: CUBE-LOVERS at MIT-MC
Let me suggest the following corrections to your cube language
message. If any of these are not actual errors, but are actually
misunderstandings on my part, then I apologize for causing undue
confusion, but I thought it would be helpfull to people trying to
understand you system to have these corrected as soom as possible.
ACW@MIT-AI 07/19/80 01:42:56
...
One more piece of notation: For any manipulations M and N, I write
M'NM as N[M], reading this as if N were a function: "N of M".
I think you must intend N[M] to mean M N M' .
...
T[I] T'[I'] J' means "Rotate the floor-parallel center-slice a
=
=
=
=
=
=
f'
F[J]
f[J] = R'
F[j]
f[j] = L'
F[i]
u
D
d
B
b
=
=
=
=
=
f[i] = U'
F[I]
f[I] = D'
F[ii]
f[ii] = B'
It also solves the problem that I was having of saying IJK in English since ther
e
is no way to say it in "Singmaster".
I didn't follow your Pons Asinorum checkerboard solution because I didn't
understand the Z in Q, but in the mixed notation I think it is:
rrllffbbuudd
or, if Q = rrll; QjQkQ
Your "quite pretty" is (lrfb)^3 = (lr (lr)[j])^3
I propose the following grammer for RCML (Rubik Cube Manipulation
Language):
algorithm ::= definition* move*
definition ::= letter = algorithm ;
move ::= letter
| move'
| move^digit
| move[algorithm]
| (algorithm)
digit ::= 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9
where the letters RLUDFBrludfbIJKijk are predefined as noted above.
--Bill
Date:
From:
Subject:
To:
<Greenberg at MIT-Multics>
<Greenberg at MIT-Multics>
Ok, everybody wants a way to tell :CUBE "I have a cube like
so and so, solve it". The reason I have avoided doing this
(other than laze) is that you may specify a non-reachable
state. (Duplicate cubies, or purple faces, etc. are easy
to check for). I am not convinced that there is any
better or faster check for an inconsistent/unreachable
cube than trying to solve it and blowing out: lest
we find such a check (other possibility: run the program once
silently and once loudly; if the first time fails
(no, i will not make it list all cubes it cant solve)
give an error), all internal breakpoints become
user errors.
The second issue is what is the best language
to specify a configuration in? Comments?
Date: 19 JUL 1980 1548-EDT
From: ALAN at MIT-MC (Alan Bawden)
To: Greenberg at MIT-MULTICS
CC: CUBE-HACKERS at MIT-MC
Date:
From:
<Greenberg at MIT-Multics>
...
I am not convinced that there is any
better or faster check for an inconsistent/unreachable
cube than trying to solve it and blowing out: ...
You mean I haven't convinced you of that yet? Show me how your cube
is represented and I will write one for you. What would be so bad
about blowing out anyway? What happens when you hand a person an
impossible cube? his algorithm "blows out" eventually! There is no
shame in being tricked into trying to solve a bad cube.
Date: 19 July 1980 15:53 edt
From: Greenberg.Multics at MIT-Multics
To: ALAN at MIT-MC (Alan Bawden)
cc: CUBE-HACKERS at MIT-MC
In-Reply-To: Message of 19 July 1980 15:48 edt from Alan Bawden
I certainly believe you have shown me sufficient conditions for
inconsistency, but until now I wa not aware that you claimed
that they were necessary...
Date: 19 July 1980 16:31-EDT
From: Ed Schwalenberg <ED at MIT-AI>
Subject: Nope, I guess we don't understand
To: ALAN at MIT-MC
cc: CUBE-LOVERS at MIT-MC
First of all, I was sitting here inventing a GOOD cube notation,
when McKeeman's message about his notation came in. I begin
to fear that ACW was right, and we will all die before agreeing
on a notation. I dislike all proposed notations so far, for the
following reasons:
ALAN's labeling of the cube for describing a POSITION is excellent;
however it is not useful for describing transforms, which are
operations and not positions.
ACW's language I find baffling, principally because it lacks
verbosity. I would much rather have a notation like
[doubleswap] than F[X,[Y']] since it is descriptive. An analogy
with Life is useful here; I think that the adoption of names
like "traffic lights" is far more usable in the long run than
any of [the-oscillator-of-period-two-composed-of-three-blots-placedorthogonally-adjacent] or
|
.
... |
.
|
.
or "if you put 3 blots in a row, the center dot remains alive forever
while the two outer dots appear first horizontally then vertically".
Creating names like "The Christman Cross" is fun, and makes for
interesting wordplay, even if you don't resort to Latin. So my
proposed language is Augmented English, which has the great feature
of being able to put in comments (a feature notably absent in the
others).
I urge people to describe the transform and its result in any
message describing a nifty transform or pattern (provided both are
known, of course!). RP's pretty pattern may not be what I think it
is, since "pretty" is not a good description. I propose that this
be called "swapping-centers-in-triplets" (procedural notation) or
Twelve Squares (which is not a movie by Mel Brooks, but the positional
notation).
Bernie's comment about decision-making I think is important: it
------
Date: 23 Jul 1980 4:10 am PDT (Wednesday)
From: Woods at PARC-MAXC
Subject: Xerox cube-lovers
To: Cube-Lovers at MIT-MC
This message is of (possible) interest only to Xerox members of Cube-Lovers, but
there's no easy way for me to send it only to them; apologies to the rest of you
.
I've created a Laurel-format message file containing the Cube-Lovers mail (from
the archive at MC) up through approximately last weekend. The message is in
chronological order (instead of reverse chronological the way MIT's stupid maile
r
defaults) and has otherwise been cleaned up slightly (e.g., to get rid of lines
just
barely too wide). If you're interested in looking through this archive, it's in
[maxc]<Woods>Cube.mail. It'll probably be around for several days at least.
-- Don.
Date: 23 JUL 1980 1044-EDT
From: RP at MIT-MC (Richard Pavelle)
To: CUBE-LOVERS at MIT-MC
There is a short article on the cube written by Singmaster. The
reference is Mathematical Intelligencer, Vol.2, #1, pp.29, 1979.
Date: 23 Jul 1980 1640-PDT (Wednesday)
From: Mike at UCLA-SECURITY (Michael Urban)
Subject: Clarification
To: cube-lovers at mit-mc
My description of the checkerboard pattern was clearly
inadequate, due to haste and the fact that my cube was "borrowed"
and scrambled before I had a chance to see precisely what was
going on.
The pattern in question is formed from Crux Plummeri by
applying the "Pons Asinorum" transformation; from the
resulting almost-checkerboard, a simple "12-squares" transformation
will provide the 6 checkerboards; it isn't hard to inspect
the almost-checkerboard to find a trio of center cubies to
rotate.
The resulting checkerboards are a completely interlocked
set. In the following unfolded cube, A/B means that the center
and corners are color A, and the edges are color B.
A/B
C/A
B/D
D/E
F/C
E/F
I don't THINK anyone has mentioned this pattern; the checkerboard
pattern mentioned by Greenberg consists of two cycles of three colors
each, something like
A/B
B/C
C/A
D/E
E/F
F/D
although I may have the handedness wrong.
Mike
------
Date: 23 Jul 1980 5:23 pm PDT (Wednesday)
Sender: Woods at PARC-MAXC
From: Woods at PARC-MAXC, Boyce at PARC-MAXC
Subject: 216 vs 1260 vs . . .
To: Cube-Lovers at MIT-MC
Regarding finding a sequence of twists that requires a maximal number of
repetitions to restore the original state, it is indeed true that 1260 is the
maximum, but the sequence suggested by ED (on July 18) is not it. (He
thought its period was 216; ALAN thought it was 1260. It's actually 315.)
This assumes that we don't consider the orientation of the center-face cubies
nor the orientation of the cube as a whole.
First off, note that the transformation given by ED was ill-specified, since he
didn't say which way to rotate the cube relative to the rotations of the faces.
In Singmaster's notation the two operations would be, say, LBRF and LFRB.
It turns out these are each period-315, though it's not immediately obvious
that the two should be at all similar. Having done (LBRF)^1, you'll find
there are 7 edge cubies that have cycled through each other's positions, and
the other 5 edge cubies have done likewise, 5 corner cubies have rotated 120
degrees in place, and the other 3 corner cubies have cycled through each
other's positions and one of them has rotated. Hence, if we do (LBRF) a
multiple of 9 times, the corner cubies will be back where they started, and if
we do it a multiple of 35 times, the edge cubies will be restored, so if we do
(LBRF)^315, all of the cubies will be back to the solved state.
Perhaps the reason ALAN thought this was period 1260 was because it takes
1260 twists to restore the original cube. But since the transformation that we
are repeating is a sequence of four twists, it actually has period 315. If you
count total twists, you can obviously get "identity" sequences with lengths
much greater than 1260. Or perhaps ALAN was considering the transformation
to be LJ (twist left face, rotate cube about vertical axis). This indeed has
period 1260, but it "cheats" in that we're not supposed to consider reorienting
the entire cube as a legitimate operation. If we do, then it is again possible
to get periods greater than 1260.
To get a period-1260 transformation, we need to observe whence the period
arises. If we do a sequence of twists, and then look at the cycles of the form
"face X moved to where face Y used to be, face Y moved to where face Z was,
. . . , face <mumble> moved to where face X was", we take the least common
multiple of the lengths of those cycles. If we repeat the sequence of twists
that many times, the faces will moved around those cycles an integral number
of times and end up where they started; if we stop short of that many
repetitions, at least one of the cycles will be left stuck in the middle.
(Apologies to any readers who are familiar with group theory and are bored
by this verbosity.)
To get a period of 1260, we need to have a cycle of 7 edge cubies, another
cycle of 2 edge cubies where one of them is also rotated (so that the four
faces on those cubies form a single period-4 cycle), a cycle of 5 corner cubies,
and a cycle of 3 corner cubies where again one of them is rotated (forming
a face-cycle of length 9). If, instead of the 2 edge cubies, we could get 4
edge cubies cycling with one of them rotated, that would be a period-8 cycle
and the overall transformation would have period 2520, but due to the fact
that it's impossible to swap exactly two cubies it turns out you can't get that
combination of subcycles.
Based on this analysis, it's not hard to construct a period-1260 sequence,
although this one is almost certainly not the shortest such (it's 24 twists):
F U F^2 u r f b l r f D^2 F b R U (R^2 F^2)^3 u r B
Date: 23 Jul 1980 1757-PDT
From: DDYER
Subject: impossible cubes
To: cubr-lovers at MIT-MC
Remailed-date: 23 Jul 1980 1758-PDT
Remailed-from: Dave Dyer <DDYER at USC-ISIB>
Remailed-to: cube-lovers at MIT-MC
According to Singmaster, there are 12 disjoint orbits (his term) of cube
positions. The article I read doesn't give a derivation, but the basic
idea is clear. It should be possible to characterize the orbit of a given
position as a function of the transformations that would have had to
be done on each cubie ( as if it could be moved alone ) to bring it
to its proposed position. Given such a charaterization, one could
declare positions fed to :CUBE illegal without trying to solve the
problem.
This relates to my thoughts on the problem of inversion of an unknown
transformation. Suppose you are given a Cube that is N unit moves
from home, determine the correct sequence to get it there. I think
a representation of the Cube in terms of combined tranformations on
the cubies is the place to start.
------
Date: 24 JUL 1980 2115-EDT
From: RP at MIT-MC (Richard Pavelle)
To: CUBE-LOVERS at MIT-MC
Many of you have seen this message before. However, since the mailing
list of CUBISTS is growing rapidly I shall repeat this on occasion.
A file of past cube mail is on ALAN;CUBE MAIL on MC.
Date: 25 JUL 1980 0845-EDT
From: ZIM at MIT-MC (Mark Zimmermann)
Subject: "blindfold" cube?
To: CUBE-HACKERS at MIT-MC
How long does it typically take for one to be able to work with only
a mental image of the cube? (As in"blindfold" chess, soma, pentominoes....)
Date: 25 JUL 1980 0906-EDT
From: RP at MIT-MC (Richard Pavelle)
Subject: "blindfold" cube?
To: ZIM at MIT-MC
Hi,
You're welcome to look at the code, although it is hardly fit for
human consumption. I wrote it essentially in one sitting, and stopped
as soon as I solved the cube. There are some poorly thought out
design features, etc. Anyway, you can get it at SAIL. It is written
in PASSCAL, and is called MCUBE.PAS[1,TRD]. I scribbled in a few
comments this morning, so maybe there is enough information to
figure out how to use it. Unfortunately I am at home on a typewriter
terminal, and I refuse to try to edit without a display. Good luck.
By the way, I have never used :cube, and have no idea exactly what
it does. Is there some documentation associated with it somewhere?
Thanks,
-- Tom DAvis
------
Date: 31 JUL 1980 1006-EDT
From: RP at MIT-MC (Richard Pavelle)
To: CUBE-LOVERS at MIT-MC
IS IT POSSIBLE?
I just got a note from Martin Gardner who plans an article about the
cube in Sci.Am. sometime. But he claims that a person names
THISTLETHWAITE has proved a restoration procedure of 50 moves
and believes he can reduce it to 41.
Gardner also says that a 2x3x3 solid is selling in Hungary, also
by Rubic.
Date: 31 July 1980 10:52 edt
From: Greenberg.Multics at MIT-Multics
To: RP at MIT-MC (Richard Pavelle)
cc: CUBE-LOVERS at MIT-MC
In-Reply-To: Message of 31 July 1980 10:06 edt from Richard Pavelle
OK, your'e on the spot..... you'd better produce these artifacts
or we're all not gonna let you log out....
Date: 31 July 1980 13:06-EDT
From: Alan Bawden <ALAN at MIT-MC>
To: RP at MIT-MC
cc: CUBE-HACKERS at MIT-MC
,
30 and 19.
Date: 31 Jul 1980 5:13 pm PDT (Thursday)
From: Woods at PARC-MAXC
Subject: Re: The shortest solution?
In-reply-to: McKeeman's message of 31 Jul 1980 16:44 PDT
To: CUBE-HACKERS at MIT-MC
You can do better than that for a lower bound! Say you consider all
single-hand-motion twists to be okay. Then yes, there are 18 such,
but there's no point in twisting the same face twice consecutively, so
after the first twist the tree branching factor is only 15. In fact, there's
no point in twisting a face twice if the only intervening twist was done
on the opposite face; if we look at the operations of the form "twist face
X thusly and the opposite face thusly", there are 45 initial such operations,
and 30 at each succeeding branch, but since some branches now represent
two twists and some only one twist, it's much harder to compute the
minimum depth of the tree.
-- Don.
Date: 31 JUL 1980 2159-EDT
From: ALAN at MIT-MC (Alan Bawden)
Subject: The shortest solution?
To: McKeeman at PARC-MAXC
CC: CUBE-HACKERS at MIT-MC
Date: 31 Jul 1980 16:44 PDT
Sender: McKeeman.PA at PARC-MAXC
In-reply-to: ALAN's message of 31 July 1980 13:06-EDT
From: (Bill) McKeeman
A lower bound on the number of twists can be derived as follows: There
are 4.3*10^19 distinct reachable arrangments of the cube. Suppose the
moves are restricted to the (more than sufficient) set RLFBUD. Then
there are at most six independent choices at each step and the number
of reachable places is bounded by 6^n. That gives
6^25 < 4.3*10^19 < 6^26,
or 26 moves as the (probably unachievable) minimum.
This is not an improvement on my result. I (and the rest of the cube
hackers I know) consider a unit move to be a 90 degree twist in EITHER
direction. You are only considering CLOCKWISE 90 degree twists.
Let me point out that if we were to count twists your way, we would
no longer have a metric. Both the quarter twist method and Singmaster's
method result in a measure of distance that is a true metric.
Date: 31 Jul 1980 5:13 pm PDT (Thursday)
From: Woods at PARC-MAXC
Subject: Re: The shortest solution?
In-reply-to: McKeeman's message of 31 Jul 1980 16:44 PDT
To: CUBE-HACKERS at MIT-MC
You can do better than that for a lower bound! Say you consider all
single-hand-motion twists to be okay. Then yes, there are 18 such,
but there's no point in twisting the same face twice consecutively, so
after the first twist the tree branching factor is only 15. In fact, there'
s
no point in twisting a face twice if the only intervening twist was done
on the opposite face; if we look at the operations of the form "twist face
X thusly and the opposite face thusly", there are 45 initial such operations
,
and 30 at each succeeding branch, but since some branches now represent
two twists and some only one twist, it's much harder to compute the
minimum depth of the tree.
-- Don.
Singmaster's notes are aware of these factors, that is how he improves
on the 16 count computed by McKeeman to arrive at 18. Similarly I
used the same factors to improve on the 12^n argument for quarter
twists (which gives 19 as a lower bound) to arrive at the number 21.
I also compute 24 as the quarter twist lower bound for the extended
cube (considering orentations of the center faces).
Date: 1 Aug 1980 11:53 PDT
From: Lynn.ES at PARC-MAXC
Subject: Re: cube group theory.
In-reply-to: Sandeep.Johar's message of 31 July 1980 1413-EDT (Thursday),
<31Jul80 141352 SJ61@CMU-10A>
To: Sandeep.Johar at CMU-10A
cc: cube-lovers at mit-mc
Singmaster ("Notes on the 'Magic Cube'", p. 12 of edition 4) gives 12 classes or
"orbits" of assemblies, each one such that the other 11 cannot be reached withou
t
disassembling the cube. To quote Singmaster, "This also means that if we
reassemble the cube at random, there is only a 1/12 chance of being able to get
back to START," and, "It is advisable to reassemble in the starting pattern."
Singmaster's notes have several pages of relevant group theory for those
interested.
/Don Lynn
Date: 1 Aug 1980 12:20 PDT
From: Lynn.ES at PARC-MAXC
Subject: Re: Sorry to spoil your day, but.....
In-reply-to: RP's message of 31 JUL 1980 1431-EDT
To: RP at MIT-MC (Richard Pavelle)
cc: CUBE-LOVERS at MIT-MC
2) Another TRUE story of rubiking:
The May Company department stores in the Los Angeles area held contests at
four of their stores when they started carrying the cube, a couple of months
ago. The object was to solve one face in three minutes with their scrambled
cube. Reward was a 50 buck gift certificate. They allowed about 6 people every
fifteen minutes to enter for two days. The store where I won (and my 10 year
old daughter, my wife, and three of my next door neighbors) had 21 winners. I
assume the other three stores had similar numbers. They also gave away lots of
cube tee shirts for "good tries".
I got my cube the night before the contest. But it took me another month to
solve the whole thing. I might have worked faster at it if the megabuck prize
had been offered. But then you all would have also.
4) Singmaster (edition 4, p. 34) reports on the Rubik domino (2x3x3 cubies), and
says each cubie has spots like a domino (1 up to 9). They are to be lined up in
numeric order. Top and bottom (the 3x3 faces) are two different colors.
Unfortunately, Singmaster laments, magic square type patterns (all directions ad
d
to same number) are not possible, since all edges are even, all corners odd. Th
e
mechanics of the device are said to be "more complicated than for the Magic
Cube."
/Don Lynn
Date: 1 Aug 1980 15:47 PDT
From: McKeeman at PARC-MAXC
Subject: Re: The shortest solution?
In-reply-to: ALAN's message of 31 JUL 1980 2159-EDT
To: ALAN at MIT-MC (Alan Bawden)
cc: McKeeman, CUBE-HACKERS at MIT-MC
Alan,
Sorry I missed your earlier words of wisdom on the subject. Anyway, I am
interested in how you show any of these is, or is not, a metric. Can you
forward same?
Thanks, Bill
Date: 2 August 1980 01:55-EDT
From: Alan Bawden <ALAN at MIT-MC>
Subject: a metric for the cube group.
To: CUBE-HACKERS at MIT-MC, McKeeman at PARC-MAXC
First off, a metric is a function (call it D) that assigns a
non-negitive number to every pair of points in some set. This number
is to be thought off as the distance between those two points. The
function must satisfy the following:
For all a, b and c
1) D(a,b) >= 0
2) D(a,b) = D(b,a)
3) D(a,b) = 0 if and only if a = b
4) D(a,b) + D(b,c) >= D(a,c)
(Number 4 is usually called the "triangle inequality". It is the
constraint that most makes D act like a distance, and not something
random.)
We wish to construct a metric on the set of all attainable cube
configurations. So from now on, those lower case letters will
represent cube configurations.
Now we have recently been talking a lot about methods of counting the
number of "twists" that it takes to perform certain manipulations of
the cube. We have been looking for a function (call it T) that
assigns a non-negitive integer to each manipulation. I claim that it
is obvious that any such function should satisfy the following:
For all M and N
1) T(M) >= 0
how some of the properties you might expect of a twist measure could
be used to generate a metric, so I didn't actually need strong enough
properties to ensure reasonable twist measures. The additional property
I have been using to assure reasonability is the following:
5) For all M, if T(M) > 1, then there exists an N such that
0 < T(N) < T(M) and T(N) + T(N'M) = T(M).
Note that N'M has the property that 0 < T(N'M) < T(M) (easy to show)
so the two manipulations N and N'M are both "simpler" than M. We can
thus easily show that any manipulation M can be expressed as the
product of T(M) twists (where a twist is defined as a manipulation
such that T(N) = 1).
Date: 4 Aug 1980 1156-PDT
From: Dave Dyer <DDYER at USC-ISIB>
Subject: cube permutations
To: cube-lovers at MIT-MC
I'm not quite satisfied with the numbers that have been quoted
for various cube groups. I believe they are probably correct,
but I haven't seen anything resembling a satisfactory sketch
of a proof.
Number of Reachable Positions:
The standard calculation runs like : we have 8 corners that can
be in all 8! arrangements, and 12 sides that can be in all 12!
arrangements, except that the total permutation must be even.
( I'll talk about orientations later)
I can accept this figure as an upper bound, but can anyone demonstrate
that all the positions not ruled out can actually be reached? The only
way I can think of is an actual construction, which means finding
a generator for the group (for instance of corner cube positions) and
showing that its order is 8!. This is somewhat less than elegant,
and requires an unspecified bit of magic to 'find' a generator
for the group.
I also don't like the recourse to geometric arguments ( the corners
and sides can't be interchanged ) but I am willing to accept it.
Number of Reachable orientations:
A similar line of reasoning is used to argue the number of reachable
orientations : the amount of twist on all corners and (independantly)
on all sides is a multiple of 360 degrees. I haven't seen a demonstration
that all the not-forbidden orientations are actually reachable.
Finally, I haven't seen a demonstration that the orientation and permutation
subgroups are indepentant, that is, that you can get to an arbitrarily
selected location and an arbitrarily selected orientation at the same
time. This assumption is the basis for Singmasters claim that there
are 12 orbits of cubes : ( even-spacial-permutation X even-side-orientation
X one-of-three-corner-orientation = 2 X 2 X 3 = 12 )
------
Date: 4 August 1980 20:36-EDT
From: Alan Bawden <ALAN at MIT-MC>
Subject: cube permutations
To: DDYER at USC-ISIB
Using the transformation which flips any two in place, and the two discussed
above, it is not hard to see that the factor is at most and at least 12.
Boy, It sure is hard to prove things on a computer. On myy notes with
diagrams and all, this is perfectly clear, but I get confused trying to read my
online proof. I hope that someone can make some sense out of it.
------
Date: 6 AUG 1980 0939-EDT
From: JURGEN at MIT-MC (Jon David Callas)
Subject: Random Notes on Cubism
To: CUBE-LOVERS at MIT-MC
CC: JURGEN at MIT-MC
I am beginning to feel competant enough to jump into the fray, so ...
1)
language: I guess that RLUDFB has become the de facto lg. for cubing.
(by the way, how IS 'RLUDFB' pronounced?) Is this to include the IJK rotations
that ACW described on 7-19? They are very handy, but I know that there is a desi
re tokeep the lq. brief. Also what (if anything is being done with the center
slices that Bernie used while describing the higher crosses? Given RLUDFB+IJK
notation, they are easy algorithms (subroutines?) that can be described as:
FPC {floor-parallel center} := udK
BPC {body-|| center}
:= fbI
BSC {body-slicing center} := rlJ
without IJK, they're much more complex. Perhaps this is a case for using
RLUDFB & IJK.
2) I have been playing a little with the group of rotations, twists
and so forth. I thought that it mighht be very wieldy until I remembered that D
on Woods has given us a transform with period 1260. (Is it just my
imagination or is 1260 important? it seems like a familiar number) This means
(by la Grange's thm ) that the order of this group is some integral multiple
of 1260 (ugh!) That eliminates hand-enumerating (at least by me) it. How
much does anyone else know about this, and are thre any interesting subgroups?
It also seems that this thing might be related in some way to the dihedral gps.
but I'm not sure.
3)Somewhat connected with (2) above, it would be nice if we compiled
"Famous Transforms I Have Known" or something to that effect. Not only would
it help in finding my desired sbgp, but it would help keep some of the
folklore out of cubing (while there is nothing wrong with folklore,
there are problems when falsities (especially VERY inobvious ones) creep
into the body of lore. Folklore also does tend to cause duplication of
effort and that most of us have probably been wasting time re-inventing
the Pons Ansinorum (which, by the way, IS 'rrllffbbuudd', isn't it?)
Since I'm suggesting this nonesense, I guess it's only fair to
volunteer to do this compendium (unless someone else is dying to).
4) Would it be possible to modify :cube to work on a
printing tty? I really don't care if it produces large volumes of paper.
I am avoiding implementing my own form of a Cube program on my
APPLE ][ (see my notes above on duplication of effort), if I can
get away with using :cube, but at least the Apple display will be pretty.
5) There MUST be a better way to check for an impossible cube
than solving for it. While there is no shame in solving a bad cube,
there IS a little embarassment, and there certainly is no honor in it , either.
back your candidacy for the job. Several references of both types are buried in
the previous messages to cube-lovers. I would favor using Singmaster's
extensions of the RLUDFB to describe the results of a transform. Combinations o
f
two of the set RLUDFB specify edges, and three give the corners. Then
statements like "this transform moves URF to DBL" (or briefly "URF => DBL")
imply not only the position within the cube, but the rotation of the cubie also.
That is, the U face of URF corner moves to the D face of position DBL.
/Don Lynn
Date: 6 Aug 1980 1131-PDT (Wednesday)
From: Mike at UCLA-SECURITY (Michael Urban)
Subject: Ideal Toys
To: cube-lovers at mc
Has anyone learned whether Ideal will be "supporting" Rubik's Cube
with some vehicle like Parker's "Soma Addict" newsletter of some years
past? They will certainly be missing a bet if they don't. Just the
number of people who would pay for a solution...
...hmm...maybe we shouldn't suggest it to Ideal after all..
Mike
------
Date:
From:
Subject:
To:
<Greenberg at MIT-Multics>
0
0
=
=
or 1)
or 1)
0 or 1)
0, 1, or 2)
<Greenberg at MIT-Multics>
Hungarian and had bought two for his children after visiting Hungary.
The children didn't like them, but he did, and he imported 1000 more and
sent them around (including 10 to Stanford Book Store). When Ideal took
over the rights (by offering the Hungarians more money, I assume), he
started manufacturing them here - the white plastic versions. He is also
a source for the Singmaster pamphlet, and an 'Anginvine' solution.
------
Date: 9 Aug 1980 1610-PDT
From: CSD.VANDERSCHEL at SU-SCORE
Subject: Two-Faced Cubing
To: CUBE-LOVERS at MIT-MC
Having posed the question about Two-Faced cubes, I could not resist
thinking about it some more and I have pretty well resolved it. If
you want to try your hand at Two-Faced Cubing, DON'T READ THIS !
Suppose we only allow twists of the Right and Back faces.
Edge Orientations: You cannot change the orientation of any edge
cubie. So there are 2^7 different ways to orient them.
Edge Permutations: Any permutation is possible. (Must be even if
corner permutation is.)
Corner Orientations: Just like the complete cube.
Corner Permutations: The twists generate a subgroup containing only
one-sixth of the 6! possible permutations of the 6 corners. I do not
know a simple way to see why this has to be so. (I have a messy
argument that is no more persuasive than enumerating them.)
Solving It Do whatever it takes to put the two right-front corners in their home
positions (without regard to orientation). Rotate the back face to
bring (say) the URB corner to correct position. Of the six possible
permutations of the remaining three corners, you must have the correct
one. (That is, assuming you started from a reachable configuration.
Otherwise, this is a way to find a unique representative of the
equivalence class of corner permuations.)
(RB')^3(R'B)^3 may be used to fix the orientations of the corners.
(RRBB)^3 may be used to permute the edges. (You can always get the
two edge cubie pairs you want to swap in vertical opposition.)
(FQ)^4(QF)^4
Yes, indeed, you can express Q as U'DJ', that's legit. I can't see any way
you can count (FQ)^4 as more than 24 quarter turns, for thats all it is.
In the eyes of the people who count turns, not quarter turns, the
QQ in the middle (4 quarter turns) can be expressed as U^2D^2,
where U^2 and D^2 each count as one, which could be thus contributory
to counting the formula as 22 turns, but surely no more than 24 by
any kind of metric.
By the way, my description of the left-handed wrench in the
annotations was in error, it should have been (F'Q)^4, not (FQ')^4.
(The formula (FQ)^4(QF)^4 is correct as stands).
Date: 12 Aug 1980 0823-PDT
From: Davis at OFFICE-3
Subject: The Spratt Wrench
To: greenberg at MIT-MULTICS
cc: cube-lovers at MIT-MC
Thanks for sharing the wrench with us. It is a truly wonderful tool.
Another question I had had is now answered. It was basically: Once I
get near the end of a solution, I often have more than one set of
edge flips to do. Is there some scheme which will easily allow me to
do more than one. After staring at how the wrench works it is clear
that in many cases, with one or two preliminary flips, or flips between
the applications of the two halves of the wrench, wonderous things can
be accomplished.
Also, having worked on my own for quite awhile, and being more of a
mathematician than a computer scientist, I decided that a standard
rotation was CCW, and most of my algorithms favor turns in that direction.
In converting the wrench to my own particular quirks (I can't turn things
CW very well anymore), I discovered that essentially any combination of
F and F', and of Q and Q' work just fine. In other words, if "f" stands
for F or F' and "q" stands for Q or Q', then (fq)^4 (qf)^4 is a wrench.
(Better leave f and q fixed through the whole transformation, however.)
This next comment is off the subject, but I had been meaning to ask it
for some time. I have one of the white cubes, and after reading the
message some time ago about underwater cubing, my cube "accidentally"
fell into a pool too. Well, the white ones float, which I found very
annoying. Does anyone know if the specific gravity of the hungarian
or Ideal cubes is greater than 1? I'm also not so sure that I would
recommend that others try to teach their cubes to swim -- mine now
has a disturbing squeak as it turns -- I fear that something has rusted
inside.
Tom Davis
------
Date: 12 Aug 1980 15:10 PDT
From: McKeeman at PARC-MAXC
Subject: Singmaster's talk at Stanford
To: Cube-lovers at MIT-MC
David talked at ICME (Int. Conf. Math. Ed.) in Berkeley on 8/11 about using the
cube to teach group theory. He talked to some Rubniks at Stanford this noon.
Among the tidbits:
Notes on Rubik's 'Magic Cube', Fifth Edition, Preliminary Version, 75 pgs., $5.
Including: A Detailed Step-By-Step Solution, Thistlewaite's Best Algorithm (52
moves), Conway's Monoswop, Rubik's Duotwist and much more. Write:
David Singmaster, Polytechnic of the South Bank, London, SE1 0AA, UK.
He brought a 3x3x2 domino version, and a 2x2x2 Stanford homebrew which is
apparently nearly identical to a Japanese patent showed up. The 2x2x2 is
conceptually just a pasting of big overlapping corners on the standard 3x3x3
version although the one we saw was nicely machined in brass and some kind of
ivory-like material.
Singmaster counts double twists, e.g., R^2, as single moves. He doesn't see muc
h
use for the IJK whole-cube moves.
The Thistlewaite algorithm goes from subgroup to subgroup as follows:
Starting with a random cube, reachable by
closure(F,B,R,L,U,D) = the full group
7 moves to a cube reachable by
closure(F,B,R,L,U^2,D^2)
13 moves to a cube reachable by
closure(F,B,R^2,L^2,U^2,D^2)
15 moves to a cube reachable by
closure(F^2,B^2,R^2,L^2,U^2,D^2)
17 moves to a cube reachable by
the identity
---52 moves total. Singmaster expects the 17 to be 15 by the time he returns to
London.
The Hungarians have cube races. A contestant take his/her cube out of its box
and unscrambles the judges' randomizing in about 50 seconds. Apparently they
file and lubricate their cubes with loving care to reach such speeds.
The U.S. made white cubes violate no patents because Rubik never applied for
foreign rights. Hmmmm. Is that ethical?
Bill
Date: 15 Aug 1980 1558-PDT
From: CSD.VANDERSCHEL at SU-SCORE
Subject: Re: A solution subroutine
To: Davis at OFFICE-3, cube-lovers at MIT-MC
cc: CSD.VANDERSCHEL at SU-SCORE
In-Reply-To: Your message of 10-Aug-80 1709-PDT
I am surprised about all this talk about edge flipping algorithms.
Once I learned the concept of mono-ops, I immediately proceeded to
invent a very simple, and quite obvious, edge monoflip. You can do it
by manipulating only the front and the horizontal center slice
(hereinafter referred to as the "slice"). A virtue of the procedure I
am going to describe is that you can think of it in geometrical terms
and explain it that way. The result is that it is easy to remember
and invert, and you do not need any notation.
Hold the cube with the two edge cubies you want to flip in the U face.
Have one of them, the current "target", in the UF position. You can
think of its "socket" as moving with the F face. Now turn the target
ACW@MIT-AI 08/15/80 19:27:57 Re: "Monoflips"
To: CUBE-LOVERS at MIT-MC
M. van der Schel has come up with a very elegant move,
indeed. Easy to do, easy to think about. But he sets
no records. A 22-qtw diflip was reported some time
ago to this mailing list. It is true that the monoflip
takes only 10 qtw. Unfortunately, this does not allow you
to do a 20-qtw diflip, since the monoflip is only useful
as a tranform: in other words, when accompanied by its inverse.
Monoflip U Monoflip' U' still has 22 qtw. I suspect this to be
a minimum.
---Wechsler
Date: 15 Aug 1980 2347-PDT
From: Stuart McLure Cracraft <McLure at SRI-KL>
Subject: Undocumented position?
To: cube-lovers at MIT-MC
I've been attempting to use the 'method' Singmaster offers in his 5th
edition pamphlet for solving my cube, and have arrived at an
undocumented position. Perhaps I'm overlooking something and a
cubeophile out there can offer an explanation of where I'm going
wrong. I have reached the position described after step 6 just before
proceeding to step 7. The bottom and top faces are one color, the
center slice cubies are corrrectly colored and the upper edge cubies
(the + pattern) are correctly oriented. Two of the upper adjacent
corners are correctly oriented but the opposite two are not. At this
point Singmaster deals with putting the rest of the U corners in their
correct slots although not orienting them correctly necessarily. If we
represent my U corners with
1
Briefly, he first puts the U edges in place and then the U corners.
One face is a single color now. He turns over the cube so that it
is on the bottom and puts the middle layer edges correcty in place
and then makes the U edges into an even permutation.
Then he puts the U edges in place and then the U corners.
Finally he orients the U corners. And supposedly it is solved
at that point. I haven't had much luck with the method though.
He claims that this method takes less then 200 moves.
------
BSG@MIT-AI 08/17/80 17:22:24
To: CUBE-LOVERS at MIT-MC
OK, you guys. You asked for it, you got it. :CUBE will
now accept typed-in cube configurations. The ^F command
(control f) to installed :cube prompts for a standard
ITS file name. If you give ? as the file name, it tells you
the thing I'm about to tell you.
The hack is as follows. You copy AI:BSG;CUBE TEMPLT into
some choice name of your own, and follow what it says
inside of it, editing it to look like your cube configuration.
Then enter :cube, and give the file name in response to ^F.
We do only the most rudimentary checking, for syntax,
center duplications, only six colors, overambitious file munging.
We have not yet put in conservation of cubies, let alone
ALAN's necessary and sufficient solubility conditions. So
it's quite possible to make :cube go blarmy in this way, so
be careful.
Occasionally, :CUBE may give an ambiguous description of a 180-degree
center-slice twist: to disambiguate it, let it do it, and see
what it did, and do it on your cube.
Go nuts in mid-August, -bsg
BSG@MIT-AI 08/18/80 06:58:28
To: cube-hackers at MIT-MC
The ^F command (read in file) of :CUBE has now been augmented to make
cubie conservation checks; it will tell you by way of what cubie
is missing if there are duplications. The Bawdenian solvability
criteria have not yet been implemented. There is also a brief
form of the input file in AI:BSG;CUBE BFTPLT.
BSG@MIT-AI 08/18/80 19:47:49
To: cube-lovers at MIT-MC
As per request of CLIVE, :CUBE ^F now accepts R for RED, Y for
YELLOW and so forth. See AI:BSG;CUBE TEMPLT for reference.
Date: 19 AUG 1980 0927-EDT
From: ZIM at MIT-MC (Mark Zimmermann)
Subject: Singmaster 5th ed. via LOGICAL GAMES
To: CUBE-HACKERS at MIT-MC
Bela Fzalai at LOGICAL GAMES (Haymarket, VA) told me last night that he
is out of the 4th edition booklets, but that he has a bunch of the 5th on
order. Maybe it would be quicker to get them via him than from
England directly; I don't know....
Mark
Date: 21 Aug 1980 17:14 PDT
From: Wastebasket at PARC-MAXC (not authenticated)
Subject: Re: please add me to your mailing list
In-reply-to: CSD.CVW's message of 21 Aug 1980 1536-PDT
To: CSD.CVW at SU-SCORE
cc: cube-lovers at MIT-MC
Please send requests like "please add me to your mailing list" only to the
maintainer of the list, not to everyone on the list. Thank you.
Date: 24 AUG 1980 1623-EDT
From: RP at MIT-MC (Richard Pavelle)
Subject: Plumbers Cross?
To: CUBE-LOVERS at MIT-MC
There have been discussions of Plummer's Cross which I believed to be
unique. Lets have an almost visual representation such as the following
for one face:
X Y X
def
Y Y Y === (Y,X)
X Y X
Then Plummer's Cross looks like
(O,B)
(W,G) (G,R) (B,Y) (Y,O)
(R,W)
(and this gives the coloring of my cube as well). The point about this
configuration which I am stressing is that opposite sides have no colors
in common.
Now I find there is a second cross for which the point above is not valid.
This cross takes a form
(Y,G)
(R,B) (O,W) (W,O) (B,R)
(G,Y)
Is this known by any Cubists out there?
Date: 24 August 1980 17:52 edt
From: Greenberg.Multics at MIT-Multics
Subject: Re: Plumbers Cross?
Sender: BSG1.SIPB at MIT-Multics
To: RP at MIT-MC (Richard Pavelle)
cc: CUBE-LOVERS at MIT-MC
In-Reply-To: Message of 24 August 1980 16:23 edt from Richard Pavelle
That is Christmans crross, describedin my earlier letter in which
I described Plummers cross.
Date: 25 AUG 1980 0907-EDT
From: ZIM at MIT-MC (Mark Zimmermann)
Subject: anecdotes from visit to Logical Games Inc.
To: CUBE-HACKERS at MIT-MC
I visited Bela Szalai on Saturday; his country home near Manassas battlefield
("Bull Run") is LGI, and he and his family are the employees. I learned a few
interesting things:
--he saw the cube in 1978 Aug, during a trip to visit relatives in Hungary;
after many delays was able to get some wholesale from the Hungarian gov't., but
they wanted $1 million for exclusive rights to distribute the things in the
Western world. Ideal may have learned of the cube from Bernie DeKovon,
GAMES magazine editor who is also a toy consultant for them; Ideal paid the
$10^6 in 1979 Sep.
--the cube is not patentable in the USA bnecause it was sold publicly for
over a year in Hungary before patents were applied for; in England, however,
it is "copyrighted" (equivalent to US patent+trademark) and Ideal has a legal
monopoly.
--Ideal will run nationwide TV advertisements for the cube beginning in a month
or so; something to do with Newton, and including an animated cube which solves
itself....
--Bela took out a second mortgage on his home to pay for the plastic molds for
his cube parts. He uses white plastic so that it will be possible to print the
colors on ("pad printing", the same process whereby labels are put on some
shampoo bottles). If all goes well, LGI will start making printed-color cubes
within a month.
--Bela ordered 300 copies of the 4th edition of Singmaster's booklet in 1980 Jun
;
Singmaster informed him in July that the 4th edition was out of print, but that
he could have 300 of the 5th edition for the same price as soon as they come
out. LGI has >90 orders already pending, and while the remaining <210
copies of the 5th edition last, Bela is willing to sell them for $4 each...
whenever they arrive.
--LGI wholesale prices start at orders for 12 cubes, $6 each plus shipping;
individuals may want to consolidate their orders to save money.
--Rubik developed the cube partly as an aid to teaching 3-dimensional
visualization in students.
--cube manufacturing is VERY labor-intensive:
4 minutes to tap in and glue the caps to cover the internal 6th face of edge
and corner cubes for the 20 pieces necessary to make one 3x3x3
1 minute to assemble with screws and springs 5 out of six swivels (central
cube faces) onto the middle cross
1 minute to assemble the pieces and screw in the last central face
4 minutes to perform final adjustment: silicone grease, torque up all screws
evenly, rotate every which way to test each cube out, tap in and gl
ue
6 central face caps over screws.
6 minutes to apply the color squares to the faces
Bela can read while performing most of the final assembly, now that his hands ha
ve
had several thousand cubes' practice. The color application step will be
eliminated if/when they begin to use pad printing for face coloring.
--about 4% of cubes are rejected for mechanical reasons, so far
--could work without gluing in internal faces of sub-cubes, but then about 1 cub
e
in 20 would fail...so, they don't
--READERS' DIGEST phoned to confirm some data, presumably for a story someday...
.
Date: 26 Aug 1980 1729-PDT
From: CSD.VANDERSCHEL at SU-SCORE
Subject: [CSD.VANDERSCHEL: Re: "Monoflips"]
To: cube-lovers at MIT-MC
Date: 25 Aug 1980 2041-PDT
From: CSD.VANDERSCHEL
Subject: Re: "Monoflips"
To: ACW at MIT-AI
cc: CSD.VANDERSCHEL
in here.
Two notes about my particular cube ...
it worked fine until I decided to lubricate it (as is recommended in
one of these cube-lovers messages ... they say to use a graphite lubricant
(I believe) ... do so.)
DO NOT use "dri-slide". It is a Molybdenum
Disulfide lubricant for use with metals and it tends to eat plastic, making
it harder to turn. Also (this may be unrelated), three weeks or so after I
lubricated mine (an Ideal from Toys-R-Us in the San Fernando Valley (in LA))
one of the facies popped off and now it's out of commission until I try to
glue it back together.
I found that the descriptions of its inner workings do not do it justice.
It is indeed a work of 3-dimensional mechanical genius. To elaborate on
the descriptions ... the face cubies are about 7 mm thick, and they are
what holds the other pieces in. The corner cubies have a growth on their
"opposite" corners which is roughly cubical but the "inside" corner is
rounded to allow them to rotate.
The side cubies (sidies?) have a 7-mm thick disk-like thing sticking out
of the center of their least visible edge (the one opposite the one that
the sidie's two colors border on).
It's actually more like a thick
square with one rounded corner (the inside one).
Anyway, here's a description of my notation followed by a set of
transformations I use; this is indeed sufficient to fix a cube and
if you wish to remain self-sufficient in this matter, read no further;
but then what're you doing on this list anyway?
happy cubing ....
-dalgorf-
... Details:
they're wrong. I'd go edit them but my Unix is out
of space on /tmp and on my filesystem and I only have 300-baud access to
it anyway).
Have fun
DalGorf
Date: 27 August 1980 1:58 pm PDT (Wednesday)
From: Woods at PARC-MAXC
Subject: Lexicon
To: Cube-Lovers@MIT-MC
We need a term to refer to the fear that someone will randomise your cube. Any
suggestions?
-- Don.
ACW@MIT-AI 08/29/80 13:24:04
To: Cube-hackers at MIT-MC
Date: 29 AUG 1980 1316-EDT
From: ACW at MIT-AI (Allan C. Wechsler)
On this coast we have a colorful slang term for randomizing the
cube. Hint: the command to randomize the cube in Bernie's
:CUBE program is "F".
---Wechsler
Date: 29 Aug 1980 10:55 PDT
From: Lynn.ES at PARC-MAXC
Subject: Re: Lexicon
In-reply-to: Woods's message of 27 August 1980 1:58 pm PDT (Wednesday)
To: Woods.PA
cc: Cube-Lovers@MIT-MC
Incongruphobia seems nice, but may imply more of a fear of lack of order, rather
than of the change from order to lack thereof. It doesn't specify the objects t
hat
are disordered either. Cubic discombobulaphobia might be a little better in the
se
respects.
/Don Lynn
MJA@MIT-AI 08/29/80 21:45:41 Re: Knowledge-based Rubik's Cube solver?
To: cube-hackers at MIT-MC
Please excuse the previous badly munged versions of this message i accidently
sent.
I am taking an artificial intelligence course at the University of
Illinois at Urbana-Champaign and I need to do a term project for the
course that involves developing and/or using a knowledge based system.
Since I am interested in the Rubik's cube and since it seems like
there are human expert cube solvers (as is evident from msgs on this mailing
list) that use heuristic knowledge (for example the various macro's (as I like
to call them, they are more properly called composite elements of the
group, i guess) that people have come up with to do some particular operations
on the cube), it seems natural to me that a knowledge based system would
be just right for this task. I would like to hear:
(1) Pros/cons on why this can/can't be done and how effective
such a system would be.
(2) Considering that I have but one semester to work on this project
(while concurently taking 5 courses, incl. this AI course) is it reasonable to
attempt this for a term project? (The actual system would not have to be
100% complete at the end of the term, but I would at least like to have
something to show for a semester's worth of work.)
(3) If this is too ambitious for a term project for a course, what
about the use of this as a topic for a master's thesis. (I would have until
the summer of '81 to finish since thats how long my company will wait for me
to get a master's degree if im going to continue working for them.)
(4) Of course in addition to all of the above trivia, I would like
some ideas for such a system (even if it isnt feasable to do in this limited
amount for time, im still interested in at least thinking about such a system
with the idea in mind that someday it can be implemented).
I dont know how interested the majority of the people on this mailing
list are in knowledge based systems, so we might consider a seperate mailing
list for knowledge based cube solvers to spare those on the current mailing
list from mail they dont want.
/ MJA
ZILCH@MIT-AI 08/30/80 01:39:38
To: CUBE-LOVERS at MIT-MC
Subjects covered:random notes on the cube and cubing
1> To enable new and old cube-lovers to communicate on an equal basis
I propose that a file be established that describes FBLRUD
(or whatever),the I,J,K and centerslice move and any other concepts
that might be usefull in describing transformations.(I can not do this as I know
nothing about establishing files or editting them, and as a tourist I am
unsure of my status as file-creater.)
2> Today having nothing better to do, I fiddled with the 3x3x2 version
of the cube (actually I just didn't allow certain moves on my
3x3x3). At this point I have two transforms which would enable me
to solve the 3x3x2 version if and when I ever see it. I derived the number of or
bits and the reasons behind them, but will not describe them here because
the 3x3x2 is a novel idea and I don't want to ruin all the fun.
Also I came up with the idea of a 2x2x3 version which basically operates on
the same principles as the 3x3x2 , but it looks different.
solving the 3x3x2 can give a slight amount of insight about
Thistlewaite's algorithim ,described by
McKeeman on 12 Aug. @15:10 PDT.
3> On July 15 @1413 EDT, ALAN asked if anyone had learned to solve the extended
cube problem independently. I had known the faces could assume different
orientations when the cube was solved but hadn't done any work on this
and didn't know what the possible transforms might be. In a later note
cmb described what his transforms accomplished and from that idea
I recently derived similar (if not equal) transforms.With a little foresight
I find that only one of these transforms might be needed.
This evening a friend and I came up with a modification to the cube
construction that would facilitate the extended problem. this is simply affixing
another cubie to each face cubie and coloring appropriatly.
4> As suggested by CSD.VANDERSCHEL on 9 Aug. @1610 pdt has anyone given any
thoughts to 3-faced cubing (2 types) or 4-faced cubing (again 2
types). As far as I can tell 5-faced cubing is not profitable.
Also does CSD.VANDERSCHEL know a good way (at this time) to show
why only 1/6 of the 6! possible permutations of the corners are possible?
There is also an extended problem for the 2-faced problem
involving face orientations.
5>An addendum to WOOD's message of 23 JUL. @5:23pm
regarding repetitive sequences to get to the identity. As it turns out,
by my calculations the number 1260 is sufficient even including the prblem
where the faces are permitted to move. The only subcycles of the faces
themselves have lengths 4,3,2,and of course 1.
These subccles may immediatly be generated by I (4 times), I^2
(2 times),<top-to-front>^2<left-to-front> (also 2 times)
and finally <top-to-front><right-to-top> (3 times).
(I do not use IJK notation on these last two because I have noticed a little
disagreement on exactly what these mean.)
These subcycles make no difference in the total number because subcycles of
lengths 4,3,and 2 have already been taken into consideration in the computation
of 1260.
I have no idea whether th e face cubies might change their orientations in a
sequence of length 1260,when the faces are allowed to move,
but I know it does nothe number above 1260 when the faces are not allowed to
move.
6> Singmaster's solution from his version 5, as reported by McLure
on 16 Aug. @1053 PDT sounds almost exactly like my solution
except that I keep the first face completed in the up poition at
all times. He reported that his method takes less than 200
(of his) moves. My transformations yield a solution in a maximum of 190
quarter-twists. My actual solving length (from when I bothered counting)
averages about 125 qtws , and my time (when I bother) is almost
consistantly 2.5 minutes, occasionaly under 2, with worst case about 3 min. 10se
c when I messed
up. Usually my fast times do not use all of my algorithms techniques ,
because they are new to me and I don't know them by heart.
Breakdown of moves:
1. Top edges in proper position and orientation 20 (3+6+6+5)
2. " corners " "
"
"
"
36 (9+9+9+9)
3> 3 middle edges " "
"
"
"
45 (15+15+15)
4. 4th middle edge in proper position and orientation
and bottom edges in proper orientation 23
5. bottom edges in proper position
18
6.bottom corners in proper position
20
7. "
" "
"
"
28
Total
190
Note: in step 4 proper orientation means that say if the down face is
white , after step 4 all bottom edges will be showing white on their
down sides.
This algorhitim has not been optimized much and uses lookahead only in step
4. Step 2 gives the worst case for any corner, but if only worst cases are
present then they cancel each other out somewhat.
Replys, questions, and comments are welcome.
Chris
DMM@MIT-ML 08/30/80 18:41:24 Re: DAL@ucla-security's msg of 27-Aug
To: cube-lovers at MIT-MC
Genereally, I have found themacros supplied by DAL to be quite useful,
but t|ere are a few bugs i've noticed. First of all, oC8&C9,
I believe DAL has the cw & ccw notations reversed.
Regarding C7: What is M? I have not seen it referenced anywhere.
and last, in maco EB: the last move is f, not F.
(I`guess it was the 5 beers and 2 Drambuies.)
Date: 3 September 1980 2108-EDT
From: James.Saxe at CMU-10A (C410JS30)
Subject: Orbit classification revisited
To: cube-lovers at MIT-MC
Message-ID: <03Sep80 210846 JS30@CMU-10A>
Hi, folks.
Having read Vanderschel's msg of Aug. 6, it appears to me that the
explanations of the orientation parities are unnecessarily complex,
though the material on permutation parities is presented in a way
that should be immediately convincing to anyone familiar with the
notion of even and odd permutations from elementary group theory (and
anyone who isn't should be!). Here's my attempt at a more elegant
demonstration.
In what follows, I will use the term "facelet" to denote any (visible)
face of a cubie. Thus, each edge cubie has two facelets and each
corner cubie has three facelets.
I will address Edge Orientation Parity (EOP) first. Consider the
following diagram:
+-------+
| 1 |
|0 U 0|
| 1 |
+-------+-------+-------+-------+
| 1 | 0 | 1 | 0 |
|0 L 0|1 F 1|0 R 0|1 B 1|
| 1 | 0 | 1 | 0 |
+-------+-------+-------+-------+
| 1 |
|0 D 0|
| 1 |
+-------+
The numbers label absolute positions, not facelets, and therefore
------
Date: 8 SEP 1980 0853-EDT
From: ZIM at MIT-MC (Mark Zimmermann)
Subject: Singmaster 5th; 2x3x3 "domino"
To: CUBE-HACKERS at MIT-MC
I received the 5th edition of Singmaster's booklet a few days ago, from Logical
Games, Inc.--presumably they have some $4 copies left (they sent mine 1st
class mail)....Bela Szalai (LGI) loaned me a 2x3x3 to play with. Conceptually,
of course, it's the same as the 3x3x3 with moves restricted to <U,D,F^2,R^2,
B^2,L^2>...physically, however, it was MUCH less aesthetic to play with...
turned poorly, and looked ugly. Bela had a couple of them, both of which
were rather more "delicate" than the 3x3x3 (in that they tended to get jammed
or come apart).
-- Mark
Date: 10 Sep 1980 1500-PDT
From: Alan R. Katz <KATZ at USC-ISIF>
Subject: randomizing
To: cube-lovers at MIT-MC
Here's a question for all of you. How do you maximally randomize a cube?
In other words, suppose you were going to have a cube solving compitition
and wanted to make it as hard as possible to solve a cube, what precautions
would you take?
One thing I would do would be to make sure the corners are not correct in
relation to each other. Anyone have other ideas??
Alan (Katz@isif)
------
Date: 10 September 1980 18:12 edt
From: Greenberg at MIT-Multics (Bernard S. Greenberg)
Subject: Re: randomizing
Sender: Greenberg.Multics at MIT-Multics
To: Alan R. Katz <KATZ at USC-ISIf>
cc: cube-lovers at MIT-MC
In-Reply-To: Message of 10 September 1980 18:00 edt from Alan R. Katz
My personal theory on making a cube as hard to solve as possible
is as follows: Put the cube in
some very complex pretty, regular, pattern, such as the Plummer's
Cross with centers trebly rotated, or the Pons Asinorum on that
which looks like the simple Pons: the challenged cubemeister
will INVARIABLY say, "Oh,
this, this is just one of those and solved like this" and go through
a long hairy procedure, hopefully the wrong one,
or make an error, or try hard at understanding it, and so forth.
In short, s/he will take MORE time trying to undo what s/he perceives
as a "simple" hairy hack than a straightforward application of
solve-from-random technology.
Date: Wed, 10 Sep 1980 1854-CDT
Message-id: <337474464.2@DTI>
From: aramini at DTI (Michael Aramini)
To: cube-hackers@mit-mc
Subject: randomizing
Date: 11 SEP 1980 0016-EDT
From: ALAN at MIT-MC (Alan Bawden)
Subject: How do you maximally randomize a cube?
To: CUBE-HACKERS at MIT-MC
I am interested in maximally distant states of the cube. I have often
wondered just what a maximally distant state would look like. I also
wonder HOW MANY of them there are!
Interesting fact (offered without proof (it's not hard)):
Assuming we are counting quarter-twists. If I hand you a cube in a
maximally distant state, and ask you to solve it in as few twists as
possible, you don't have to think at all in order to know what to do
first! ANY first twist will bring it closer to home (after that it
gets harder).
Call a state with this property a "local maximum". Any maximally
distant state is also a local maximum. Also, any symmetric state is a
local maximum. This doesn't mean that a maximally distant state is
symmetric, but it does get you thinking along those lines!
Date: 10 Sep 1980 9:37 pm PDT (Wednesday)
From: Woods at PARC-MAXC
Subject: Re: How do you maximally randomize a cube?
In-reply-to: ALAN's message of 11 SEP 1980 0016-EDT
To: ALAN at MIT-MC (Alan Bawden)
cc: CUBE-HACKERS at MIT-MC
It is not obvious to me that any twist in a maximally distant state
brings you closer to the home position. How do you know there
aren't two equally distant states that are a qtw apart? There is
probably a parity argument that proves this can't be so, but if you
count half-twists as single operations I'd be much surprised if you
could find a simple proof.
-- Don.
Date: 11 September 1980 01:04-EDT
From: Alan Bawden <ALAN at MIT-MC>
Subject: Re:"Assuming we are counting quarter-twists."
To: Woods at PARC-MAXC
cc: CUBE-HACKERS at MIT-MC
I said:
"Assuming we are counting quarter-twists."
I realize that there are people out there who like to count
half-twists as a single twist. I don't. I'm not trying to force my
way of counting twists on anyone. I always try to be carefull to make
this assumption explicit out of courtesy to those who might want to
count otherwise. (Sort of like the Axiom of Choice.)
In fact I DON'T know if that result is true if you count half-twists
3*4
you will get 4.3 * 10^19.
DPC@MIT-AI 09/17/80 17:02:00 Re: cube mode on lisp machines
To: cube-lovers at MIT-MC
cube mode has been fixed (at least temporarily) to work in
the new window system on the lisp machines. to invoke it:
(load "bsg;cubpkg")
(cube)
it is self explanitory once you get in. have fun.
-dpc
Date: Wed, 17 Sep 1980 2047-CDT
Message-id: <338086051.13@DTI>
From: aramini at DTI (Michael Aramini)
To: cube-hackers@mc
Cc: boken@mc
Subject: cubing hazordous to your health?
I saw someone cubing near the digital computer lab at the Univ. of
Ill. today. Considering how much traffic there is on springfeild av
(the street in front of said bldg.) i though it might be a good idea
to be crossing the street while intently solving a cube (i didnt see the
person try crossing the street, but he was at least walking without looking
where he was going).
----
Date:
From:
Subject:
To:
:cube's Lisp Machine instantiation has now been fixed to work in color
on Color LISPM's again.
To invoke, (load "bsg;cubpkg"), wait till it loads, and (cube).
Should be self-documenting.
Date: 23 SEP 1980 0851-EDT
From: ZIM at MIT-MC (Mark Zimmermann)
Subject: report on DC area cube contest
To: CUBE-HACKERS at MIT-MC
-------