[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference rusure::math

Title:Mathematics at DEC
Moderator:RUSURE::EDP
Created:Mon Feb 03 1986
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2083
Total number of notes:14613

1243.0. "Test your intuition in higher dimensions" by GUESS::DERAMO (Dan D'Eramo) Mon May 21 1990 12:33

Path: shlump.nac.dec.com!decuac!haven!aplcen!uakari.primate.wisc.edu!zaphod.mps.ohio-state.edu!tut.cis.ohio-state.edu!ucbvax!vms.huji.ac.il!KLEIN
From: [email protected]
Newsgroups: sci.math
Subject: Intuition in higher dimensions (Summary)
Message-ID: <[email protected]>
Date: 8 May 90 10:45:42 GMT
Sender: [email protected]
Reply-To: [email protected]
Organization: The Hebrew University, Jerusalem, Israel
Lines: 199
 
 
 Following is a summary of responses I received to the question
below. Also included is a quick solution to the question.
For those who haven't haven't seen the question yet and wish to try
their hand at it, be careful not to read the solution.
 
 
    How good is a mathematicians raw intuition of high dimensional
spaces ? I would like to survey responses to the following question.
I am not interested in solutions as the problem is easy to solve. I
want people to "guess" what the right answer is using only their "feel"
for the problem.
    The question is as follows :
 
    In two dimensions we draw the unit square and divide it into four
quadrants. In each quadrant we draw the largest circle we can inscribe
 
   ---------------------      IGNORING now the unit square and its
      *****     *****         subsquares, we draw in the center
     *     *   *     *        region the largest circle bounded by
    *       * *       *       the four previous circles and call its
     *     *   *     *        radius R2.
      ***** *** *****
   --------*   *--------
      ***** *** *****
     *     *   *     *
    *       * *       *
     *     *   *     *
      *****     *****
   ---------------------
 
    In a similiar fashion we define R3 by dividing the 3-dimensional unit
cube into 8 octants, inscribing the largest sphere we can in each octant,
and then, ignoring the cubes, inscribing the largest circle we can in the
center bounded by the previous circles.
 
    Continuing in the same fashion we define Rn for all n.
 
    The question is to guess which of the following
 
         lim    Rn     equals.
        n->00
 
    a) 0          b) PI/8
 
    c) infinity   d) 1/2
 
 
    Again, to actually solve the problem is simple. What I want is for you
to get a feel for the problem and then guess the correct answer without
explicitly finding the solution.
 
    Please send your replies to KLEIN@HUJIVMS and I will post a summary.
Please DON'T reply if you have already seen this problem before or if
you got your answer by actually solving the problem.
 
    Thank you,
 
 
       David Klein
 
       KLEIN@HUJIVMS
 
 
 
    And here is the solution
    ------------------------

	[I split out this section.  It appears as reply .1. (Dan)]
T.RTitleUserPersonal
Name
DateLines
1243.1continued from .0GUESS::DERAMODan D&#039;EramoMon May 21 1990 12:37143
    Thank you,
 
 
       David Klein
 
       KLEIN@HUJIVMS
 
 
 
    And here is the solution
    ------------------------

    The correct answer is  lim Rn  = infinity. This counter intuitive
result is due to the fact that the spheres in the individual quadrants
only bound the center sphere along the direction of the cubes diagonal.
Since the length of the diagonal goes to infinity as root(n) we see
that the central sphere really isn't very well bounded.
 
    To be specific, the radius of each bounding sphere is clearly 1/4.
The center of the sphere is at a distance of root(n)/4 from the origin.
Along the diagonal of the cube the center sphere just touches the corner
spheres so its radius is root(n)/4 - 1/4. In 9-dimensional space, Rn=1/2
and the central sphere is tangent to the surface of the unit cube. For
any higher dimension the sphere protrudes outside of the unit cube.
 
 
 
   The final vote was divided as follows
 
a) Lim = 0         4 votes
b) Lim = PI/8      5 votes
c) Lim = infinity  6 votes
d) Lim = 1/2       3 votes
 
   Thus one out of three had the right intuition (Which is better
than I would have thought).
 
   Following are some of the comments I received along with the
answers. They are interesting as they show some insight into
peoples thought processes. The last comment includes another
question on insight, this time in only four dimensions (which
is hard enough as is.) Beware that the question is closely followed
by the solution.
 
1)
 
My intuition was the answer is 1/2.  I have calculated
the correct answer and found it surprising .  I think
this is more a reflection of poor intuition about
dimension tending to infinity rather than poor intuition
about dimension n for n large.
 
 
 
2)
 
To my opinion b) is the right solution.
 
a) should not be right because it seems to me that the sequence Rn is increasing
(consider R1=0).
c) is also wrong because Rn <= 1.
and d) means that the inner circle would reach the boudaries of the surrounding
square (cube, ...) what seems impossible in my intuition.
 
 
 
3)
 
Thanks for an entertaining problem.  I'm afraid my intuition is in
need of brushing up however; I incorrectly guessed (a) 0.
 
I suspect that infinite-dimensional intuition (or an asymptotic
version thereof, as in your example) is hard to come by.  (Who can
visualize Hilbert space, even one of countably infinite
dimensionality?)
 
You might try posing questions on Euclidean geomentry in four
dimensions, as an alternative.  (To what extent does mathematicians'
spatial perception [the ability to rotate figures, etc.] carry over to
four dimensions?)  I think I'd do a bit better.  At least, I hope so.
 
 
 
4)
 
In response to your survey:
First, I thought the answer is 1/2.
But that is because I misunderstood the problem.
(I thought that ignoring the little cubes
meant also ignoring the little spheres
inscribed in the little cubes.)
Then, I thought that the answer is pi/8.
Then, I could not make up my mind between 1/2 and pi/8.
Then, it occurred to me that the answer might be infinity
because the length of the diagonal gets long.
I was about to guess infinity when it occurred to me
that if the radius is greater than 1/2,
then the sphere sticks out of the cube.
Therefore, I am now totally confused.
But there is no law against the sphere
sticking out of the cube provided
it does not stick into the 2-to-the-n-th spheres,
so I guess infinity anyway.
 
 
5)
 
The question was too easy
In particular you could `see' the answer algebraically rather than
geometrically, since the distance of a corner to the center tends to
infinity, while the distance between adjacent corners is constant.
 
If you REALLY want to test higher dimensional intuition try asking the
following.
 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
QUESTION:
~~~~~~~~~
If you slice a cube along a plane perpendicular to an interior diagonal
the intersection of the cube with the plane is shaped like a regular hexagon.
Now take a 4-dimensional hypercube, and slice it along a three dimensional
hyperplane which perpendicularly bisects an interior diagonal. The three
dimensional solid which results is a ...... ?
 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Ask people to try to do it in their heads without writing anything, and ask
them to indicate how long it took.
 
(answer below)
 
 
octahedron
 
Proof: It is a regular platonic solid. It touches all `faces' of the hypercube
and thus has 8 faces. QED.
 
 
    Thanks for your attention
 
    David Klein
    KLEIN@HUJIVMS
 
Disclaimer: These opinions aren't even my own, never mind anybody else's.
1243.2HPSTEK::XIAIn my beginning is my end.Mon May 21 1990 14:345
    This problem is discussed in a book "coding theory" or something like
    that by THE Hamming of the Hamming metric.  Anyway, I don't think a
    correct answer to this problem is indicative of anything.
    
    Eugene