home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!convex!darwin.sura.net!jvnc.net!yale.edu!think.com!news!columbus
- From: columbus@strident.think.com (Michael Weiss)
- Newsgroups: sci.math
- Subject: space-eating curves
- Date: 14 Aug 92 16:51:16
- Organization: Thinking Machines Corporation, Cambridge MA, USA
- Lines: 62
- Distribution: sci
- Message-ID: <COLUMBUS.92Aug14165116@strident.think.com>
- NNTP-Posting-Host: strident.think.com
-
- As recent posts have pointed out, there is a continuous map from the unit
- interval, I, onto the unit square, IxI. The first one was discovered by
- Peano; the best known is due to Hilbert. Pictures of the first few
- approximations to the Hilbert curve can be found in many math
- popularizations, e.g., Kasner and Newman's "Mathematics and the
- Imagination". It's a cute and fairly easy exercise to write a
- program that inputs n and outputs the order n approximation to the Hilbert
- curve. (Just a little more interesting than the cliched Towers of Hanoi.)
-
- The Hahn-Mazurkiewicz theorem asserts that a Hausdorff space T is a
- continuous image of I if and only if T is connected, locally connected,
- compact, and second countable. I^2 and more generally I^n (n any integer)
- qualify.
-
- Special case of interest: the Hilbert cube I^{aleph_null} also qualifies
- (call it H). Topologically, H is the product of countably many copies of I
- with the product topology. Equivalently, for x in R^{aleph_null}, define
- ||x|| to be sqrt(sum (x_i)^2) (setting ||x|| = infinity of course if the
- series diverges; otherwise this is the usual l^2 norm); then sets of the
- form N_e(x) = {y : ||y-x|| < e} form a neighborhood base (with e>0, of
- course.)
-
- A few years ago I noticed that this special case of the H-M theorem enjoys
- a short and (IMHO) elegant proof-- once you have the continuous map
- from I onto IxI. Here goes:
-
- Let x --> (f(x), g(x)) be the continuous map from I onto IxI.
- Construct the map from I to I^{aleph_null} as follows:
-
- x--> (f(x), f(g(x)), f(g(g(x))), ...)
-
- Motivation: think of (f,g) as a tool for splitting a element x of I
- into two elements. Starting with the list (x), we obtain longer and
- longer lists by repeatedly splitting the last element. The following
- diagram illustrates the process:
-
- x
- |\
- | \
- | \
- f(x) g(x)
- | \
- | \
- | \
- f(g(x)) g(g(x))
- | \
- | \
- | \
- f(g(g(x)) g(g(g(x)))
- | \
- . .
- . .
- . .
-
- It follows immediately from the definition of the product topology that
- this map is continuous and its image is dense in H. However, its image
- is the continuous image of a compact set and is hence compact, hence
- closed, hence all of H!
-
- Returning the general H-M theorem: if T satisfies 1)-4), then by the
- Urysohn imbedding theorem, T can be imbedded as a subspace of H. One can
- in fact turn this observation into a proof the H-M theorem.
-