home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math
- Path: sparky!uunet!caen!sol.ctr.columbia.edu!shire.math.columbia.edu!dy
- From: dy@shire.math.columbia.edu (Deane Yang)
- Subject: Re: proof wanted 2
- References: <1iqcp7INNoph@skeena.ucs.ubc.ca> <1993Jan11.030012.26208@Princeton.EDU> <1993Jan11.210012.18587@Princeton.EDU>
- Sender: nobody@ctr.columbia.edu
- Organization: Mathematics Department, Columbia University
- Date: Mon, 11 Jan 1993 22:23:38 GMT
- Message-ID: <1993Jan11.222338.17682@sol.ctr.columbia.edu>
- X-Posted-From: shire.math.columbia.edu
- NNTP-Posting-Host: sol.ctr.columbia.edu
- Lines: 44
-
- In article <1993Jan11.210012.18587@Princeton.EDU> tao@fine.princeton.edu (Terry Tao) writes:
- >>>In article <ARA.93Jan10172314@camelot.ai.mit.edu> ara@zurich.ai.mit.edu (Allan Adler) writes:
- >>>>
- >>>>
- >>>>True or false: A metric space (X,d) is locally compact if and only if
- >>>>for every point p of X and every closed subset Y of X, there is a
- >>>>point q of Y such that d(p,q) = inf {d(p,r) | r in Y}.
- >>>>
- >>>
- >
- >False. A friend suggested the following counter example:
- >
- >Let X be the space of infinitely many copies of {0} union {1/n: n = 1, 2,
- >...} with all the 0's glued together. the metric is the natural one
- >between two points on the same copy, or through 0 on different copies. The
- >space is not locally compact at 0, yet every closed set and every point has
- >a point of closest approach. Basically the reason for this is that the
- >only problematic point is 0, and the fact that Y is closed then guarantees
- >that 0 is in Y.
- >
- >Terry
- >
- >
-
- The point here is that, except for 0, X is a discrete and therefore
- locally compact space. The conjecture needs a further assumption,
- something that makes the space everywhere nondiscrete.
-
- If I'm not mistaken, a reasonable assumption is that X is a length space.
- This means that the distance between two points in X is equal to the infimum
- of the lengths of continuous curves joining the two points. In particular,
- this means that given two points that are a finite distance apart,
- then you can always move one point a little bit and make the distance a little
- smaller.
-
- Now suppose X is not locally compact. Then we can find a bounded sequence
- that has no convergent subsequence. In particular, this is a closed set.
- Pick a point outside the sequence that is a finite distance from the sequence.
- By nudging the points in the sequence, you should be able to obtain a closed
- set and a point such that the distance between the two is not realized by
- the distance between the given point and a point in the closed set.
-
-
-
-