home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math
- Path: sparky!uunet!decwrl!parc!xerox!friedman
- From: friedman@ieor.berkeley.edu (Eric Friedman)
- Subject: Convex Dual Question
- Message-ID: <1992Jul21.164426.12627@parc.xerox.com>
- Sender: news@parc.xerox.com
- Reply-To: friedman@ieor.berkeley.edu (Eric Friedman)
- Organization: U. C. Berkeley
- Date: Tue, 21 Jul 1992 16:44:26 GMT
- Lines: 24
-
- Given a convex set $ G \in R^n$ we can define a map
- $T : G \rightarrow S^{n-1}$ the n-1-sphere by mapping the normal to the surface
- of the convex set in the obvious manner.
- This is mentioned in the standard books on convexity under the name
- of convex duality.
-
- I'd like to know if any serious study of this map has been done.
- For example:
-
- 1) The map is typically injective but not surjective. It seems straightforward
- to prove that we can construct a G' which is arbitrarily close to G (in some
- natural metric) such that the map is surjective.
-
- 2) Consider the intersection of a hyperplane H with G. The map of H\intersection G
- (which is a convex set) forms a subset of S^n. What are the properties of this
- subset? For example let n=3, then the intersection of a plane with
- G maps into a loop on S^2. I don't think that this loop can be too complicated,
- but i`m not sure exactly how to define this.
-
- Presumably, somewhere this has been studied. Any pointers, hints, or
- proofs would be greatly appreciated.
-
- thanks
- eric
-