home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math.symbolic
- Path: sparky!uunet!news.smith.edu!orourke
- From: orourke@sophia.smith.edu (Joseph O'Rourke)
- Subject: Re: Finding convex hull ?
- Message-ID: <1992Jul23.115609.16096@sophia.smith.edu>
- Organization: Smith College, Northampton, MA, US
- References: <1992Jul21.161451.8912@taloa.unice.fr> <1992Jul22.113118.24947@ai.univie.ac.at> <PHR.92Jul22170347@soda.berkeley.edu>
- Date: Thu, 23 Jul 1992 11:56:09 GMT
- Lines: 12
-
- In article <PHR.92Jul22170347@soda.berkeley.edu> phr@soda.berkeley.edu (Paul Rubin) writes:
- >The best known algorithms for convex hull are very complicated.[...]
- >I believe "Compuational Geometry" by Preparata and Shamos has some
- >info on convex hull algorithms.
-
- Yes. Graham's algorithm does not generalize to d > 2. There
- are several algorithms that do generalize, however. One is gift-wrapping.
- Another is beneath-beyond. Both are described in P&S. Edelsbrunner's
- book has a detailed description of beneath-beyond that is close to
- code. The worstcase size of the hull of n points in d dimensions is
- n^{[d/2]} (that's floor of d/2), and the beneath-beyond algorithm
- nearly achieves that complexity.
-