On Wed, Oct 8, 2008 at 2:37 AM, Patrick Mihelich <patrick.mihelich@gmail.com
wrote:
In my opinion, a Boost geometry library must have at least basic support for n-dimensional computational geometry. Let's start with able to handle points of arbitrary dimension and calculate distances between them (given some metric). This is easy to do and already sufficient to implement many useful spatial indices. I have not looked at Barend's geometry library in detail, but on the surface it looks generic enough to support this.
I very much need this for the generic DM/KDD algorithms that I am working on. We started work on a clustering algorithms library at boostcon this year, and support for n-dimensional spatial indexes/queries is still needed to improve the performance of some of the algorithms. WRT the clustering library, I dropped the ball over the summer when things heated up at work and school, but will be picking it up again Real Soon Now. If you have any interest in clustering or classification algorithms, then please contact me! </shameless_plug>. Jon