Can Qhull triangulate a hundred 16-d points?
No. This is an immense structure. A triangulation of 19, 16-d points has 43 simplices. If you add one point at a time, the triangulation increased as follows: 43, 189, 523, 1289, 2830, 6071, 11410, 20487. The last triangulation for 26 points used 13 megabytes of memory. When Qhull uses virtual memory, it becomes too slow to use. »Approximation questions »How do I approximate data with a simplex Qhull may be used to help select a simplex that approximates a data set. It will take experimentation. Geomview will help to visualize the results. This task may be difficult to do in 5-d and higher. Use rbox options ‘x’ and ‘y’ to produce random distributions within a simplex. Your methods work if you can recover the simplex. Use Qhull’s precision options to get a first approximation to the hull, say with 10 to 50 facets. For example, try ‘C0.05’ to remove small facets after constructing the hull. Use ‘W0.05’ to ignore points within 0.05 of a facet. Use ‘PA5’ to print the five largest facets by
