Douglas,
Thanks for your reply. I took your suggestion and tried the Cholesky
factorization with dppsv, the positive definite version of dspsv. I
found dppsv to be a great deal faster than dspsv, as might be expected
since dspsv uses a more complicated factorization. However, I ran into
trouble
On Mon, Apr 12, 2010 at 10:27 PM, shotwelm shotw...@musc.edu wrote:
r-devel list,
I have recently written an R package that solves a linear least squares
problem, and computes the multivariate normal density function.
For both of those applications you can use a Cholesky decomposition of
the
r-devel list,
I have recently written an R package that solves a linear least squares
problem, and computes the multivariate normal density function. The bulk
of the code is written in C, with interfacing code to the BLAS and
Lapack libraries. The motivation here is speed. I ran into a problem