Ok, I've found the answer - the dgesvd function from the lapack addon. Yoel
On Jan 14, 2008 9:12 AM, Yoel Jacobsen <[EMAIL PROTECTED]> wrote: > Hello, > > It seems to me that the SVD implementation from > ~system\packages\math\svd.ijs is inefficient for large matrices (i.e. > calculating the SVD of the color matrices of an 640x500 image never > ends..). > > You may even try: $ svd i. 200 200 > (in NumPy it takes less than a second) > > Before I reinvent the wheel, is there an efficient implementation of > SVD available? > > Yoel > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
