If you need efficient numeric matrix operations, your best bet is to
move away from higher level abstractions like seqs and (especially)
refs and just use a one-dimensional Java array of primitives of size
(* m n), and indexed as (+ (* i m) j) or (+ (* j n) i) using primitive
arithmetic for speed.
Could you possibly put up a minimal example of code that shows the
problem? I'm having a hard time following exactly what you're doing
but would like to help. :)
sincerely,
--Robert McIntyre
On Fri, Feb 4, 2011 at 10:20 AM, Nick wrote:
> I've got a matrix transformation that I'm doing which i
I've got a matrix transformation that I'm doing which is really slow.
I'm looking for ways to speed it up.
source is a NxD matrix (seqs of seqs of values). I will set and read
this matrix.
dest is a NxN transformation of f in which indices are mapped to
specific indices of f. I will only read th
