Hello,
I have to solve a typical Ax=b problem, with an sparse A matrix, and I
want to use the LINBCG function.
I am using the sparse IDL functions, but due to the huge size of the
my A matrix (almost 65000*65000 float values) I have created my own
version of the sprsin function whitch is able to create correctly the
sparse matrix of A (A_sp) avoiding memory problems. The result vectors
A_sp.sa and A_sp.ija have a length of 313953 positions each.
SPRSAX and SPRSTP functions are able to process A_sp, but when I use
result=LINBCG(A_sp,b,x,ITER=iterations)
the function LINBCG apparently "does nothing": iterations=0, result=x,
and no error message. The code jumps to the next line as if nothing
wrong had happened.
I have used this code with a small A_sp, and works correctly, and when
using huge arrays always appear the error message "Array has too many
elements."
Why LINBCG apparently does nothing? It is possible that this function
has no error handling?
Thanks,
Marc
|
