Can I solve the memory problem by using Intel Fortran by Linux???

[ali512342000]

Dear all,

I am developing a FORTRAN code in which a 225000 X 225000, real symmetric, bandwidth positive definite Matrix must be inverted. When I use IMSL and its corresponding subroutine, "DLSXD". I get the following Error message:
"Common block exceeded".
As a remedy I wish to use "Intel FORTRAN compiler 11 professional edition for Linux" with its associated "Math Kernel library". I will be so glad if anyone help me to know whether "Intel FORTRAN compiler 11 professional edition for Linux"
would help me??

 

[mikrom]

Use routines for numerical treatment of sparse matrices.
For example, search for sparse at

http://www.netlib.org/cgi-bin/search.pl

 

[xwb]

Your array is too big. As mikrom said, try sparse matrices.

What are you using the inversion for? If it is for solving simultaneous equations, it would be better to use Choleski LU decomposition instead of inversion.