Howdy!
I am using SuperLU_dist to solve a sparse linear system Ax = b in a massively parallel CFD solver. However, sometimes my matrix 'A' is highly ill-conditioned with the condition number approaching 1.e18-1.e19 beyond and hence regular LU decomposition does give an accurate solution. I was wondering if there any in-built methods in SuperLU_dist to handle matrices with large condition numbers (upto 1.e19).
Thank you very much.
Howdy!
I am using SuperLU_dist to solve a sparse linear system Ax = b in a massively parallel CFD solver. However, sometimes my matrix 'A' is highly ill-conditioned with the condition number approaching 1.e18-1.e19 beyond and hence regular LU decomposition does give an accurate solution. I was wondering if there any in-built methods in SuperLU_dist to handle matrices with large condition numbers (upto 1.e19).
Thank you very much.