ztdep
|
August 11, 2006 13:16 |
problem about direct solver for sparse matrix
Hi: I just compiled sucessfully a free avalable sparse direct solver under linux OS, I validate my procedure by a simple 2-D heat conduction problem in a square domain with all its four BC=2, so the exact solution is the terperature filed=2 in the square domain! But i have the following problem, when i use the same number of grid points in the x and y direction , i have the exact solution ,but when i set different number of gridpoints in the x and y direction 100by 50, or 5by10, the solution has a large discrepancy between the exact value! the package use a multi frontal method with pivot method! Regards
|