I'm trying to use Pardiso to solve a Stokes problem using weak galerkin finite element method.
All matrix and code are attached and tested at MKL 2019.5.281, Visual Studio 2019, WIN10
The linear system is a saddle point system which can be solved in Full sysem or Schur complement system. Both linear system are symmetric indefinite.
//2nd order FEM with 384 tets
//System1. Full system, where u={u0,ut,v0,vt,w0,wt} and u0,v0,w0 is not shared by neighboring element
[K G * {u = {b_u
G' 0] p} b_p}
system size = 4080 x 4080
//System2. Schur complement , where u_bar={ut,vt,wt} as u0,v0,w0 is eliminated
[A B * {u_bar = {b_u
B' C] p} b_p}
system size = 2640 x 2640
left: Full matrix right: Schur complement matrix
Obviously, System2 can save a lot of memory for a large problme.
Q: I can use Pardiso to solve the System1 very well (residual = 1e-16).
But When I solve the System2, it returns the very high residual (1e-2 -> 1e4).
Any suggestions to solve this symmetric indefinite system?
I already tried the solution of setting following iparm but it is not helps
iparm[9] = 8;
iparm[10] = 1;
iparm[12] = 1;