[Getdp] Re-2: Magnetic force

[Christophe Geuzaine]

Colignon David wrote:
> Hi,
> 
> in the file solver.par, try to increase the Nb_Fill value from 20 to 40 or
> 60 or ... until the solver converge to a solution.
> 

With the new version released last week, you can also compile GetDP with
PETSc support (./configure --enable-petsc), and then use the pretty
efficient direct sparse solvers provided by (or interfaced with) PETSc,
e.g. "getdp file.pro -solve -ksp_type preonly -pc_type lu".

For problems of moderate size, this works surprisingly well! For
example, we used it recently to solve fairly complex 2D problems in
complex arithmetic: 175,000 complex unknowns required about 1 Gb of RAM
and 1 minute of CPU time; 700,000 complex unknowns required 4 Gb and 9
minutes.

Take care,

Christophe

-- 
Christophe Geuzaine
Assistant Professor, Case University, Mathematics
christophe.geuzaine at case.edu