[petsc-users] CG or GMRES

[Jed Brown]

Alan <zhenglun.wei at gmail.com> writes:

> Thanks, Dr. Knepley,
>       Can I treat AMG and MG like this way. To solve problems with 
> non-symmetric matrices, MG needs users to provide the coarse mesh 
> information. This take more complexity on coding than using AMG; but MG, 
> generally, do better job than MG for non-symmetric problems.

It's hard to generalize.  FAS multigrid with conservative
discretizations and element agglomeration is popular for
transport-dominated problems like steady-state transonic flows.  Other
geometric MG schemes could perform worse than a given algebraic
multigrid.