[Getdp] Magnetodynamics a-v formulation 3D ?

David Colignon David.Colignon at ulg.ac.be
Tue Apr 4 16:30:26 CEST 2006


> I tried lots o f different combinations of solvers, preconditioners and other 
> solver parameters, but get typically something like this:
> 
> 
> Loading   : Problem definition 'InductionTestBox.pro'
> Loading   : Problem definition 'Jacobian_Lib.pro'
> Loading   : Problem definition 'Integration_Lib.pro'
> Loading   : Problem definition 'MagDyn_av_3D.pro'
> Info      : Selected Resolution 'MagDyn_vf'
> Loading   : Geometric data 'InductionTestBox.msh'
> Info      : System 'Sys_MagDyn' : Complex, Frequency = 10 Hz
> P r e - P r o c e s s i n g . . .
> Operation : Treatment Formulation 'MagDyn_avf'
> Info      :   Generate ExtendedGroup '_CO_Entity_14' (EdgesOf)
> Info      :   Generate ExtendedGroup '_CO_Entity_11' (NodesOf)
> Info      :   Generate ExtendedGroup '_CO_Entity_12' (NodesOf)
> Resources : cpu 5.812363 s / mem 0 kb
> E n d   P r e - P r o c e s s i n g
> P r o c e s s i n g . . .
> Operation : Generate[Sys_MagDyn]
> Solver    : Loading parameter file 'solver.par'
> Info      : Setting System {A,b} to zero
> Resources : cpu 16.5000 s / mem 0 kb
> Operation : Solve[Sys_MagDyn]
> Solver    : Scaling system of equations
> Solver    : RCMK algebraic renumbering
> Solver    : N: 264674, NZ: 4949840, BW max/avg: 110/18, SW max: 260437
> Resources : cpu 21.325332 s / mem 0 kb

Hi,

can you try with Nb_Fill set to at least 120 in solver.par ( greater than the max Band Width of the matrix) and with GMRES ( Algorithm = 8 ) and Krylov_Size = 100 ?

But given the size of your mesh, it will take a fair amount of time and RAM. Can I suggest that you test your formulation with a much more smaller mesh ?

Cheers,

Dave



More information about the getdp mailing list