[Getdp] non symetric tensor - Hall effect

[alain]

Dear all,

I discovered recently this very interesting softwares, GMSH+Getdp.

To solve an electric 2D problems involving Hall effect, I started from 
the example www.geuz.org/getdp/wiki/Anisotropy2D, but I set  
non-symmetric off-diagonal terms in the tensor, ex : 
"k[Sur]=Tensor[1,-0.5,0, 0.5,1,0, 0,0, 1];".

So doing the resolution did not converge with :
"Galerkin { [k[]*Dof{d n} ,{d n} ]; ..."

However multiplying the test function by the transposed matrix instead 
of the Dof in
"Galerkin { [Dof{d n} ,kt[]*{d n} ]; ..."
or simply multiplying the test function by the identity matrix :
"Galerkin { [k[]*Dof{d n} ,id[]*{d n} ]; ..."
resulted in the convergence of the resolution.

Have you an idea why the test function has to be multiplied by a matrix 
when non-symmetric non-diagonal terms  ?

Thank's a lot

Alain M.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: aniso.pro
URL: <http://www.geuz.org/pipermail/getdp/attachments/20090804/d18abba8/attachment.pro>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: aniso.geo
URL: <http://www.geuz.org/pipermail/getdp/attachments/20090804/d18abba8/attachment.geo>