[Getdp] Difficult BC

[Christophe Geuzaine]

Nacho Andres de la Fuente wrote:
> My geometry consists of two concentric cylinders inside of a sphere
> (i.e. geometrically is axialsymmetric) .. the BC however, are NOT
> axialsymmetric and depend on the colatitude angle:
> 	* which type of Jacobian should I use? If I understood correctly, I
> have a geometry in gmsh, I apply the normal Laplacian(phi) = 0 and by
> using Jacobian, any Riemann geometry can be solved, am I right? could
> you comment a little on this please?

The Jacobian object in GetDP is related to the transformation of
the reference geometrical elements (defined in some parametric space)
into the computational space. If the geometry you define in Gmsh
directly represents your "real" geometry, you don't have to worry about
jacobians--just use the Vol jacobian for N dim->N dim transformations,
or the Sur jacobian for (N-1) dim->N dim transformations.

> 	* for those Neumann BC on he surface of the outer sphere I need to
> define a function that depends on the point of the boundary in which we
> are and the value of phi (the exponential of the value of the potential
> at that point and the coltitude at that point), how can I do that? I
> mean to recover the value of the position of the node that it is being
> used in that moment? because it is better to use nodes here isn't it? (I
> mean for the FunctionSpace ..). How can I include the value of what I am
> going to solve into the function?

You should use

- $X, $Y and $Z in the Constraint object (for Dirichlet BC). These
constraints are evaluated during the pre-processing phase ("getdp -pre").

- X[], Y[] and Z[] in the Formulation object, e.g. for Neumann BC. These
constraints are taken into account during the processing phase ("getdp
-cal").

Christophe

-- 
Christophe A. Geuzaine
Applied and Computational Mathematics, Caltech
geuzaine at acm.caltech.edu - http://geuz.org