[Getdp] Glueing together functionspaces

[Kubicek Bernhard]

Dear all,

I have a (basic?) question on FEM: Having a mesh, I have two distinct
sets of regions. For each set, one creates an individual nodal GetDP
functionspace. The two sets are physically connected via an inferface
surface. On each set, one solves basically the same PDE, and on the
interface, the nodal values of one function space should equal the
values on the second space, respectively.

By trial and error, I found this (electric potential) formulation:
Quantity { 
 { Name v ; Type Local  ; NameOfSpace fs_vglob ; }
 { Name vs ; Type Local  ; NameOfSpace fs_vglob2 ; }
}
Equation {
 Galerkin { [ c[]*Dof{d v} , {d v} ] ; In Region[{Vol1}]; Jacobian Vol ;
Integration Int ; }
 Galerkin { [ c[]*Dof{d vs} , {d vs} ] ; In Region[{Vol2}]; Jacobian Vol
; Integration Int ; }
		
 Galerkin { [ c2[]*Dof{ v} , { v} ]; In interface; Integration Int;
Jacobian Vol;  }
 Galerkin { [ -c2[]*Dof{ vs} , { v} ]; In interface; Integration Int;
Jacobian Vol;  }
      
 Galerkin { [ -c2[]*Dof{ v} , { vs} ]; In interface; Integration Int;
Jacobian Vol;  }
 Galerkin { [ c2[]*Dof{ vs} , { vs} ]; In interface; Integration Int;
Jacobian Vol;  }

However, this only gives the same interfacial nodal values is c2[]=c[].
If the result nodal values are stored in a vector, this should give a
matrix simmilar to this (* are some "random" values linear to c[], due
to the gradient):

(  .  .  *  .  | .  .  .  . ) ( v1[1] )
(  .  .  *  .  | .  .  .  . ) ( v1[2] )
(  *  *  c2 .  | . -c2 .  . ) ( v1[3] )
(  .  .  .  .  | .  .  .  . ) ( v1[4] )
(--------------+------------).           = mostly zero
(  .  .  .  .  | .  .  .  . ) ( v2[1] )
(  .  . -c2 .  | .  c2 *  . ) ( v2[2] )
(  .  .  .  .  | .  *  .  . ) ( v2[3] )
(  .  .  .  .  | .  .  .  . ) ( v2[4] )

v1[3]:= v2[2]
I understand, that the absolut value of c2 is of importance, as multiply
a whole row/colum with 2 also changes the "*" values (sorry for this
relatively weak formualtion). 
My struggle is the physical meaning of c2: What is it, and how does it
need it to be choosen, if there e.g. is a different conductivity c3[] in
Vol2?

(Just for explanation, this questions occur by trying to couple a 2d
conducting surface with a 3d conducting volume)

very nice greetings,
 Bernhard