3D analogue of intalledges etc

Feature Request
1

Hi,
I’m trying to explore some stabilized FE schemes involving the integrals of gradient jumps over the mesh facets. This works fine in 2D with intalledges. I know that the 3D analogue intallfaces exists but it does not seem to work. Here is the minimal example :

load “msh3”
int n=10;
mesh3 Th = cube(n, n, n);
fespace Vh(Th,P1);
Vh u,v;
u=1;
cout <<intallfaces(Th)(u)<<endl;

This gives the obviously wrong answer 0 (in version 4.2.1)

If I continue with

macro dN(u) (dx(u)*N.x+dy(u)*N.y+dz(u)*N.z) //
solve test(u,v)=intallfaces(Th)(jump(dN(u))*jump(dN(v)) )
-int3d(Th)(v);

then it gives
Assertion fail : (0)
line :202, in file problem.cpp

Could you please correct this issue?
Moreover, it would be useful to create a 3D analogue of nTonEdge and to check if jump and mean work properly in 3D.

Thanks in advance,
Alexei

2

Dear All,

I share the same request than Alexei; everything is fine in 2d with intalledges, but the command intallfaces(jump(dn(u))*jump(dn(v))) with v,u in P^2 produces a “Segmentation fault: 11”.

Regards, Arnaud

3

Yes, I know, the Discontinuous Galerkin is not coded in 3d, so all this stuff given wrong result.

Sorry, Frederic Hecht