Boundary condition in Stokes equation incorrect
In the stokes example, the boundary conditions are implemented in two steps:
- All boundary (velocity) DOFs are identified
- For all those DOFs a function is interpolated on the rhs vector that is either
(0,1)
on the left boundary, or(0,0)
on all other parts of the boundary. The top/bottom-left corners are still set to(0,1)
.
Looking into the solution of the Taylor-Hood example, this boundary conditions is not fulfilled. Especially the y-components of the velocity at bottom and top boundary is not 0.
Is there an error in my thinking? (It is already later in the evening :-) ), Or is there maybe an error in the implementation? I couldn't spot any obvious error.
Note, the same problem shows up when using other stable finite-element pairs.
Not also that the rhs
vector is not properly set to 0 in the inside domain. There also seems to be something wrong.