It seems there are currently some issues with linearizing component operators when e.g. doing stress-free boundary conditions in Cartesian domains. This can currently be worked around by creating constant unit-vector fields in Cartesian problems, and taking dot products with these, but the underlying issue should be addressed.
It seems there are currently some issues with linearizing component operators when e.g. doing stress-free boundary conditions in Cartesian domains. This can currently be worked around by creating constant unit-vector fields in Cartesian problems, and taking dot products with these, but the underlying issue should be addressed.