For post-processing purpose (e.g., monitoring residuals) I find the function get_dirichlet_dof_values very useful.
However, I noticed a possible inconsistency in the API: some related functions such as get_free_dof_values or get_cell_is_dirichlet provide methods that accept a ::FEFunction, while get_dirichlet_dof_values must be called on the finite element space associated with the function. The code below illustrates the difference:
fv = get_free_dof_values(xh)
dv = get_dirichlet_dof_values(get_fe_space(xh))
I'm wondering if it makes sense to add an overload accepting a FEFunction, following the pattern of get_cell_is_dirichlet defined in:
|
function get_cell_is_dirichlet(f::FEFunction) |
See #1186 for a possible implementation. Thanks in advance
For post-processing purpose (e.g., monitoring residuals) I find the function
get_dirichlet_dof_valuesvery useful.However, I noticed a possible inconsistency in the API: some related functions such as
get_free_dof_valuesorget_cell_is_dirichletprovide methods that accept a::FEFunction, whileget_dirichlet_dof_valuesmust be called on the finite element space associated with the function. The code below illustrates the difference:I'm wondering if it makes sense to add an overload accepting a
FEFunction, following the pattern ofget_cell_is_dirichletdefined in:Gridap.jl/src/FESpaces/FESpaceInterface.jl
Line 53 in 9cf59d7
See #1186 for a possible implementation. Thanks in advance