Francesc Verdugo
Francesc Verdugo
I would say this is correct. When you attach Dirichlet BCs to a matrix, you get a matrix + a vector. Is this causing trouble in some user code?
AutoDiff of interpolate_everywhere fails. Mismatch among FE space vector type and dual number types.
@amartinhuertas IMO the workaround already looks better than anything else I can think of.
AutoDiff of interpolate_everywhere fails. Mismatch among FE space vector type and dual number types.
> Ok. So we should find a way to automatically trigger the workaround whenever autodiff tries to differentiate interpolate_everywhere. Correct? Or perhaps use the workaround directly. Since I guess it...
In collaboration with @amartinhuertas
@tmigot I can reproduce the error.
@Balaje any thoughts on this?
Hi @ConnorMallon I can reproduce this in the `release_0.17` branch with the following error: ```julia ERROR: LoadError: AssertionError: A check failed Stacktrace: [1] macro expansion @ ~/Code/jl/Gridap.jl/src/Helpers/Macros.jl:60 [inlined] [2] get_triangulation(f::Gridap.MultiField.MultiFieldCellField{Gridap.CellData.ReferenceDomain})...
Unfortunately this is not a dummy bug that can be solved easily. It is a whole functionality still not implemented in the library. To implement this one needs to understand:...
The `CLagrangianFESpace` has been implemented in commit 5d9b892c9df4aa78ad43e98c5932b08d662d35ed
Good point @eschnett ! In general, we want a way of passing all options supported by `WriteVTK.vtk_grid`. The signature of `Gridap.writevtk` has to be modified with care in order to...