Edouard Oudet
Edouard Oudet
The following example produces the same kind of error even in 1D whereas the example of @dlfivefifty above is ok. ```julia julia> S = Fourier(0..2π) ; julia> f = Fun(x->...
Nice catch, thank you! I updated extend_interpolation.jl with the ordering required by Ferrite. The file `extend_interpolations_Ferriteorder.jl` is not needed anymore. It seems to be ok up to order 5. I...
.. I just read in `MixDofHandler.jl` ```julia @assert nfacedofs == 1 "Currently only supports interpolations with nfacedofs = 1" ``` Is this limitation also for H^1 problems?
I made some tests using another library to compare. I think the assembling part for order 4 and 5 may be incorrect. Perhaps due to `ncelldofs > 1`?
I used Getfem++ to assemble mass and stiffness matrices to solve the eigenvalue test problem we already discussed. With a small mesh I obtain exactly the same error using Ferrite...
Still the same mistake! Yes you are right. When I use Getfem++ I chose an exact quadrature order..which may explain the difference
I think order 8 is exact for degree 4. Order 10 would be required to be exact for order 5. It could be the occasion to extend quadrature rules up...
You are right, I was thinking to the stiffness matrix
I tried to follow the style of `gaussquad_tri_table.jl` to extend Gauss quadrature formula up to order 20 using the python package `quadpy`. Does it look reasonable? More generally, it could...