ahojukka5
commented
5 years ago Defining fields element-wise means that one should not use a dictionary to define fields, that is, instead of having
T = Dict(1 => 1.0, 2 => 2.0, 3 => 3.0)
element1 = Element(Seg2, (1, 2))
element2 = Element(Seg2, (2, 3))
update!([element1, element2], "temperature", T)We should do something like
update!(element1, "temperature", (1.0, 2.0))
update!(element2, "temperature", (3.0, 4.0))So both elements are sharing node 2, but the first element "local node temperature" is 2.0 and second element 3.0.
But then, thinking of postprocessing, I think it's not possible to visualize this kind of nodal jump easily using Paraview. In principle, there's only one value per node. You could have two nodes with the same coordinates so that you can have two different values, but that means the elements are physically disconnected.
T = Dict(1 => 1.0, 2 => 2.0, 3 => 3.0, 4 => 4.0)
element1 = Element(Seg2, (1, 2))
element2 = Element(Seg2, (3, 4))
update!([element1, element2], "temperature", T)If that is not a problem, it should be possible to write a function which alters the connectivity map such a way that the elements are physically disconnected. And that should be done after the actual analysis unless you are doing some kind of DG-FEM. An interesting problem indeed, but definitely solvable.
http://www.xdmf.org/index.php/XDMF_Model_and_Format#Attribute
Actually, there are some DG things which can be defined using ElementFamily, but I'm not familiar with how this should be defined.
The manual states that
How do I define a field element-wise? Do I need to define a different name for the field in every element?
What I wan't to do is viewing the results in paraview without using the lsq interpolation, which means strains, stresses ... jump between nodes.