Was "intall0d" intended for any purpose?

General Discussion
1

I was trying to find a way to just get the matrix for a differential equation ( no integration/ bilinear form ). That is just the matrix of an expression containing various derivatives on some fespace.

I found some things like “int0d” and most recently “intall0d” which
sounds like what I’m looking for but it is not used anywhere.

grep -n  "intall0d" `find src -type f -name "*"`  | grep enum 
src/fflib/problem.hpp:285:  enum typeofkind  { int2d=0, int1d=1, intalledges=2,intallVFedges=3, int3d = 4, intallfaces= 5,intallVFfaces=6,int0d=7,intall0d=8 } ; //3d

I was going to add something like “intzed” for this purpose since I can’t find another way
to get there. Thanks. Just making some matricies from varf I think I mostly understand
what is going on and at least for P0 elements things look about right by inspection :slight_smile:

2

Can you given a small exact of matrix construction, because you need to explain howto convert function in number to build the matrix.

3

I just wanted to get the matrix representing a discretized differential equation- not the
variaional integral . I was wondering if that had something to to with support for strong form.
Thanks.

4

So you want to make finite difference scheme!

5

Well, that is intermingled lol. I have a case I’m doing for learning but also a start
on a practical application. It is just diffusion but with a spatially varying diffusivity
that depends on the species concentations. In essence, it is like a two component
glue such as epoxy with one component circulating in an inert state activated
at a surface. When the activated hardener meets the flue, the diffusivity drops.

Right now I’m doing finite difference along the time axis . And yes I wanted to get
the depednence of the time derivatives on the contirbutors to the laplacian.
But, it should be independent of that. That is, you can get a matrix in
FEM right? Its just now you only support the weak form.

Thanks.

6

OK, for me the more simple to have the equation and start in 2d. with time,

you have 2 concentrations c1,c2 depending of space and time

and the minimal equation are something like :

d c1/dt - div( k1(c1,c2) grad(c1)) = ???
d c2/dt - div( k2(c1,c2) grad(c2)) = ???

tHis problem is hard a level of modélisation ,
I found a thesis on this subject in french : Modelization of the joint formation phase of a solvent-based adhesive in a porous medium - Archive ouverte HAL

7

Thanks. yes this is just diffusion. If there is interest and I get most of the
junk code removed, I can post the code. I realized I may be able to do this
even with the integration, I’m sure its just something dumb I was missing.