Elmer Discussion Forum

Hi, Sorry for late reply. Scotch is a dependency of MUMPS. Do you need MUMPS? If not you could edit https://github.com/ElmerCSC/homebrew-elmerfem/blob/master/elmer.rb and set there (line 40 or so) -DWITH_Mumps:BOOL=FALSE The problem with Mac is that nobody in ElmerTeam uses it and homobrew is not te...

Hi All,

There has recently been some work to enable Elmer to use internal partitioning. More about this in a blog post:

http://www.elmerfem.org/blog/general/in ... titioning/

Discussion on the features can take place here, if needed.

-Peter

Hi Your picture makes me wonder whether you're still on the linear regime. What if you reduce the load significantly. Now your result looks very distorted. You would like to click the load in with the mouse. That would of course be convenient. Unfortunately Elmer is a project mostly driven by projec...

Hi Connectivity of nodes is actually available in the sparse matrix graph. Hence if you have scalar fields in the code you can do TYPE(Matrix_t), POINTER :: A INTEGER :: i,j,k A => Solver % Matrix DO i=1,A % NumberOfRows DO j=A % Rows(i),A % Rows(i+1)-1 k = A % Cols(j)) print *,'nodes are connected:...

Hi Yes, some solvers use the feature of namespace when you have to combine different linear solver strategies (or material parameters etc.) in one simulation. Another example is block preconditioning where different setting may be needed for different blocks in the linear system. Then the namespace ...

Hi Matthias

I had a look at our Windows compilation process and was able to at least get fresh .exe installers. The setup start to be somewhat outdated. Hope this helps. The installers are based on todays devel branch.

-Peter

Hi

I updated today the Windows binaries. Unfortunately there seemed to be some problem creating the .zip files. I hope the new executables work. Testing was limited.

-Peter

Hi, Its difficult to comment without the whole case. However, it is clear that if you you have a jump in currents over elements this cannot be presented by continuous function. This is difficult to ovecome without a discontinuous visualization space. -Peter

Hi

The problem is that whereas Az is continuous sigma*Az is not. So what you are seeing is the effort to map discontinuous field to continuous basis functions. The remedy is to use discontinuous Galerkin (DG) basis as may be done with MagnetoDynamicsCalcFields (instead of BSolver).

-Peter

Hi Joe, How do you create your mesh. I hope internally with "Mesh Levels"? Then the preprocessing with ElmerGrid should not lead to any problems. There are a number of linear solver libraries and I don't know which of them would work out-of-box. Probably it would be easiest to start with the interna...