This is a read only copy of the old FEniCS QA forum. Please visit the new QA forum to ask questions

Roadmap for the project?

+2 votes

Is there a 'roadmap' for the FEniCS project?

asked Jul 24, 2015 by Charles FEniCS User (4,220 points)

1 Answer

+3 votes
 
Best answer

There is no definitive 'roadmap', but there are quite a few projects which are on the radar of the developers, e.g.

  • parametric geometry - i.e. curved surfaces
  • mesh views
  • block preconditioners for multiphysics
  • making everything work in parallel
  • complex number support
  • dynamic mesh improvement/repair for moving meshes
  • quadratic and hexahedral elements

If there's anything you want to add to the list, you can post here or on bitbucket.

answered Jul 24, 2015 by chris_richardson FEniCS Expert (31,740 points)
selected Jul 24, 2015 by Charles

parametric geometry - i.e. curved surfaces

Parametric geometry would be really nice! I'm using higher order elements with curved boundaries... Not to be greedy, but is this months/years away?

mesh views

I'm unfamiliar with what this might be?

block preconditioners for multiphysics

Could you explain a bit here?

making everything work in parallel

I've been running in parallel for the most part just fine. Scaling falls off for me around 64 processes with ghost elements though.

If I could make a wish though... hybrid spectral DG elements

Parametric geometry is likely to happen in the next year... but I can promise nothing, as it is a team effort.

Mesh View means, taking a view into a subset of the Mesh, e.g. often you may wish to define a Function on the surface, and couple it to a different Function on the interior.

Block preconditioning is for efficient solving of large blocked problems, e.g. Stokes equation, with u and p in separate spaces, but coupled.

Parametric geometry is likely to happen in the next year... but I can promise nothing, as it is a team effort.

Exciting! In the next year is actual fast :)

Block preconditioning is for efficient solving of large blocked problems, e.g. Stokes equation, with u and p in separate spaces, but coupled.

Sorry to be naive here, but, with the incompressible Navier-Stokes projection method as an example... would the block preconditioning consider the intermediate momentum field and the pressure field and solve the two at once?

I'm reading up on preconditioning of multiphysics problems (doing my homework :) )

I also wanted to mention mesh coarsening would be a nice feature to have...

By the way, is there a way to get involved with the fenics development team? I've been a user for a few years, abusing it in implementing a CFD scheme. I'm a very strong programmer, my profession pre-cfd, and would love to be involved.

...