[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [femlisp-user] Constructing 2D domain

From: Nicolas Neuss
Subject: Re: [femlisp-user] Constructing 2D domain
Date: Sat, 30 Oct 2010 09:43:09 +0200
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/23.2 (gnu/linux)

Sebastian Sturm <address@hidden> writes:

> Hello all,
> I have recently installed Femlisp and I would like to solve a simple
> electrostatic problem div(sigma*grad phi) = 0 on a non-rectangular
> domain consisting of parts with different conductivities (Neumann BC
> on some boundaries, Dirichlet BC on the others). Although I have
> browsed the examples and large parts of the source code, I do not see
> how to construct a domain different from the presupplied ones. I guess
> there should be a fairly easy way to supply a set of corners as the
> domain boundary, send that to triangle and work on the resulting mesh?

You can look at the tests in "src/mesh/triangulate.lisp" where a circle
domain is defined only by its boundary and then triangulated using

> Also, can I assign some kind of id to different domains ( / patches ?)
> and use that to have different material parameters (in this case, the
> A tensor of an elliptic problem) for different spatial regions?

If you want this, then you should specify the boundaries (line segments)
for each region as linking the corner points (this is a 1d "skeleton"),
and use these to define "region blobs" with those boundaries.  Those can
be triangulated using Triangle.

If you can specify the domain and problem more precisely, I can also
send you more information.  I have some similar things in my private

> On a side note, I had problems installing OpenDX using macports and
> since it seems to be abandonware anyway, I'd rather do without it. 

OK, I have not used a Mac with Femlisp graphics in the last two years.
But I have one with Mac OS 10.6 available.  I'll try it next week.

> For now I'm using cl-cairo and Mathematica to visualize meshes and
> corresponding solutions, but as there seems to be a VTK version in the
> works, I'd like to know if there is already some preliminary version
> up for testing?

You should be able to write data out in VTK file format using

(plot .... :program :vtk) 
[This has as default output a file "images/output.vtk".  The name is
also printed for convenience.]

which can be looked at using Mayavi/Mayavi2 or others.

> best regards,
> Sebastian Sturm

Best regards,
Nicolas Neuss

PD Dr. Nicolas Neuss   KIT Karlsruhe   Tel: 0049-721-608-7634
Email: address@hidden   WWW:  <http://www.math.kit.edu/~neuss>

reply via email to

[Prev in Thread] Current Thread [Next in Thread]