Re: [femlisp-user] Constructing 2D domain

[Nicolas Neuss]

Sebastian Sturm <address@hidden> writes:

> First of all, thank you very much for your detailed solution! Looking
> at the source code, I would definitely not have been able to achieve
> this by guessing.
>
> My apologies for replying so late; the reason is that I'm encountering
> a probably embarrassingly trivial problem that, up to now, prevents me
> from looking at the solution. When using (plot :program :vtk) to
> export the solution, SBCL crashes inside
> fl.graphic::graphic-write-data. 

Please do another "cvs update" - the VTK interface is (at least, was,
because I don't intend to work on it in the next time) under development
and I may have removed a bug some days ago.

Doing so has also the bonus that you should have something like
interactive VTK graphics: Simply you open a shell, go to the
"femlisp/images" directory and issue the command

 "mayavi2 -x mayavi_poll.py"

there.


> For the ellsys-model-problem, (plot :program :vtk) works as expected,
> though. I thought I would be able to access the solution manually by
> calling something like (fe-value (getbb *result* :solution) (vector
> 0.5 0.5)), but this produces the error message 'There is no applicable
> method for the generic function #<standard-generic-function
> fl.mesh:global->local (3)> when called with arguments (nil #(0.5
> 0.5))'. 

This error occurs also for me.  I'm not completely sure what is the
reason.  In principle, it can happen that some point lying on an edge
between two cells is not found to be inside any cell due to rounding
errors.  But I am a little astonished that it happens for this
unstructured Triangle-generated mesh.

Probably I should introduce a threshold here (the code for this is
already there), but it is not completely clear to me how large it should
be.

Anyway,

  (fe-value (getbb *result* :solution) #d(0.50000001 0.50000001))

works.

> Since the test has passed on your machine, probably something
> is wrong with my femlisp installation. I'll try to figure out what's
> broken, just wanted to let you know why I'm taking so long to answer.
>
> thanks again,
> Sebastian

Best,
Nicolas