[firedrake] problems with importing a mesh

[Francis Poulin]

Yes, that did the trick.  I used your idea in many different tests that I have done and basically added the second of the following two lines and that seemed to do the trick.


Thank you!


...

Plane Surface(1) = {1};
Physical Surface(2) = {1};




------------------
Francis Poulin
Associate Dean, Undergraduate Studies
Associate Professor
Department of Applied Mathematics
University of Waterloo

email:           fpoulin at uwaterloo.ca
Web:            https://uwaterloo.ca/poulin-research-group/
Telephone:  +1 519 888 4567 x32637

________________________________
From: firedrake-bounces at imperial.ac.uk <firedrake-bounces at imperial.ac.uk> on behalf of G. D. McBain <gdmcbain at protonmail.com>
Sent: Thursday, November 2, 2017 8:07:51 PM
To: firedrake at imperial.ac.uk
Subject: Re: [firedrake] problems with importing a mesh



2) For splines, I have also removed the point (updated code below) and when I look at it using gmsh I see that some points aren't actually on the boundary.  This does seem odd but I am not sure how to fix it.   Should I use a different kind of spline?

The problem may be this: the Gmsh concept of Physical versus Elementary entities.

http://gmsh.info/doc/texinfo/gmsh.html#Elementary-vs-physical-entities

Your Gmsh file has only Elementary entities (Point, Line/Spline, Line Loop, Plane Surface).  The trick is to mark the domain (for this two-dimensional problem) as a Physical Surface and each part of the boundary as a Physical Line.  In three dimensions, those would be Physical Volume and Physical Surface, respectively.

When a Gmsh file has no Physical entities, Gmsh puts everything into the mesh, including construction points (centres of arcs, control-points of splines), which often end up as disconnected, but if there is a least one Physical entity, only those finite elements belonging to Physical entities are included in the output .msh.

A quirk of Firedrake, maybe inherited from PETSc which I think it uses to read Gmsh meshes, is that every part of the boundary has to be declared as part of a Physical Line/Surface, even if it's only going to get homogeneous natural boundary conditions in the finite element problem.

Hope this helps; I can supply a little example, if the above is unclear.

-------------- next part --------------
HTML attachment scrubbed and removed