[Gmsh] Apply background mesh to a portion of the geometry

Guangran Kevin Zhu sket16 at gmail.com
Tue Nov 13 16:51:19 CET 2007


Hi Mr.  Geuzaine,

Thank you for the response. I tried out the tar ball of the nightly build, but it does not appear clean. The following is the dump of "make". It looks for FourierModel/FM_Patch.h. I do  not know which package this  file belongs. Also, do you mind creating the tar ball without opencascade, whose installation is too big for my machine?

I am using finite element method to solve an eletromagnetic scatterring problem. The interior region is generated from a background mesh and the exterior will be the perfectly matched layer and does not have a background mesh. That's  how I run into this problem.


Kevin

P.s. There are a couple of minor mistake in the makefile under Geo, Numeric, and Mesh directories, in which the path to gsl is set to  "/sw/include/gsl/...", (being machine dependent.).


kzhu at kbt-symmetry:~/devcode/gmsh-2.1.0-nightly-20071112$ make
make[1]: Entering directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Common'
make[1]: `../lib/libGmshCommon.a' is up to date.
make[1]: Leaving directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Common'
make[1]: Entering directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/DataStr'
make[1]: `../lib/libGmshDataStr.a' is up to date.
make[1]: Leaving directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/DataStr'
make[1]: Entering directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Geo'
make[1]: *** No rule to make target `/usr/local/opencascade//inc/config.h', needed by `OCCVertex.o'.  Stop.
make[1]: Leaving directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Geo'
make[1]: Entering directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Mesh'
make[1]: `../lib/libGmshMesh.a' is up to date.
make[1]: Leaving directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Mesh'
make[1]: Entering directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Post'
make[1]: `../lib/libGmshPost.a' is up to date.
make[1]: Leaving directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Post'
make[1]: Entering directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Numeric'
make[1]: `../lib/libGmshNumeric.a' is up to date.
make[1]: Leaving directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Numeric'
make[1]: Entering directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Parallel'
make[1]: `../lib/libGmshParallel.a' is up to date.
make[1]: Leaving directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Parallel'
make[1]: Entering directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Parser'
make[1]: `../lib/libGmshParser.a' is up to date.
make[1]: Leaving directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Parser'
make[1]: Entering directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Plugin'
make[1]: `../lib/libGmshPlugin.a' is up to date.
make[1]: Leaving directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Plugin'
make[1]: Entering directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Graphics'
make[1]: `../lib/libGmshGraphics.a' is up to date.
make[1]: Leaving directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Graphics'
make[1]: Entering directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Fltk'
make[1]: *** No rule to make target `/Users/remacle/SOURCES/FourierModel/FM_Patch.h', needed by `GUI_Projection.o'.  Stop.
make[1]: Leaving directory `/home/kzhu/devcode/gmsh-2.1.0-nightly-20071112/Fltk'
make: *** [compile] Error 2
kzhu at kbt-symmetry:~/devcode/gmsh-2.1.0-nightly-20071112$ 



On Nov 9, 2007 9:38 AM, Christophe Geuzaine <cgeuzaine at ulg.ac.be> wrote:
> Guangran Kevin Zhu wrote:
> > Hello Folks,
> >
> > I am working on a problem, in which I need to apply to apply a
> > background mesh to a portion of a 2D geometry. GMSH could properly
> > generate a non-uniform mesh over the region where the background mesh
> > is applied. However, in the region which does not have a background
> > mesh, triangular elements with accute angles are generated. These
> > elements will introduce errors when I solve the wave equation.
> >
> > I also tried to insert some "dummy" geometrical entities over the
> > regions without the background mesh. The problem is not alleviated. Is
> > this a limitation of GMSH when handling this kind of problem? Is there
> > a workaround?
> >
> > I have modified t7.geo from the tutorial to illustrate my point below.
> > When you generate the mesh, you will see some ill-shaped elements in
> > the outer layer.
> >
> 
> 
> Hi - It's not documented yet, but with the latest nightly snapshot
> (http://geuz.org/gmsh/src/) you can do the following:
> 
> 
> // Merge the first tutorial
> Merge "t1.geo";
> 
> // Merge a post-processing view containing the target mesh sizes
> Merge "bgmesh.pos";
> 
> PostView Field(1) = 0; // first lc field is View[0]
> Function Field(2) = "0.03"; // second field is constant
> Characteristic Length Field {1,2}; // lc field is combination (min) of both
> 
> ...
> 
> 
> This will lead to much nicer meshes.
> 
> 
> 
> 
> 
> 
> 
> > =================================================
> > // Merge the first tutorial
> > Merge "t1.geo";
> >
> > // Merge a post-processing view containing the target mesh sizes
> > Merge "bgmesh.pos";
> >
> > // Apply the view as the current background mesh
> > Background Mesh View[0];
> >
> > Point(5) = {-0.1,-0.1,0,0.01};
> > Point(6) = {0.2,-0.1,0,0.01};
> > Point(7) = {0.2,0.4,0,0.01};
> > Point(8) = {-0.1,0.4,0,0.01};
> > Line(101) = {8,5};
> > Line(102) = {5,6};
> > Line(103) = {6,7};
> > Line(104) = {7,8};
> > Line Loop(105) = {104,101,102,103};
> > Plane Surface(106) = {105,5};
> >
> > Point(11) = {-0.2, -0.2,0, 0.03};
> > Point(12) = {0.3, -0.2, 0, 0.03};
> > Point(13) = {0.3, 0.5, 0, 0.03};
> > Point(14) = {-0.2, 0.5, 0, 0.03};
> >
> >
> > Line(107) = {14,13};
> > Line(108) = {13,12};
> > Line(109) = {12,11};
> > Line(110) = {11,14};
> > Line Loop(111) = {108,109,110,107};
> > Plane Surface(112) = {105,111};
> >
> > ================================================
> >
> > _______________________________________________
> > gmsh mailing list
> > gmsh at geuz.org
> > http://www.geuz.org/mailman/listinfo/gmsh
> >
> >
> 
> 
> --
> Prof. Christophe Geuzaine
> University of Liege, Electrical Engineering and Computer Science
> http://www.montefiore.ulg.ac.be/~geuzaine
> 
> 
>