<div dir="ltr">Dear Gmsh team,<div><br></div><div>Recently I encountered an issue when meshing some geometries with ellipse arc, and after some time working on this issue, I think currently gmsh cannot deal with ellipses that start and end points are symmetric with respect to major axis (or minor axis). All line numbers and reference to code are based on 3.0.6 source code. example.geo is a small example that can reproduce this issue on 3.0.6 windows version, and GmshTest.py is a python script based on gmsh ellipse handling code that can print variable values during processing (I do not have gmsh build environment on my machine..). Just a note that I was using the problem I encountered as test case (instead of example.geo) in GmshTest.py.</div><div><br></div><div>Basically what I found is that in Geo/Geo.cpp, near line 364, if for (x1,x3) and (y1,y3), one pair has the same value and the other has the same absolute value but opposite sign, sys matrix will have two identical rows and sys2x2() on line 371 will fail.</div><div><br></div><div>If I understand the code correctly, the code here is primarily trying to calculate the semi-major axis length f1 and
<span style="font-size:small;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline">semi-m</span>inor axis length f2. In case that two points are symmetric wrt axis after rotation, I think we have to use the major axis point coordinate to get f1, and only solve f2 using the coordinate of an end point.</div><div><br></div><div>I can work around this issue in my work, though it will be nice if Gmsh is able to handle this case..</div><div><br></div><div>Thanks,</div><div>Shengjie</div></div>