[Gmsh] Trouble with 1D meshing in gmsh-4.0.2

[Christophe Geuzaine]

Hi Pat,

Thanks for the report and the test case. For some reason the smaller step size we introduced in Gmsh 4 in the finite difference approximation of the derivative of curve parametrizations leads to some numerical instability. I have reverted the value back to the one used in Gmsh 3: the next snapshot will contain the change.

@all : let us know if you see any adverse side effects from this change on other cases. (It will only affect geometries built with the "built-in" CAD kernel.)

Christophe


> On 15 Oct 2018, at 17:09, Pat Prodanovic <pprodano at gmx.com> wrote:
> 
> Hello,
>  
> I am having a bit of trouble with 1D meshing using gmsh-4.0.2. It seems like there is a new 1D meshing algorithm that is much slower than what was available in previous versions. Allow me to illustrate:
> 
> Consider a rectangle, 700 m by 1200 m whose edges are resampled to a 10 m spacing. The example also has a line embedded in a surface, with an edge spacing of 25 m. This simple *.geo file is attached.
> 
> Then I carry out mesh generation using gmsh, which consists of 1D meshing, following by 2D meshing:
> 
> gmsh-4.0.2: 1D meshing task takes 31.04 sec, and 2D meshing task takes 0.196 sec.
> gmsh-3.0.6: 1D meshing task takes 0.28 sec, and 2d meshing task takes 0.276 sec.
>  
> Please realize this is a trivial example only to illustrate the differences between the two versions.
>  
> On real life examples where mesh boundary consists of thousands of nodes (in my examples of highly irregular coastlines), I have not been able to use gmsh-4.0.2 to produce 2D meshes at all. This is because the 1D meshing task takes simple too long to complete.
>  
> Has something changed in how gmsh-4.0.2 carries out 1D meshing? Perhaps there is a command to recover the previous 1D meshing algorithm?
>  
> I look forward to any comments and insights into the above,
>  
> Pat Prodanovic
>  
> <mesh.geo>_______________________________________________
> gmsh mailing list
> gmsh at onelab.info
> http://onelab.info/mailman/listinfo/gmsh

— 
Prof. Christophe Geuzaine
University of Liege, Electrical Engineering and Computer Science 
http://www.montefiore.ulg.ac.be/~geuzaine

Free software: http://gmsh.info | http://getdp.info | http://onelab.info