Dear mmg community,
I am doing time-dependent problems on a disk with anisotropic mesh adaptation. However, after many steps of remeshing, the disk slowly shrinks. I understand this is due to the fact that mmg cannot know I am meshing a disk if I do not give this information. I do not want to fix the boundary of my initial mesh, as I would like to allow refinement on the edge.
As a solution, I have decided to mesh the boundary myself using the metric, ensuring the nodes are located on the geometric boundary (\sqrt{x^2+y^2}=r the radius of the circle). I have used a method described in ‘Maillages’ (Frey and George).
I am left with the following quantities :
- The background mesh
- The desired metric defined on the background mesh
- The boundary nodes of the new mesh
Is it possible to run mmg to achieve anisotropic mesh adaptation in the interior, with imposition of the given boundary nodes, which are not the ones of the background mesh? Or is it possible to specify a geometry to mmg, so the boundary nodes are forced to be on the geometric boundary?
Thank you very much for your help, and your beautiful software!
Théophile