It seems buildDmesh() can decompose a global mesh and distribute it to different MPI processes, but is there a similar function that can combine meshes from different processes to create a global mesh?
An initial backup of the global mesh does now work in my case, because the meshes are updated on every single MPI process.
It seems savevtk(…) can automatically combine results and meshes from different MPI cores. Where can I find the implementation of savevtk (), so that I can learn an idea?