h for distributed triangulations

[Michael Harmon]

Hi,

I can GridTools::maximal_cell_diameter(triangulation) when the triangulation is distributed, but under global refinements the value stays the same.  Is there anyone else that has had this issue?

Thanks,

Mike

[Wolfgang Bangerth]

The function is not implemented for parallel triangulations right now.
It looks like this:


template <int dim, int spacedim>
double
maximal_cell_diameter (const Triangulation<dim, spacedim> &triangulation)
{
double max_diameter = triangulation.begin_active()->diameter();
for (typename Triangulation<dim, spacedim>::active_cell_iterator
cell = triangulation.begin_active(); cell != triangulation.end();
++cell)
max_diameter = std::max (max_diameter,
cell->diameter());
return max_diameter;
}

It needs to learn to ignore cells that are not locally owned, and if the
triangulation is a shared triangulation, to take the max over all
processors. Want to give it a try to write a patch in this direction?

Best
W.

--
------------------------------------------------------------------------
Wolfgang Bangerth email: bang...@math.tamu.edu
www: http://www.math.tamu.edu/~bangerth/

[Michael Harmon]

Thanks for the reply.

I needed h for the penalty term in the LDG method and wasnt getting the correct convergence rates. I changed the fluxes to get the correct convergence rates, so its not as pressing, but I will still look into it!