[aspect-devel] Treatment of artificial diffusion and convection terms

[Wolfgang Bangerth]

Sylvia,

> Thanks for your quick reply. Are there any additional changes like this in
> the numerical methods?

I don't recall any. The mesh refinement has been made significantly more 
general, but the general principle still holds.


> Well, I was just wondering, as I couldn't find where in the assembly.cc
> the output of 'bdf2_extrapolate' was used. In the paper these terms only
> occurred in the rhs of the energy equation (page 18, def of G_i), I think.

Good point. I don't recall the details (one could probably find this out from 
the revision history in git) but my suspicion is that we figured that that if 
one sets time_step and old_time_step appropriately, that many things work just 
by themselves in all of the places where we used the function. There are still 
two instances where we use a variable use_bdf2_scheme.

In any case, I've removed the function:
   https://github.com/geodynamics/aspect/pull/556

Best
  Wolfgang

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