[CIG-LONG] Convergence problem

[Walter Landry]

Dian He <hedian18 at gmail.com> wrote:
> Hi there,
> 
> When I run models, I have always encountered the problems that the solutions
> can not be converged at some (usually hundreds of) time steps, even I change
> kinds of parameters. I want to see some results after tens Myr, but they were
> stopping at just several Myr.

There is probably something happening at that point that is causing
problems.  If you look at the simulation shortly before it stops
converging, it may become obvious what the problem is.

> The more strange thing is that, for one case, it can only run for a
> certain resolution of the mesh. It doesn't run one step at all if I
> change the resolution of the mesh or particles.
>
> Another question is that the temperature structure is reset if I restart the
> running from some step, but the geometry and velocity seem to continue.

That is odd.  It may be a bug in Gale. Can I see your input file?

Cheers,
Walter Landry
walter at geodynamics.org