[aspect-devel] AGU meeting feedback

[Timo Heister]

Okay, sometimes it helps to look at the output:
http://i.imgur.com/0o9ojFa.png
these are two runs, on the left with 5 global refinements and 6 on the right.

So, not only does the second peak occur at a different time, it is a
plume that rises from a different location (see t=600).

Why does this happen? It looks like the plume on the left comes from
the first staircase (on the right) in the initial condition. Now it
makes sense that the solution is dependent on the mesh size. We
obviously do a poor job with a good initial condition for the
compositional field.

I guess we could solve this by:
- asking the user to only input "smooth" initial conditions
- implementing smoothing/reinitialization for fields



On Tue, Dec 31, 2013 at 5:52 AM, Timo Heister <heister at clemson.edu> wrote:
>> 2. Could the second peak be initiated by an instability that depends
>> on the mesh? All the runs are on a mesh with an even number of
>> elements. It would be easy to test this by changing the geometry model
>> to call subdivided_hyper_rectangle() with an odd number of repitions
>> instead of hyper_rectangle().
>
> ./ I tried that out and the time of the 2nd peak is still changing a
> lot with refinement:
> 48^2: t=500
> 96^2: t=800
> 192^2: t=620
> 384^2: t=870
>
> ./ A fixed time step makes no difference.
>
> ./ Surprisingly, an adaptive computation with 3+3 levels gives the
> same peak as a fixed mesh with 6 refinements, etc.
>
> ./ Is this just because of the initial condition? For example the
> amount of mass in the composition at t=0? No, if I only do adaptive
> refinement in the first timestep, the 2nd peak is at a different time
> (similar to the coarser fixed mesh).
>
> So, at some time t>0 at some location the size of the cells determines
> the timing of the 2nd peak. I have to think about this some more. Any
> ideas welcome.
>
> --
> Timo Heister
> http://www.math.clemson.edu/~heister/



-- 
Timo Heister
http://www.math.clemson.edu/~heister/