use seissol v1.3/Rusanov fluxes/ vtkhdf/ extend tpv13

[Thomas-Ulrich]

No description provided.

[Thomas-Ulrich]

I think it is worth updating to the vtkhdf output:


vs the older output:

[AliceGabriel]

Thank you Thomas!
I agree about updating the output.
Question: how different is the output after switching the flux in this coarse mesh?

[Thomas-Ulrich]

the figure above shows a qualitative answer.

[AliceGabriel]

Do you mind plotting them in the same figure

[AliceGabriel]

But overall I agree- we should update this

[AliceGabriel]

Do you mind plotting them in the same figure

[Thomas-Ulrich]

Ok, it turns out the
reader.UpdateTimeStep(5.) that we had first meant read at time simulation time 5+1
(makes not much sense but that what I got when testing).
Anyway, if I put reader.UpdateTimeStep(9.) and compare with the file 'output/{prefix}-free-surface-4.vtkhdf', I get matching data:

[Thomas-Ulrich]

Latest changes:

• I realized that pv.Plotter() needs to be closed to avoid memory leaks (leading to the RAM getting filled if you play to much with the iterative plots. Now that's fixed.
• Since the last exercise is based on the xdmf output, I reverted to fault xdmf, also because this does not brings much improvement.
• I've added one more step in the rupture speed analysis, in which we decrease initial stress to suppress supershear.

[Thomas-Ulrich]

I added a small text detailing the initial stress and choice of mu_s in the nucleation area.

[davschneller]

LGTM from looking over the source.

(do we even need the Dockerfile, i.e. the Frontera container, at the moment? It's not built right now anyways)

[Thomas-Ulrich]

(do we even need the Dockerfile, i.e. the Frontera container, at the moment? It's not built right now anyways)

you mean, frontera does not need a specific file anymore and we can remove the file?

[davschneller]

(do we even need the Dockerfile, i.e. the Frontera container, at the moment? It's not built right now anyways)

you mean, frontera does not need a specific file anymore and we can remove the file?

It still does need a special one. However, our CI doesn't build it at the moment (that can be fixed quite easily though)—the question would be: do we need it for the trainings still? (or at least, at the moment?)