Resilience in Schrödinger equation

[brownbaerchen]

This PR adds some stuff for simulating faults in the Schrödinger equation and a file to generate plots that are supposed to go in some publication at some point.

I didn't want to test the function for running the Schrödinger equation because that doesn't contain much new stuff. I did add a test for the flipping of bits in complex numbers, which is also new in this PR.

Also, I changed the way I generate stats in parallel so I can do this on the cluster to get some statistics for PDEs in reasonable time.

Finally, I removed some stuff from the test to check the order of error estimates. Since they produce a plot that goes on the website, I couldn't remove too much stuff, but I am hoping to cut the time in half.

This PR should not change anything related to other peoples projects.

[codecov]

Codecov Report

Base: 71.46% // Head: 71.16% // Decreases project coverage by -0.30% ⚠️

Coverage data is based on head (bfe353e) compared to base (58cff59).
Patch coverage: 46.62% of modified lines in pull request are covered.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #276      +/-   ##
==========================================
- Coverage   71.46%   71.16%   -0.30%     
==========================================
  Files         239      242       +3     
  Lines       20183    20640     +457     
==========================================
+ Hits        14424    14689     +265     
- Misses       5759     5951     +192     

Impacted Files	Coverage Δ
pySDC/projects/Resilience/paper_plots.py	0.00% <0.00%> (ø)
pySDC/projects/Resilience/fault_stats.py	48.20% <76.31%> (+1.30%)	⬆️
pySDC/projects/Resilience/Schroedinger.py	87.50% <87.50%> (ø)
...mentations/controller_classes/controller_nonMPI.py	98.64% <100.00%> (ø)
...ns/problem_classes/NonlinearSchroedinger_MPIFFT.py	96.20% <100.00%> (ø)
pySDC/projects/Resilience/Lorenz.py	80.00% <100.00%> (ø)
pySDC/projects/Resilience/accuracy_check.py	97.18% <100.00%> (ø)
pySDC/projects/Resilience/fault_injection.py	90.74% <100.00%> (+2.54%)	⬆️
...ementations/problem_classes/LeakySuperconductor.py	88.77% <0.00%> (ø)
pySDC/projects/Resilience/leaky_superconductor.py	68.36% <0.00%> (ø)
... and 3 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[pancetta]

That's quite a hit in coverage. How much does running the example cost?

[brownbaerchen]

That's quite a hit in coverage. How much does running the example cost?

I guess it doesn't cost too much, but I don't really know what to test here. All I do is setup the problem with certain parameters.

[pancetta]

Just run it and see that it does meaningful things in terms of iteration counts, etc.. You don't want to end up with a setup that crashes because.. reasons.

[pancetta]

Looks like the faulty case is not tested, correct?

[brownbaerchen]

I still don't know why this flakeheaven complaint does not show up on my local machine. I even tried setting up a new conda environment to no avail. Anyways, the "fix" changes the behaviour of the code. Now the except behaviour is entered if there is dill.PicklingError or TypeError, not necessarily both as before. To me it looked like this is actually desired, if not please complain. Anyways, thanks to these guys the previous line is no longer allowed. Another solution would be, of course, to exclude B030 from flakeheaven. Your choice.

This is the line in question, by the way.