Test for AllenCahn_1D_FD.py
#370
Conversation
There was a problem hiding this comment.
Thanks a lot! Looks like you fixed quite a lot of things!
Generally, a test that I would find useful for problems with a Newton solver is to compare the solver to Newton Krylov from scipy like implemented here.
It should be possible to write a generic function that takes as input a problem class and some parameters and then solves using the solver in the problem class and scipy and compares. It is some effort, though. So don't feel like you need to do this. But if you are eager to write some more tests, this could be useful.
|
|
||
| if np.isnan(res) and self.stop_at_nan: | ||
| raise ProblemError('Newton got nan after %i iterations, aborting...' % n) | ||
| elif np.isnan(res): | ||
| self.logger.warning('Newton got nan after %i iterations...' % n) | ||
|
|
||
| print(f'{type(self).__name__}') |
There was a problem hiding this comment.
Did you forget to remove this?
There was a problem hiding this comment.
Yes, did forget that.
| @@ -735,7 +717,7 @@ def __init__( | |||
| stop_at_nan=True, | |||
| ): | |||
| super().__init__(nvars, dw, eps, newton_maxiter, newton_tol, interval, radius, stop_at_nan) | |||
| self.A -= sp.eye(self.init) * 0.0 / self.eps**2 | |||
| self.A -= sp.eye(self.nvars) * 0.0 / self.eps**2 | |||
There was a problem hiding this comment.
Does this line do anything? It just subtracts zero. At least with your fix this does not raise an error, but still. Is there a reason for this line @pancetta?
There was a problem hiding this comment.
There was, but not anymore. I don't know why this is still in there. Can be removed!
| @@ -822,7 +805,7 @@ def __init__( | |||
| stop_at_nan=True, | |||
| ): | |||
| super().__init__(nvars, dw, eps, newton_maxiter, newton_tol, interval, radius, stop_at_nan) | |||
| self.A -= sp.eye(self.init) * 0.0 / self.eps**2 | |||
| self.A -= sp.eye(nvars) * 0.0 / self.eps**2 | |||
There was a problem hiding this comment.
Again, what does this do?
Codecov ReportAttention:
Additional details and impacted files@@ Coverage Diff @@
## master #370 +/- ##
==========================================
+ Coverage 72.66% 73.62% +0.96%
==========================================
Files 270 270
Lines 22387 22383 -4
==========================================
+ Hits 16267 16480 +213
+ Misses 6120 5903 -217
☔ View full report in Codecov by Sentry. |
|
Thanks for your comment! That sounds meaningful! I also noticed that I forgot several lines to test. I'll add some more testing for |
|
I advocate the use of git hooks to automate black as described here. |
I added a test for the one-dimensional AllenCahn equation. The test is more or less based on
test_NonlinearSchroedinger_MPIFFT.py, testing things like:ProblemErrorraised correctly whennanarises?,logger.warningalso called correctly in case ofnanSuch basic things. In my opinion there could be also tests for some problem specific things. But at this moment I don't know yet which things this could be. For example, one of the AC-classes uses Finel's trick, which is only a rewriting of the right-hand side. Here could be also a specific test added, but I don't have any idea what test would make sense.
Let me know if anyone has some idea what I can add.