Add FiniteDifferences based scalar rule tests #87
Conversation
…t the incorrect acshc asec acsc, asecd acscd rules
willtebbutt
left a comment
willtebbutt
left a comment
There was a problem hiding this comment.
I've left one minor comment that I'm happy to remain unaddressed if you don't fancy addressing it. Assuming that we now have 100% coverage of the scalars rules through the use of test_scalar, I'm happy for this to be merged.
| # Check that we get the derivative right: | ||
| @test isapprox( | ||
| ∂x(1), fdm(f, x); | ||
| rtol=rtol, atol=atol, kwargs... |
There was a problem hiding this comment.
Do we need to have atol and rtol outside of kwargs? If we don't set the defaults to 1e-9, do we start to get errors?
There was a problem hiding this comment.
idk, I was copying the test_frule and test_rrule setup
We don't yet. |
|
This is great |
|
we got the derviative of |
lol |
|
OK. |
|
we should re-run CI for Julia 1.0? |
|
@nickrobinson251 It seems that Travis needs to be reconfigured to not abort if tests take longer than 10 minutes |
Yes, we should extend that time limit - I don't think I have permissions to do that, but hopefully @oxinabox does? (Or maybe i just don't know how to...) |
|
It is not possible to increase the time without log output in Travis. |
|
Is the solution just to insert random print statements? |
Apparently. I wonder if we should just get rid of the top-level TestSet so it prints each testset as it completes. |
Closes #76
and the mistakes found in #86 for incorrect acshc asec acsc, asecd acscd rules