integrals.transforms: Fix for issue 2486 #2498
Conversation
|
@asmeurer The Travis CI build failed while running doctests on PyPy. Can you tell me how to fix it? |
|
Ignore the Travis builds on PyPy (that's why they're in the "allowed failures" section). They won't pass until Travis updates their version of PyPy (travis-ci/travis-ci#1106). |
|
Please add a test. |
|
Hope this works fine. |
| @@ -547,6 +547,8 @@ def simp_hyp(expr): | |||
| assert ILT(1/(s*sqrt(s + 1)), s, t) == Heaviside(t)*erf(sqrt(t)) | |||
| # TODO can we make erf(t) work? | |||
|
|
|||
| assert ILT(1/(s**2*(s**2 + 1)),s,t) == (t - sin(t))*Heaviside(t) | |||
There was a problem hiding this comment.
There's a trailing whitespace character on this line.
| try: | ||
| h = hyperexpand(G) | ||
| except NotImplementedError as detail: | ||
| raise IntegralTransformError( |
There was a problem hiding this comment.
I'm assuming this causes it to create an unevaluated object when it can't do it.
|
Looks OK to me. |
|
I found that to compute inverse_laplace_transform sympy first tries to do the inverse_mellin_transform and then if it fails it the inversion integral. So, it doesn't always return an unevaluated object. In the particular cases added in the tests, sympy successfully solves the transform using the inversion integral after failing with inverse_mellin_transform . |
|
OK, well that will need to be fixed eventually as well. |
integrals.transforms: Fix for issue 2486
This tries to fix issue 2486.
#2486