Hi Diane and welcome, I'm looking forward to seeing hypothesis used with SymPy. It seems like it could be really useful and I looked at it a few times but I just couldn't quite get my head round what the workflow would be, like exactly how we could use it...
If you can figure that out (and explain it to everyone) then I am sure that we can put it to good use! Oscar On Wed, 5 Jul 2023 at 17:36, Aaron Meurer <[email protected]> wrote: > > Hi everyone. > > As many of you may know, I work at Quansight, a company that works > with and funds a lot of open source work in the Python ecosystem. > Quansight Labs has a yearly internship program where interns work on > various open source projects. > > I'm happy to announce that this summer, Diane Tchuindjo will be > interning at Quansight Labs to work on SymPy. Her project will be to > introduce Hypothesis into the SymPy test suite. Everyone join me in > welcoming Diane to the project. > > You can read more about the project here > https://github.com/sympy/sympy/issues/20914. Basically, hypothesis > (https://hypothesis.readthedocs.io/en/latest/) is a Python testing > library that uses property based tests. This lets you write tests that > take generic inputs and test functions using exact mathematical > properties, rather than only testing explicit inputs and outputs. For > example, to take an example from the issue, a Hypothesis test for > factorint() might look like > > from hypothesis import given > from hypothesis.strategies import integers > > @given(integers()) > def test_factorint(x): > f = factorint(x) > assert Mul(*[b**e for b, e in f.items()]) == x > for b in f: > assert abs(b) in [0, 1] or isprime(b) > > (as opposed to the existing factorint tests, which just test explicit > inputs and outputs > https://github.com/sympy/sympy/blob/master/sympy/ntheory/tests/test_factor_.py#L168). > This test generates an integer, runs factorint() on it, and tests that > the result is mathematically correct (i.e., that the factors are prime > and multiply back together to the original integer). > > This is somewhat like random testing, except Hypothesis is actually a > lot more sophisticated than a purely random test, because it always > tries to generate interesting examples, and it also does things like > shrinking test inputs, and makes input strategies easy to compose. In > my experience, Hypothesis is *really* good at finding bugs that you > would otherwise never find. > > Hypothesis has always been a good fit for SymPy, but we've as of yet > never used it. Our plan is to start small, to prove its usefulness, > but I'm confident we will be able to convince the SymPy community that > Hypothesis is a tool that we should be using regularly in the SymPy > test suite. > > Aaron Meurer > > -- > You received this message because you are subscribed to the Google Groups > "sympy" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/sympy/CAKgW%3D6Js_3keq-UcmRQDaVtu2mhagBgVauTBpS%3DKyO4vYSWQ3A%40mail.gmail.com. -- You received this message because you are subscribed to the Google Groups "sympy" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/sympy/CAHVvXxQm%3D7jh2%2BDsU3P7mc-8weF24saVuY4t9qHZ%2BJnbGBWO2g%40mail.gmail.com.
