Failing tests on macos
Objective of issue: Fix failing test results on macos for TestStochasticIntegrateAndFire.test_noisy_firing and TestStochasticIntegrateAndFire.test_noisy_state_progression.
Lava version:
- [x] 0.3.0
I'm submitting a ...
- [x] bug report
Current behavior:
_______________ TestStochasticIntegrateAndFire.test_noisy_firing _______________
33
34self = <tests.lava.lib.optimization.solvers.generic.test_models.TestStochasticIntegrateAndFire testMethod=test_noisy_firing>
35
36 def test_noisy_firing(self):
37 expected = np.array([[0., 0., 1., 0., 0., 1., 0., 1., 0., 0., 1., 0.,
38 0., 1., 0., 0., 1., 0., 1., 0., 0., 1.]])
39 self.kwargs.update(dict(noise_amplitude=1, name="Process_2"))
40 np.random.seed(1)
41 self.data = set_up(self, var="messages", **self.kwargs)
42 spike_vector = self.data[self.bit.name]["messages"]
43> self.assertTrue((spike_vector.T == expected).all())
44E AssertionError: False is not true
45
46tests/lava/lib/optimization/solvers/generic/test_models.py:159: AssertionError
47_________ TestStochasticIntegrateAndFire.test_noisy_state_progression __________
48
49self = <tests.lava.lib.optimization.solvers.generic.test_models.TestStochasticIntegrateAndFire testMethod=test_noisy_state_progression>
50
51 def test_noisy_state_progression(self):
52 np.random.seed(1)
53 expected = np.cumsum(np.random.randint(0, 200, (self.steps,)))
54 expected += np.arange(100, (self.steps + 1) * 100, 100)
55
56 self.kwargs.update(
57 dict(noise_amplitude=1, name="Process_2", steps_to_fire=1000)
58 )
59 np.random.seed(1)
60 self.data = set_up(self, "state", **self.kwargs)
61 state = self.data[self.bit.name]["state"]
62> self.assertTrue((state.T == expected).all())
63E AssertionError: False is not true
64
65tests/lava/lib/optimization/solvers/generic/test_models.py:150: AssertionError
66=============================== warnings summary ==========================
Expected behavior: They do not fail.
Hi @mgkwill, Is this issue still outstanding or already obsolete?
@GaboFGuerra Assigning this to you. Please request a MacOS. Perhaps consult with @awintel and @mgkwill to figure out what is the best way to verify this bug or just close it if we do not want to support MacOS. This probably falls under the purview of @mathisrichter now as the overall owner of the lava software.
@GaboFGuerra You should be able to reactivate CI jobs on MacOS for lava-optimization and debug it this way. Check with @mgkwill on how to do that.
In general, we support Lava on MacOS, we just don't have CI jobs for lava-loihi. And for some libraries, it is deactivated because of failing unit tests.