Tests fail on NumPy 2 #17
Description
Upgrading the version of numpy to >=2 results in the following failures to the tests:
Test output
pytest .
================================================================================= test session starts ==================================================================================
platform darwin -- Python 3.9.7, pytest-8.4.1, pluggy-1.6.0
rootdir: ***/annchor
configfile: pyproject.toml
collected 16 items
annchor/tests/test_annchor.py .....F. [ 43%]
annchor/tests/test_datasets.py .... [ 68%]
annchor/tests/test_distances.py .. [ 81%]
annchor/tests/test_examples.py .F. [100%]
======================================================================================= FAILURES =======================================================================================
_________________________________________________________________________________ test_function_input __________________________________________________________________________________
get_exact_ijs = <function get_exact_ijs_.<locals>.get_exact at 0x127c3bdc0>, f = <function test_function_input.<locals>.wasserstein1 at 0x128681b80>
X = array([[ 0., 0., 5., ..., 0., 0., 0.],
[ 0., 0., 0., ..., 10., 0., 0.],
[ 0., 0., 0., ..., 16... 0., 1., ..., 6., 0., 0.],
[ 0., 0., 2., ..., 12., 0., 0.],
[ 0., 0., 10., ..., 12., 1., 0.]])
nx = 1797, backend = 'loky', s = 20
def test_parallelisation(get_exact_ijs, f, X, nx, backend, s=20):
try:
> get_exact_ijs(f, X, np.random.randint(nx, size=(s, 2)))
.venv/lib/python3.9/site-packages/annchor/utils.py:250:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
.venv/lib/python3.9/site-packages/annchor/utils.py:170: in get_exact
Parallel(n_jobs=CPU_COUNT, backend=backend, timeout=30)(
.venv/lib/python3.9/site-packages/joblib/parallel.py:2072: in __call__
return output if self.return_generator else list(output)
.venv/lib/python3.9/site-packages/joblib/parallel.py:1682: in _get_outputs
yield from self._retrieve()
.venv/lib/python3.9/site-packages/joblib/parallel.py:1841: in _retrieve
batched_results = batched_results.get_result(self.timeout)
.venv/lib/python3.9/site-packages/joblib/parallel.py:758: in get_result
return self._return_or_raise()
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <joblib.parallel.BatchCompletionCallBack object at 0x12aa81190>
def _return_or_raise(self):
try:
if self.status == TASK_ERROR:
> raise self._result
E multiprocessing.context.TimeoutError
.venv/lib/python3.9/site-packages/joblib/parallel.py:773: TimeoutError
During handling of the above exception, another exception occurred:
def test_function_input():
# Load digits
data = load_digits()
X = data["X"]
M = data["cost_matrix"]
nx = len(X)
# not njit, no kwargs
def wasserstein1(x, y):
return kantorovich(x, y, cost=M)
# not njit, kwargs
def wasserstein2(x, y, cost=M):
return kantorovich(x, y, cost=M)
# njit, no kwargs
@njit()
def wasserstein3(x, y):
return kantorovich(x, y, cost=M)
# njit, kwargs
@njit()
def wasserstein4(x, y, cost=M):
return kantorovich(x, y, cost=M)
> ann1 = Annchor(X, wasserstein1)
annchor/tests/test_annchor.py:189:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
.venv/lib/python3.9/site-packages/annchor/annchor.py:185: in __init__
test_parallelisation(
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
get_exact_ijs = <function get_exact_ijs_.<locals>.get_exact at 0x127c3bdc0>, f = <function test_function_input.<locals>.wasserstein1 at 0x128681b80>
X = array([[ 0., 0., 5., ..., 0., 0., 0.],
[ 0., 0., 0., ..., 10., 0., 0.],
[ 0., 0., 0., ..., 16... 0., 1., ..., 6., 0., 0.],
[ 0., 0., 2., ..., 12., 0., 0.],
[ 0., 0., 10., ..., 12., 1., 0.]])
nx = 1797, backend = 'loky', s = 20
def test_parallelisation(get_exact_ijs, f, X, nx, backend, s=20):
try:
get_exact_ijs(f, X, np.random.randint(nx, size=(s, 2)))
except TimeoutError:
print("TimeoutError: Parallelisation failed.")
if isinstance(f, CPUDispatcher):
print(
"Currently using numba parallelisation, try"
+ " specifying custom parallelistation with"
+ " get_exact_ijs keyword argument."
)
elif backend == "loky":
print(
"Current backend is 'loky', try backend='multiprocessing',"
+ " or specifying custom parallelistation with"
+ " get_exact_ijs keyword argument."
)
elif backend == "multiprocessing":
print(
"Current backend is 'multiprocessing', try backend='loky',"
+ " or specifying custom parallelistation with"
+ " get_exact_ijs keyword argument."
)
> raise TimeoutError()
E multiprocessing.context.TimeoutError
.venv/lib/python3.9/site-packages/annchor/utils.py:271: TimeoutError
--------------------------------------------------------------------------------- Captured stdout call ---------------------------------------------------------------------------------
TimeoutError: Parallelisation failed.
Current backend is 'loky', try backend='multiprocessing', or specifying custom parallelistation with get_exact_ijs keyword argument.
____________________________________________________________________________ test_annchor_selective_subset _____________________________________________________________________________
def test_annchor_selective_subset():
np.random.seed(1)
X, y = make_blobs(n_samples=1000, centers=5) # First data set X,y
U, v = make_moons(n_samples=1000, noise=0.1) # Second data set U,v
U = np.fliplr(U)
# First data set
knn = 15
annX = Annchor(X, "euclidean", n_neighbors=knn, p_work=0.2)
annX.fit()
# Second data set
knn = 15
annU = Annchor(U, "euclidean", n_neighbors=knn, p_work=0.2)
annU.fit()
# First data set
ssx = annX.annchor_selective_subset(y=y, alpha=0)
> assert len(ssx) == 90
E assert 89 == 90
E + where 89 = len(array([ 8, 22, 61, 104, 140, 168, 200, 207, 224, 265, 277, 287, 298,\n 303, 307, 314, 317, 338, 397, 420, 467,..., 35, 174, 195, 253, 324, 365, 529, 572, 26, 28, 41,\n 46, 95, 97, 118, 142, 326, 502, 516, 565, 582, 691]))
annchor/tests/test_examples.py:81: AssertionError
--------------------------------------------------------------------------------- Captured stderr call ---------------------------------------------------------------------------------
100%|██████████| 1000/1000 [00:00<00:00, 5203.17it/s]
100%|██████████| 1000/1000 [00:00<00:00, 10529.88it/s]
100%|██████████| 138/138 [00:00<00:00, 236.23it/s]
=============================================================================== short test summary info ================================================================================
FAILED annchor/tests/test_annchor.py::test_function_input - multiprocessing.context.TimeoutError
FAILED annchor/tests/test_examples.py::test_annchor_selective_subset - assert 89 == 90
======================================================================= 2 failed, 14 passed in 482.82s (0:08:02) =======================================================================
The second test (annchor/tests/test_examples.py::test_annchor_selective_subset) might just be due to "new random", but the timeout error will need addressing.