def test_average_9():
"""test_average_9
"""
data = torch.arange(6).reshape((3, 2))
weights = torch.Tensor([1. / 4, -1. / 4])
with pytest.raises(ZeroDivisionError) as error:
pm.average(data, weights=weights, axis=1)
assert error.match("Weights sum to zero, can't be normalized")
def test_average_8():
"""test_average_8
"""
data = torch.arange(6).reshape((3, 2))
weights = torch.Tensor([1. / 4, 3. / 4])
with pytest.raises(ValueError) as error:
pm.average(data, weights=weights, axis=0)
assert error.match("Length of weights not compatible with specified axis.")
def test_average_3():
"""test_average_3
"""
data = torch.arange(6).reshape((3, 2))
weights = torch.Tensor([1. / 4, 3. / 4])
with pytest.raises(TypeError) as error:
pm.average(data, weights=weights)
assert error.match(
"Axis must be specified when shapes of a and weights differ.")
def test_average_6():
"""test_average_6
"""
data = torch.arange(6).reshape((3, 2))
weights = torch.Tensor([[1, 2], [3, 4]])
with pytest.raises(TypeError) as error:
pm.average(data, weights=weights, axis=1)
assert error.match(
"1D weights expected when shapes of a and weights differ.")
def test_average_5():
"""test_average_5
"""
data = torch.arange(6).reshape((3, 2))
result = pm.average(data, axis=1)
expected_result = torch.Tensor([0.5, 2.5, 4.5])
assert (result == expected_result).all()
def test_average_4():
"""test_average_4
"""
a = torch.ones(5, dtype=torch.int16)
w = torch.ones(5, dtype=torch.float32)
avg = pm.average(a, weights=w)
assert avg.dtype == torch.float64
def test_average_11():
"""test_average_11
"""
data = torch.arange(6, dtype=torch.float).reshape((3, 2))
weights = torch.Tensor([1. / 4, 3. / 4])
result_avg = pm.average(data, axis=1, weights=weights)
expected_avg = torch.Tensor([0.75, 2.75, 4.75])
assert (result_avg == expected_avg).all()
def test_average_7():
"""test_average_7
"""
data = torch.arange(6).reshape((3, 2))
weights = torch.Tensor([1. / 4, 3. / 4])
result = pm.average(data, axis=1, weights=weights)
expected_result = torch.Tensor([0.75, 2.75, 4.75])
assert (result == expected_result).all()
def test_average_10():
"""test_average_10
"""
data = torch.arange(6).reshape((3, 2))
weights = torch.Tensor([1. / 4, 3. / 4])
result_avg, result_scl = pm.average(data,
axis=1,
weights=weights,
returned=True)
expected_avg = torch.Tensor([0.75, 2.75, 4.75])
expected_scl = torch.Tensor([1., 1., 1.])
assert (result_avg == expected_avg).all()
assert (result_scl == expected_scl).all()
def test_average_0():
"""test_average_0
"""
X = np.random.rand(5, 5)
avg_np, _ = np.average(X, axis=1, returned=True)
X_th = torch.tensor(X)
avg_th, _ = pm.average(X_th, axis=1, returned=True)
assert (X == X_th.numpy()).all()
assert (avg_np == avg_th.numpy()).all()
X -= avg_np[:, None]
X_th -= avg_th[:, None]
assert (X == X_th.numpy()).all()
def test_average_2():
"""test_average_2
"""
assert pm.average(torch.arange(1, 11), weights=torch.arange(10, 0,
-1)) == 4
def test_average_1():
"""test_average_1
"""
data = torch.arange(1, 5)
assert pm.average(data) == 2.5