class TestLog2(object):
def setup_class(self):
self.ops = LogOperations(2)
def test_add(self):
X = self.ops.log(np.array([0, 1, 0, 2, 0, 1, 2, 1, 2]))
Y = self.ops.log(np.array([0, 0, 1, 0, 2, 2, 1, 1, 2]))
Z = self.ops.log(np.array([0, 1, 1, 2, 2, 3, 3, 2, 4]))
for x, y, z in zip(X, Y, Z):
assert np.allclose(self.ops.add(x, y), z)
assert np.allclose(self.ops.add(X, Y), Z)
def test_add_inplace(self):
X = self.ops.log(np.array([0, 1, 0, 2, 0, 1, 2, 1, 2]))
Y = self.ops.log(np.array([0, 0, 1, 0, 2, 2, 1, 1, 2]))
Z = self.ops.log(np.array([0, 1, 1, 2, 2, 3, 3, 2, 4]))
self.ops.add_inplace(X, Y)
assert np.allclose(X, Z)
def test_add_reduce(self):
X = self.ops.log(np.array([[0, 0, 0], [0, 1, 2], [1, 1, 1]]))
Y = self.ops.log(np.array([0, 3, 3]))
for x, y in zip(X, Y):
assert np.allclose(self.ops.add_reduce(x), y)
assert np.allclose(self.ops.add_reduce(np.array([])), self.ops.zero)
def test_mult(self):
X = self.ops.log(np.array([0, 1, 0, 0.5, 0, 1, 0.5, 1, 0.5, 2, 2, 2]))
Y = self.ops.log(np.array([0, 0, 1, 0, 5, 0.5, 1, 1, 0.5, 1, 2, 0.5]))
Z = self.ops.log(np.array([0, 0, 0, 0, 0, 0.5, 0.5, 1, 0.25, 2, 4, 1]))
for x, y, z in zip(X, Y, Z):
assert np.allclose(self.ops.mult(x, y), z)
assert np.allclose(self.ops.mult(X, Y), Z)
def test_mult_inplace(self):
X = self.ops.log(np.array([0, 1, 0, 0.5, 0, 1, 0.5, 1, 0.5, 2, 2, 2]))
Y = self.ops.log(np.array([0, 0, 1, 0, 5, 0.5, 1, 1, 0.5, 1, 2, 0.5]))
Z = self.ops.log(np.array([0, 0, 0, 0, 0, 0.5, 0.5, 1, 0.25, 2, 4, 1]))
self.ops.mult_inplace(X, Y)
assert np.allclose(X, Z)
def test_invert(self):
X = self.ops.log(np.array([1, 2, 0.5, 10], dtype=float))
Y = self.ops.log(np.array([1, 1 / 2, 2, 1 / 10]))
for x, y in zip(X, Y):
assert np.allclose(self.ops.invert(x), y)
def test_mult_reduce(self):
nums = np.arange(1, 5 + 1)
prods = np.cumprod(nums)
prods = self.ops.log(prods)
nums = self.ops.log(nums)
for i, p in enumerate(prods):
assert np.allclose(self.ops.mult_reduce(nums[:i + 1]), p)
assert np.allclose(self.ops.mult_reduce(np.array([])), self.ops.one)
def test_normalize(self):
W = np.ones(3)
X = self.ops.log(W)
Y = self.ops.normalize(X)
Z = self.ops.exp(Y)
Z_ = W / 3
assert np.allclose(Z, Z_)
def setup_class(self):
self.ops = LogOperations(2)
def test_exp_func2():
ops = LogOperations(0.5)
assert np.allclose([0.5**1, 0.5**2, 0.5**3], ops.exp([1, 2, 3]))
def test_log_func2():
ops = LogOperations(2)
assert np.allclose(
[np.log2(1), np.log2(2), np.log2(3)], ops.log([1, 2, 3]))
def setUp(self):
self.ops = LogOperations(2)
class TestLog2(object):
def setUp(self):
self.ops = LogOperations(2)
def test_add(self):
X = self.ops.log(np.array([0, 1, 0, 2, 0, 1, 2, 1, 2]))
Y = self.ops.log(np.array([0, 0, 1, 0, 2, 2, 1, 1, 2]))
Z = self.ops.log(np.array([0, 1, 1, 2, 2, 3, 3, 2, 4]))
for x, y, z in zip(X, Y, Z):
npt.assert_allclose(self.ops.add(x, y), z)
npt.assert_allclose(self.ops.add(X, Y), Z)
def test_add_inplace(self):
X = self.ops.log(np.array([0, 1, 0, 2, 0, 1, 2, 1, 2]))
Y = self.ops.log(np.array([0, 0, 1, 0, 2, 2, 1, 1, 2]))
Z = self.ops.log(np.array([0, 1, 1, 2, 2, 3, 3, 2, 4]))
self.ops.add_inplace(X, Y)
npt.assert_allclose(X, Z)
def test_add_reduce(self):
X = self.ops.log(np.array([[0, 0, 0], [0, 1, 2], [1, 1, 1]]))
Y = self.ops.log(np.array([0, 3, 3]))
for x, y in zip(X, Y):
npt.assert_allclose(self.ops.add_reduce(x), y)
npt.assert_allclose(self.ops.add_reduce(np.array([])), self.ops.zero)
def test_mult(self):
X = self.ops.log(np.array([0, 1, 0, 0.5, 0, 1, 0.5, 1, 0.5, 2, 2, 2]))
Y = self.ops.log(np.array([0, 0, 1, 0, 5, 0.5, 1, 1, 0.5, 1, 2, 0.5]))
Z = self.ops.log(np.array([0, 0, 0, 0, 0, 0.5, 0.5, 1, 0.25, 2, 4, 1]))
for x, y, z in zip(X, Y, Z):
npt.assert_allclose(self.ops.mult(x, y), z)
npt.assert_allclose(self.ops.mult(X, Y), Z)
def test_mult_inplace(self):
X = self.ops.log(np.array([0, 1, 0, 0.5, 0, 1, 0.5, 1, 0.5, 2, 2, 2]))
Y = self.ops.log(np.array([0, 0, 1, 0, 5, 0.5, 1, 1, 0.5, 1, 2, 0.5]))
Z = self.ops.log(np.array([0, 0, 0, 0, 0, 0.5, 0.5, 1, 0.25, 2, 4, 1]))
self.ops.mult_inplace(X, Y)
npt.assert_allclose(X, Z)
def test_invert(self):
X = self.ops.log(np.array([1, 2, 0.5, 10], dtype=float))
Y = self.ops.log(np.array([1, 1/2, 2, 1/10]))
for x, y in zip(X, Y):
npt.assert_allclose(self.ops.invert(x), y)
def test_mult_reduce(self):
nums = np.arange(1, 5+1)
prods = np.cumprod(nums)
prods = self.ops.log(prods)
nums = self.ops.log(nums)
for i, p in enumerate(prods):
npt.assert_allclose(self.ops.mult_reduce(nums[:i+1]), p)
npt.assert_allclose(self.ops.mult_reduce(np.array([])), self.ops.one)
def test_normalize(self):
W = np.ones(3)
X = self.ops.log(W)
Y = self.ops.normalize(X)
Z = self.ops.exp(Y)
Z_ = W / 3
npt.assert_allclose(Z, Z_)
def test_log_func2():
ops = LogOperations(2)
npt.assert_allclose([np.log2(1), np.log2(2), np.log2(3)],
ops.log([1, 2, 3]))
def test_exp_func2():
ops = LogOperations(0.5)
npt.assert_allclose([0.5**1, 0.5**2, 0.5**3], ops.exp([1, 2, 3]))
def setup_class(self):
self.ops = LogOperations(2)
