def testCdistExecution(self):
from scipy.spatial.distance import cdist as sp_cdist
raw_a = np.random.rand(100, 10)
raw_b = np.random.rand(89, 10)
# test 1 chunk
xa = tensor(raw_a, chunk_size=100)
xb = tensor(raw_b, chunk_size=100)
dist = distance.cdist(xa, xb)
result = self._executor.execute_tensor(dist, concat=True)[0]
expected = sp_cdist(raw_a, raw_b)
np.testing.assert_array_equal(result, expected)
dist = distance.cdist(xa, xb, metric='hamming')
result = self._executor.execute_tensor(dist, concat=True)[0]
expected = sp_cdist(raw_a, raw_b, metric='hamming')
np.testing.assert_array_equal(result, expected)
f = lambda u, v: np.sqrt(((u - v)**2).sum())
dist = distance.cdist(xa, xb, metric=f)
result = self._executor.execute_tensor(dist, concat=True)[0]
expected = sp_cdist(raw_a, raw_b, metric=f)
np.testing.assert_array_equal(result, expected)
# test more than 1 chunk
xa = tensor(raw_a, chunk_size=12)
xb = tensor(raw_b, chunk_size=13)
dist = distance.cdist(xa, xb)
result = self._executor.execute_tensor(dist, concat=True)[0]
expected = sp_cdist(raw_a, raw_b)
np.testing.assert_array_equal(result, expected)
dist = distance.cdist(xa, xb, metric='hamming')
result = self._executor.execute_tensor(dist, concat=True)[0]
expected = sp_cdist(raw_a, raw_b, metric='hamming')
np.testing.assert_array_equal(result, expected)
f = lambda u, v: np.sqrt(((u - v)**2).sum())
dist = distance.cdist(xa, xb, metric=f)
result = self._executor.execute_tensor(dist, concat=True)[0]
expected = sp_cdist(raw_a, raw_b, metric=f)
np.testing.assert_array_equal(result, expected)
for xa, xb in [(tensor(raw_a), tensor(raw_b)),
(tensor(raw_a,
chunk_size=12), tensor(raw_b, chunk_size=13))]:
# test w
weight = np.random.rand(10)
w = tensor(weight, chunk_size=7)
dist = distance.cdist(xa, xb, metric='wminkowski', p=3, w=w)
result = self._executor.execute_tensor(dist, concat=True)[0]
expected = sp_cdist(raw_a,
raw_b,
metric='wminkowski',
p=3,
w=weight)
np.testing.assert_array_equal(result, expected)
# test V
v = np.random.rand(10)
V = tensor(v, chunk_size=7)
dist = distance.cdist(xa, xb, metric='seuclidean', V=V)
result = self._executor.execute_tensor(dist, concat=True)[0]
expected = sp_cdist(raw_a, raw_b, metric='seuclidean', V=v)
np.testing.assert_array_equal(result, expected)
# test VI
vi = np.random.rand(10, 10)
VI = tensor(vi, chunk_size=8)
dist = distance.cdist(xa, xb, metric='mahalanobis', VI=VI)
result = self._executor.execute_tensor(dist, concat=True)[0]
expected = sp_cdist(raw_a, raw_b, metric='mahalanobis', VI=vi)
np.testing.assert_array_equal(result, expected)
def test_cdist():
raw_a = np.random.rand(100, 10)
raw_b = np.random.rand(90, 10)
# test 1 chunk
a = tensor(raw_a, chunk_size=100)
b = tensor(raw_b, chunk_size=100)
dist = distance.cdist(a, b)
assert dist.shape == (100, 90)
dist = tile(dist)
assert len(dist.chunks) == 1
for c in dist.chunks:
assert c.shape == dist.shape
# test multiple chunks
a = tensor(raw_a, chunk_size=15)
b = tensor(raw_b, chunk_size=16)
dist = distance.cdist(a, b)
assert dist.shape == (100, 90)
ta, tb, dist = tile(a, b, dist)
assert len(dist.chunks) == (100 // 15 + 1) * (90 // 16 + 1)
assert dist.nsplits == (ta.nsplits[0], tb.nsplits[0])
for c in dist.chunks:
assert c.shape == (ta.cix[c.index[0], 0].shape[0],
tb.cix[c.index[1], 0].shape[0])
# XA can only be 2-d
with pytest.raises(ValueError):
distance.cdist(np.random.rand(3, 3, 3), np.random.rand(3, 3))
# XB can only be 2-d
with pytest.raises(ValueError):
distance.cdist(np.random.rand(3, 3), np.random.rand(3, 3, 3))
# XA cannot be sparse
raw = sps.csr_matrix(np.zeros((4, 3)))
a = tensor(raw)
with pytest.raises(ValueError):
distance.cdist(a, np.random.rand(10, 3))
# XB cannot be sparse
raw = sps.csr_matrix(np.zeros((4, 3)))
b = tensor(raw)
with pytest.raises(ValueError):
distance.cdist(np.random.rand(10, 3), b)
# out type wrong
with pytest.raises(TypeError):
distance.cdist(raw_a, raw_b, out=2)
# out shape wrong
with pytest.raises(ValueError):
distance.cdist(raw_a, raw_b, out=tensor(np.random.rand(100, 91)))
# out dtype wrong
with pytest.raises(ValueError):
distance.cdist(raw_a, raw_b,
out=tensor(np.random.randint(2, size=(100, 90))))
# test extra param
with pytest.raises(TypeError):
distance.cdist(raw_a, raw_b, unknown_kw='unknown_kw')
def testCdist(self):
raw_a = np.random.rand(100, 10)
raw_b = np.random.rand(90, 10)
# test 1 chunk
a = tensor(raw_a, chunk_size=100)
b = tensor(raw_b, chunk_size=100)
dist = distance.cdist(a, b)
self.assertEqual(dist.shape, (100, 90))
dist = dist.tiles()
self.assertEqual(len(dist.chunks), 1)
for c in dist.chunks:
self.assertEqual(c.shape, dist.shape)
# test multiple chunks
a = tensor(raw_a, chunk_size=15)
b = tensor(raw_b, chunk_size=16)
dist = distance.cdist(a, b)
self.assertEqual(dist.shape, (100, 90))
dist = dist.tiles()
self.assertEqual(len(dist.chunks), (100 // 15 + 1) * (90 // 16 + 1))
self.assertEqual(dist.nsplits,
(get_tiled(a).nsplits[0], get_tiled(b).nsplits[0]))
for c in dist.chunks:
ta = get_tiled(a)
tb = get_tiled(b)
self.assertEqual(c.shape, (ta.cix[c.index[0], 0].shape[0],
tb.cix[c.index[1], 0].shape[0]))
# XA can only be 2-d
with self.assertRaises(ValueError):
distance.cdist(np.random.rand(3, 3, 3), np.random.rand(3, 3))
# XB can only be 2-d
with self.assertRaises(ValueError):
distance.cdist(np.random.rand(3, 3), np.random.rand(3, 3, 3))
# XA cannot be sparse
raw = sps.csr_matrix(np.zeros((4, 3)))
a = tensor(raw)
with self.assertRaises(ValueError):
distance.cdist(a, np.random.rand(10, 3))
# XB cannot be sparse
raw = sps.csr_matrix(np.zeros((4, 3)))
b = tensor(raw)
with self.assertRaises(ValueError):
distance.cdist(np.random.rand(10, 3), b)
# out type wrong
with self.assertRaises(TypeError):
distance.cdist(raw_a, raw_b, out=2)
# out shape wrong
with self.assertRaises(ValueError):
distance.cdist(raw_a, raw_b, out=tensor(np.random.rand(100, 91)))
# out dtype wrong
with self.assertRaises(ValueError):
distance.cdist(raw_a,
raw_b,
out=tensor(np.random.randint(2, size=(100, 90))))
# test extra param
with self.assertRaises(TypeError):
distance.cdist(raw_a, raw_b, unknown_kw='unknown_kw')