def test_distance_matrix():
metric = dipymetric.SumPointwiseEuclideanMetric()
for dtype in [np.int32, np.int64, np.float32, np.float64]:
# Compute distances of all tuples spawn by the Cartesian product
# of `data` with itself.
data = (np.random.rand(4, 10, 3)*10).astype(dtype)
D = dipymetric.distance_matrix(metric, data)
assert_equal(D.shape, (len(data), len(data)))
assert_array_equal(np.diag(D), np.zeros(len(data)))
if metric.is_order_invariant:
# Distance matrix should be symmetric
assert_array_equal(D, D.T)
for i in range(len(data)):
for j in range(len(data)):
assert_equal(D[i, j], dipymetric.dist(metric, data[i], data[j]))
# Compute distances of all tuples spawn by the Cartesian product
# of `data` with `data2`.
data2 = (np.random.rand(3, 10, 3)*10).astype(dtype)
D = dipymetric.distance_matrix(metric, data, data2)
assert_equal(D.shape, (len(data), len(data2)))
for i in range(len(data)):
for j in range(len(data2)):
assert_equal(D[i, j], dipymetric.dist(metric, data[i], data2[j]))
def test_distance_matrix():
metric = dipymetric.SumPointwiseEuclideanMetric()
for dtype in [np.int32, np.int64, np.float32, np.float64]:
# Compute distances of all tuples spawn by the Cartesian product
# of `data` with itself.
data = (np.random.rand(4, 10, 3) * 10).astype(dtype)
D = dipymetric.distance_matrix(metric, data)
assert_equal(D.shape, (len(data), len(data)))
assert_array_equal(np.diag(D), np.zeros(len(data)))
if metric.is_order_invariant:
# Distance matrix should be symmetric
assert_array_equal(D, D.T)
for i in range(len(data)):
for j in range(len(data)):
assert_equal(D[i, j], dipymetric.dist(metric, data[i],
data[j]))
# Compute distances of all tuples spawn by the Cartesian product
# of `data` with `data2`.
data2 = (np.random.rand(3, 10, 3) * 10).astype(dtype)
D = dipymetric.distance_matrix(metric, data, data2)
assert_equal(D.shape, (len(data), len(data2)))
for i in range(len(data)):
for j in range(len(data2)):
assert_equal(D[i, j], dipymetric.dist(metric, data[i],
data2[j]))