def test_sliding_tensor():
N = 1000
V = 5
width = 10
ts = np.random.rand(N, V)
for step in 1 + np.arange(width):
sts = transform.sliding_tensor(ts, width, step)
assert sts.shape[1] == width
assert sts.shape[2] == V
Nsts = 1 + (N - width) // step
assert Nsts == sts.shape[0]
for j in range(V):
assert np.all(np.isin(sts[:, :, j], ts[:, j]))
# todo: reconstruct tensor ts
final_tensor = []
for step in 1 + np.arange(width):
sts = transform.sliding_tensor(ts, width, step, 'C')
final_tensor.append(sts)
assert sts.flags.c_contiguous
assert sts.shape[1] == width
assert sts.shape[2] == V
Nsts = 1 + (N - width) // step
assert Nsts == sts.shape[0]
for j in range(V):
assert np.all(np.isin(sts[:, :, j], ts[:, j]))
assert np.concatenate(final_tensor).flags.c_contiguous
def test_sliding_tensor():
N = 1000
V = 5
width = 10
ts = np.ones((N, V))
for step in (1 + np.arange(width)):
sts = transform.sliding_tensor(ts, width, step)
assert sts.shape[1] == width
assert sts.shape[2] == V
Nsts = 1 + (N - width) // step
assert Nsts == sts.shape[0]
def test_sliding_tensor():
N = 1000
V = 5
width = 10
ts = np.random.rand(N, V)
for step in 1 + np.arange(width):
sts = transform.sliding_tensor(ts, width, step)
assert sts.shape[1] == width
assert sts.shape[2] == V
Nsts = 1 + (N - width) // step
assert Nsts == sts.shape[0]
for j in range(V):
assert np.all(np.isin(sts[:, :, j], ts[:, j]))