def test_window(rank1, solver_d, uv_constraint):
# Smoke test that the parameter window does something
n_trials, n_channels, n_times = 2, 3, 100
n_times_atom, n_atoms = 10, 4
rng = check_random_state(42)
X = rng.randn(n_trials, n_channels, n_times)
*_, uv_constraint_ = check_solver_and_constraints(rank1, solver_d,
uv_constraint)
D_init = init_dictionary(X,
n_atoms,
n_times_atom,
rank1=rank1,
uv_constraint=uv_constraint_,
random_state=0)
kwargs = dict(X=X,
n_atoms=n_atoms,
n_times_atom=n_times_atom,
verbose=0,
uv_constraint=uv_constraint,
solver_d=solver_d,
rank1=rank1,
random_state=0,
n_iter=1,
solver_z='l-bfgs',
D_init=D_init)
res_False = learn_d_z_multi(window=False, **kwargs)
res_True = learn_d_z_multi(window=True, **kwargs)
assert not np.allclose(res_False[2], res_True[2])
def test_init_random(rank1, solver_d, uv_constraint):
""""""
n_trials, n_channels, n_times = 5, 3, 100
n_times_atom, n_atoms = 10, 4
_, uv_constraint_ = check_solver_and_constraints(rank1, solver_d,
uv_constraint)
if rank1:
expected_shape = (n_atoms, n_channels + n_times_atom)
prox = functools.partial(prox_uv,
uv_constraint=uv_constraint_,
n_channels=n_channels)
else:
expected_shape = (n_atoms, n_channels, n_times_atom)
prox = prox_d
X = np.random.randn(n_trials, n_channels, n_times)
# Test that init_dictionary is doing what we expect for D_init random
random_state = 42
D_hat = init_dictionary(X,
n_atoms,
n_times_atom,
D_init='random',
rank1=rank1,
uv_constraint=uv_constraint_,
random_state=random_state)
rng = check_random_state(random_state)
D_init = rng.randn(*expected_shape)
D_init = prox(D_init)
assert_allclose(D_hat,
D_init,
err_msg="The random state is not correctly "
"used in init_dictionary .")
# Test that learn_d_z_multi is doing what we expect for D_init random
random_state = 27
_, _, D_hat, _, _ = learn_d_z_multi(X,
n_atoms,
n_times_atom,
D_init='random',
n_iter=0,
rank1=rank1,
solver_d=solver_d,
uv_constraint=uv_constraint,
random_state=random_state)
rng = check_random_state(random_state)
D_init = rng.randn(*expected_shape)
D_init = prox(D_init)
assert_allclose(D_hat,
D_init,
err_msg="The random state is not correctly "
"used in learn_d_z_multi.")
def one_run(X, X_shape, random_state, method, n_atoms, n_times_atom, reg):
assert X.shape == X_shape
func, label, n_iter = method
current_time = time.time() - START
msg = ('%s - %s: started at T=%.0f sec' %
(random_state, label, current_time))
print(colorify(msg, BLUE))
if len(X_shape) == 2:
n_trials, n_times = X.shape
n_channels = 1
X_init = X[:, None, :]
else:
n_trials, n_channels, n_times = X.shape
X_init = X
# use the same init for all methods
ds_init = init_dictionary(X_init,
n_atoms,
n_times_atom,
D_init='chunk',
rank1=False,
uv_constraint='separate',
D_init_params=dict(),
random_state=random_state)
if len(X_shape) == 2:
ds_init = ds_init[:, 0, :]
# run the selected algorithm with one iter to remove compilation overhead
_, _, _, _ = func(X, ds_init, reg, 1, random_state, label)
# run the selected algorithm
pobj, times, d_hat, z_hat = func(X, ds_init, reg, n_iter, random_state,
label)
# store z_hat in a sparse matrix to reduce size
for z in z_hat:
z[z < 1e-3] = 0
z_hat = [sp.csr_matrix(z) for z in z_hat]
duration = time.time() - START - current_time
current_time = time.time() - START
msg = ('%s - %s: done in %.0f sec at T=%.0f sec' %
(random_state, label, duration, current_time))
print(colorify(msg, GREEN))
return (random_state, label, np.asarray(pobj), np.asarray(times),
np.asarray(d_hat), np.asarray(z_hat), n_atoms, n_times_atom,
n_trials, n_times, n_channels, reg)
def test_init_array(rank1, solver_d, uv_constraint):
n_trials, n_channels, n_times = 5, 3, 100
n_times_atom, n_atoms = 10, 4
_, uv_constraint_ = check_solver_and_constraints(rank1, solver_d,
uv_constraint)
if rank1:
expected_shape = (n_atoms, n_channels + n_times_atom)
prox = functools.partial(prox_uv,
uv_constraint=uv_constraint_,
n_channels=n_channels)
else:
expected_shape = (n_atoms, n_channels, n_times_atom)
prox = prox_d
X = np.random.randn(n_trials, n_channels, n_times)
# Test that init_dictionary is doing what we expect for D_init array
D_init = np.random.randn(*expected_shape)
D_hat = init_dictionary(X,
n_atoms,
n_times_atom,
D_init=D_init,
rank1=rank1,
uv_constraint=uv_constraint_)
D_init = prox(D_init)
assert_allclose(D_hat, D_init)
assert id(D_hat) != id(D_init)
# Test that learn_d_z_multi is doing what we expect for D_init array
D_init = np.random.randn(*expected_shape)
_, _, D_hat, _, _ = learn_d_z_multi(X,
n_atoms,
n_times_atom,
D_init=D_init,
n_iter=0,
rank1=rank1,
solver_d=solver_d,
uv_constraint=uv_constraint)
D_init = prox(D_init)
assert_allclose(D_hat, D_init)
def test_init_shape(D_init, rank1):
n_trials, n_channels, n_times = 5, 3, 100
n_times_atom, n_atoms = 10, 4
X = np.random.randn(n_trials, n_channels, n_times)
expected_shape = (n_atoms, n_channels, n_times_atom)
if rank1:
expected_shape = (n_atoms, n_channels + n_times_atom)
# Test that init_dictionary returns correct shape
uv_hat = init_dictionary(X,
n_atoms,
n_times_atom,
D_init=D_init,
rank1=rank1,
uv_constraint='separate',
random_state=42)
assert uv_hat.shape == expected_shape
def D_hat(X, rank1):
return init_dictionary(X,
N_ATOMS,
N_TIMES_ATOM,
random_state=0,
rank1=rank1)
# Note the use of reshape to shape the signal as per alphacsc requirements: the
# shape of the signal should be (n_trials, n_channels, n_times). Here, we have
# a single-channel time series so it is (1, 1, n_times).
from alphacsc.init_dict import init_dictionary
# set dictionary size
n_atoms = 8
# set individual atom (patch) size.
n_times_atom = 200
D_init = init_dictionary(X,
n_atoms=8,
n_times_atom=200,
rank1=False,
window=True,
D_init='chunk',
random_state=60)
print(D_init.shape)
""
from alphacsc import BatchCDL
cdl = BatchCDL(
# Shape of the dictionary
n_atoms,
n_times_atom,
rank1=False,
uv_constraint='auto',
