def test_10(self):
x = np.random.randn(9, 10)
y = np.random.randn(9, 10)
u = signal.grad(x, 0)
v = signal.gradT(y, 0)
err = np.dot(x.ravel(), v.ravel()) - np.dot(y.ravel(), u.ravel())
assert np.abs(err) < 5e-14
u = signal.grad(x, 1)
v = signal.gradT(y, 1)
err = np.dot(x.ravel(), v.ravel()) - np.dot(y.ravel(), u.ravel())
assert np.abs(err) < 5e-14
def test_11(self):
x = np.random.randn(9, 10)
y = np.random.randn(10, 10)
u = signal.grad(x, 0, zero_pad=True)
v = signal.gradT(y, 0, zero_pad=True)
err = np.dot(x.ravel(), v.ravel()) - np.dot(y.ravel(), u.ravel())
assert np.abs(err) < 1e-14
y = np.random.randn(9, 11)
u = signal.grad(x, 1, zero_pad=True)
v = signal.gradT(y, 1, zero_pad=True)
err = np.dot(x.ravel(), v.ravel()) - np.dot(y.ravel(), u.ravel())
assert np.abs(err) < 1e-14
def cnst_AT(self, X):
r"""Compute :math:`A^T \mathbf{x}` where :math:`A \mathbf{x}` is
a component of ADMM problem constraint. In this case
:math:`A^T \mathbf{x} = (G_r^T \;\; G_c^T) \mathbf{x}`.
"""
return np.sum(np.concatenate(
[gradT(X[..., ax], ax)[..., np.newaxis] for ax in self.axes],
axis=X.ndim-1), axis=X.ndim-1)