def test_infeasible(self):
np.random.seed(1)
leadfield = np.random.random((5, 10, 3))
np.random.seed(None)
targets = [0, 1]
target_direction = [[1, 0, 0], [1, 0, 0]]
target_mean = 1e4
max_total_current = 0.2
max_el_current = 0.1
tes_problem = optimization_methods.TESLinearConstrained(
leadfield, max_total_current, max_el_current
)
tes_problem.add_linear_constraint(targets, target_direction, target_mean)
x = tes_problem.solve()
x_sp = optimize_comp(
np.squeeze(tes_problem.l),
np.ones((1, 6)),
max_el_current,
max_total_current
)
assert np.linalg.norm(x, 1) <= 2 * max_total_current
assert np.all(np.abs(x) <= max_el_current)
assert np.isclose(np.sum(x), 0)
assert np.allclose(tes_problem.l.dot(x), tes_problem.l.dot(x_sp),
rtol=1e-4, atol=1e-5)
def test_feasible(self, target_mean):
np.random.seed(1)
leadfield = np.random.random((5, 10, 3))
np.random.seed(None)
targets = [0, 1]
target_direction = [[1, 0, 0], [1, 0, 0]]
max_total_current = 0.2
max_el_current = 0.1
tes_problem = optimization_methods.TESLinearConstrained(
leadfield, max_total_current, max_el_current
)
tes_problem.add_linear_constraint(targets, target_direction, target_mean)
x = tes_problem.solve()
x_sp = optimize_focality(
tes_problem.l, tes_problem.Q,
target_mean, max_el_current,
max_total_current
)
assert np.all(np.abs(x) <= max_el_current)
assert np.all(np.linalg.norm(x) <= 2*max_total_current)
assert np.isclose(np.sum(x), 0)
assert np.isclose(tes_problem.l.dot(x), tes_problem.target_means)
assert np.allclose(
x.dot(tes_problem.Q.dot(x)),
x_sp.dot(tes_problem.Q.dot(x_sp)),
rtol=1e-4, atol=1e-5
)
def test_calc_many_targets(self):
A = np.random.random((2, 5, 3))
volumes = np.array([1, 2, 2, 2, 2])
tes_problem = optimization_methods.TESLinearConstrained(
A, weights=volumes
)
tes_problem.add_linear_constraint([0], [1, 0, 0], 0.2)
tes_problem.add_linear_constraint([1], [1, 0, 0], 0.4)
currents = [-2, 1, 1]
field_x = A[..., 0].T.dot(currents[1:])
assert np.allclose(field_x[[0, 1]], tes_problem.l.dot(currents))
assert np.allclose([0.2, 0.4], tes_problem.target_means)
def test_calc_different_directions(self):
A = np.random.random((2, 5, 3))
targets = [0, 1]
target_direction = np.array([[1, 0, 0], [0, 1, 0]])
volumes = np.array([1, 2, 2, 2, 2])
tes_problem = optimization_methods.TESLinearConstrained(
A, weights=volumes
)
tes_problem.add_linear_constraint(targets, target_direction, 0.2)
currents = [1, 1]
field_x = A[..., 0].T.dot(currents)
field_y = A[..., 1].T.dot(currents)
avg = (field_x[0]*1 + field_y[1]*2)/3
assert np.allclose(avg, tes_problem.l.dot([-2, 1, 1]))
def test_calc_l(self):
A = np.random.random((2, 5, 3))
targets = [0, 1]
target_direction = [[1, 0, 0], [1, 0, 0]]
volumes = np.array([1, 2, 2, 2, 2])
tes_problem = optimization_methods.TESLinearConstrained(
A, weights=volumes
)
tes_problem.add_linear_constraint(targets, target_direction, 0.2)
currents = [-2, 1, 1]
field_x = A[..., 0].T.dot(currents[1:])
avg = np.average(field_x[[0, 1]], weights=[1, 2])
assert np.allclose(avg, tes_problem.l.dot(currents))
assert np.allclose(0.2, tes_problem.target_means)
def test_multi_target_feasible(self):
np.random.seed(1)
leadfield = np.random.random((5, 10, 3))
np.random.seed(None)
max_total_current = 0.2
max_el_current = 0.1
tes_problem = optimization_methods.TESLinearConstrained(
leadfield, max_total_current, max_el_current
)
tes_problem.add_linear_constraint([0], [1, 0, 0], 1e-3)
tes_problem.add_linear_constraint([1], [1, 0, 0], 2e-3)
x = tes_problem.solve()
assert np.all(np.abs(x) <= max_el_current)
assert np.all(np.linalg.norm(x) <= 2*max_total_current)
assert np.isclose(np.sum(x), 0)
assert np.allclose(tes_problem.l.dot(x), tes_problem.target_means)