def test_aug_lag(self):
x0 = np.array([[1], [1]])
start = time.time()
x_opt = opt.augmented_lagrange(self.p,
x0,
tol=1e-4,
tol_const=1e-4,
hist=True)
end = time.time()
g = np.array(
[np.linalg.norm(self.p.grad(x_opt[i])) for i in range(len(x_opt))])
c_e = np.array([
np.linalg.norm(np.minimum(self.p.eq_const(x_opt[i]), 0))
for i in range(len(x_opt))
])
fig = plt.figure()
plt.plot(np.arange(len(x_opt)), g, label='Gradient Norm')
plt.plot(np.arange(len(x_opt)), c_e, label='Equality Constraint Norm')
plt.xticks(np.arange(len(x_opt)))
plt.xlabel('Iteration')
plt.legend()
fig.savefig('./fig/al-pEEq.eps', format='eps')
print('\nProblem E-Eq, Augmented Lagrange')
print('arg min v(x) =\n', x_opt[-1])
print('time =\n', end - start, 's')
def test_eq_const_al(self):
x0 = np.array([[1], [0]])
x = opt.augmented_lagrange(self.p, x0, tol=1e-4, tol_const=1e-4)
x_opt = np.array([[0.7071318], [0.7071093]])
self.assertTrue(np.linalg.norm(x_opt - x) < 1e-4)
def test_aug_lag(self):
x0 = np.array([[2.1], [0.1]])
x = opt.augmented_lagrange(self.p, x0, tol=1e-4, tol_const=1e-4)
def test_aug_lag(self):
x0 = np.array([[10], [1]])
x = opt.augmented_lagrange(self.p, x0, tol=1e-4, tol_const=1e-4)
print(x)
