def test_novak():
# expect (1.5, 1.5)
f = Function(lambda x: (x - 4).dot(x - 4), lambda x: 2 * (x - 4))
g1 = Function(lambda x: x[0] - 5, lambda x: np.array([1, 0]))
g2 = Function(lambda x: x.sum() - 3, lambda x: np.array([1, 1]))
x0 = np.zeros(2)
return penalty_optim(f, [g1], [g2], solver, x0, alpha_gen(), disp=True)
def test_medium():
# expect (0, 0)
f = Function(lambda x: x[0]**2 + x[0] * 6 + x[1]**2 + x[1] * 9,
lambda x: 2 * x + np.array([6, 9]))
g1 = Function(lambda x: -x[0], lambda x: np.array([-1, 0]))
g2 = Function(lambda x: -x[1], lambda x: np.array([0, -1]))
x0 = np.array([-1, 0.5])
return penalty_optim(f, [g1, g2], [], solver, x0, alpha_gen(), disp=True)
def test_impossible():
# expect to diverge
f = Function(lambda x: np.exp(-x[0]), lambda x: -np.exp(-x[0]))
g = Function(lambda x: x[0] * np.exp(-x[0]), lambda x:
(1 - x[0]) * np.exp(-x[0]))
x0 = np.array([1])
return penalty_optim(f, [], [g],
solver,
x0,
alpha_gen(),
eps=1e-20,
disp=True)
def test_what():
# expect to reach limit in function calls
f = Function(lambda x: x[0]**9, lambda x: 9 * x[0]**8)
g = Function(lambda x: 100000 - x[0], lambda x: np.array([-1]))
x0 = np.array([0])
return penalty_optim(f, [g], [], solver, x0, alpha_gen(), disp=True)
def test_simple():
# expect (1)
f = Function(lambda x: x[0]**2, lambda x: x[0] * 2)
g = Function(lambda x: 1 - x[0], lambda x: np.array([-1]))
x0 = 10
return penalty_optim(f, [g], [], solver, x0, alpha_gen(), disp=True)