def test_newton_optimization(self):
opt = StandardOptimizer(function = self.function,
x0 = self.x0,
step = NewtonStep(),
line_search = SimpleLineSearch(),
criterion = criterion(ftol = 0.0001, iterations_max = 1000))
optimum = opt.optimize()
assert_array_almost_equal(optimum, numpy.array((4., 4., 3., -2.)))
def test_gradient_optimization(self):
opt = StandardOptimizer(function = self.function,
x0 = self.x0,
step = GradientStep(),
line_search = FibonacciSectionSearch(min_alpha_step=0.000001),
criterion = criterion(ftol = 0.0001, iterations_max = 1000))
optimum = opt.optimize()
assert_array_almost_equal(optimum, numpy.array((4., 4., 3., -2.)))
def test_gradient_optimization(self):
startPoint = numpy.zeros(2, numpy.float)
optimi = StandardOptimizer(function = Function2(),
step = FRConjugateGradientStep(),
criterion = criterion(ftol = 0.0000001, gtol=0.0001),
x0 = startPoint,
line_search = StrongWolfePowellRule())
assert_almost_equal(optimi.optimize(), numpy.array((2., -2)))
def test_hessian_optimization(self):
startPoint = numpy.zeros(2, numpy.float)
optimi = StandardOptimizer(function = Function(),
step = NewtonStep(),
criterion = criterion(iterations_max = 100, ftol = 0.0000001),
x0 = startPoint,
line_search = SimpleLineSearch())
assert_almost_equal(optimi.optimize(), numpy.array((2., -2)))
def test_newton_optimization(self):
opt = StandardOptimizer(function=self.function,
x0=self.x0,
step=MarquardtStep(gamma=10.),
line_search=SimpleLineSearch(),
criterion=criterion(ftol=0.00001,
iterations_max=1000))
optimum = opt.optimize()
assert_array_almost_equal(optimum, numpy.array((4., 4., 3., -2.)))
def test_gradient_optimization(self):
opt = StandardOptimizer(
function=self.function,
x0=self.x0,
step=GradientStep(),
line_search=FibonacciSectionSearch(min_alpha_step=0.000001),
criterion=criterion(ftol=0.00001, iterations_max=1000))
optimum = opt.optimize()
assert_array_almost_equal(optimum, numpy.array((4., 4., 3., -2.)))
def test_gradient_optimization(self):
startPoint = numpy.zeros(2, numpy.float)
optimi = StandardOptimizer(function=Function2(),
step=FRConjugateGradientStep(),
criterion=criterion(ftol=0.0000001,
gtol=0.0001),
x0=startPoint,
line_search=StrongWolfePowellRule())
assert_almost_equal(optimi.optimize(), numpy.array((2., -2)))
def test_hessian_optimization(self):
startPoint = numpy.zeros(2, numpy.float)
optimi = StandardOptimizer(function=Function(),
step=NewtonStep(),
criterion=criterion(iterations_max=100,
ftol=0.0000001),
x0=startPoint,
line_search=SimpleLineSearch())
assert_almost_equal(optimi.optimize(), numpy.array((2., -2)))