def quadratic_optimization(M,
b,
Ds=[],
hypers=[],
maxiter=None,
tol=1e-6,
min_alpha_step=1e-10,
**kwargs):
"""
Use scikits.optimization to perform least-square inversion.
"""
from scikits.optimization import step, line_search, criterion, optimizer
if maxiter is None:
maxiter = M.shape[0]
model = QuadraticModel(M, b, Ds, hypers)
mystep = step.PRPConjugateGradientStep()
mylinesearch = QuadraticOptimalStep(M, b, Ds, hypers)
#mylinesearch = line_search.QuadraticInterpolationSearch(
# min_alpha_step=min_alpha_step,alpha_step = 1e-4)
mycriterion = criterion.criterion(gtol=tol, iterations_max=maxiter)
myoptimizer = VerboseStandardOptimizer(function=model,
step=mystep,
line_search=mylinesearch,
criterion=mycriterion,
x0=np.zeros(M.shape[1]))
return myoptimizer.optimize()
def huber_optimization(M,
b,
Ds=[],
hypers=[],
maxiter=None,
tol=1e-6,
min_alpha_step=1e-10,
**kwargs):
"""
Use scikits.optimization to perform least-square inversion.
"""
from scikits.optimization import step, line_search, criterion, optimizer
if maxiter is None:
maxiter = M.shape[0]
model = HuberModel(M, b, Ds, hypers, **kwargs)
mystep = step.PRPConjugateGradientStep()
mylinesearch = BacktrackingFromOptimal(M, b, Ds, hypers)
mycriterion = criterion.criterion(ftol=tol, iterations_max=maxiter)
myoptimizer = VerboseStandardOptimizer(function=model,
step=mystep,
line_search=mylinesearch,
criterion=mycriterion,
x0=np.zeros(M.shape[1]))
return myoptimizer.optimize()
def start_optimization(): startPoint = np.empty((3, 2), np.float) startPoint[:,0] = 1. startPoint[:,1] = 0. startPoint[1,0] -= .1 startPoint[2,1] += .1 recorder = Recorder() optimi = optimizer.PolytopeOptimizer(function = Rosenbrock(), criterion = criterion.criterion(ftol = .01, iterations_max=50), x0 = startPoint, record=recorder) print optimi.optimize()
def huber_optimization(M, b, Ds=[], hypers=[], maxiter=None, tol=1e-6,
min_alpha_step=1e-10, **kwargs):
"""
Use scikits.optimization to perform least-square inversion.
"""
from scikits.optimization import step, line_search, criterion, optimizer
if maxiter is None:
maxiter = M.shape[0]
model = HuberModel(M, b, Ds, hypers, **kwargs)
mystep = step.PRPConjugateGradientStep()
mylinesearch = BacktrackingFromOptimal(M, b, Ds, hypers)
mycriterion = criterion.criterion(ftol=tol, iterations_max=maxiter)
myoptimizer = VerboseStandardOptimizer(function=model,
step=mystep,
line_search=mylinesearch,
criterion=mycriterion,
x0 = np.zeros(M.shape[1]))
return myoptimizer.optimize()
def quadratic_optimization(M, b, Ds=[], hypers=[], maxiter=None, tol=1e-6,
min_alpha_step=1e-10, **kwargs):
"""
Use scikits.optimization to perform least-square inversion.
"""
from scikits.optimization import step, line_search, criterion, optimizer
if maxiter is None:
maxiter = M.shape[0]
model = QuadraticModel(M, b, Ds, hypers)
mystep = step.PRPConjugateGradientStep()
mylinesearch = QuadraticOptimalStep(M, b, Ds, hypers)
#mylinesearch = line_search.QuadraticInterpolationSearch(
# min_alpha_step=min_alpha_step,alpha_step = 1e-4)
mycriterion = criterion.criterion(gtol=tol, iterations_max=maxiter)
myoptimizer = VerboseStandardOptimizer(function=model,
step=mystep,
line_search=mylinesearch,
criterion=mycriterion,
x0 = np.zeros(M.shape[1]))
return myoptimizer.optimize()
"""
Use the optimization scikits to define a set of optimizer and models
using LinearOperators.
"""
import numpy as np
from copy import copy
from scikits.optimization import step, line_search, criterion, optimizer
from scikits.optimization.optimizer import StandardOptimizer
import criterions
import norms
# defaults
default_step = step.PRPConjugateGradientStep()
default_line_search = line_search.QuadraticInterpolationSearch(0)
default_stop_criterion = criterion.criterion()
# a class wrapping StandardOptimizer
class Optimizer(object):
def __init__(self,
criterion,
x0=None,
step=default_step,
line_search=default_line_search,
stop_criterion=default_stop_criterion):
self.criterion = criterion
self.n_variables = criterion.n_variables
self.step = step
self.line_search = line_search
self.stop_criterion = stop_criterion
self.first_guess()
"""
Use the optimization scikits to define a set of optimizer and models
using LinearOperators.
"""
import numpy as np
from copy import copy
from scikits.optimization import step, line_search, criterion, optimizer
from scikits.optimization.optimizer import StandardOptimizer
import criterions
import norms
# defaults
default_step = step.PRPConjugateGradientStep()
default_line_search = line_search.QuadraticInterpolationSearch(0)
default_stop_criterion = criterion.criterion()
# a class wrapping StandardOptimizer
class Optimizer(object):
def __init__(self, criterion,
x0=None,
step=default_step,
line_search=default_line_search,
stop_criterion=default_stop_criterion):
self.criterion = criterion
self.n_variables = criterion.n_variables
self.step = step
self.line_search = line_search
self.stop_criterion = stop_criterion
self.first_guess()
self.optimizer = StandardOptimizer(function=self.criterion,
step=self.step,