def CostFactory2(self,pts,datapts,nparams):
"""generates a cost function instance from datapoints
and evaluation points"""
F = CF()
F.addModel(self.ForwardFactory,self.__name__,nparams)
self.__cost__ = F.getCostFunction(evalpts=pts,observations=datapts,sigma=self.__sigma__,metric=self.__metric__)
return self.__cost__
def PolyCostFactory(evalpts,datapts,ndim):
"""generate a cost function instance from evaluation points and data"""
from numpy import sum
F = CostFactory()
F.addModel(ForwardPolyFactory,ndim,"noisy_polynomial")
return F.getCostFunction(evalpts=evalpts,observations=datapts, \
metric=lambda x: sum(x*x),sigma=1000.)
def CostFactory(self,target,pts):
"""generates a cost function instance from list of coefficients & evaluation points"""
datapts = self.evaluate(target,pts)
F = CF()
F.addModel(self.ForwardFactory,len(target),self.__name__)
self.__cost__ = F.getCostFunction(evalpts=pts,observations=datapts,sigma=sqrt(datapts),metric=self.__metric__)
return self.__cost__
def PolyCostFactory(evalpts,datapts,ndim):
"""generate a cost function instance from evaluation points and data"""
from numpy import sum
F = CostFactory()
F.addModel(ForwardPolyFactory,ndim,"noisy_polynomial")
return F.getCostFunction(evalpts=evalpts,observations=datapts, \
metric=lambda x: sum(x*x),sigma=1000.)
def CostFactory(self,target,pts):
"""generates a cost function instance from list of coefficients & evaluation points"""
datapts = self.evaluate(target,pts)
F = CF()
F.addModel(self.ForwardFactory,self.__name__,len(target))
self.__cost__ = F.getCostFunction(evalpts=pts,observations=datapts,sigma=sqrt(datapts),metric=self.__metric__)
return self.__cost__
def CostFactory2(self,pts,datapts,nparams):
"""generates a cost function instance from datapoints
and evaluation points"""
F = CF()
F.addModel(self.ForwardFactory,nparams,self.__name__)
self.__cost__ = F.getCostFunction(evalpts=pts,observations=datapts,sigma=self.__sigma__,metric=self.__metric__)
return self.__cost__
minrange = [-1000., -1000., -100., -10.]
maxrange = [1000., 1000., 100., 10.]
solver.SetRandomInitialPoints(min=minrange, max=maxrange)
solver.SetEvaluationLimits(generations=MAX_GENERATIONS)
solver.SetGenerationMonitor(stepmon)
solver.Solve(CF, termination = ChangeOverGeneration(generations=100), \
CrossProbability=0.5, ScalingFactor=0.5)
solution = solver.Solution()
return solution, stepmon
if __name__ == '__main__':
F = CostFactory()
F.addModel(ForwardMogiFactory, 4, 'mogi1', outputFilter=component(2))
myCostFunction = F.getCostFunction(evalpts=stations, observations=data_z)
print(F)
rp = F.getRandomParams()
print("new Cost Function : %s " % myCostFunction(rp))
print("orig Cost Function: %s " % cost_function(rp))
f1 = ForwardMogiFactory(rp)
f2 = ForwardMogiFactory(rp)
print('start cf')
for i in range(3000):
xx = cost_function(rp)
print('end cf')
maxrange = [1000., 1000., 100., 1.]*2;
solver.SetRandomInitialPoints(min = minrange, max = maxrange)
solver.SetEvaluationLimits(generations=MAX_GENERATIONS)
solver.SetGenerationMonitor(stepmon)
solver.Solve(CF, termination = ChangeOverGeneration(generations=300), \
CrossProbability=0.5, ScalingFactor=0.5, \
sigint_callback = plot_sol)
solution = solver.Solution()
return solution, stepmon
if __name__ == '__main__':
F = CostFactory()
F.addModel(ForwardMogiFactory, 4, 'mogi1', outputFilter = PickComponent(2, -1))
F.addModel(ForwardMogiFactory, 4, 'mogi2', outputFilter = PickComponent(2, -1))
myCostFunction = F.getCostFunction(evalpts = stations, observations = data_z)
print F
def C2(x):
"This is the new version"
return 100000 * myCostFunction(x)
def C3(x):
"Cost function constructed by hand"
from test_mogi2 import cost_function
return cost_function(x)
def test():
from mystic.models.lorentzian import gendata, histogram
npts = 4000; binwidth = 0.1
N = npts * binwidth
xmin, xmax = 0.0, 3.0
pdf = F(target) # normalized
print "pdf(1): ", pdf(1)
data = gendata(target, xmin, xmax, npts) # data is 'unnormalized'
#pylab.plot(data[1:N],0*data[1:N],'k.')
#pylab.title('Samples drawn from density to be estimated.')
#show()
#pylab.clf()
binsc, histo = histogram(data, binwidth, xmin,xmax)
print "binsc: ", binsc
print "count: ", histo
print "ncount: ", histo/N
print "exact : ", pdf(binsc)
print "now with DE..."
from mystic.forward_model import CostFactory
CF = CostFactory()
CF.addModel(F, 'lorentz', ND)
myCF = CF.getCostFunction(binsc, histo/N)
sol, steps = de_solve(myCF)
plot_sol()(sol)
#print "steps: ", steps.x, steps.y
# end of file
npts = 4000
binwidth = 0.1
N = npts * binwidth
xmin, xmax = 0.0, 3.0
pdf = F(target) # normalized
print "pdf(1): ", pdf(1)
data = gendata(target, xmin, xmax, npts) # data is 'unnormalized'
# pylab.plot(data[1:N],0*data[1:N],'k.')
# pylab.title('Samples drawn from density to be estimated.')
# show()
# pylab.clf()
binsc, histo = histogram(data, binwidth, xmin, xmax)
print "binsc: ", binsc
print "count: ", histo
print "ncount: ", histo / N
print "exact : ", pdf(binsc)
print "now with DE..."
from mystic.forward_model import CostFactory
CF = CostFactory()
CF.addModel(F, ND, "lorentz")
myCF = CF.getCostFunction(binsc, histo / N)
sol, steps = de_solve(myCF)
plot_sol()(sol)
# print "steps: ", steps.x, steps.y
# end of file
from mystic.models.lorentzian import gendata, histogram
npts = 4000; binwidth = 0.1
N = npts * binwidth
xmin, xmax = 0.0, 3.0
pdf = F(target) # normalized
print("pdf(1): %s" % pdf(1))
data = gendata(target, xmin, xmax, npts) # data is 'unnormalized'
#pylab.plot(data[1:N],0*data[1:N],'k.')
#pylab.title('Samples drawn from density to be estimated.')
#show()
#pylab.clf()
binsc, histo = histogram(data, binwidth, xmin,xmax)
print("binsc: %s" % binsc)
print("count: %s" % histo)
print("ncount: %s" % (histo//N))
print("exact : %s" % pdf(binsc))
print("now with DE...")
from mystic.forward_model import CostFactory
CF = CostFactory()
CF.addModel(F, ND, 'lorentz')
myCF = CF.getCostFunction(binsc, histo//N)
sol, steps = de_solve(myCF)
plot_sol()(sol)
#print("steps: %s %s" % (steps.x, steps.y))
# end of file
stepmon = Monitor()
minrange = [-1000., -1000., -100., -10.];
maxrange = [1000., 1000., 100., 10.];
solver.SetRandomInitialPoints(min = minrange, max = maxrange)
solver.SetEvaluationLimits(generations=MAX_GENERATIONS)
solver.SetGenerationMonitor(stepmon)
solver.Solve(CF, termination = ChangeOverGeneration(generations=100), \
CrossProbability=0.5, ScalingFactor=0.5)
solution = solver.Solution()
return solution, stepmon
if __name__ == '__main__':
F = CostFactory()
F.addModel(ForwardMogiFactory, 'mogi1', 4, outputFilter = PickComponent(2))
myCostFunction = F.getCostFunction(evalpts = stations, observations = data_z)
print F
rp = F.getRandomParams()
print "new Cost Function : %s " % myCostFunction(rp)
print "orig Cost Function: %s " % cost_function(rp)
f1 = ForwardMogiFactory(rp)
f2 = ForwardMogiFactory(rp)
print 'start cf'
for i in range(3000):
xx = cost_function(rp)
print 'end cf'
solver.Solve(
CF,
termination=ChangeOverGeneration(generations=300),
CrossProbability=0.5,
ScalingFactor=0.5,
sigint_callback=plot_sol,
)
solution = solver.Solution()
return solution, stepmon
if __name__ == "__main__":
F = CostFactory()
F.addModel(ForwardMogiFactory, 4, "mogi1", outputFilter=PickComponent(2, -1))
F.addModel(ForwardMogiFactory, 4, "mogi2", outputFilter=PickComponent(2, -1))
myCostFunction = F.getCostFunction(evalpts=stations, observations=data_z)
print F
def C2(x):
"This is the new version"
return 100000 * myCostFunction(x)
def C3(x):
"Cost function constructed by hand"
from test_mogi2 import cost_function
return cost_function(x)
