def __init__(self,
params,
num,
ds=False,
response=True,
iteration_cb=None):
PSweep.__init__(self, iteration_cb)
self.params = params
num = int(num)
self.num = num
self.ds = ds
self.response = response
self._start_at = 0
if self.response:
# To generate a complete response surface, use Uniform distributions
# with the same range as the original distributions.
for p in self.params:
if ds:
p.values = UniformPDF(*p.pdf.range).ds(num)
else:
p.values = UniformPDF(*p.pdf.range).lhs(num)
else:
for p in self.params:
if ds:
p.values = p.pdf.ds(num)
else:
p.values = p.pdf.lhs(num)
class UniformParameter(Parameter):
'''
Class implementing a Parameter with a Uniform distribution.
Args:
name: Name of the parameter. This should be a short name,
like a variable.
description: A longer description of the parameter.
kwargs: Keyword args. Valid args are:
======== ============================
Property Description
======== ============================
mean The mean of the distribution
max The maximum
min The minimum
======== ============================
You **must** specify two of the above properties. If you
give all three, they will be checked for consistency.
:math:`mean = (min + max)/2`
'''
def __init__(self, name, description, **kwargs):
Parameter.__init__(self, name, description, **kwargs)
self.pdf = UniformPDF(**self.kwargs)
def __str__(self):
return "UniformParameter %s (%s)\n\t%s" %\
(self.name, self.description, self.pdf.__str__())
def extend(self, num):
if num <= 0:
print "Monte Carlo extend requires a valid num argument."
raise ValueError
for p in self.params:
if self.response:
p.values = np.concatenate(
(p.values, UniformPDF(*p.pdf.range).random(num)))
else:
p.values = np.concatenate((p.values, p.pdf.random(num)))
self._start_at = self.num
self.num += num
def extend(self, num=None):
if not hasattr(self, 'ds') or not self.ds:
print("You must use Descriptive Sampling to extend.")
sys.exit(1)
print("Extending Descriptive Sampling run to %s samples." % (self.num * 3))
for p in self.params:
if self.response:
v = np.sort(UniformPDF(*p.pdf.range).ds(self.num * 3))
else:
v = np.sort(p.pdf.ds(self.num * 3))
# remove the ones we already did
v = np.concatenate((v[0::3], v[2::3]))
p.values = np.concatenate((p.values, np.random.permutation(v)))
self._start_at = self.num
self.num *= 3
def __init__(self, name, description, **kwargs):
Parameter.__init__(self, name, description, **kwargs)
self.pdf = UniformPDF(**self.kwargs)
def __init__(self, name, description, **kwargs):
self.name = self.check_name(name)
self.description = description
self.caldata = kwargs.pop('caldata', None)
self.pdf = UniformPDF(**kwargs)