def __init__(self, loc, scale, skew): try: self._sknorm = pdf_moments((loc, scale**2, skew)) self.st = (loc, scale**2, skew) except ValueError: self._sknorm = pdf_moments((loc, 1e-8, skew)) self.st = (loc, 1e-8, skew) self.pdf = self._sknorm
def __init__(self, *args):
"""
Initializes the class. Should be given a list of tuples:
[('norm', (norm1_mean, norm1_std), 0.4),
('norm', (norm2_mean, norm2_std), 0.6)]
Can be given:
'norm', (mean, std), weight
'skewnorm', (mean, std, skewness), weight
'uniform', (min, max), weight
"""
self.dists = []
self.weights = []
self.cutoff = None
self.use_bc = False
if len(args) == 0:
return
for name, params, weight in args[0]:
if name == 'norm':
self.dists.append(norm(*params))
self.weights.append(weight)
elif name == 'skewnorm':
#pdf_moments takes the VARIANCE ... so must square the std
params = (params[0], params[1]**2, params[2])
self.dists.append(pdf_moments(params))
self.weights.append(weight)
elif name == 'uniform':
self.dists.append(uniform(*params))
self.weights.append(weight)
else:
raise KeyError, 'Unknown distribution %s' % name
assert N.sum(self.weights) == 1, 'sum(weight) != 1'