def __init__(self, sampler, a=2, b=2):
"""
Args:
sampler (DiscreteDistribution/TrueMeasure): A
discrete distribution from which to transform samples or a
true measure by which to compose a transform
a (ndarray): a
b (ndarray): b
"""
self.parameters = ['a', 'b']
self.domain = array([[0, 1]])
self.range = array([[0, 1]])
self._parse_sampler(sampler)
self.a = a
self.b = b
if isscalar(self.a):
a = tile(self.a, self.d)
if isscalar(self.b):
b = tile(self.b, self.d)
self.alpha = array(a)
self.beta = array(b)
if len(self.alpha) != self.d or len(self.beta) != self.d:
raise DimensionError(
'a and b must be scalar or have length equal to dimension.')
super(Kumaraswamy, self).__init__()
def _set_dimension(self, dimension):
l0 = self.l[0]
if (self.l!=l0).any():
raise DimensionError('''
In order to change dimension of a Continuous Bernoulli measure
lam must all be the same.''')
self.d = dimension
self.l = tile(l0,self.d)
def _set_dimension(self, dimension):
a = self.alpha[0]
b = self.beta[0]
if not (all(self.alpha == a) and all(self.beta == b)):
raise DimensionError('''
In order to change dimension of a Kumaraswamy measure
a must all be the same and
b must all be the same''')
self.d = dimension
self.alpha = tile(a, self.d)
self.beta = tile(b, self.d)
def _set_dimension(self, dimension):
gamma = self._gamma[0]
xi = self._xi[0]
delta = self._delta[0]
lam = self._lam[0]
if not (all(self._gamma == gamma) and all(self._xi == xi)
and all(self._delta == delta) and all(self._lam == lam)):
raise DimensionError('''
In order to change dimension of a Johnson's S_U measure
gamma must all be the same and
xi must all be the same and
delta must all be the same and
lam (lambda) must all be the same.''')
self.d = dimension
self._gamma = tile(gamma, self.d)
self._xi = tile(xi, self.d)
self._delta = tile(delta, self.d)
self._lam = tile(lam, self.d)
def __init__(self, sampler, lam=1/2, b=2):
"""
Args:
sampler (DiscreteDistribution/TrueMeasure): A
discrete distribution from which to transform samples or a
true measure by which to compose a transform
lam (ndarray): lambda, a shape parameter, independent for each dimension
"""
self.parameters = ['lam']
self.domain = array([[0,1]])
self.range = array([[0,1]])
self._parse_sampler(sampler)
self.lam = lam
if isscalar(self.lam):
lam = tile(self.lam,self.d)
self.l = array(lam)
if len(self.l)!=self.d or (self.l<=0).any() or (self.l>=1).any():
raise DimensionError('lam must be scalar or have length equal to dimension and must be in (0,1).')
super(BernoulliCont,self).__init__()
def __init__(self, sampler, gamma=1, xi=1, delta=2, lam=2):
"""
Args:
sampler (DiscreteDistribution/TrueMeasure): A
discrete distribution from which to transform samples or a
true measure by which to compose a transform
gamma (ndarray): gamma
xi (ndarray): xi
delta (ndarray): delta
lam (ndarray): lambda
"""
self.parameters = ['gamma', 'xi', 'delta', 'lam']
self.domain = array([[0, 1]])
self.range = array([[-inf, inf]])
self._parse_sampler(sampler)
self.gamma = gamma
self.xi = xi
self.delta = delta
self.lam = lam
if isscalar(self.gamma):
gamma = tile(self.gamma, self.d)
if isscalar(self.xi):
xi = tile(self.xi, self.d)
if isscalar(self.delta):
delta = tile(self.delta, self.d)
if isscalar(self.lam):
lam = tile(self.lam, self.d)
self._gamma = array(gamma)
self._xi = array(xi)
self._delta = array(delta)
self._lam = array(lam)
if len(self._gamma) != self.d or len(self._xi) != self.d or len(
self._delta) != self.d or len(self._lam) != self.d:
raise DimensionError(
"all Johnson's S_U parameters be scalar or have length equal to dimension."
)
if (self._delta <= 0).any() or (self._lam <= 0).any():
raise ParameterError(
"delta and lam (lambda) must be greater than 0")
super(JohnsonsSU, self).__init__()