def testLgamma(self):
items = (
(0.1, 2.25271265),
(0.2, 1.52406382),
(0.3, 1.09579799),
(0.4, 0.79667782),
(0.5, 0.57236494),
(0.6, 0.39823386),
(0.7, 0.26086725),
(0.8, 0.15205968),
(0.9, 0.06637624),
(1.0, 0.00000000),
(1.1, -0.04987244),
(1.2, -0.08537409),
(1.3, -0.10817481),
(1.4, -0.11961291),
(1.5, -0.12078224),
(1.6, -0.11259177),
(1.7, -0.09580770),
(1.8, -0.07108387),
(1.9, -0.03898428),
(2.0, 0.00000000),
(2.1, 0.04543774),
(2.2, 0.09694747),
(2.3, 0.15418945),
(2.4, 0.21685932),
(2.5, 0.28468287),
(2.6, 0.35741186),
(2.7, 0.43482055),
(2.8, 0.51670279),
(2.9, 0.60286961),
(3.0, 0.69314718),
)
for v, lg in items:
print(v, lg, lgamma(v))
self.assertLessEqual(
abs(lgamma(v) - lg), 1.0e-8,
"log Gamma(%f) = %f; lgamma(%f) -> %f" % (v, lg, v, lgamma(v)))
def testLgamma(self):
items = (
(0.1, 2.25271265),
(0.2, 1.52406382),
(0.3, 1.09579799),
(0.4, 0.79667782),
(0.5, 0.57236494),
(0.6, 0.39823386),
(0.7, 0.26086725),
(0.8, 0.15205968),
(0.9, 0.06637624),
(1.0, 0.00000000),
(1.1, -0.04987244),
(1.2, -0.08537409),
(1.3, -0.10817481),
(1.4, -0.11961291),
(1.5, -0.12078224),
(1.6, -0.11259177),
(1.7, -0.09580770),
(1.8, -0.07108387),
(1.9, -0.03898428),
(2.0, 0.00000000),
(2.1, 0.04543774),
(2.2, 0.09694747),
(2.3, 0.15418945),
(2.4, 0.21685932),
(2.5, 0.28468287),
(2.6, 0.35741186),
(2.7, 0.43482055),
(2.8, 0.51670279),
(2.9, 0.60286961),
(3.0, 0.69314718),
)
for v, lg in items:
print v, lg, lgamma(v)
self.assertLessEqual(abs(lgamma(v) - lg), 1.0e-8,
"log Gamma(%f) = %f; lgamma(%f) -> %f" % (
v, lg, v, lgamma(v)))
def __init__(self, shape, scale=None, beta=None):
self.alpha = shape
if scale is not None:
self.scale = scale
self.beta = 1.0 / scale
assert beta is None, "Specify exactly one of 'scale' or 'beta'."
elif beta is not None:
self.scale = 1.0 / beta
self.beta = beta
else:
assert beta is None, "Specify exactly one of 'scale' or 'beta'."
self.lgammaAlpha = lgamma(shape)
self.logBeta = -numpy.log(scale)
def __init__(self, shape, scale=None, beta=None):
self.alpha = shape
if scale is not None:
self.scale = scale
self.beta = 1.0 / scale
assert beta is None, "Specify exactly one of 'scale' or 'beta'."
elif beta is not None:
self.scale = 1.0 / beta
self.beta = beta
else:
assert beta is None, "Specify exactly one of 'scale' or 'beta'."
self.lgammaAlpha = lgamma(shape)
self.logBeta = -numpy.log(scale)
def logFactorial(x):
"""Approximation to the log of the factorial function."""
return lgamma(x + 1.0)
def logBeta(alpha):
alpha = numpy.asarray(alpha)
return numpy.sum([lgamma(a) for a in alpha]) - lgamma(alpha.sum())
def logFactorial(x): """Approximation to the log of the factorial function.""" return lgamma(x + 1.0)
def logBeta(alpha): alpha = numpy.asarray(alpha) return numpy.sum([lgamma(a) for a in alpha]) - lgamma(alpha.sum())