Lines Matching refs:scale
47 """The Laplace distribution with location `loc` and `scale` parameters.
58 where `loc = mu`, `scale = sigma`, and `Z` is the normalization constant.
63 The Lpalce distribution is a member of the [location-scale family](
68 X ~ Laplace(loc=0, scale=1)
69 Y = loc + scale * X
76 scale,
80 """Construct Laplace distribution with parameters `loc` and `scale`.
82 The parameters `loc` and `scale` must be shaped in a way that supports
83 broadcasting (e.g., `loc / scale` is a valid operation).
88 scale: Positive floating point tensor which characterizes the spread of
101 TypeError: if `loc` and `scale` are of different dtype.
104 with ops.name_scope(name, values=[loc, scale]):
105 with ops.control_dependencies([check_ops.assert_positive(scale)] if
108 self._scale = array_ops.identity(scale, name="scale")
122 zip(("loc", "scale"), ([ops.convert_to_tensor(
131 def scale(self):
132 """Distribution parameter for scale."""
137 array_ops.shape(self.loc), array_ops.shape(self.scale))
141 self.loc.get_shape(), self.scale.get_shape())
164 return (self.loc - self.scale * math_ops.sign(uniform_samples) *
188 return math.log(2.) + math_ops.log(self.scale)
191 # Use broadcasting rules to calculate the full broadcast scale.
192 scale = self.scale + array_ops.zeros_like(self.loc)
193 return math.log(2.) + 1. + math_ops.log(scale)
196 return self.loc + array_ops.zeros_like(self.scale)
199 return math.sqrt(2.) * self.scale + array_ops.zeros_like(self.loc)
208 return (x - self.loc) / self.scale
212 """Laplace with softplus applied to `scale`."""
216 scale,
221 with ops.name_scope(name, values=[loc, scale]):
224 scale=nn.softplus(scale, name="softplus_scale"),