nifty8.re.prior module#

class InvGammaPrior(a, scale, loc=0.0, step=0.01, **kwargs)[source]#

__init__(a, scale, loc=0.0, step=0.01, **kwargs)[source]#

Transforms standard normally distributed random variables to an inverse gamma distribution.

Parameters:

a (float) – Shape parameter.
scale (float) – Scale parameter.
loc (float) – Location parameter.
step (float) – Step size for numerical integration.
name (hashable, optional) – Name within the new input on which call acts.
shape (tuple or tree-like structure of ShapeWithDtype) – Shape of the latent parameter(s) that are transformed to the desired distribution. This can also be an arbitrary shape-dtype structure in which case dtype is ignored. Defaults to a scalar.
dtype (dtype) – Data type of the latent parameter(s) that are transformed to the desired distribution.

Notes

Broadcasting over tree-like structure is not yet implemented. Please file an issue if you need this feature.

class LaplacePrior(alpha, **kwargs)[source]#

__init__(alpha, **kwargs)[source]#

Transforms standard normally distributed random variables to a Laplace distribution.

Parameters:

alpha (tree-like structure with arithmetics) – Scale parameter.
name (hashable, optional) – Name within the new input on which call acts.
shape (tuple or tree-like structure of ShapeWithDtype) – Shape of the latent parameter(s) that are transformed to the desired distribution. This can also be an arbitrary shape-dtype structure in which case dtype is ignored. Defaults to a scalar.
dtype (dtype) – Data type of the latent parameter(s) that are transformed to the desired distribution.

class LogNormalPrior(mean, std, **kwargs)[source]#

__init__(mean, std, **kwargs)[source]#

Transforms standard normally distributed random variables to a log-normal distribution.

Parameters:

mean (tree-like structure with arithmetics) – Mean of the log-normal distribution.
std (tree-like structure with arithmetics) – Standard deviation of the log-normal distribution.
name (hashable, optional) – Name within the new input on which call acts.
shape (tuple or tree-like structure of ShapeWithDtype) – Shape of the latent parameter(s) that are transformed to the desired distribution. This can also be an arbitrary shape-dtype structure in which case dtype is ignored. Defaults to a scalar.
dtype (dtype) – Data type of the latent parameter(s) that are transformed to the desired distribution.

class NormalPrior(mean, std, **kwargs)[source]#

__init__(mean, std, **kwargs)[source]#

Transforms standard normally distributed random variables to a normal distribution.

Parameters:

mean (tree-like structure with arithmetics) – Mean of the normal distribution.
std (tree-like structure with arithmetics) – Standard deviation of the normal distribution.
name (hashable, optional) – Name within the new input on which call acts.
shape (tuple or tree-like structure of ShapeWithDtype) – Shape of the latent parameter(s) that are transformed to the desired distribution. This can also be an arbitrary shape-dtype structure in which case dtype is ignored. Defaults to a scalar.
dtype (dtype) – Data type of the latent parameter(s) that are transformed to the desired distribution.

class UniformPrior(a_min, a_max, **kwargs)[source]#

__init__(a_min, a_max, **kwargs)[source]#

Transforms standard normally distributed random variables to a uniform distribution.

Parameters:

a_min (tree-like structure with arithmetics) – Minimum value.
a_max (tree-like structure with arithmetics) – Maximum value.
name (hashable, optional) – Name within the new input on which call acts.
shape (tuple or tree-like structure of ShapeWithDtype) – Shape of the latent parameter(s) that are transformed to the desired distribution. This can also be an arbitrary shape-dtype structure in which case dtype is ignored. Defaults to a scalar.
dtype (dtype) – Data type of the latent parameter(s) that are transformed to the desired distribution.