pycmtensor.expressions#

PyCMTensor expressions module

Module Contents#

class pycmtensor.expressions.Param(name: str, value=None)[source]#

Bases: Expressions

Base class for expression objects

Constructor for model param object

property name[source]#

Returns the name of the object

property init_value[source]#
property shape[source]#
get_value()[source]#

Returns the numpy representation of the parameter value

reset_value()[source]#

Resets the value of the shared variable to the initial value

class pycmtensor.expressions.Beta(name, value=0.0, lb=None, ub=None, status=0)[source]#

Bases: Param

Base class for expression objects

Class object for Beta parameters

Parameters:

  • name (str) – name of the Beta class object

  • value (float) – initial starting value. Defaults to 0

  • lb (float) – lowerbound value. Defaults to None

  • ub (float) – upperbound value. Defaults to None

  • status (int) – whether to estimate (0) this Beta expression or not (1).

property status[source]#
class pycmtensor.expressions.Sigma(name, value=1.0, ub=None, status=0, dist='NORMAL')[source]#

Bases: Beta

Base class for expression objects

Class object for Beta parameters

Parameters:

  • name (str) – name of the Beta class object

  • value (float) – initial starting value. Defaults to 0

  • lb (float) – lowerbound value. Defaults to None

  • ub (float) – upperbound value. Defaults to None

  • status (int) – whether to estimate (0) this Beta expression or not (1).

property dist[source]#
class pycmtensor.expressions.Bias(name: str, size: tuple, init_value=None)[source]#

Bases: Param

Base class for expression objects

Constructor for model param object

class pycmtensor.expressions.Weight(name: str, size: tuple, init_type=None, init_value=None, rng=None)[source]#

Bases: Param

Base class for expression objects

Class object for Neural Network weights

Parameters:

  • name (str) – name of the parameter

  • size (tuple, list) – array size of the parameter

  • init_type (str) – initialization type, see notes

  • init_value (numpy.ndarray, optional) – initial values of the parameter

  • rng (numpy.random.Generator, optional) – random number generator

Note

Initialization types are one of the following:

  • ‘zeros’: a 2-D array of zeros

  • ‘he’: initialization method for neural networks that takes into account the non-linearity of activation functions, e.g. ReLU or Softplus [1]

  • ‘glorot’: initialization method that maintains the variance for symmetric activation functions, e.g. sigm, tanh [2]

Hint

Initialization of Weights:

w = expressions.Weight(name="w_1", size=(3, 10), init_type="he")
property init_type[source]#