pymc.distributions.transforms.ZeroSumTransform#

class pymc.distributions.transforms.ZeroSumTransform(zerosum_axes)[source]#

Constrains any random samples to sum to zero along the user-provided zerosum_axes.

Parameters:
zerosum_axeslist of ints

Must be a list of integers (positive or negative).

Methods

ZeroSumTransform.__init__(zerosum_axes)

ZeroSumTransform.backward(value, *rv_inputs)

Invert the transformation.

ZeroSumTransform.extend_axis(array, axis)

ZeroSumTransform.extend_axis_rev(array, axis)

ZeroSumTransform.forward(value, *rv_inputs)

Apply the transformation.

ZeroSumTransform.log_jac_det(value, *rv_inputs)

Construct the log of the absolute value of the Jacobian determinant.

Attributes

name

ndim_supp