LocScaleTransformationDist

LocScaleTransformationDist#

class liesel_ptm.LocScaleTransformationDist(coef, loc, scale, bspline, parametric_distribution=<class 'tensorflow_probability.substrates.jax.distributions.normal.Normal'>, validate_args=False, allow_nan_stats=True, name='LocScaleTransformationDist', centered=False, scaled=False, batched=True, reference_distribution=<tfp.distributions.Normal 'Normal' batch_shape=[] event_shape=[] dtype=float32>)[source]#

Bases: TransformationDist

Location–scale specialization of TransformationDist.

Uses a Normal parametric layer with location loc and scale scale, combined with a spline transformation and reference Normal(0, 1).

Parameters:

  • coef (Any) – Coefficients for the spline basis.

  • loc (Any) – Location parameter for the Normal layer.

  • scale (Any) – Scale parameter for the Normal layer.

  • bspline (PTMSpline | OnionSpline) – Spline object providing the transformation.

  • validate_args (bool) – Whether to validate input arguments. (default: False)

  • allow_nan_stats (bool) – Whether to allow NaN statistics. (default: True)

  • name (str) – Name of the distribution. (default: 'LocScaleTransformationDist')

  • centered (bool) – If True, the transformation is centered. (default: False)

  • scaled (bool) – If True, the transformation is scaled. (default: False)

  • batched (bool) – If True, use batched computations. (default: True)

Notes

Inherits public attributes from TransformationDist.

Methods

inverse_transformation_parametric

Invert the location–scale normalization.

transformation_and_logdet_parametric

Apply location–scale normalization and its log-determinant.

Attributes

Contents