PTMCoef

class liesel_ptm.PTMCoef(scale, penalty, inference=None, name='', scale_penalty=True, diagonalize_penalty=True, Z=None, role='transformation_coef', noncentered=False)

Bases: Var

Coefficient for PTM transformation splines.

This class wraps latent spline coefficients with penalty handling and provides several factory constructors for common penalty/prior setups.

Parameters:

• scale (Var) – A scale variable used for the coefficient prior.

• penalty (Any) – Penalty matrix applied to the spline coefficients.

Z

Reparameterization matrix applied to latent coefficients.

penalty

The (possibly scaled) penalty matrix used for priors.

latent_coef

Underlying latent coefficient variable.

nparam

Number of spline parameters (after transformations).

Methods

new_ridge	Create a ridge-penalized coefficient set.
new_rw1_fromzero	Create RW1 coefficients anchored at zero.
new_rw1_fromzero_ig	Create from-zero RW1 coefficients with an inverse-gamma prior on the variance parameter.
new_rw1_fromzero_wb	Create RW1 sum-to-zero coefficients with a Weibull prior on the random walk variance.
new_rw1_sumzero	Create RW1 (first-order random walk) coefficients with sum-to-zero.
new_rw1_sumzero_ig	Create RW1 sum-to-zero coefficients with an inverse-gamma prior on the random walk variance.
new_rw1_sumzero_wb	Create RW1 sum-to-zero coefficients with a Weibull prior on the random walk variance.

Attributes