damped_newton_r: Damped Newton-Raphson Parameter Optimization

In lgspline: Lagrangian Multiplier Smoothing Splines for Smooth Function Estimation

Damped Newton-Raphson Parameter Optimization

Description

Performs iterative parameter estimation with adaptive step-size damping

Internal function for fitting unconstrained model components using damped Newton-Raphson updates.

Usage

damped_newton_r(
  parameters,
  loglikelihood,
  gradient,
  neghessian,
  tol = 1e-07,
  max_cnt = 64,
  max_dmp_steps = 16
)

Arguments

parameters	Initial parameter vector to be optimized
loglikelihood	Function computing log-likelihood for current parameters
gradient	Function computing parameter gradients
neghessian	Function computing negative Hessian matrix
tol	Numeric convergence tolerance (default 1e-7)
max_cnt	Maximum number of optimization iterations (default 64)
max_dmp_steps	Maximum damping step attempts (default 16)

Details

Implements a robust damped Newton-Raphson optimization algorithm. The Newton direction is computed once per outer iteration and reused across damping half-steps.

Value

Final parameter vector returned at termination.

See Also

- nr_iterate for parameter update computation

lgspline documentation built on Aug. 5, 2026, 1:10 a.m.