Performs iterative parameter estimation with adaptive step-size dampening
Internal function for fitting unconstrained GLM models using damped Newton-Raphson optimization technique.
damped_newton_r(
parameters,
loglikelihood,
gradient,
neghessian,
tol = 1e-07,
max_cnt = 64,
max_dmp_steps = 16
)Optimized parameter estimates after convergence or reaching iteration limit
Initial parameter vector to be optimized
Function computing log-likelihood for current parameters
Function computing parameter gradients
Function computing negative Hessian matrix
Numeric convergence tolerance (default 1e-7)
Maximum number of optimization iterations (default 64)
Maximum damping step attempts (default 16)
Implements a robust damped Newton-Raphson optimization algorithm.
- nr_iterate for parameter update computation