Penalized log-spline density estimation (Rust backend)

Reimplementation of density_mpl_legacy with the Fisher-scoring loop and Romberg integrators executed in Rust. Same statistical model, same output shape; typically 10–50x faster on B-spline bases with nbasis in the usual range.

Usage

density_mpl_rust(x, WfdParobj, conv = 1e-04, iterlim = 20)

Arguments

x

Observation vector (length n) or two-column matrix (x, f).

WfdParobj

An fdPar object (or fd / basisfd, in which case the penalty is zero).

conv

Convergence tolerance on the objective. Default 1e-4.

iterlim

Maximum Fisher-scoring iterations. Default 20.

Value

A list with the same fields as density_mpl_legacy: Wfdobj, C, Flist, iternum, iterhist.