fit_ksc_mixture: Fit K-Component Gaussian Mixture to Measurement Noise Density

In wARMASVp: Winsorized ARMA Estimation for Higher-Order Stochastic Volatility Models

Fit K-Component Gaussian Mixture to Measurement Noise Density

Description

Uses EM algorithm to approximate the measurement noise density with a Gaussian mixture. For Gaussian SV, returns the pre-computed KSC (1998) 7-component table.

Usage

fit_ksc_mixture(
  distribution = c("gaussian", "student_t", "ged"),
  nu = NULL,
  K = 7,
  n_sample = 10000,
  max_iter = 500,
  tol = 1e-08,
  seed = 42
)

Arguments

distribution	Character: "gaussian", "student_t", or "ged".
nu	Numeric. Shape parameter (ignored for Gaussian).
K	Integer. Number of mixture components. Default 7.
n_sample	Integer. Sample size for EM fitting. Default 10000.
max_iter	Integer. Maximum EM iterations. Default 500.
tol	Numeric. Convergence tolerance. Default 1e-8.
seed	Integer. Random seed. Default 42.

Value

List with weights, means, vars, KL_div.

wARMASVp documentation built on May 15, 2026, 5:07 p.m.