mkPenaltyFun: Make penalty function

In stevencarlislewalker/lme4ord: Generalized Linear Mixed Models with Flexible (Co)Variance Structures

Description

Make penalty function

Usage

mkPenaltyFun(penCovar, penFixef, penLoads, env = environment())

mkPenLpNorm(p = 2, lambda = 1)

mkPenPareto(alpha = 3, beta = 1)

Arguments

penCovar,penFixef,penLoads	Penalty functions for covariance, fixed effects, and loadings parameters. Can be NULL.
env	Environment in which to evaluate the resulting function.
p	exponent of the L-p norm
lambda	multiplier for the penalty term
alpha,beta	parameters of the generalized double Pareto distribution (defaults from Murray et al 2013, JASA)

Value

A function with two arguments:

pars	Parameter vector
parInds	Named list of indices pointing to the elements of pars corresponding to one of the three parameter types. The names of this list must include covar, fixef, and/or loads if the correponding type has a supplied (i.e. non-null) penalty function.

Examples

penCovar <- function(x) x^2
penFixef <- NULL
penLoads <- function(gg) abs(gg) + 10
mkPenaltyFun(penCovar, penFixef, penLoads)(1:10, list(covar = 1:5, loads = 6:10))

stevencarlislewalker/lme4ord documentation built on May 30, 2019, 4:43 p.m.