log_compdens_conv_mix: Find log-component density convolution when error...

In dcgerard/stramash: STRAMASH: Specifying The Right Amount of Mixture-likelihood for Adaptive SHrinkage

Description

This function will find the log of the coefficients of the prior mixing proportions in the ASH-likelihood when the model is betahat = g + errordist, where g and errordist are both mixtures of either normals or uniforms.

Usage

log_compdens_conv_mix(g, betahat, errordist)

Arguments

g	A mixture density. Either of class normalmix or of class unimix. This is the prior.
betahat	A vector of numerics. The locations at which to evalutate the density.
errordist	A list of objects of either class normalmix or unimix. The length of this list must be the length of betahat. errordist[[i]] is the ith error distribution of betahat[i].

Details

There are four cases: (1) g and errordist are both of class normalmix, (2) g is of class normalmix and errordist is of class unifmix, (3) g is of class unifmix and errordist is of class normalmix, and (4) g and errordist are both of class unifmix. All of these are supported. Though (2) and (3) differ only in indexing.

Value

A matrix with row dimension length(betahat) and column dimension length(errordist).

Author(s)

David Gerard

dcgerard/stramash documentation built on May 15, 2019, 1:24 a.m.