log.prop.den: Caluclate the proposal dens on the log scale Computes...

In jrlewi/brlm: Bayesian Restricted Likelihood Methods

Description

Caluclate the proposal dens on the log scale Computes proposal density by computing the necessary Jacobian of the transformation from y* to y**

Usage

## S3 method for class 'prop.den'
log(y.prop, X, proj, l1obs, s1obs, fn.psi, fn.chi, n, p)

## S3 method for class 'prop.den2'
log(y.prop, X, proj, Qt, l1obs, s1obs, fn.psi, fn.chi, n, p)

Arguments

y.prop	proposed data vector satisfying T(y.prop)=T(y_obs)
X	design matrix
proj	the projection matrix onto the deviation space (aka the least squares residual space)
l1obs,	s1obs observed statsitics
fn.psi	custom psi function returning psi(x) or psi'(x) defining M-estimator for location ; default is fn.psi.huber
fn.chi	custom chi function returning chi(x) or chi'(x) defining M-estimator for scale; default is fn.chi.prop2
n	length(y_obs)=length(y.prop)
p	number of regression coefficients
Qt	Q transpose where Q is the orthonormalized X

Details

Designed for use within fn.one.rep.y. Two equivalent versions, the second uses fn.attenuation2 and is faster.

Value

The proposal density for y.prop on the log scale (assuming original vector sampled uniformly on unit sphere)

jrlewi/brlm documentation built on March 17, 2021, 1:10 a.m.