fn.attenuation: Computes attenuation factor

In jrlewi/brlm: Bayesian Restricted Likelihood Methods

Description

Compute the attenuation factor of the Jacobian in the restricted likelihood MCMC step

Usage

fn.attenuation(
  ystst,
  X,
  proj,
  l1obs,
  s1obs,
  fn.psi = fn.psi.huber,
  fn.chi = fn.chi.prop2
)

fn.attenuation2(
  ystst,
  X,
  proj,
  Qt,
  l1obs,
  s1obs,
  fn.psi = fn.psi.huber,
  fn.chi = fn.chi.prop2
)

Arguments

ystst	a data vector such that T(ystst)=T(y_obs)
X	design matrix
proj	the projection matrix onto the deviation space (aka the least squares residual space)
l1obs	s1obs: observed statsitics
Qt	Q transpose where Q is the orthonormalized X
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

Details

Two equivalent functions computing the attenuation piece. The second one uses the transpose of the orthonormalized X as an input. This makes it somewhat quicker.

Value

Attenuations of the transformation to ystst. These attenuations make up part of the Jacobian.

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