ebnv.pm: Solve EBNV problem

In stephenslab/ebmr.alpha: Empirical Bayes Multiple Regression

Solve EBNV problem

Description

Functions for solving the Empirical Bayes Normal Variances (EBNV) problem for various prior families

Usage

ebnv.pm(b, s2)

ebnv.exp(b, s2)

ebnv.np(
  b,
  s2,
  g.init,
  update.mixprop = c("em", "mixsqp", "none"),
  update.w = c("em", "none")
)

ebnv.np.em(b, s2, g.init)

ebnv.np.fixgrid(b, s2, g.init)

ebnv.exp_mix(
  b,
  s2,
  g.init,
  update.mixprop = c("em", "mixsqp", "none"),
  update.w = c("em", "none")
)

ebnv.exp_mix.em(b, s2, g.init)

ebnv.exp_mix.fixgrid(b, s2, g.init)

Arguments

b	vector of n observations
s2	a positive real number
g.init	(present only in some cases) an object with the same structure as the prior to be fit used to initialize the algorithm (and so, implicitly, to sometimes specify parameters like the grid to be used)
update.mixprop	string indicating how to estimate/update the mixture proportions; if "none" then mixture proportions are supplied by g$mixprop
update.w	string indicating how to update w parameters; ; if "none" then mixture proportions are supplied by g$w

Details

These functions fit the model

b_j~N(0,s2 w_j)

w_j ~ g()

by empirical Bayes. They estimate the prior g, and (the inverse of the) posterior mean for each 1/w_j.

Value

A list with elements

g	A list containing the details of the estimated prior
wbar	A vector containing inverses of posterior mean for 1/w
loglik	The log-likelihood p(b | s2, ghat)

Functions

• ebnv.pm: Solve EBNV problem with point mass prior

• ebnv.exp: Solve EBNV problem with exponential prior

• ebnv.np: Solve EBNV problem with non-parametric prior

• ebnv.np.em: Solve EBNV problem with non-parametric prior with EM update of both grid and mixture proportions

• ebnv.np.fixgrid: Solve EBNV problem with non-parametric prior with fixed grid

• ebnv.exp_mix: Solve EBNV problem with mixture of exponentials prior

• ebnv.exp_mix.em: Solve EBNV problem with mixture of exponentials prior with EM update of both grid and mixture proportions

• ebnv.exp_mix.fixgrid: Solve EBNV problem with mixture of exponentials prior with fixed grid

stephenslab/ebmr.alpha documentation built on March 30, 2022, 3:49 a.m.