dlhyper {dlbayes}

Tune the hyperparameter in the prior distribtuion

Description

This function is to tune the value of hyperparameter in the prior, which can be [1/max(n,p),1/2]. We use the method proposed by Zhang et al. (2018). This method tune the hyperparameter by incorporating a prior on R^2. And they give a direct way to minimize KL directed divergence for special condition.

Usage

dlhyper(x, y)

Arguments

x

input matrix, each row is an observation vector, dimension n*p. Same as the argument in dlmain

y

Response variable, a n*1 vector. Same as the argument in dlmain

Value

hyper

A value that can use in the following posterior computation

Examples

p=50
n=6
#generate x
x=matrix(rnorm(n*p),nrow=n)
#generate beta
beta=c(rep(0,10),runif(n=5,min=-1,max=1),rep(0,10),runif(n=5,min=-1,max=1),rep(0,p-30))
#generate y
y=x%*%beta+rnorm(n)
hyper=dlhyper(x,y)