R软件Normal variates using uniform.R 内有具体内容

replace the function calls dnorm(newPosition) and dnorm(position) with
dnormmix(newPosition) and dnormmix(position), where
dnormmix( ) is a function that you'll write that averages two dnorm( )s,
one with mean = -3 and sd = 1, the other with mean = 3 and sd = 1.

Here's the listing:

# Simulate standard normal random numbers using Gibbs and uniform

step <- function (radius) {
  return(runif(1,-radius,radius))
}

# Parameters of random walk
radius = 1
position = 0.45
length = 100000

# initialize
values = rep(0,length)

for (i in 1:length) {
  newPosition = position + step(radius)
  if ( dnorm(newPosition) >= dnorm(position) ) {
    position = newPosition
  } else {
    if ( runif(1) < dnorm(newPosition)/dnorm(position) ) {
      position = newPosition
    } # else stay at original position
  }
  values = position
}

hist(values)
ks.test(values, "pnorm")

最后得到的结果根据starting positions的不同有不同的直方图

应该怎么修改代码？已经安装了JAGS

没搞清楚你的目的  你是想干啥？

我表示你的代码好像有点问题     你的倒数第三行 应该是    values[i]= position （values应该是一个向量才对啊）
运行下这个试试：
step <- function (radius) {
  return(runif(1,-radius,radius))
}

# Parameters of random walk
radius = 1
position = 0.45
length = 100000

# initialize
values = rep(0,length)

for (i in 1:length) {
  newPosition = position + step(radius)
  if ( dnorm(newPosition) >= dnorm(position) ) {
    position = newPosition
  } else {
    if ( runif(1) < dnorm(newPosition)/dnorm(position) ) {
      position = newPosition
    } # else stay at original position
  }
  values[i]= position
}

hist(values,freq=F)
ks.test(values, "pnorm")
windows()
plot.ecdf(values)

改的话就改那个position的初始值就好了