library(mvtnorm)
library(MASS)

# primtiver ecdf-Sampler

# Beispiel: Zufallszahlen aus einer N(mu,sigma^2)
mu <- 5
sigma <- sqrt(100)

# setze "passenden" Grid

#stepsize
stepsize <- 0.0001
gridx <- seq(from=mu-6*sigma, to=mu+6*sigma, by=stepsize )

x.unnorm<- dnorm( gridx, mean=mu, sd=sigma )
x.norm <- x.unnorm/ ( stepsize*sum(x.unnorm) )

n<-10000
nzv <- sample( gridx, n, stepsize*x.norm, replace=T )
hist(nzv)

# jetzt normale Methode
nzv2 <- rnorm(n=n, mean=mu, sd=sigma)
hist(nzv2, add=T, border="red")

plot(ecdf(nzv), col.hor="blue", col.vert="blue", do.points=F)
plot(ecdf(nzv2), add=T, col.hor="red", col.vert="red", do.points=F)