samplr: Single and Two Variable Rejection Sampler

In bubj/samplr:

Description

This function implements both one and two variable rejection sampling for probability density functions which are bounded, but the support does not need to be bounded.

Usage

samplr(N, f, twod = FALSE)

Arguments

N	The number of samples returned by the rejection sampler.
f	The pdf that we are sampling from, input as a predefined function. For one variable, the input variable must be declared as x. For two variables, the parameter of the function must be a vector containing the variables x and y.
twod	Specification if the input pdf is two dimensional if TRUE. Default is FALSE, meaning that the input function is one dimensional.

Value

If using a single variable, the output is a vector containing the samples. If two variables, the output is a matrix with N rows and 2 columns containing the samples.

Examples

Single variable

f <- function(x) {
  ifelse(0 < x & x < 1, 2*x, 0)
}
hist(samplr(10000,f))

g <- function(x) {
  ifelse(0 < x & x < 2, 1/2*x, 0)
}
hist(samplr(10000,g))

h <- function(x) {
  ifelse(0 < x & x < 6.2832, 1/2/pi*(sin(x) + 1), 0)
}
hist(samplr(10000,h))

u <- function(x) {
  ifelse(x < 0, 1/2*exp(x), 1/2*exp(-x))
}
hist(samplr(10000,u))

Two dimensional (requires ggplot2)

v <- function(z) {
  x <- z[1]
  y <- z[2]
  ifelse(0 <= x & x <= 1 & 0 <= y & y <= 1, x + y, 0)
}
samps <- data.frame(samplr(10000, v, twod = TRUE))
colnames(samps) <- c("x","y")
ggplot(samps, aes(x, y)) +
  geom_density_2d()

w <- function(z) {
  x <- z[1]
  y <- z[2]
  ifelse(0 <= x & x <= 1 & 0 <= y & y <= 1 & 0 <= x + y & x + y <= 1, 24*x*y,0)
}
samps <- data.frame(samplr(10000, w, twod = TRUE))
colnames(samps) <- c("x","y")
ggplot(samps, aes(x, y)) +
  geom_density_2d()

l <- function(z) {
  x <- z[1]
  y <- z[2]
  ifelse(0 < x & 0 < y, exp(-x)*exp(-y), 0)
}
samps <- data.frame(samplr(10000, l, twod = TRUE))
colnames(samps) <- c("x","y")
ggplot(samps, aes(x, y)) +
  geom_density_2d()

bubj/samplr documentation built on May 28, 2019, 7:14 p.m.