random_walk function

Metropolis sampling of a discrete distribution

Implements Metropolis sampling for an arbitrary discrete probability distribution

random_walk(pd, start, num_steps)

Arguments

• pd: function containing discrete probability function on the integers 1, 2, ...
• start: starting value of algorithm
• num_steps: number of iterations of algorithm

Returns

A vector of simulated values

Author(s)

Jim Albert

Examples

# random walk through a binomial distribution
pd <- function(x){
  dbinom(x, size = 10, prob = 0.5)
}
start <- 4
num_steps <- 50
out <- random_walk(pd, start, num_steps)