rejection sampling

A technique used to generate observations from a distribution by repeatedly sampling an approximation function that bounds the distribution everywhere from above and rejecting the samples based on the quotient of the two functions at the sample's location.