A random number generation technique that retains the unused equally likely outcomes after a rejected trial in rejection sampling. Instead of discarding those outcomes, it uses their conditional probability to obtain a new uniform distribution on a finite set for the input state of the next trial.
New to topics? Read the docs here!