How do you calculate the standard deviation of a bounded probability distribution function?