A Simple and Efficient Algorithm to Compute Tail Probabilities from Transforms

We present an algorithm to approximately compute the “tail” probability that a random variable exceeds a specified number, given only an expression for its transform. We also show that the problem is #P-hard (more difficult than NP-hard), suggesting that no efficient procedure can solve it exactly. Our method consists essentially of summing a power series, and thus is easy to perform and requires little memory. Furthermore, its computational effort is nearly linear in the reciprocal of a prespecified worst-case error bound.