You can usually get away with storing the log-factorials, and then just taking the exponent at the end. (If you don't have an arbitrary precision library handy

You can usually get away with storing the log-factorials, and then just taking the exponent at the end. (If you don't have an arbitrary precision library handy, or computational speed is an issue.)