Evaluating TreeAge’s Pseudo Random Number Generator and Implementation of Statistical Distributions

To establish if the uniform number generator implemented in the TreeAge software delivers the expected outcome we investigated the probabilistic and statistical properties of pseudo uniform random numbers generated with TreeAge in several ways. We computed the first four empirical central and noncentral moments of a uniform sequence generated with TreeAge and compared them to their theoretical counterparts. We also conducted a global goodness of fit test known as the KolmogorovSmirnov test (K-S test) 2 and looked at a Quantile-Quantile plot (Q-Q plot) to inspect deviation from the expected outcome graphically. Finally, we employed a whole battery of tests designed to detect deviations from the expected outcome, known as the Die Hard battery of tests . The collection of 17 different tests we used checks whether the generated uniform sequence from TreeAge has the probabilistic and statistical properties expected of such a sequence.