Reply to “Comment on ‘Algorithm for normal random numbers’ ”

We have recently proposed [Phys. Rev. E 60, 3361 (1999)] an algorithm for the generation of normal pseudorandom numbers that is nearly 10 times faster than the Box-Muller algorithm. A flaw in this algorithm is pointed out in the preceding Comment [Probert, preceding Comment, Phys. Rev. E 63, 058701 (2001)]. It turns out that significant correlations show up in strings of pseudorandom numbers generated by our algorithm if such strings are sufficiently long. A slightly modified algorithm that is free from this defect is proposed.