Abstract
It is easy to observe that standard pseudo-random generators of compilers of C, Pascal and many other most often used programming languages are not perfect and faithful. In particular, some of them are not equiprobable. This observation remains valid even for “professional” pseudo-random generators considered as very good (cf. [2]). In this paper we propose a method for improving the quality of pseudo-random generators. However, the proposed method will be efficacious if the pseudo-random generator satisfies the following postulate formulated in terms of probability theory. Namely, it is required that the results of realization of two (successive) assignments x:= random, y:= random, treated as random events, are independent.