A method of testing a randomizer having at least one source of probabilistic data and a device for mixing the probabilistic data to form random data, where the probabilistic data includes at least two states, by receiving a probability of occurrence of each of the at least two states in the probabilistic data; squaring each probability of occurrence; generating a row vector a of the results of the last step; receiving a probability of transition between each of the at least two states in the probabilistic data; squaring the transition probabilities; generating a matrix B of the results of the last step; generating a column vector c having entries equal in number to the at least two states in the probabilistic data, where each entry is 1; computing S(P)=a(B (L-1))c for each at least one source of probabilistic data, where L is a number of outputs of probabilistic data processed by the randomizer to produce m bits of output; if the at least one source of probabilistic data is greater than 1 then generating a product of the results of the last step and designating the product as S(P); computing S(Q)=(1/2 m)[1+(2 m-1)S(P)]; and declaring the randomizer acceptable if S(Q) is below a user-definable value, otherwise declaring the randomizer unacceptable.Method of testing a randomizer679888328/09/200418/09/200020041,471Buller; Darryl M.US Patent and Trademark OfficeGoogle Patent Searchpatentimages.storage.googleapis.com/US6798883B1/US06798883-20040928-D00000.pngUnique