Generiranje pseudoslučajnih brojeva i testovi slučajnosti

In modern times, algorithms for generating random numbers are becoming the main source of random numbers. However, they are deterministic and periodic. An algorithm must pass strict tests of randomness in order to be acceptable for cryptographic and similar purposes, as unpredictability of generated numbers is the main requirement. There is no such algorithm that can pass all tests of randomness, but passing a number of tests can boost our faith in a certain algorithm. Also, if an algorithm has passed all tests of randomness, that does not mean it is flawless. Only random number generators that use some physical phenomenon can generate true random numbers. But, those generators are so slow that they cannot be used for modern purposes. In practice, hybrid approach has shown the best results, where the seed for a pseudorandom number generator is determined with a true random number generator.