No announcement yet.

Constants for Random Numbers Generator

  • Filter
  • Time
  • Show
Clear All
new posts

  • #21
    I agree with David. Statistically speaking, there is no such thing as an "ultimate final test" or proof of randomness.

    Then too, don't forget John von Neumann's famous injunction, which Donald Knuth quotes at the outset of his great chapter in vol 2 of "The art of computer programming" on random numbers: "Anyone who considers arithmetical methods of producing random digits is, of course in a state of sin."


    • #22
      > once you have an algorithm that tests without statistical flaws

      > once you've hit the ceiling, there's apparently no place higher to go.

      As in 'beyond reasonable doubt'? That had its boundary moved when forensic science started playing around with DNA testing.

      I believe that the inherent weakness of all prngs is that they are generated using continuous functions which do not prohibit things like infinities and singularities.

      The brakes are off if we use discrete functions and for an abundance of them we need to venture into the quantum environment.