Pseudo-Random Numbers

I have been using some programs that generate random numbers. i wanted

to know if there is a way to get the next numbers in the pseudo-random

sequence if the algorithm and seed are not known.and is there a

program that can take the numbers allready generated and guess the

next ones.



One thought on “Pseudo-Random Numbers

  1. This is one of those "a million monkeys at a million typewriters for a

    million years" type of questions. In theory, they might produce a

    copy of Hamlet. In practice, it'll never happen.

    Any decent pseudorandom number generator should have enough of a

    resemblence to true randomness as to make the type of prediction of

    you're asking about next to impossible. As noted on:

    http://webnz.com/robert/true_rng.html

    'Pseudo-random number generators…use a formula to generate numbers

    which behave very like genuine random numbers and are widely used for

    simulations of random processes and statistical methods. In most cases

    a good pseudo-random number generator seems to work as you would

    expect a genuine random generator to work."

    In other words, predicting the next number in the series shouldn't be

    an easy task, even for sophisticated software analyses. However,

    pseudorandom number generators are not perfect, as noted at another

    site on the topic: http://random.mat.sbg.ac.at/tests/

    "Every generator has its regularities which, ocassionally, may become

    deficiencies. Hence, in a given application, even reliable generators

    may fail."

    Here's where the million monkeys come in. If you could generate

    millions of numbers from your program, and subject them to some of the

    statistical tests that are freely available, then you might begin to

    zero on some of your program's irregularities. With enough

    information, you could probably, eventually, identify the actual

    algorithm in use (unless its proprietary) and then you'd have

    predictability from any given seed number. Some literature about

    tests, with links to the tests themselves, are at

    http://random.mat.sbg.ac.at/tests/.

    Bottom line: with millions or billions of numbers, tons of computing

    power, and lots of time, it might be possible to 'reverse engineer'

    your random number generator. Professional cryptographers sometimes

    do this sort of thing, but there's an awful lot of human ingenuity

    involved in this, not just a simple "plug the numbers into the

    software" type of thing. From a practical point of view, however, you

    won't be able to predict the next number (unless your random number

    generator is so poorly programmed as to be obviously predictable, even

    to the naked eye).

    Hope this answers the question to your satisfaction. Let me know if

    you need additional information.

Leave a Reply

Your email address will not be published. Required fields are marked *