Re: Weakness of Feistel ciphers

Kim G. S. Øyhus wrote:
Sure, it is weak, and it does not need to be strong in my application,
which is similar to random number generation.
[...]
Anyway, I must use a cryptographic PRNG, since it must be invertible.

Hmm. I'm lost. How do I reconcile the above two statements?

If you don't need your generator to be cryptographically strong,
then just use a Mersenne Twister or something, as others have suggested.

If you do need your generator to be cryptographically strong, then
you could just use any existing cryptographically strong pseudorandom
generator (e.g., AES in counter mode). I would not recommend trying
to design your own pseudorandom generator, as it is easy to go wrong
and hard to come up with something better than current standards.

Note: By asking on sci.crypt, most people will implicitly assume that
you are looking for something that is cryptographically strong, unless
you state explicitly otherwise. sci.crypt is primarily focused on
discussions of cryptography and related topics.
.

Relevant Pages

  • Re: SHA-based subclass for random module
    ... The problem is that you want encryption. ... This sha prng module was motivated partly by your own remark in sf bug ... So I run it again using the crypto-based generator ... things usually aren't even thinking about cryptography. ...
    (comp.lang.python)
  • Re: http://www.cerberus-sys.com/~belleisl/infosec.html
    ... random number generator picks 4 random words from a list of 4096 words. ... stuff that had nothing to do with good or bad cryptography on a computer ... > Many others and I think the NSA knew about differential cryptanalysis long ... I am also sure differential cryptanalysis and more were ...
    (sci.crypt)
  • Re: Efficient pseudo-random number generation
    ... means a toy project or a "real" software project. ... neither rand nor Mersenne are suitable for cryptography. ... to be using something such as Blum Blum Shub for example. ... A Simple Unpredictable Pseudo-Random Number Generator. ...
    (comp.lang.cpp)
  • Re: Testing a Pseudo-Random Generator
    ... The general definition for a pseudorandom generator is that there does ... not exist a polynomial distinguisher, that is, an algorithm which runs ...
    (sci.stat.math)
  • [PATCH] rcu: Fix sign bug making rcu_random always return the same sequence
    ... get_random_bytes into the state of its pseudorandom generator. ... struct rcu_random_state { ... * Crude but fast random-number generator. ...
    (Linux-Kernel)