Re: diehard and ent results quesion

From: Mok-Kong Shen (mok-kong.shen@t-online.de)
Date: 02/21/03 

From: Mok-Kong Shen <mok-kong.shen@t-online.de>
Date: Fri, 21 Feb 2003 01:05:08 +0100

"Douglas A. Gwyn" wrote:
>
> Cristiano wrote:
> > I'll do a web search for it. But do you know a link for that statistic?
> > Or can you give me a simple explanation?
>
> If you can locate the source code that I published long ago,
> which was included on Bruce Schneier's Applied Cryptography
> disks under the name "i-hat", it includes Unix manual pages
> (which are ASCII text files meant to be formatted but which
> can be read with only moderate difficulty in themselves).
> One link I found quickly through a Web search is:
> http://tirnanog.ls.fi.upm.es/Servicios/Software/ap_crypt/disk1/i-hat.zip
> I also posted formatted plain-ASCII versions of the manual
> pages to this newsgroup a couple of months ago in response to
> a long thread in which MKS kept going on about the absence of
> proof-by-testing for randomness.

I like to cite what Olson posted previously:

   Here, we're given the stream and asked if it's random.
   We may be able to find a statistic that lets us confidently
   reject the hypothesis that it's random, but no matter how
   many tests we run, we can never reject that it's non-random.

In my view, one could under circumstances be very
'confident' that one's source is of such a good quality
that the opponent can't discern any deviations from
'randomness', but that's certainly different from a
'proof' of randomness in the rigorous sensce of the word,
if I don't err.

M. K. Shen

Relevant Pages

  • Re: Is the question answered?
    ... I predict that the result of throwing a pair of dice is random. ... to be 95% or 99% or whatever confident that the result was random. ... If you remembered more statistics, or just thought about it a little, ... to attain a preset level of confidence in the hypothesis of randomness. ...
    (talk.origins)
  • Re: Test for randomness?
    ... Are you saying that A) randomness is not a problem at all (ie, ... tails" is the only real problem OR ... ... I am NOT saying you can be confident that you have a random sample. ...
    (sci.stat.consult)
  • Re: Circuit simulation software
    ... are confident will work? ... new is for that newness to be achieved by randomness. ... the predictability of its behavior can be evaluated ... Randomness is not a permanent pollutant. ...
    (sci.electronics.design)
Loading