Re: Rand implementation question

[mpender]

I realized after writing this that I phrased it poorly:


mpender wrote
> I understand why it is a bad idea to take successive bits from a PRNG and
> assume statistical independence, after all successive bits have a linear
> relationship to each other.

In general, PRNGs do not necessarily have a linear relationship between
successive bits.  However, the old Matlab random number generator looks like
some form of linear feedback shift register--and those do have a linear
relationship between successive bits.  Not only are they inappropriate for
cryptography work, but they are often inappropriate for use in
communications because of correlated higher even order statistics such as
the power spectral density.



--
View this message in context: 
http://octave.1599824.n4.nabble.com/Re-Rand-implementation-question-tp4662472p4662536.html
Sent from the Octave - General mailing list archive at Nabble.com.