[Bug-gsl] Possible error in GSL Reference Manual

[Toshikazu Aiyama, Ph.D.]

Thank you for your effort of collecting and documenting all routines.

I have read a part of GSL Reference manual version 1.15.  At page 195 somewhat 
a little bit lower than the middle of the page writes:
   If the generator is seeded with the same same value of s on two different 
runs, the same stream of random numbers will be generated by successive calls 
to the routines bellow.

If the state of generator is described by a single integer: a seed; then the 
above statement is true. The manual says that 'state' of size t describes the 
random number state.  For example, consider ranlxs description at page 200:
   The generator provides single precision output (24 bits) ... The period of 
the generator is about 10^171 ... Note that the range of allowed seeds for this 
generator is [0,2^31-1] ...

It means, in case of ranlxs:
1. The generated number is up to 2^21-1.
2. 2^21-1 < 2^31-1 < 10^171
3. Seed is 32 bit integer


Unless the state associated with ranlxs is represented by about 568 bits 
integer or bigger, the period described above is impossible.  If that's the 
case, then description at page 195 above, and that at page 200 contradicts.

Thank you for your attention to this inquiry.