[Nauty] Random doubles...

[Brendan McKay]

Yes, provided you are happy with 30 bits of mantissa.  You would
get slightly more uniformity (and no exact 0.0) with
   (2.0*NEXTRAN+1.0)/2147483648.0

However, I seem to recall that Knuth's web page has a version
of this generator using double throughout.

B.

* Gordon Royle <gordon at csse.uwa.edu.au> [050304 18:52]:
> Is rng.c suitable for generating random doubles in [0,1) using
> 
> 	(double)NEXTRAN/(1<<30);