Re: 32-bit random numbers
- Posted by "Juergen Luethje" <j.lue at gmx.de> Jul 05, 2004
- 549 views
Christian Cuvier wrote: >> *Concatenation* of two uniformly distributed 16-bit values -- as Derek >> and Rob suggested -- should IMHO result in a uniform distribution, too. >> >> Rob said, that *addition* of two uniformly distributed values does >> result in a distribution, that is not uniform. >> >> Regards, >> Juergen > > Rob also said that, in theory, concatenation would be nice. The problem is > concatenating adjacent random numbers, because the random generator is > actually deterministic. For this reason, only 2^16 among all possible 2^32 > integers will be generated. Maybe not all possible 2^32 integers will be generated, but why exactly 2^16? Also, when using the fancy code in combination with set_rand() that Derek posted, no *adjacent* random numbers are concatenated. My text that you quoted above, was a reply to Pete, when he wrote: "As I think was said, I would also have noted that #00010001, #00020002, etc will be less likely." Sorry, I still don't see the reason why e.g. #00010001 would be less likely than any other value, when concatenating 2 uniformly distributed 16-bit random numbers. <snip> Regards, Juergen