Good day,

There is a pseudo-random number generator written in c ++ from the range (0.1) Someone may have worked with diehard tests or with NIST STS, the question is whether they work with rational numbers or only with natural types like int. When converting a file with data into a binary form, the output file becomes 0 bytes. With int data, everything is fine, from which I concluded that the double tests do not work. The documentation for the answer to this question did not find.
As in this case, I did not find testing a large array of random numbers from the range (0.1).

  • It seems that yes, only for integers. But is this a problem? I see two solutions: - multiply all numbers by 10 to the necessary extent and discard the fractional part. In this case, you can test, first multiplying by 10, then by 100, and so on. If all the sequences are random, then probably the source also - an array where real numbers are stored, interpret simply as an array of bytes and apply algorithms to it. - KoVadim
  • Well, the main attempt to translate an array of data in ASCII into a binary form with the help of DieHard did not succeed, maybe you know some means that will help to represent ASCII with real numbers as a binary? - Stanis
  • I think that by transferring to ascii, you will degrade the quality of the random house. If you look closely at the binary representation of numbers, you will understand. - KoVadim
  • I apologize, it was sealed in my comment, my output is an ASCII file with a sequence of data in the form: 0.12 0.0053 0.03, etc. How can I overtake such a format in binary and what would diehard or NIST STS eat it? - Stanis

0