The question is about the document: NIST SP 800-22, which can be obtained, for example, here
The issue is: in the linear-complexity test it is assumed that the expected probabilities of classes are: [0.010417, 0.03125, 0.125, 0.5, 0.25, 0.0625, 0.20833]. You can easily verify that the sum of probabilities equals 1, which is OK.
But all of the implementations I have found use another vector of expected probabilities, which is: [0.01047, 0.03125, 0.125, 0.5, 0.25, 0.0625, 0.20833].
The sum of the coordinates of this vector is definitely not equal 1, which means that this is a replicated typo.
NIST provides several examples of results that are expected (for instance, for e, pi, sha-1 sequence and two others). It can be checked that if you want to obtain the same results, the vector should be with this typo. Is it possible, that there is a mistake in this standard or this is a mistake in all available implementations?
Aucun commentaire:
Enregistrer un commentaire