In mixnet/arithm/JECPGroup.magic, the functions randomElement() and randomElementArray() do not behave as one would expect.

[pgaudry]

The interfaces of these functions take statDist, so that one would think that the output is uniform random, up to this statistical distance. But for a given x, the y=sqrt(x) is computed in a deterministic way, so that half of the points are never reached. Maybe I got something wrong, here.

But anyway, my understanding is that this function is only used for generating independent generators for Pedersen commitments, and for tests. Therefore, this should be harmless, but I would recommend to make the y pure random, to make things closer to what one would expect from the naming of the functions, and from the associated theory.