integral result shifted by random offset / random translation by constant "c"

The current implementation of the integral function produces a random offset / translation by a constant value. This is a legitimate behaviour, since indeed integration cannot determine an offset, all offsets are equally possible.
But it is not clear where this random offset comes from and it should be considered if an implementation is possible which gives a less random behaviour.
