Would I create "more" randomness if if I had the library random function selecting randomly a value from the logistic function,with initial values showing chaotic behaviour, which I then would use as a seed to the random function provided by the library? Phew, whole mouthfull I guess . . .
My guess is no, but my math skills are not that big to confirm this.
No, ultimately the sequence you get is still bound by the randomness of the library. But if that process produces a reasonably random seed for the library, it's a net win, though I'm not sure it's worth the effort over something simpler and probably faster like taking a hash of the current system time.