FWIW that algorithm for uniform floating-point sampling in [0,1] is actually originally due to: Walker, Alastair J. “Fast generation of uniformly distributed pseudorandom numbers with floating-point representation.” Electronics Letters 10 (1974): 533-534.
