wessma@uofm-cv.UUCP (Prof. Wayne Stark) (10/29/84)
What is the difference between using floating point representation of the samples and first passing the samples through a nonlinear device and then using fixed point. I think there is no difference if the nonlinear device is suitable chosen. To minimize the representation error (however that is defined) either a optimum nolinear device and then a uniform (fixed point) representation or a nonuniform (e.g. floating point) representation (chosen optimally) will give the same performance.