There is given a set of n ordered pairs of time functions (ri(t), xi;(t)). A network N is to be designed, whose output is xi(t − τ) in response to input ri(t − τ). A solution is presented consisting of n parallel branches, each with its own unique sampling function followed by a linear, time invariant network. The inputs ri(i) must, however, be linearly independent. If so, N is linear with
... [Show full abstract] respect to linear combinations of the nri(t). A general formula is found for the sensitivity of each response function to variations in the input signal. A slight modification gives a more general network, applicable to inputs ri not linearly independent. It is shown how to design N for stochastic, stationary inputs so that it presents n desired statistical weighting functions to the n inputs.