The functions that we will consider represent digital signals. Thus the natural model to present them are sequences of real numbers.
In practice we only have a inite sequence of terms x1 , . . . , xk but to analyze it we may consider it ininite by padding it with ininite zeros before and after it. Sometimes, it is also convenient to consider complex valued sequences although they may have not a physical interpretation.
We are interested in transforming this signal, to eliminate noise from it for instance. The type of transformations that we are going to
consider are linear operators y = T (x), , i.e
T (λx + μz) = λT (x) + μT (z).
This operators are usually denoted in the engineering literature as Lin- ear Systems.
Another important property that allows us to use Fourier Analysis techniques is that we will only consider time invariant linear systems (LTI for short).
Definition 1. A system T is invariant if it commutes with the translations, i.e. T 。τm = τm 。T, where τm is a delay of the sequence: τm (x[n]) = x[n 一 m].
That is, in a LTI “if we delay the input, we delay the output” . If Tx = y then T (τm (x)) = τm (y).