MIT-SCIENCE-Lectures-lecture_notes2.pdfVIP

9 2. FOURIER SERIES AND TIME-LIMITED FUNCTIONS 2. Fourier Series and Time-Limited Functions Suppose x (t) is periodic: x (t) = x (t + T ) (2.1) Define the complex Fourier coefficients as ? T/2 ? ? α 1 ?2πint n = T ?T/2 x (t) exp dt (2.2) T Then under very general conditions, one can represent x (t) in a Fourier Series: ? ? ? ∞ 2πint x (t) = α n exp . (2.3) T n=?∞ Exercise. Write x (t) as a Fourier cosine and sine series. The Parseval Theorem for Fourier series is ∞ ? T/2 ? 1 2 T ?T/2 x (t) 2 dt = |a n | , (2.4) n=?∞ and which follows immediately from the orthogonality of the complex exponentials over interval T. Exercise. Prove the Fourier Series versions of the shift, differentiation, scaling, and time- reversal theorems. Part of the utility of δ?functions is that they permit us to do a variety of calculations which are not classically permitted. Consider for example, the Fourier transform of a periodic function, e.g., any x (t) as in Eq. (2.3) ∞ ? ∞ ? n=?∞ n=?∞ ignoring all convergence issues. We thus have the nice result that a periodic function has a Fourier transform; it has the property of vanishing except precisely at the usual Fourier series frequencies where its value is a δ?function with amplitude equal to the complex Fourier series coefficient at that frequency. Suppose that instead, x (t) = 0, |t| ≥ T/2 (2.6) that is, x (t) is non-zero except in the finite interval ?T/2 ≤ t ≤ T/2 (this is called a “time- limited” function). The following elementary statement proves to be very useful. Write x (t) as a Fourier series in |t| T/2, and as zero elsewhere: ? ? ∞ α n exp (2πint/T ) , |t| ≤ T/2 n=?∞ x (t) = (2.7) 0, |t| T/2 ? ∞ ?∞ ?x (s) = α n e (2πint/T ) e ?2πist dt = α n δ (s ? n/T ) , (2.5) 10 1. FREQUENCY DOMAIN FORMULATION where ? T/2 ? ? α 1 n = ?T/2 x (t) exp T 2πint ? dt, (2.8) T as though it were actually periodic. Thus as defined, x (t) corresponds to some different, per