LECTURE 21: CONVOLUTION MINGFENG ZHAO October 24, 2014 Definition 1. Let f (t) be a function on [0, ∞), then I. The Laplace transform of f , denoted by L[f ](s), is defined as: Z L[f ](s) = ∞ f (t)e−st dt, for all s > 0. 0 II. If F (s) = L[f ](s), the inverse Laplace transform of F , denoted by L−1 [F ](t), is defined as: L−1 [F ](t) = f (t), ∀for all t > 0. Proposition 1. There holds that I. Linearity: L[af (t) + bg(t)](s) = aL[f (t)](s) + bL[g(t)](s). That is, L−1 [aF (s) + bG(s)](t) = aL−1 [F (s)](t) + bL−1 [G(s)](t). II. First Shifting Property: L e−at f (t) (s) = L[f (t)](s + a). That is, L−1 [F (s + a)] (t) = e−at L−1 [F (s)](t). III. Transforms of derivatives: L[f 0 ](s) = sL[f ](s) − f (0) L[f 00 ](s) = s2 L[f ](s) − sf (0) − f 0 (0). 1 2 MINGFENG ZHAO IV. Second Shifting Property: L[ua (t)f (t − a)] = e−as L[f (t)](s), for all s > 0. That is, L−1 e−as G(s) (t) = ua (t)L−1 [G(s)](t − a). V. Transform of Integrals: t Z L f (τ ) dτ (s) = 0 L[f (t)](s) . s That is, L−1 Z t F (s) = L−1 [G(s)](τ ) dτ. s 0 VI. Transform of Convolution: L[(f ∗ g)(t)] = L[f (t)] · L[g(t)]. That is, L−1 [L[f (t)] · L[g(t)]] = f ∗ g(t) Definition 2. Let f (t) and g(t) be two functions on [0, ∞), the convolution of f and g is defined as: Z t (f ∗ g)(t) = f (τ )g(t − τ ) dτ. 0 Remark 1. The Property VI implies Property V: let g(t) = 1, then Z t (f ∗ g)(t) = f (τ ) du. 0 By the transform of convolution, we have Z t [{(t)](∫ ) L . f (τ ) (s) = L[f (t)](s) · L[1](s) = s 0 Example 1. Let f (t) = et and g(t) = t for t ≥ 0, find f ∗ g and L[f ∗ g]. By the definition of f ∗ g, we have Z (f ∗ g)(t) t f (τ )g(t − τ ) dτ = 0 Z t eτ (t − τ ) dτ = 0 Z = t τ Z e dτ − t 0 0 t τ eτ dτ LECTURE 21: CONVOLUTION = t t(et − 1) − τ eτ |0 + Z 3 t eτ dτ Use the integration by parts 0 = t(et − 1) − tet + et − 1 = et − t − 1. So we know that the convolution of f (t) = et and g(t) = t is: (f ∗ g)(t) = et − t − 1 . By looking up the table, we have L[f (t)] = L[et ] = 1 , s−1 and L[g(t)] = L[t] = 1 . s2 By the transform of convolution, we have L[f ∗ g] = L[f (t)] · L[g(t)] = 1 . s2 (s − 1) 1 . Example 2. Find L s2 (s + 1) By looking up the table, we have −1 −1 L 1 = t, s2 By the transform of convolution, then 1 −1 L = s2 (s + 1) = = −1 and L 1 = e−t . s+1 1 1 L · s2 s + 1 1 1 L−1 2 ∗ L−1 s s+1 Z t τ e−(t−τ ) dτ = e−t +t − 1. −1 0 Therefore, we have L−1 1 = e−t +t − 1 . s2 (s + 1) Volterra integral equation (NOT in Exams) The Volterra integral equation is: Z x(t) = f (t) + (x ∗ g)(t) = f (t) + t g(t − τ )x(τ ) dτ 0 4 MINGFENG ZHAO Let X(s) = L[x(t)], apply the Laplace transform on the both sides of the above equation, we get X(s) = L[f (t)] + L[x ∗ g]. By the transform of convolution, we get X(s) = L[f (t)] + X(s) · L[g(t)]. Then X(s) = L[f (t)] . 1 − L[g(t)] So x(t) = L−1 [X(s)] = L−1 L[f (t)] . 1 − L[g(t)] Dirac delta and impulse response Definition 3. For any continuous function f (t) on (−∞, ∞), we have Z ∞ δ(t)f (t) dt = f (0). −∞ Proposition 2. For any constant a > 0, we have L[δ(t − a)] = e−as . That is, L−1 [e−as ] = δ(t − a). Proof. By the definition of the Laplace transform, we have Z ∞ L[δ(t − a)] = e−st δ(t − a) dt 0 Z ∞ = 0 t−s(t +a) dt0 Let t0 = t − a −a = e−as , Since a > 0, by the definition of δ. Department of Mathematics, The University of British Columbia, Room 121, 1984 Mathematics Road, Vancouver, B.C. Canada V6T 1Z2 E-mail address: mingfeng@math.ubc.ca