6.3 The impulse response and Green’s function

Theorem 6.3.1. Let \(G\) be the solution of \(p(D)G = 0\) with \(G(0) = G'(0) = \dots = G^{(n-2)}(0) = 0\) and \(G^{(n-1)}(0) = 1\). For continuous \(g\), the solution of \(p(D)y = g\) with zero initial data at \(t = 0\) is \[ y(t) = \int _0^tG(t - s)g(s)\,ds = (G * g)(t) . \]

Proof. Differentiating under the integral sign once, \(y'(t) = G(0)g(t) + \int _0^tG'(t - s)g(s)\,ds\), and the first term vanishes if \(n \geq 2\). Repeating, for \(k \leq n - 1\) \[ y^{(k)}(t) = \int _0^tG^{(k)}(t - s)g(s)\,ds , \] because each differentiation produces a boundary term \(G^{(j)}(0)g(t)\) with \(j \leq n - 2\), which is zero; in particular \(y^{(k)}(0) = 0\). One more differentiation gives the boundary term \(G^{(n-1)}(0)g(t) = g(t)\): \[ y^{(n)}(t) = g(t) + \int _0^tG^{(n)}(t - s)g(s)\,ds . \] Then \(p(D)y = g(t) + \int _0^t\bigl (p(D)G\bigr )(t - s)g(s)\,ds = g(t)\). □

\(G\) is the impulse response: the response of the system at rest to a unit blow at time \(0\), which leaves the position and lower derivatives unchanged and gives the \((n - 1)\)st derivative a unit jump. The convolution formula superposes the responses to the impulses \(g(s)\,ds\) delivered at the earlier times \(s\), and \(G(t - s)\) is the Green’s function of the initial value problem. For the oscillator \(y'' + 2\zeta \omega _0y' + \omega _0^2y = g\) with \(0 < \zeta < 1\), \[ G(t) = \frac {e^{-\zeta \omega _0t}\sin \omega _dt}{\omega _d}, \qquad \omega _d = \omega _0\sqrt {1 - \zeta ^2}, \] a damped oscillation at the damped frequency \(\omega _d\). Figure 9 shows \(G\) and the response to a rectangular pulse.

t246810foimrerpsciupnlongses ge (re yss =hpadonGesde∗) Gg(t)

Figure 9: The oscillator \(y'' + 0.4y' + 4y = g(t)\) (\(\omega _0 = 2\), \(\zeta = 0.1\)) at rest. Dashed: the impulse response \(G\). Shaded: a unit pulse of force on \(1 \leq t \leq 3\), drawn at a quarter of its height. Solid: the response, the convolution \(G * g\). While the force acts, the oscillator rings about the new equilibrium \(\frac 14\); after it is removed, it rings about \(0\), with an amplitude that depends on the phase at which the pulse ended.

Questions on this section

Stuck on something here? Ask below and it stays attached to this topic.