Example 6.3.1.
Take \(f(t) = e^t\) and \(g(t) = t\) for \(t \geq 0\text{.}\) Then
\begin{equation*}
(f*g)(t)
=
\int_0^t e^\tau (t-\tau) \, d\tau
=
e^t - t - 1 .
\end{equation*}
To solve the integral we did one integration by parts.
Section 6.3 Convolution
Subsection 6.3.1 The convolution
The Laplace transformation of a product is not the product of the transforms. All hope is not lost, however. We simply have to use a different type of a “product.” Take two functions \(f(t)\) and \(g(t)\) defined for \(t \geq 0\text{,}\) and define the convolution of \(f(t)\) and \(g(t)\) as
1
For those that have seen convolution before, you may have seen it defined as \((f * g)(t) =
\int_{-\infty}^\infty f(\tau) g(t-\tau) \, d\tau\text{.}\) This definition agrees with (6.2) if you define \(f(t)\) and \(g(t)\) to be zero for \(t < 0\text{.}\) When discussing the Laplace transform, the definition we gave is sufficient. Convolution does occur in many other applications, however, where you may have to use the more general definition with infinities.
\begin{equation}
\mybxbg{~~
(f * g)(t) \overset{\text{def}}{=}
\int_0^t f(\tau) g(t-\tau) \, d\tau .
~~}\tag{6.2}
\end{equation}
As you can see, the convolution of two functions of \(t\) is another function of \(t\text{.}\)
Example 6.3.2.
Take \(f(t) = \sin (\omega t)\) and \(g(t) = \cos (\omega t)\) for \(t \geq 0\text{.}\) Then
\begin{equation*}
(f*g)(t)
=
\int_0^t \sin ( \omega \tau ) \,
\cos \bigl( \omega (t-\tau) \bigr) \, d\tau .
\end{equation*}
Apply the identity
\begin{equation*}
\cos (\theta) \sin (\psi) =
\frac{1}{2} \, \bigl( \sin (\theta + \psi) - \sin (\theta - \psi) \bigr) ,
\end{equation*}
to get
\begin{equation*}
\begin{split}
(f*g)(t)
& =
\int_0^t
\frac{1}{2} \, \bigl( \sin (\omega t) - \sin (\omega t - 2 \omega \tau
) \bigr) \, d\tau
\\
& =
\left[ \frac{1}{2} \, \tau \sin (\omega t) - \frac{1}{4\omega} \, \cos
(\omega t - 2 \omega \tau) \right]_{\tau=0}^t
\\
& = \frac{1}{2} \, t \sin (\omega t) .
\end{split}
\end{equation*}
The formula holds only for \(t \geq 0\text{.}\) The functions \(f\text{,}\) \(g\text{,}\) and \(f*g\) are undefined for \(t < 0\text{.}\)
Convolution has many properties that make it behave like a product. Let \(c\) be a constant and \(f\text{,}\) \(g\text{,}\) and \(h\) be functions. It is a calculus exercise to verify that
\begin{equation*}
\begin{aligned}
& f * g = g * f , \\
& (c f) * g = f * (c g) = c (f*g) , \\
& (f+g) * h = f * h + g * h , \\
& ( f * g ) * h = f * ( g * h ) .
\end{aligned}
\end{equation*}
The most interesting property for us is the following theorem.
Theorem 6.3.1.
If \(f(t)\) and \(g(t)\) are of exponential order, then so is \((f*g)(t)\) and
\begin{equation*}
\mybxbg{~~
\mathcal{L} \bigl\{ (f*g)(t) \bigr\}
=
\mathcal{L} \left\{ \int_0^t f(\tau) g(t-\tau) \, d\tau \right\}
=
\mathcal{L} \bigl\{ f(t) \bigr\} \mathcal{L} \bigl\{ g(t) \bigr\} .
~~}
\end{equation*}
In other words, the Laplace transform of a convolution is the product of the Laplace transforms: \({\mathcal{L}} \bigl\{ (f*g)(t) \bigr\} = F(s)G(s)\text{,}\) or in reverse, \({\mathcal{L}}^{-1} \bigl\{ F(s) G(s) \bigr\} = (f*g)(t)\text{.}\)
Example 6.3.3.
Suppose we wish to find the inverse Laplace transform of
\begin{equation*}
\frac{1}{s^2(s+1)} =
\frac{1}{s^2}
\,
\frac{1}{s+1}
.
\end{equation*}
We recognize the two entries of Table 6.2:
\begin{equation*}
\mathcal{L}^{-1}
\left\{
\frac{1}{s^2} \right\}
= t
\qquad \text{and} \qquad
\mathcal{L}^{-1}
\left\{
\frac{1}{s+1} \right\}
= e^{-t}
.
\end{equation*}
Therefore, we convolve \(t\) and \(e^{-t}\text{,}\)
\begin{equation*}
\mathcal{L}^{-1}
\left\{
\frac{1}{s^2}
\,
\frac{1}{s+1}
\right\}
=
\int_0^t
\tau e^{-(t-\tau)} \,d\tau
=
e^{-t}+t-1 .
\end{equation*}
The calculation of the integral involved an integration by parts.
Subsection 6.3.2 Solving ODEs
The next example demonstrates the full power of the convolution and the Laplace transform. We can give the solution to the forced oscillation problem for any forcing function as a definite integral.
Example 6.3.4.
Find the solution to
\begin{equation*}
x'' + \omega_0^2 x = f(t) , \quad x(0) = 0, \quad x'(0) = 0 ,
\end{equation*}
for an arbitrary function \(f(t)\text{.}\)
We first apply the Laplace transform to the equation. Denote the transform of \(x(t)\) by \(X(s)\) and the transform of \(f(t)\) by \(F(s)\) as usual. We get
\begin{equation*}
s^2 X(s) + \omega_0^2 X(s) = F(s) ,
\end{equation*}
or in other words,
\begin{equation*}
X(s) = \frac{1}{s^2+ \omega_0^2} F(s).
\end{equation*}
Recall that \(H(s) = \frac{1}{s^2+ \omega_0^2}\) is the transfer function. We know
\begin{equation*}
{\mathcal{L}}^{-1} \left\{
\frac{1}{s^2+ \omega_0^2}
\right\} =
\frac{\sin (\omega_0 t)}{\omega_0} .
\end{equation*}
Therefore,
\begin{equation*}
x(t) =
\int_0^t
\frac{\sin (\omega_0 \tau)}{\omega_0}
f(t-\tau) \, d\tau ,
\end{equation*}
or, if we reverse the order,
\begin{equation*}
x(t) =
\int_0^t
f(\tau)
\frac{\sin \bigl( \omega_0 (t-\tau) \bigr)}{\omega_0} \, d\tau .
\end{equation*}
Notice one more feature of the example above. We can now see how the Laplace transform handles resonance. Suppose that \(f(t) =
\cos (\omega_0 t)\text{.}\) Then
\begin{equation*}
x(t) =
\int_0^t
\frac{\sin (\omega_0 \tau)}{\omega_0} \,
\cos \bigl( \omega_0 (t-\tau) \bigr) \, d\tau
=
\frac{1}{\omega_0}
\int_0^t
\sin ( \omega_0 \tau ) \,
\cos \bigl(\omega_0 (t-\tau) \bigr) \, d\tau .
\end{equation*}
We have computed the convolution of sine and cosine in Example 6.3.2. Hence
\begin{equation*}
x(t) =
\left(
\frac{1}{\omega_0}
\right) \,
\left(
\frac{1}{2} \,
t
\sin ( \omega_0 t )
\right)
=
\frac{1}{2 \omega_0} \,
t
\sin ( \omega_0 t ).
\end{equation*}
Note the \(t\) in front of the sine. The solution, therefore, grows without bound as \(t\) gets large, meaning we get resonance.
The general idea here is that if \(H(s)\) is the transfer function, then \(X(s)=H(s)F(s)\text{.}\) If we find the \(h(t) = {\mathcal{L}}^{-1}\bigl\{ H(s) \bigr\}\text{,}\) then
\begin{equation*}
x(t)
= {\mathcal{L}}^{-1}\bigl\{ X(s) \bigr\}
= {\mathcal{L}}^{-1}\bigl\{ F(s)H(s) \bigr\}
= (f * h)(t)
= \int_0^t f(\tau) h(t-\tau) \, d\tau .
\end{equation*}
Hence, we can solve any constant-coefficient equation with an arbitrary forcing function \(f(t)\) as a definite integral using convolution. A definite integral, rather than a closed form solution, is usually enough for most practical purposes. It is not hard to numerically evaluate a definite integral.
Subsection 6.3.3 Volterra integral equation
A common integral equation is the Volterra integral equation
2
Named for the Italian mathematician Vito Volterra (1860–1940).
\begin{equation*}
x(t) = f(t) + \int_0^t g(t-\tau) x(\tau) \, d\tau ,
\end{equation*}
where \(f(t)\) and \(g(t)\) are known functions and \(x(t)\) is an unknown we wish to solve for. To find \(x(t)\text{,}\) we apply the Laplace transform to the equation to obtain
\begin{equation*}
X(s) = F(s) + G(s) X(s) ,
\end{equation*}
where \(X(s)\text{,}\) \(F(s)\text{,}\) and \(G(s)\) are the Laplace transforms of \(x(t)\text{,}\) \(f(t)\text{,}\) and \(g(t)\) respectively. We find
\begin{equation*}
X(s) = \frac{F(s)}{1-G(s)} .
\end{equation*}
To find \(x(t)\text{,}\) we now need to find the inverse Laplace transform of \(X(s)\text{.}\)
Example 6.3.5.
We apply the Laplace transform to obtain
\begin{equation*}
X(s) = \frac{1}{s+1} + \frac{1}{s^2-1} X(s) ,
\end{equation*}
or
\begin{equation*}
X(s) = \frac{\frac{1}{s+1}}{1- \frac{1}{s^2-1}}
=
\frac{s-1}{s^2 - 2}
=
\frac{s}{s^2 - 2}
-
\frac{1}{s^2 - 2} .
\end{equation*}
It is not hard to apply Table 6.1 to find
\begin{equation*}
x(t) = \cosh \bigl( \sqrt{2} \, t \bigr) -
\frac{1}{\sqrt{2}} \sinh \bigl( \sqrt{2}\, t \bigr).
\end{equation*}
Exercises 6.3.4 Exercises
6.3.1.
Let \(f(t) = t^2\) for \(t \geq 0\text{,}\) and \(g(t) = u(t-1)\text{.}\) Compute \(f * g\text{.}\)
6.3.2.
Let \(f(t) = t\) for \(t \geq 0\text{,}\) and \(g(t) = \sin t \) for \(t \geq 0\text{.}\) Compute \(f * g\text{.}\)
6.3.3.
Find the solution to
\begin{equation*}
m x'' + c x' + k x = f(t) , \quad x(0) = 0, \quad x'(0) = 0 ,
\end{equation*}
for an arbitrary function \(f(t)\text{,}\) where \(m > 0\text{,}\) \(c > 0\text{,}\) \(k > 0\text{,}\) and \(c^2 - 4km > 0\) (the system is overdamped). Write the solution as a definite integral.
6.3.4.
Find the solution to
\begin{equation*}
m x'' + c x' + k x = f(t) , \quad x(0) = 0, \quad x'(0) = 0 ,
\end{equation*}
for an arbitrary function \(f(t)\text{,}\) where \(m > 0\text{,}\) \(c > 0\text{,}\) \(k > 0\text{,}\) and \(c^2 - 4km < 0\) (the system is underdamped). Write the solution as a definite integral.
6.3.5.
Find the solution to
\begin{equation*}
m x'' + c x' + k x = f(t) , \quad x(0) = 0, \quad x'(0) = 0 ,
\end{equation*}
for an arbitrary function \(f(t)\text{,}\) where \(m > 0\text{,}\) \(c > 0\text{,}\) \(k > 0\text{,}\) and \(c^2 = 4km\) (the system is critically damped). Write the solution as a definite integral.
6.3.6.
6.3.7.
6.3.8.
Compute \({\mathcal{L}}^{-1} \left\{ \frac{s}{{(s^2+4)}^2} \right\}\) using convolution.
6.3.9.
Write down the solution to \(x''-2x=e^{-t^2}\text{,}\) \(x(0)=0\text{,}\) \(x'(0)=0\) as a definite integral. Hint: Do not try to compute the Laplace transform of \(e^{-t^2}\text{.}\)
6.3.101.
Let \(f(t) = \cos t\) for \(t \geq 0\text{,}\) and \(g(t) = e^{-t}\text{.}\) Compute \(f * g\text{.}\)
6.3.102.
Compute \({\mathcal{L}}^{-1} \left\{ \frac{5}{s^4+s^2} \right\}\) using convolution.
6.3.103.
6.3.104.
Solve \(x'''+x' = f(t)\text{,}\) \(x(0) = 0\text{,}\) \(x'(0)=0\text{,}\) \(x''(0)=0\) using convolution. Write the result as a definite integral.
