Skip to main content

Complex Analysis: an Open Source Textbook

Section 11.5 The Laplace Transform

Subsection 11.5.1 From Fourier Transforms to Laplace Transforms

We have seen that certain real-valued functions \(f(t)\) have a Fourier transform and that the integral
\begin{equation*} g(\omega) = \int_{-\infty}^{\infty}f(t)e^{-i\omega t},dt \end{equation*}
defines the complex function \(g(\omega)\) of the real variable \(\omega\text{.}\) If we multiply the integrand \(f(t) e^{-i\omega t}\) by \(e^{-\sigma t}\text{,}\) we create a complex function \(G(\sigma +i\omega)\) of the complex variable \(\sigma +i\omega\text{:}\)
\begin{equation*} G(\sigma +i\omega) =\int\nolimits_{-\infty}^{\infty}f( t) e^{-\sigma t}e^{-i\omega t}dt=\int\nolimits_{-\infty}^{\infty }f(t) e^{-(\sigma +i\omega) t}dt\text{.} \end{equation*}
The function \(G(\sigma +i\omega)\) is called the two-sided Laplace transform of \(f(t)\text{,}\) and it exists when the Fourier transform of the function \(f(t) e^{-\sigma t}\) exists. From the Fourier transform theory, we can state that a sufficient condition for \(G(\sigma +iw)\) to exist is that
\begin{equation*} \int_{-\infty}^{\infty}|f(t)|e^{-\sigma t}\,dt \lt \infty \end{equation*}
exists. For a given function \(f(t)\text{,}\) this integral is finite for values of \(\sigma\) that lie in some interval \(a\lt \sigma b\text{.}\)
The two-sided Laplace transform uses the lower limit of integration, that is, \(t=-\infty\text{,}\) and hence requires a knowledge of the past history of the function \(f(t)\text{,}\) i.e., \(t\lt 0\text{.}\) For most physical applications, one is interested in the behavior of a system only for \(t \ge 0\text{.}\) Mathematically speaking, the initial conditions \(f(0), \, f\,'(0), \, f\,''(0), \ldots\text{,}\) are a consequence of the past history of the system and are often all that is necessary to know. For this reason, it is useful to define the one-sided Laplace transform of \(f(t)\text{,}\) which is commonly referred to simply as the Laplace transform of \(f(t)\text{,}\) which is also defined as an integral:
\begin{equation} \mathcal{L}\big(f(t)\big)=F(s)=\int_0^{\infty}f(t) e^{-st}\,dt, \text{ where } s=\sigma +i\omega\text{.}\tag{11.5.1} \end{equation}
If the defining integral (11.5.1) for the Laplace transform exists for \(s_0=\sigma_0+i\omega\text{,}\) then values of \(\sigma\) with \(\sigma > \sigma_0\) imply that \(e^{-\sigma t} \lt e^{-\sigma_0t}\text{,}\) and thus
\begin{equation*} \int_0^{\infty}\left| \,f(t) \right| e^{-\sigma t}\,dt \lt \int_0^{\infty}\left| \,f(t) \right| e^{-\sigma_0 t}\,dt\lt \infty\text{,} \end{equation*}
and it follows that \(F(s)\) exists for \(s=\sigma +i\omega\text{.}\) Therefore, the Laplace transform \(\mathcal{L}(f(t))\) is defined for all points \(s\) in the right half-plane \(\mathrm{Re}(s) > \sigma_0\text{.}\)
Another way to view the relationship between the Fourier transform and the Laplace transform is to consider the function \(U(t)\) given by
\begin{equation*} U(t) = \begin{cases}f(t) \amp \text{ for } t \ge 0,\\ \;\; 0 \amp \text{ for } t \lt 0. \end{cases} \end{equation*}
Then the Fourier transform theory shows us that
\begin{equation*} U(t)=\frac{1}{2\pi}\int_{-\infty}^{\infty}\left[\int_{-\infty}^{\infty}U(t)e^{-i\omega t}\,dt\right]e^{i\omega t}\,d\omega\text{,} \end{equation*}
and since the integrand \(U(t)\) is zero for \(t\lt 0\text{,}\) this equation can be written as
\begin{equation*} f(t)=\frac{1}{2\pi}\int_{-\infty}^{\infty}\left[\int_0^{\infty}f(t) e^{-i\omega t}\,dt\right]e^{i\omega t}\,d\omega\text{.} \end{equation*}
Use the change of variable \(s=\sigma +i\omega\) and \(d\omega=\frac{ds}{i}\text{,}\) where \(\sigma>\sigma_0\) is held fixed, then the new limits of integration are from \(s=\sigma -i\omega\) to \(s=\sigma +i\omega\text{.}\) The resulting equation is
\begin{equation*} f(t)=\frac{1}{2\pi}\int_{\sigma -i\infty}^{\sigma +i\infty}\left[\int_0^{\infty}f(t) e^{-st}\,dt\right]e^{st}\,ds\text{.} \end{equation*}
Therefore, the Laplace transform is
\begin{equation*} \mathcal{L}\big(f(t)\big) = F(s) = \int_0^{\infty}f(t) e^{-st}\,dt, \text{ where } s=\sigma +i\omega\text{,} \end{equation*}
and the inverse Laplace transform is
\begin{equation} \mathcal{L}^{-1}\big(F(s)\big)=f(t =\frac{1}{2\pi}\int_{\sigma-i\infty}^{\sigma +i\infty}F(s) e^{st}\,ds\text{.}\tag{11.5.2} \end{equation}

Subsection 11.5.2 Properties of the Laplace Transform

Although a function \(f(t)\) may be defined for all values of \(t\text{,}\) its Laplace transform is not influenced by values of \(f(t)\text{,}\) where \(t\lt 0\text{.}\) The Laplace transform of \(f(t)\) is actually defined for the function \(U(t)\) given by
\begin{equation*} U(t) = \begin{cases}f(t) \amp \text{ for } t \ge 0,\\ \;\;0 \amp \text{ for } t \lt 0. \end{cases} \end{equation*}
A sufficient condition for the existence of the Laplace transform is that \(|f(t)|\) does not grow too rapidly as \(t \to +\infty\text{.}\) We say that the function \(f\) is of exponential order if there exists real constants \(M>0\) and \(K\text{,}\) such that
\begin{equation*} |f(t)| \le Me^{Kt} \text{ holds for all } t \ge 0\text{.} \end{equation*}
All functions in this chapter are assumed to be of exponential order. The next theorem shows that their Laplace transform \(F(\sigma +i\tau)\) exists for values of \(s\) in a domain that includes the right half-plane \(\mathrm{Re}(s) > K\text{.}\)

Proof.

Using \(s=\sigma +i\tau\) we see that \(F(s)\) can be expressed as
\begin{equation*} F(s) = \int_0^{\infty}f(t) e^{-\sigma t}\cos \tau t\,dt - i\int_0^{\infty}f(t) e^{-\sigma t}\sin \tau t\,dt\text{.} \end{equation*}
Then for values of \(\sigma >K\text{,}\) we have
\begin{align*} \int_0^{\infty}|f(t)| e^{-\sigma t}|\cos \tau t|\,dt \amp \le M\int_0^{\infty}e^{(K-\sigma)t}\,dt \le \frac{M}{\sigma -K}, \text{ and }\\ \int_0^{\infty}|f(t)| e^{-\sigma t}|\sin \tau t|\,dt \amp \le M\int_0^{\infty}e^{(K-\sigma)t}\,dt \le \frac{M}{\sigma -K}\text{,} \end{align*}
which imply that the integrals defining the real and imaginary parts of \(F\) exist for values of \(\mathrm{Re}(s) >K\text{.}\)
Note: The domain of definition of the defining integral for the Laplace transform \(\mathcal{L}\big(f(t)\big)\) seems to be restricted to a half plane. However, the resulting formula \(F(s)\) might have a domain much larger than this half plane. Later we will show that \(F(s)\) is an analytic function of the complex variable \(s\text{.}\) For most applications involving Laplace transforms that we will study, the Laplace transforms are rational functions that have the form \(\frac{P(s)}{Q(s)}\text{,}\) where \(P\) and \(Q\) are polynomials, and some other important ones will have the form \(\frac{e^{as}P(s)}{Q(s)}\text{.}\)

Proof.

Let \(K\) be chosen so that both \(F\) and \(G\) are defined for \(\mathrm{Re}(s)>K\text{,}\) then
\begin{align*} \mathcal{L}\big(af(t)+bg(t)\big) \amp = \int_0^{\infty}[af(t)+bg(t)]e^{-st}\,dt\\ \amp = a\int_0^{\infty}f(t) e^{-st}dt+b\int_0^{\infty}g(t)e^{-st}\,dt\\ \amp = aF(s) + bG(s)\text{.} \end{align*}

Proof.

If \(\sigma\) is sufficiently large, then the integral representation, Equation (11.5.2), for the inverse Laplace transform can be used to obtain
\begin{align*} f(t) \amp = \mathcal{L}^{-1}\big(F(s)\big) = \frac{1}{2\pi i}\int_{\sigma -i\infty}^{\sigma +i\infty}F(s) e^{st}\,ds = \frac{1}{2\pi i}\int_{s-i\infty}^{s+i\infty}G(s) e^{st}\,ds\\ \amp = \mathcal{L}^{-1}i\big(G(s)\big) = g(t) \text{.} \end{align*}

Example 11.5.4.

Show that the Laplace transform of the step function given by
\begin{equation*} f(t) = \begin{cases}1 \amp \text{ for } 0 \le t\lt c,\\ 0 \amp \text{ for } c\lt t \end{cases} \text{ is } \mathcal{L}\big(f(t)\big) =\frac{1-e^{-cs}}{s}\text{.} \end{equation*}
Solution.
Using the integral definition for \(\mathcal{L}\big(f(t)\big)\text{,}\) we obtain
\begin{align*} \mathcal{L}\big(f(t)\big) \amp = \int_0^{\infty}f(t) e^{-st}\,dt\\ \amp = \int_0^{c}e^{-st}dt+\int_c^{\infty}e^{-st}(0)\,dt\\ \amp = \left.\frac{-e^{-st}}{s}\right|_{t=0}^{t=c}\\ \amp = \frac{1-e^{-cs}}{s}\text{.} \end{align*}

Example 11.5.5.

Show that \(\mathcal{L}(e^{at}) =\frac{1}{s-a}\text{,}\) where \(a\) is a real constant.
Solution.
We will actually show that the integral defining \(\mathcal{L}(e^{at})\) is equal to the formula \(F(s) =\frac{1}{s-a}\) for values of \(s\) with \(\mathrm{Re}(s) >a\text{,}\) and the extension to other values of \(s\) is inferred by our knowledge about the domain of a rational function. Using straightforward integration techniques we find that
\begin{align*} \mathcal{L}(e^{at}) \amp = \int\nolimits_0^{\infty}e^{at}e^{-st}\,dt\\ \amp = \lim_{R \to +\infty}\int_0^{R}e^{(a-s)t}\,dt\\ \amp = \lim_{R \to +\infty}\frac{e^{(a-s) R}}{a-s}+\frac{1}{s-a}\text{.} \end{align*}
Let \(s=\sigma +i\tau\) be held fixed, or where \(\sigma >a\text{.}\) Then since \(a-\sigma\) is a negative real number we have \(\lim\limits_{R \to +\infty}e^{(a-s) R}=0\) and this result can be used in the latter equation to obtain the desired conclusion.
The property of linearity can be used to find new Laplace transforms from known ones.

Example 11.5.6.

Show that \(\mathcal{L}(\sinh at) =\frac{a}{s^2-a^2}\text{.}\)
Solution.
Since \(\sinh at = \frac{1}{2}e^{at}-\frac{1}{2}e^{-at}\text{,}\) we obtain
\begin{equation*} \mathcal{L}(\sinh at) = \frac{1}{2}\mathcal{L}(e^{at}) - \frac{1}{2}\mathcal{L}(e^{-at})=\frac{1}{2}\left(\frac{1}{s-a}\right)-\frac{1}{2}\left(\frac{1}{s+a}\right)=\frac{a}{s^2-a^2}\text{.} \end{equation*}
The technique of integration by parts is also helpful in finding new Laplace transforms.

Example 11.5.7.

Show that \(\mathcal{L}(t) = \frac{1}{s^2}\text{.}\)
Solution.
Using integration by parts we obtain
\begin{align*} \mathcal{L}(t) \amp = \lim_{R \to +\infty}\int_0^{R}te^{-st}\,dt\\ \amp = \lim_{R \to +\infty}\left.\left(\frac{-t}{s}e^{-st}-\frac{1}{s^2}e^{-st}\right)\right|_{t=0}^{t=R}\\ \amp = \lim_{R \to +\infty}\left(\frac{-R}{s}e^{-sR}-\frac{1}{s^2}e^{-sR}\right) + 0 + \frac{1}{s^2}\text{.} \end{align*}
For values of \(s\) in the right half plane \(\mathrm{Re}(s) >0\text{,}\) an argument similar to that in Example 11.5.5 shows that the last limit approaches zero. This observation establishes the result.

Example 11.5.8.

Show that \(\mathcal{L}(\cos bt) =\frac{s}{s^2+b^2}\text{.}\)
Solution.
A direct approach using the definition is tedious. Let us assume that the complex constants \(\pm ib\) are permitted and hence following the Laplace transforms exist:
\begin{equation*} \mathcal{L}(e^{ibt}) = \frac{1}{s-ib}, \text{ and } \mathcal{L}(e^{-ibt}) =\frac{1}{s-ib}\text{.} \end{equation*}
Using the linearity of the Laplace transform we obtain
\begin{equation*} \mathcal{L}(\cos bt) = \frac{1}{2}\mathcal{L}(e^{ibt}) + \frac{1}{2}\mathcal{L}(e^{-ibt}) = \frac{1}{2}\frac{1}{s-ib} + \frac{1}{2}\frac{1}{s+ib} = \frac{s}{s^2+b^2}\text{.} \end{equation*}
Inverting the Laplace transform is usually accomplished with the aid of a table of known Laplace transforms and the technique of partial fraction expansion.

Example 11.5.9.

Find \(\mathcal{L}^{-1}(\frac{3s+6}{s^2+9})\text{.}\)
Solution.
Using linearity and lines 6 and 7 of Table 11.5.10, we obtain
\begin{equation*} \mathcal{L}^{-1}\left(\frac{3s+6}{s^2+9}\right) = 3\mathcal{L}^{-1}\left(\frac{s}{s^2+9}\right) + 2\mathcal{L}^{-1}\left(\frac{3}{s^2+9}\right)= 3\cos 3t+2\sin 3t\text{.} \end{equation*}
Table 11.5.10 gives the Laplace transforms of some well-known functions, and Table 11.5.11 highlights some important properties of Laplace transforms.
Table 11.5.10. Well-Known Laplace Transforms
\(f(t)\) \(F(s) =\int_0^{\infty}f(t)e^{-st}\,dt\)
1. \(1\) \(\frac{1}{s}\)
\hdashline 2. \(t^n\) \(\frac{n!}{s^{n+1}}\)
\hdashline 3. \(U_c(t)\) unit step function \(\frac{e^{-cs}}{s}\)
\hdashline 4. \(e^{at}\) \(\frac{1}{s-a}\)
\hdashline 5. \(t^ne^{at}\) \(\frac{n!}{(s-a)^{n+1}}\)
\hdashline 6. \(\cos bt\) \(\frac{s}{s^2+b^2}\)
\hdashline 7. \(\sin bt\) \(\frac{b}{s^2+b^2}\)
\hdashline 8. \(e^{at}\cos bt\) \(\frac{s-a}{(s-a)^2+b^2}\)
\hdashline 9. \(e^{at}\sin bt\) \(\frac{b}{(s-a)^2+b^2}\)
\hdashline 10. \(t\cos bt\) \(\frac{s^2-b^2}{(s^2+b^2)^2}\)
\hdashline 11. \(t\sin bt\) \(\frac{2bs}{(s^2+b^2)^2}\)
\hdashline 12. \(\cosh at\) \(\frac{s}{s^2-a^2}\)
\hdashline 13. \(\sinh at\) \(\frac{a}{s^2-a^2}\)
Table 11.5.11. Properties of the Laplace Transform
Definition \(\mathcal{L}(\,f(t)) =F(s)\)
\hdashline First derivative \(\mathcal{L}(\,f\,'(t)) = sF(s) - f(0)\)
\hdashline Second derivative \(\mathcal{L}(\,f\,''(t)) = s^2F(s) - sf(0) - f\,'(0)\)
\hdashline Integral \(\mathcal{L}(\int_0^tf(\tau)\,d\tau) = \frac{F(s)}{s}\)
\hdashline Multiplication by \(t\) \(\mathcal{L}(tf(t)) = -F\,'(s)\)
\hdashline Division by \(t\) \(\mathcal{L}(\frac{f(t)}{t}) = \int\nolimits_s^{\infty}F(\sigma)\,d\sigma\)
\hdashline \(s\) axis Shifting \(\mathcal{L}(e^{at}f(t))=F(s-a)\)
\hdashline \(t\) axis Shifting \(\mathcal{L}(U_a(t)f(t-a)) = e^{-as}F(s)\)    for    \(a>0\)
\hdashline Convolution \(\mathcal{L}\big(h(t)\big) = F(s) G(s)\)    where    \(h(t) =\int_0^tf(t-\tau ) g(\tau)\,d\tau\)

Exercises Exercises

1.
Show that \(\mathcal{L}(1) =\frac{1}{s}\) by using the integral definition for the Laplace transform. Assume that \(\mathrm{Re}(s)>0\text{.}\)
Solution.
Use \(s=\sigma +i\tau\) and the integral \(\int e^{-(\sigma +i\tau)t}\,dt = \frac{e^{-\sigma t}[-\sigma\cos(\tau t)+\tau\sin(\tau t)]}{\sigma^2+\tau^2} + i\frac{e^{-\sigma t}[\tau\cos(\tau t)+\sigma\sin(\tau t)]}{\sigma ^2+\tau^2} = u(t)+iv(t)\text{.}\) Supply the details showing that \(\lim\limits_{t \to +\infty }u(t)=0\) and \(\lim\limits_{t \to +\infty}v(t)=0\text{.}\) From that conclusion it follows that \(\mathcal{L}(1)=\int_0^{\infty}e^{-(\sigma +i\tau)t}\,dt = 0 + 0i - \frac{-1}{\sigma +i\tau }= \frac{1}{s}\text{.}\)
2.
Show that \(\mathcal{L}(t^2) =\frac{2}{s^3}\) by using the integral definition for the Laplace transform. Assume that \(\mathrm{Re}(s) >0\text{.}\)
3.
Find \(\mathcal{L}\big(f(t)\big)\) for each of the following.
(a)
\(U(t) = \begin{cases}t \amp \text{ for } 0 \le t\lt c, \\ 0 \amp \text{ otherwise. } \end{cases}\)
Solution.
\(\mathcal{L}\big(f(t)\big) =\frac{1}{s^2}- \frac{ce^{-cs}}{s}-\frac{e^{-cs}}{s^2}\text{.}\)
(b)
\(U(t) = \begin{cases}1 \amp \text{ for } 1\lt t\lt 2, \\ 0 \amp \text{ otherwise } . \end{cases}\)
(c)
\(U(t) = \begin{cases}e^{at} \amp \text{ for } 0 \le t\lt 1, \\ \; 0 \amp \text{ otherwise. } \end{cases}\)
Solution.
\(\mathcal{L}\big(f(t)\big) =\frac{1}{s-a}-\frac{ e^{a-s}}{s-a}\text{.}\)
(d)
\(U(t) = \begin{cases}\sin t \amp \text{ for } 0 \le t \le \pi , \\ \;\; 0 \amp \text{ otherwise. } \end{cases}\)
4.
Use the linearity of Laplace transform and Table 11.5.10 to find:
(a)
\(\mathcal{L}(3t^2-4t+5)\text{.}\)
Solution.
\(\mathcal{L}(3t^2-4t+5) = \frac{6}{s^3} - \frac{4}{s^2} + \frac{5}{s}\text{.}\)
(b)
\(\mathcal{L}(2\cos 4t)\text{.}\)
(c)
\(\mathcal{L}(e^{2t-3})\text{.}\)
Solution.
\(\mathcal{L}(e^{2t-3}) = \frac{e^{-3}}{s-2}\text{.}\)
(d)
\(\mathcal{L}(6e^{-t}+3\sin 5t)\text{.}\)
(e)
\(\mathcal{L}((t+1)^4)\text{.}\)
Solution.
\(\mathcal{L}\big((t+1)^4\big) = \frac{24}{s^5}+\frac{24}{s^4}+\frac{12}{s^3}+\frac{4}{s^2}+\frac{1}{s}\text{.}\)
(f)
\(\mathcal{L}(\cosh 2t)\text{.}\)
5.
Use the linearity of the inverse Laplace transform and Table 11.5.11 to find:
(a)
\(\mathcal{L}^{-1}(\frac{1}{s^2+25})\text{.}\)
Solution.
\(\mathcal{L}^{-1}\big(\frac{1}{s^2+25}\big) = \frac{1}{5} \sin 5t\text{.}\)
(b)
\(\mathcal{L}^{-1}(\frac{4}{2}-\frac{6}{s^2})\text{.}\)
(c)
\(\mathcal{L}^{-1}(\frac{1+s^2-s}{s^4})\text{.}\)
Solution.
\(\mathcal{L}^{-1}\big(\frac{1+s^2-s^{3}}{s^4}\big) = -1 + t + \frac{t^{3}}{6}\text{.}\)
(d)
\(\mathcal{L}^{-1}(\frac{2s+9}{s^2+9})\text{.}\)
(e)
\(\mathcal{L}^{-1}(\frac{6s}{s^2-4})\text{.}\)
Solution.
\(\mathcal{L}^{-1}\big(\frac{6s}{s^2-4}\big) = 3e^{-2t}+3e^{2t} = 6\cosh 2t\text{.}\)
(f)
\(\mathcal{L}^{-1}\big(\frac{2s+1}{s(s+1)}\big)\text{.}\)
6.
Write a report on how complex analysis is used in the study of Laplace transforms. Include ideas and examples that are not mentioned in the text.