Differential Equations › Section D · Laplace transforms › Topic D.1

The definition of the Laplace transform

The Laplace transform turns a function of time into a function of a new variable, and in doing so turns differential equations into algebra. Everything in this section rests on one improper integral, so it pays to understand it properly.

Why transform at all?

Logarithms once made multiplication easy by turning it into addition: take logs, add, then convert back. The Laplace transform plays the same trick on calculus. It takes a function \(f(t)\) of time and produces a new function \(F(s)\), and, as later topics show, it turns differentiation in \(t\) into multiplication by \(s\). A differential equation becomes an algebraic equation, which is solved and then converted back. The method shines when the forcing term is switched on and off or arrives as a sudden blow, cases that undetermined coefficients handles badly or not at all. This topic sets up the definition itself.

Piecewise continuous functions

The transform is defined for a wide class of functions, including ones with jumps. Call a function piecewise continuous on an interval when you can cut that interval into finitely many pieces, the function is continuous on the inside of each one, and it has finite one-sided limits at every endpoint. In plain terms: finitely many breaks, and no blowing up to infinity. A switch that jumps from 0 to 12 volts at \(t = 4\) is piecewise continuous. The function \(\frac{1}{t - 2}\) is not, on any interval containing 2, because it is unbounded there.

The definition

Let \(f(t)\) be piecewise continuous on \([0, \infty)\). Its Laplace transform is

\mathcal{L}\{f(t)\} = F(s) = \int_0^\infty e^{-st} f(t)\,dt

for those values of \(s\) where the improper integral converges. The variable of integration is \(t\), so once the integral is evaluated every \(t\) disappears and only \(s\) remains. Lower-case letters name functions of \(t\) and the matching capitals name their transforms: \(g \to G\), \(y \to Y\). In this course \(s\) is a real number. Only values of \(f\) for \(t \ge 0\) matter, which suits initial value problems that start at \(t = 0\).

Improper integrals: a quick review

An integral to infinity means a limit:

\int_0^\infty g(t)\,dt = \lim_{n \to \infty} \int_0^n g(t)\,dt

It converges if this limit exists and is finite. The key case is an exponential. For \(c \ne 0\), \(\int_0^n e^{ct}\,dt = \frac{e^{cn} - 1}{c}\). If \(c < 0\), then \(e^{cn} \to 0\) and the integral equals \(-\frac{1}{c}\). If \(c > 0\), then \(e^{cn} \to \infty\) and it diverges, and for \(c = 0\) the integrand is 1, which also diverges. So \(\int_0^\infty e^{-2t}\,dt = \frac{1}{2}\), while \(\int_0^\infty e^{2t}\,dt\) does not exist.

Worked example: exponentials and constants

Find \(\mathcal{L}\{e^{at}\}\), where \(a\) is a constant. 1. Substitute into the definition and combine the exponentials:

\int_0^\infty e^{-st}e^{at}\,dt = \int_0^\infty e^{(a - s)t}\,dt

2. This is the review integral with \(c = a - s\), which converges only when \(a - s < 0\), that is, \(s > a\). 3. Its value is \(-\frac{1}{a - s} = \frac{1}{s - a}\). So \(\mathcal{L}\{e^{at}\} = \frac{1}{s - a}\) for \(s > a\). 4. Setting \(a = 0\) gives \(\mathcal{L}\{1\} = \frac{1}{s}\) for \(s > 0\). Check: for \(e^{3t}\) the answer is \(\frac{1}{s - 3}\), valid only for \(s > 3\), when \(e^{-st}\) decays faster than \(e^{3t}\) grows.

Worked example: the transform of t

Find \(\mathcal{L}\{t\}\). 1. Write \(F(s) = \lim_{n \to \infty} \int_0^n t e^{-st}\,dt\) and integrate by parts with \(u = t\) and \(dv = e^{-st}\,dt\), so \(v = -\frac{1}{s}e^{-st}\). 2. This gives

\Big[-\frac{t}{s}e^{-st}\Big]_0^n + \frac{1}{s}\int_0^n e^{-st}\,dt

3. For \(s > 0\), \(ne^{-sn} \to 0\) as \(n \to \infty\), because the exponential beats any power, so the bracket tends to 0. 4. The remaining integral tends to \(\frac{1}{s}\), so \(\mathcal{L}\{t\} = \frac{1}{s^2}\) for \(s > 0\). Check: for \(s \le 0\) the integrand \(te^{-st}\) never decays, so the integral must diverge, consistent with the condition.

Worked example: a switched-off signal

A signal holds the value 2 for \(0 \le t < 3\) and is 0 afterwards. Find its transform. 1. The function is piecewise continuous, with one jump at \(t = 3\). 2. Split the integral where the formula changes; the second piece is zero:

F(s) = \int_0^3 2e^{-st}\,dt + \int_3^\infty 0\,dt

3. Integrate, for \(s \ne 0\):

F(s) = \Big[-\frac{2}{s}e^{-st}\Big]_0^3 = \frac{2(1 - e^{-3s})}{s}

4. Check: the integral is over a finite interval, so it converges for every \(s\); at \(s = 0\) it is simply the area, 6, which is also the limit of the formula as \(s \to 0\).

A harder integral: cosine

Some transforms need integration by parts twice. For \(F(s) = \mathcal{L}\{\cos bt\}\) with \(s > 0\), one round of parts gives

F = \frac{1}{s} - \frac{b}{s}\int_0^\infty e^{-st}\sin bt\,dt

A second round turns that sine integral into \(\frac{b}{s}F\), so the original integral reappears: \(F = \frac{1}{s} - \frac{b^2}{s^2}F\). Solving this equation for \(F\) gives

\mathcal{L}\{\cos bt\} = \frac{s}{s^2 + b^2}, \quad s > 0

A similar calculation gives \(\frac{b}{s^2 + b^2}\) for \(\sin bt\). The effort involved is why the next topic, Computing Laplace transforms, replaces direct integration with a table and a few rules.

When does the transform exist?

A function \(f\) is of exponential order \(\alpha\) if there are constants \(M > 0\) and \(T\) with \(|f(t)| \le Me^{\alpha t}\) for all \(t \ge T\): it grows no faster than some exponential. The existence theorem says that if \(f\) is piecewise continuous on \([0, \infty)\) and of exponential order \(\alpha\), then \(F(s)\) exists for every \(s > \alpha\). Polynomials, sines, cosines and exponentials all qualify. The function \(e^{t^2}\) does not: \(e^{t^2}/e^{\alpha t} \to \infty\) for every \(\alpha\). In fact \(e^{-st}e^{t^2}\) grows without bound for every \(s\), so the integral always diverges and \(e^{t^2}\) has no transform. The theorem is sufficient, not necessary: \(\frac{1}{\sqrt{t}}\) blows up at 0 yet still has a transform.

Common mistakes to avoid

Dropping the condition on \(s\): every transform comes with one, such as \(s > a\) for \(e^{at}\), and it records where the integral converges. Writing \(\mathcal{L}\{e^{at}\} = \frac{1}{s + a}\): the exponent \(a - s\) gives \(s - a\) in the denominator. Treating \(s\) as the variable of integration: \(s\) is a constant inside the integral, and \(t\) is integrated out. Evaluating at infinity by substitution instead of a limit: \(te^{-st}\) tends to 0 only because the exponential wins, which needs \(s > 0\).

Key takeaways

The Laplace transform of a piecewise continuous \(f\) is \(F(s) = \int_0^\infty e^{-st}f(t)\,dt\), defined where this improper integral converges. Direct calculation gives \(\mathcal{L}\{1\} = \frac{1}{s}\), \(\mathcal{L}\{e^{at}\} = \frac{1}{s - a}\) for \(s > a\), \(\mathcal{L}\{t\} = \frac{1}{s^2}\) and \(\mathcal{L}\{\cos bt\} = \frac{s}{s^2 + b^2}\). Functions with jumps are handled by splitting the integral. A piecewise continuous function of exponential order \(\alpha\) always has a transform for \(s > \alpha\).

Flashcards

Flashcard

Laplace transform

F(s) = ∫₀^∞ e^(-st) f(t) dt

Defined wherever the improper integral converges.

Test your understanding

Question 1 of 5

What is the Laplace transform of e^(3t), and where does it exist?

Question 2 of 5

A signal is 5 for 0 ≤ t < 2 and 0 for t ≥ 2. What is its Laplace transform (s ≠ 0)?

Question 3 of 5

Which function is not of exponential order, so the existence theorem cannot be applied to it?

Question 4 of 5

For s > 0, what is the value of the integral of t e^(-st) from 0 to ∞?

Question 5 of 5

Why does the formula L{1} = 1/s carry the condition s > 0?

← Previous topicNext topic →