Differential Equations › Section D · Laplace transforms › Topic D.2

Computing Laplace transforms

Nobody integrates from zero to infinity every time. A short table of basic transforms, plus a handful of rules for sums, exponential factors, factors of t and derivatives, produces almost every transform you will need.

From integrals to a table

The previous topic, The definition of the Laplace transform, computed \(\mathcal{L}\{1\}\), \(\mathcal{L}\{e^{at}\}\), \(\mathcal{L}\{t\}\) and \(\mathcal{L}\{\cos bt\}\) straight from the integral, and the cosine took two rounds of integration by parts. In practice those results are collected in a table, and new transforms are built from old ones with a few general rules. Each rule is proved once, from the definition, and then used freely. The skill this topic builds is recognition: looking at a function, seeing which table entries it is made of, and which rule joins them.

Linearity

Because an integral of a sum is the sum of the integrals, and constants pull out of integrals, the transform is linear:

\mathcal{L}\{af(t) + bg(t)\} = aF(s) + bG(s)

whatever the constant multipliers \(a\) and \(b\) are. So a sum is transformed term by term, with each constant carried along. Linearity does not extend to products: the transform of \(f(t)g(t)\) is in general not \(F(s)G(s)\). Products need one of the special rules below, or the convolution theorem from a later topic, Convolution integrals.

The basic table

These entries, with their conditions, are worth memorising: \(\mathcal{L}\{1\} = \frac{1}{s}\) for \(s > 0\); \(\mathcal{L}\{t^n\} = \frac{n!}{s^{n+1}}\) for \(n = 0, 1, 2, \dots\) and \(s > 0\); \(\mathcal{L}\{e^{at}\} = \frac{1}{s - a}\) for \(s > a\); and, for \(s > 0\),

\mathcal{L}\{\sin bt\} = \frac{b}{s^2 + b^2}, \qquad \mathcal{L}\{\cos bt\} = \frac{s}{s^2 + b^2}

For a power that is not a whole number, \(t^p\) with \(p > -1\), the factorial is replaced by the gamma function, \(\Gamma(p + 1)\); for example \(\mathcal{L}\{\sqrt{t}\} = \frac{\sqrt{\pi}}{2s^{3/2}}\).

Hyperbolic functions from linearity

Linearity can produce new entries. Since \(\sinh bt = \frac{1}{2}(e^{bt} - e^{-bt})\), its transform is

\frac{1}{2}\Big(\frac{1}{s - b} - \frac{1}{s + b}\Big) = \frac{b}{s^2 - b^2}

for \(s > |b|\). The same steps with a plus sign give \(\mathcal{L}\{\cosh bt\} = \frac{s}{s^2 - b^2}\). These look almost identical to the sine and cosine entries: the only difference is the minus sign in \(s^2 - b^2\). That single sign is a frequent source of error, so check which function you have before reaching for the table.

Worked example: linearity in action

Find the transforms of \(f(t) = 4 - 3t^2 + 5e^{-2t}\) and \(g(t) = 2\cos 3t - 7\sin 3t\). 1. For \(f\), transform each term: \(\frac{4}{s}\), then \(-3 \cdot \frac{2!}{s^3} = -\frac{6}{s^3}\), then \(\frac{5}{s + 2}\) with \(a = -2\). 2. So \(F(s) = \frac{4}{s} - \frac{6}{s^3} + \frac{5}{s + 2}\). 3. For \(g\), with \(b = 3\): \(\frac{2s}{s^2 + 9} - \frac{21}{s^2 + 9}\), which combine over the common denominator. 4. So \(G(s) = \frac{2s - 21}{s^2 + 9}\). Check: the sine term contributes \(7 \cdot 3 = 21\), because sine carries \(b\) on top.

The first shifting theorem

Multiplying a function by \(e^{at}\) shifts its transform:

\mathcal{L}\{e^{at}f(t)\} = F(s - a)

The proof takes one line. In the definition, \(e^{-st}e^{at} = e^{-(s - a)t}\), so the integral is the transform of \(f\) evaluated at \(s - a\) instead of \(s\). To use it, find \(F(s)\) for the function without the exponential, then replace every \(s\) by \(s - a\). For \(e^{-3t}\), \(a = -3\), so \(s\) becomes \(s + 3\). The rule is how damped oscillations such as \(e^{-t}\cos 2t\) are transformed.

Worked example: shifted transforms

Find \(\mathcal{L}\{e^{-3t}\sin 4t\}\) and \(\mathcal{L}\{t^3e^{2t}\}\). 1. Without the exponential, \(\mathcal{L}\{\sin 4t\} = \frac{4}{s^2 + 16}\). 2. Here \(a = -3\), so replace \(s\) by \(s + 3\): the answer is \(\frac{4}{(s + 3)^2 + 16}\). 3. For the second, \(\mathcal{L}\{t^3\} = \frac{3!}{s^4} = \frac{6}{s^4}\). 4. With \(a = 2\), replace \(s\) by \(s - 2\): the answer is \(\frac{6}{(s - 2)^4}\). Check: the shift moves the condition too. \(\mathcal{L}\{t^3\}\) needs \(s > 0\), so the shifted transform needs \(s - 2 > 0\), that is, \(s > 2\), the condition expected for anything growing like \(e^{2t}\).

Multiplying by t

Differentiating the definition with respect to \(s\) brings down a factor of \(-t\), which gives

\mathcal{L}\{tf(t)\} = -F'(s), \qquad \mathcal{L}\{t^nf(t)\} = (-1)^nF^{(n)}(s)

For \(t\sin 2t\), start from \(F(s) = \frac{2}{s^2 + 4}\). Its derivative is \(-\frac{4s}{(s^2 + 4)^2}\), so \(\mathcal{L}\{t\sin 2t\} = \frac{4s}{(s^2 + 4)^2}\). As a check, the rule applied to \(te^{5t}\) gives \(-\frac{d}{ds}\frac{1}{s - 5} = \frac{1}{(s - 5)^2}\), the same answer the shifting theorem gives from \(\mathcal{L}\{t\} = \frac{1}{s^2}\). This rule returns in a later topic, Nonconstant coefficient problems.

Transforms of derivatives

Suppose \(f\) is continuous and of exponential order, and \(f'\) is piecewise continuous. Integrating by parts, with \(u = e^{-st}\) and \(dv = f'(t)\,dt\), gives \(\mathcal{L}\{f'\} = sF(s) - f(0)\). Applying this twice, with \(f'\) now continuous and of exponential order too, gives

\mathcal{L}\{f''\} = s^2F(s) - sf(0) - f'(0)

Differentiation in \(t\) becomes multiplication by \(s\), and the initial values appear on their own. Check with \(f = e^{2t}\): \(f' = 2e^{2t}\) has transform \(\frac{2}{s - 2}\), and \(sF - f(0) = \frac{s}{s - 2} - 1\), which also equals \(\frac{2}{s - 2}\). The topic Solving initial value problems is built on these formulas.

Common mistakes to avoid

Transforming a product factor by factor: \(\mathcal{L}\{t\sin 2t\}\) is not \(\frac{1}{s^2} \cdot \frac{2}{s^2 + 4}\). Forgetting the factorial: \(\mathcal{L}\{t^2\} = \frac{2}{s^3}\), not \(\frac{1}{s^3}\). Shifting the wrong way: \(e^{at}\) replaces \(s\) by \(s - a\), so \(e^{-3t}\) gives \(s + 3\). Shifting only part of the answer: in \(\mathcal{L}\{e^{2t}\cos 3t\}\) the numerator \(s\) must also become \(s - 2\). Mixing up \(\sinh\) and \(\sin\), whose denominators differ in one sign. Dropping \(-f(0)\) from the derivative formula.

Key takeaways

Transforms are linear, so sums are handled term by term, but products are not. Memorise the table for \(1\), \(t^n\), \(e^{at}\), \(\sin bt\), \(\cos bt\), \(\sinh bt\) and \(\cosh bt\). The first shifting theorem, \(\mathcal{L}\{e^{at}f\} = F(s - a)\), handles exponential factors; \(\mathcal{L}\{tf\} = -F'(s)\) handles factors of \(t\). Derivatives transform as \(sF - f(0)\) and \(s^2F - sf(0) - f'(0)\), with the initial values built in.

Flashcards

Flashcard

Linearity

L{af + bg} = aF(s) + bG(s)

Sums term by term; products are not transformed this way.

Test your understanding

Question 1 of 5

What is the Laplace transform of 3t² - 4e^(2t)?

Question 2 of 5

What is the Laplace transform of e^(2t) cos 3t?

Question 3 of 5

Using L{t f(t)} = -F′(s), what is the Laplace transform of t cos t?

Question 4 of 5

If y(0) = 2 and y′(0) = -1, what is the transform of y″ in terms of Y(s)?

Question 5 of 5

What is the Laplace transform of 4 sinh 3t?

← Previous topicNext topic →