Three properties turn a short table into a large one. Exponential scaling says multiplying by shifts the transform to . Time scaling says stretching time compresses frequency. And differentiating the transform corresponds to multiplying by . Together they generate almost every transform needed in practice from the handful computed in Section 7.1.
Quick Reference
| Property | Statement | Valid for |
|---|---|---|
| Linearity | both transforms exist | |
| Exponential scaling | ||
| Time scaling | , | |
| Derivative of transform | \mathscr{L}\left\{t\,f(t)\right\} = -F'(s) | |
| General form |
Transforms these generate:
Exponential Scaling
In Section 7.1 we learned that is a linear operator. We now study some further properties.
Exponential Scaling (also called the first shifting theorem). If exists for and is a constant, then
for , that is, for .
Proof
\begin{aligned} \mathscr{L}\left\{e^{at}f(t)\right\} &= \int_0^{\infty}e^{-st}e^{at}f(t)\,dt \\ &= \int_0^{\infty}e^{-(s-a)t}f(t)\,dt \\ &= F(s-a) . \end{aligned}The last step is just the definition of , read with in place of . Nothing is required beyond combining the two exponentials.
In words: multiplying by in the domain translates the transform by in the domain. This is the single most useful property in the chapter. Read backwards, it says that whenever appears only in the combination , an exponential factor is present in .
Find .
Solution
We know for . With ,
Applying the theorem with , replace by :
valid for , that is, .
Find .
Solution
Since for , taking gives
Applying the theorem with , replace by :
valid for , that is, .
Note that only the in the transform is replaced; the constant in the numerator is untouched. Replacing everywhere, including inside , is a common slip.
Time Scaling
Time Scaling. If exists for and is a positive constant, then
for , that is, .
Proof
Substitute , so and :
The requirement keeps the limits of integration in order.
The theorem makes physical sense: if time is scaled by , frequency is scaled by . Playing a recording twice as fast doubles every frequency, and the extra factor out front conserves the total.
Given and , find and .
Solution
Applying the theorem with ,
\begin{aligned} \mathscr{L}\left\{\sin kt\right\} &= \frac{1}{k}\cdot\frac{1}{(s/k)^{2}+1} \\ &= \frac{1}{k}\cdot\frac{k^{2}}{s^{2}+k^{2}} \\ &= \frac{k}{s^{2}+k^{2}} \end{aligned}where the middle step multiplied numerator and denominator by .
Similarly,
\begin{aligned} \mathscr{L}\left\{\cos kt\right\} &= \frac{1}{k}\cdot\frac{s/k}{(s/k)^{2}+1} \\ &= \frac{1}{k}\cdot\frac{sk}{s^{2}+k^{2}} \\ &= \frac{s}{s^{2}+k^{2}} \end{aligned}Both agree with the direct computations of Section 7.1, obtained here with far less work.
Derivatives of a Transform
Derivatives of a Transform. If , then
\mathscr{L}\left\{t\,f(t)\right\} = -F'(s)and in general
Proof
Differentiate both sides of with respect to , moving the derivative inside the integral:
\begin{aligned} \frac{d}{ds}F(s) &= \frac{d}{ds}\int_0^{\infty}e^{-st}f(t)\,dt \\ &= \int_0^{\infty}\frac{\partial}{\partial s}\left[e^{-st}f(t)\right]dt \\ &= \int_0^{\infty}(-t)e^{-st}f(t)\,dt \\ &= -\int_0^{\infty}e^{-st}\,t\,f(t)\,dt \\ &= -\mathscr{L}\left\{t\,f(t)\right\} \end{aligned}The interchange of differentiation and integration is justified when satisfies the conditions of the existence theorem.
Applying the result to in place of ,
\begin{aligned} \mathscr{L}\left\{t^{2}f(t)\right\} &= \mathscr{L}\left\{t\cdot t f(t)\right\} \\ &= -\frac{d}{ds}\mathscr{L}\left\{t f(t)\right\} \\ &= -\frac{d}{ds}\left(-\frac{d}{ds}\mathscr{L}\left\{f(t)\right\}\right) \\ &= \frac{d^{2}}{ds^{2}}F(s) . \end{aligned}Repeating gives
with the sign alternating because each application contributes one factor of .
Note the pleasing symmetry with Section 7.5: differentiating in corresponds (roughly) to multiplying by , and differentiating in corresponds to multiplying by . The two variables trade roles.
Find .
Solution
Since , the theorem gives
\begin{aligned} \mathscr{L}\left\{t\sin kt\right\} &= -\frac{d}{ds}\left(\frac{k}{s^{2}+k^{2}}\right) \\ &= -\frac{\dfrac{d}{ds}(k)\cdot\left(s^{2}+k^{2}\right) - k\cdot\dfrac{d}{ds}\left(s^{2}+k^{2}\right)}{\left(s^{2}+k^{2}\right)^{2}} \\ &= -\frac{0\cdot\left(s^{2}+k^{2}\right) - k(2s)}{\left(s^{2}+k^{2}\right)^{2}} \\ &= \frac{2ks}{\left(s^{2}+k^{2}\right)^{2}} \end{aligned}Remember that is a constant while is the variable, so and .
Find by two different routes.
Solution
Method (a): differentiate the transform. Start from with :
Then
\begin{aligned} \mathscr{L}\left\{t^{2}e^{-3t}\right\} &= \frac{d^{2}}{ds^{2}}\left[(s+3)^{-1}\right] \\ &= \frac{d}{ds}\left[-(s+3)^{-2}\right] \\ &= 2(s+3)^{-3} = \frac{2}{(s+3)^{3}} \end{aligned}Method (b): use exponential scaling. Start from with :
By the exponential scaling theorem, replace by :
Method (b) is much shorter, and it generalizes at once to , a formula worth memorizing.
Find .
Solution
Starting from and applying the quotient rule,
\begin{aligned} \mathscr{L}\left\{t\cos kt\right\} &= -\frac{d}{ds}\left(\frac{s}{s^{2}+k^{2}}\right) \\ &= -\frac{1\cdot\left(s^{2}+k^{2}\right) - s(2s)}{\left(s^{2}+k^{2}\right)^{2}} \\ &= -\frac{k^{2}-s^{2}}{\left(s^{2}+k^{2}\right)^{2}} = \frac{s^{2}-k^{2}}{\left(s^{2}+k^{2}\right)^{2}} \end{aligned}This transform appears constantly in resonance problems, where the factor is what the Modification Rule of Section 4.3 predicted.
Exercises
Find .
Solution
Since , exponential scaling with gives
valid for . Note that the numerator becomes , not : every in the transform is replaced.
Find .
Solution
Since , replacing by gives
Find .
Solution
Start from and differentiate twice, using so the sign is .
First derivative: F'(s) = -\dfrac{2s}{\left(s^{2}+1\right)^{2}}.
Second derivative, by the quotient rule:
F''(s) = -\frac{2\left(s^{2}+1\right)^{2} - 2s\cdot 2\left(s^{2}+1\right)(2s)}{\left(s^{2}+1\right)^{4}} = -\frac{2\left(s^{2}+1\right) - 8s^{2}}{\left(s^{2}+1\right)^{3}} = \frac{6s^{2}-2}{\left(s^{2}+1\right)^{3}} .Therefore
Given that , find without identifying .
Solution
By time scaling with ,
As a check, so , whose transform is indeed .
Frequently Asked Questions
What is the exponential scaling theorem?
It states that : multiplying by an exponential in shifts the transform in . It is often called the first shifting theorem, to distinguish it from the time-shifting theorem of Section 7.6.
How do I apply the shift correctly?
Compute for the function without the exponential, then substitute for every occurrence of . For : first get , then replace by to get . The constant in the numerator is not an , so it stays.
Why does carry a minus sign?
Because differentiating with respect to brings down a factor of . Each further differentiation brings down another, which is why the general formula carries .
Which property should I use for ?
Exponential scaling, applied to , gives in one step. Differentiating the transform times reaches the same answer with far more algebra. When two routes exist, the shifting theorem is almost always the shorter one.
Why must be positive in the time-scaling theorem?
Because the substitution must preserve the orientation of the interval . A negative would map it to , which lies outside the range the unilateral transform can see.