Differential Equations › Section D · Laplace transforms › Topic D.4
Step functions
Real inputs get switched on and off: a motor starts, a valve closes, a voltage is cut. The unit step function writes any such switching in a single formula, and its Laplace transform is remarkably simple.
Modelling a switch
Imagine a tank that receives no inflow until a pump starts at \(t = 5\) minutes, then a steady 3 litres per minute. Or a circuit fed 12 volts for two seconds and then nothing. These inputs are piecewise defined: one formula before the switch and another after. Undetermined coefficients needs a single smooth forcing function, so it would force you to solve the equation separately on each piece and match the pieces at every switching time. Laplace transforms avoid all of that, provided the switching is written in a form the transform can handle. That form is built from the unit step function.
The unit step function
For a constant \(c \ge 0\), the unit step (or Heaviside) function is
It is also written \(u(t - c)\) or \(H(t - c)\), and it is named after the engineer Oliver Heaviside. Think of it as a switch that flips on at time \(c\). Its value exactly at \(t = c\) is a convention and has no effect on any transform, because a single point does not change an integral. Two combinations are used constantly: \(1 - u_c(t)\) is a switch that turns off at \(c\), and \(u_a(t) - u_b(t)\), with \(a < b\), is a pulse that is 1 only for \(a \le t < b\).
Writing piecewise functions with steps
Multiplying by \(u_c(t)\) leaves a function untouched after \(c\) and erases it before. So to build a piecewise function, start with the first formula, and at each switching time add a step multiplied by the change in formula. For a function equal to \(g_1\) on \([0, a)\), \(g_2\) on \([a, b)\) and \(g_3\) from \(b\) on,
Test it on each interval: before \(a\) both steps are 0, between \(a\) and \(b\) the first step cancels \(g_1\), and after \(b\) only \(g_3\) survives.
The transform of a step
From the definition, \(u_c(t)\) is zero until \(t = c\), so
Compare this with \(\mathcal{L}\{1\} = \frac{1}{s}\). Switching the constant on at time \(c\) instead of 0 multiplies its transform by \(e^{-cs}\). That pattern, a delay in time becoming an exponential factor in \(s\), is the main idea of this topic, and the next card shows that it holds for every function, not only constants.
Worked example: a three-level signal
A signal is 4 for \(0 \le t < 3\), \(-2\) for \(3 \le t < 7\) and 5 for \(t \ge 7\). Write it with steps and find its transform. 1. Start with 4. At \(t = 3\) the value changes by \(-2 - 4 = -6\); at \(t = 7\) it changes by \(5 - (-2) = 7\). 2. So \(f(t) = 4 - 6u_3(t) + 7u_7(t)\). 3. Check: at \(t = 4\), \(4 - 6 = -2\); at \(t = 8\), \(4 - 6 + 7 = 5\). 4. By linearity,
Each coefficient is a jump size, and each exponential records when that jump happens.
The second shifting theorem
If \(F(s) = \mathcal{L}\{f(t)\}\) and \(c \ge 0\), then
The function \(u_c(t)f(t - c)\) is the graph of \(f\) slid \(c\) units to the right, with zero in front. To prove the rule, note the integrand vanishes before \(c\), so the transform is \(\int_c^\infty e^{-st}f(t - c)\,dt\). Substitute \(\tau = t - c\): the limits become 0 and \(\infty\), and \(e^{-st} = e^{-cs}e^{-s\tau}\). The factor \(e^{-cs}\) comes outside, and what remains is exactly \(F(s)\). Compare the first shifting theorem: there an exponential in \(t\) shifted \(s\); here a shift in \(t\) produces an exponential in \(s\).
When the function is not shifted
The theorem needs \(f(t - c)\), with the same \(c\) as the step. Often you meet \(u_c(t)g(t)\) instead, where \(g\) is written in terms of \(t\). The fix is to ask what function, shifted by \(c\), gives \(g(t)\). The answer is \(g(t + c)\), so
In practice: replace \(t\) by \(t + c\) in \(g\), simplify, transform, and multiply by \(e^{-cs}\). Equivalently, rewrite \(g\) in powers or functions of \(t - c\), for example \(t = (t - 2) + 2\).
Worked example: three transforms
1. \(\mathcal{L}\{u_2(t)(t - 2)^2\}\): this is already shifted, with \(f(t) = t^2\) and \(F(s) = \frac{2}{s^3}\), so the answer is \(\frac{2e^{-2s}}{s^3}\). 2. \(\mathcal{L}\{u_3(t)t^2\}\): not shifted. Replace \(t\) by \(t + 3\): \((t + 3)^2 = t^2 + 6t + 9\), with transform \(\frac{2}{s^3} + \frac{6}{s^2} + \frac{9}{s}\). Multiply by \(e^{-3s}\). 3. \(\mathcal{L}\{u_\pi(t)\sin t\}\): replace \(t\) by \(t + \pi\). Since \(\sin(t + \pi) = -\sin t\), the answer is \(-\frac{e^{-\pi s}}{s^2 + 1}\). Check on the third: just after \(t = \pi\), \(\sin t\) is negative, matching the minus sign.
Worked example: inverting with a delay
Find the inverses of \(\frac{e^{-4s}}{s(s + 2)}\) and \(\frac{se^{-2s}}{s^2 + 16}\). 1. Set the exponential aside and invert the rest. Partial fractions give
so \(f(t) = \frac{1}{2}(1 - e^{-2t})\). 2. The factor \(e^{-4s}\) means: shift by 4 and switch on at 4. The answer is \(\frac{1}{2}u_4(t)\big(1 - e^{-2(t - 4)}\big)\). 3. For the second, \(\frac{s}{s^2 + 16}\) inverts to \(\cos 4t\), so the answer is \(u_2(t)\cos\big(4(t - 2)\big)\). Check: both answers are zero before their switching times, as any \(e^{-cs}\) factor demands.
Common mistakes to avoid
Using \(e^{-cs}F(s)\) for \(u_c(t)f(t)\) when the function is not shifted: \(\mathcal{L}\{u_3(t)t^2\}\) is not \(\frac{2e^{-3s}}{s^3}\). Shifting only part of the answer when inverting: every \(t\) in \(f\) must become \(t - c\), including those inside exponentials. Forgetting the step in the answer, which makes the function nonzero before it should start. Writing the jump as the new value rather than the change: going from 3 to 8 at \(t = 5\) is \(3 + 5u_5(t)\), not \(3 + 8u_5(t)\).
Key takeaways
The step \(u_c(t)\) is 0 before \(c\) and 1 from \(c\) on, and \(\mathcal{L}\{u_c(t)\} = \frac{e^{-cs}}{s}\). Piecewise functions are written by adding, at each switching time, a step multiplied by the change in formula. The second shifting theorem, \(\mathcal{L}\{u_c(t)f(t - c)\} = e^{-cs}F(s)\), turns a delay into an exponential factor. For an unshifted \(u_c(t)g(t)\), transform \(g(t + c)\) instead. When inverting, set \(e^{-cs}\) aside, invert, then shift and switch on at \(c\).
Flashcards
Flashcard
0 for t < c, 1 for t ≥ c
Test your understanding
Question 1 of 5
A signal is 3 for 0 ≤ t < 5 and 8 for t ≥ 5. Which formula describes it?
Question 2 of 5
What is the Laplace transform of u₄(t)?
Question 3 of 5
What is the Laplace transform of u₃(t)(t - 3)?
Question 4 of 5
What is the Laplace transform of u₁(t) · t?
Question 5 of 5
What is the inverse Laplace transform of e^(-2s)/(s + 3)?
