Inverse Laplace Transforms — Question 5

Section 23-3 · Calculus pages 4109–4110

Show solutionHide solution

Question 5

Use the ordinary one-sided Laplace transform for real ss. Seek an inverse continuous on t≥0t\ge 0 and of exponential order; transforms agreeing for all sufficiently large ss have at most one inverse in this class.

Consider the repeated quadratic factor F(s)=1(s2+4)2.F(s)=\frac 1{(s^2+4)^2}. You may use ℒ{sin⁡2t}=2/(s2+4)\mathcal L\{\sin 2t\}=2/(s^2+4), ℒ{cos⁡2t}=s/(s2+4)\mathcal L\{\cos 2t\}=s/(s^2+4) and ℒ{tg(t)}=−G′(s)\mathcal L\{tg(t)\}=-G'(s) for s>0s>0.

Tasks

  1. Derive the transform of tcos⁡2tt\cos 2t by differentiating the cosine transform.

  2. Find constants A,BA,B such that the inverse is Asin⁡2t+Btcos⁡2tA\sin 2t+B t\cos 2t. Verify the resulting rational numerator.

  3. Show that the inverse is unbounded despite having no positive exponential factor. Give an explicit sequence witnessing growth.

  4. Find its leading term near t=0t=0 and its exact real convergence interval. Compare the local term with the large-ss scale F(s)∼s−4F(s)\sim s^{-4}.

Question 5 – Solution

Strategy. A repeated quadratic can be handled with parameter differentiation and linear combinations, without assuming a product rule for inverses.

Step 1: Produce the repeated denominator. Differentiating the cosine pair gives ℒ{tcos⁡2t}=−ddsss2+4=s2−4(s2+4)2.\mathcal L\{t\cos 2t\}=-\frac d{ds}\frac{s}{s^2+4} =\frac{s^2-4}{(s^2+4)^2}. Polynomial factors times the damping kernel are integrable for s>0s>0, justifying the differentiation.

Step 2: Cancel the unwanted numerator. The transform of Asin⁡2t+Btcos⁡2tA\sin 2t+B t\cos 2t has numerator 2A(s2+4)+B(s2−4)2A(s^2+4)+B(s^2-4). Requiring it to equal one gives 2A+B=02A+B=0, 8A−4B=18A-4B=1, hence f(t)=sin⁡2t−2tcos⁡2t16.\boxed{f(t)=\frac{\sin 2t-2t\cos 2t}{16}.} Substitution gives (s2+4)/8−(s2−4)/8=1(s^2+4)/8-(s^2-4)/8=1, verifying the numerator exactly. Inverting a product by multiplying inverses would not give this result.

Step 3: Exhibit growth, not just an envelope. The bound |f(t)|≤(1+2t)/16|f(t)|\le(1+2t)/16 is only an upper bound. Actual unboundedness follows from f(nπ)=−nπ/8→−∞.\boxed{f(n\pi)=-n\pi/8\longrightarrow-\infty.} The factor tt produced by the repeated poles causes increasing oscillation size. The graph displays several oscillations at their actual scale.

Step 4: Check onset and convergence. Taylor expansion gives sin⁡2t−2tcos⁡2t=(8/3)t3+O(t5)\sin 2t-2t\cos 2t=(8/3)t^3+O(t^5), so f(t)=t3/6+O(t5)f(t)=t^3/6+O(t^5). Since ℒ{t3/6}=s−4\mathcal L\{t^3/6\}=s^{-4}, the initial scale agrees with F(s)∼s−4F(s)\sim s^{-4}. Absolute convergence holds for s>0s>0 by the linear envelope. At s=0s=0, direct integration gives ∫0Rf(t)dt=1−cos⁡2R−Rsin⁡2R16,\int_0^R f(t)\,dt=\frac{1-\cos 2R-R\sin 2R}{16}, which has no limit. For s<0s<0, intervals near successive positive peaks of −cos⁡2t-\cos 2t have eventually positive integrand of order te|s|tt e^{|s|t}; their integrals fail the Cauchy criterion. Thus s>0\boxed{s>0} is exact.

Diagram shown in the original SVG worksheet.