The Definition — Question 7

PDF ↗

Question 7

For real ss, use F(s)=∫0∞e−stf(t)dtF(s)=\int_0^\infty e^{-st}f(t)\,dt whenever this ordinary improper integral converges.

For each integer n≥1n\ge 1, let wn=e−2n2w_n=e^{-2n^2}. Define a nonnegative signal by f(t)={en2,n≤t≤n+wn for some n≥1,0,otherwise.f(t)=\begin{cases}e^{n^2},&n\le t\le n+w_n\text{ for some }n\ge 1,\\0,&\text{otherwise}.\end{cases} A function is of exponential order if |f(t)|≤Meat|f(t)|\le Me^{at} for all sufficiently large tt, for some constants M>0M>0 and real aa.

Tasks

  1. Explain why ff is piecewise continuous on every finite interval but is not of exponential order.

  2. Prove that its Laplace integral nevertheless converges absolutely for every real ss.

  3. Express F(s)F(s) as a convergent series, treating s=0s=0 separately.

  4. Explain how the spike widths resolve the apparent conflict between arbitrarily large heights and convergence, and why the usual exponential-order theorem is not contradicted.

Original worksheet page 1: question and worked solution for 4-1-007
Show solutionHide solution

Question 7 – Solution

Strategy. The integral depends on weighted areas, not just pointwise heights. Quantify the widths before invoking a growth criterion.

Step 1: Examine local regularity and height. Since wn<1w_n<1, the spike intervals are disjoint. Any bounded time interval meets only finitely many spikes, each with finite height and finite one-sided limits. Thus ff is piecewise continuous locally. If an exponential-order bound held, evaluating at t=nt=n for large integers would give en2≤Meane^{n^2}\le Me^{an}, or n2−an≤ln⁡Mn^2-an\le\ln M. The left side tends to infinity, a contradiction.

Step 2: Estimate each weighted area. Fix real ss. On the nnth spike, t≤n+1t\le n+1 and e−st≤e|s|(n+1)e^{-st}\le e^{|s|(n+1)}. Its contribution is therefore at most en2wne|s|(n+1)=e−n2+|s|(n+1).e^{n^2}w_n e^{|s|(n+1)}=e^{-n^2+|s|(n+1)}. For sufficiently large nn, |s|(n+1)≤n2/2|s|(n+1)\le n^2/2, so these terms are at most e−n2/2≤e−n/2e^{-n^2/2}\le e^{-n/2}. The comparison series converges. Since the integrand is nonnegative, the bound proves absolute convergence for every real ss.

Step 3: Sum the exact contributions. Split finite integrals into their finitely many spike intervals, then pass to the limit using the convergent positive series. For s≠0s\ne 0, F(s)=∑n=1∞en2−sn1−e−swns.\boxed{F(s)=\sum_{n=1}^\infty e^{n^2-sn}\frac{1-e^{-sw_n}}s.} The quotient is positive even for negative ss. At s=0s=0, each contribution is simply height times width, giving F(0)=∑n=1∞e−n2.\boxed{F(0)=\sum_{n=1}^\infty e^{-n^2}.} No closed form for this convergent series is required.

Step 4: Interpret the theorem correctly. The heights grow faster than every exponential, but the widths shrink fast enough that their areas are e−n2e^{-n^2}. Any fixed exponential weight still leaves a summable sequence of areas. The common existence theorem provides sufficient conditions; exponential order is not necessary for an individual function to have a Laplace transform. This example establishes that distinction by direct estimates, rather than by trying to apply a theorem whose hypothesis fails.

Original worksheet page 2: question and worked solution for 4-1-007

Original worksheet layout. Use Enlarge or open the PDF for a closer view.