Flow Lab

Syllabus / Module 6: Laplace Transforms

Lesson 6.4

Impulses & the Dirac Delta

Idealized hammer blows: the delta function as a limit, its transform, and the impulse response.

Written byClaude Sonnet

🌊
Why this matters for simulation: A wind gust modeled as a sudden kick; point vortices are delta functions of vorticity.

By the end of this lesson you can…

  • Describe the delta as the limit of unit-area pulses of height $1/\varepsilon$ and width $\varepsilon$, and say why it is a rule for integrals and not a function you can plot.
  • Use the sifting property $\int f(t)\,\delta(t-a)\,dt = f(a)$ and derive $\mathcal L\{\delta(t-a)\} = e^{-as}$ as a limit of pulse transforms.
  • Derive the momentum jump $x'(0^+) = J/m$ by integrating the equation of motion across a short pulse, and use it to set up a hammer blow as an initial condition.
  • Find the impulse response $h(t)$ of a spring-mass-damper as the inverse transform of $1/P(s)$, and relate it to the Green's function of variation of parameters.
  • Write the response to repeated blows or a gust as a sum of shifted impulse responses.
  • Explain why a delta cannot be sampled on a grid, measure the error of a width-$\varepsilon$ pulse stand-in, and hand an impulse to a solver as a velocity jump instead.
On this page
·
  1. From a pulse to a delta
  2. Its Laplace transform
  3. The delta and the step
  4. What a hammer blow does to a spring
  5. The impulse response
  6. Narrowing the pulse
  7. Repeated blows and gusts
  8. A delta cannot be sampled
  9. Computing it
  10. Concept checks
  11. Problems
  12. Key Takeaways

In 6.3 Step Functions & Discontinuous Forcing a force switched on and off, and the shorter we made the pulse the less the solver cared. Now push that to the limit: a pulse so short that all we can still say about it is how hard it hit. A hammer on a spring, a bat on a ball, a gust that lasts a blink, a vortex squeezed into a point. The mathematical object is the Dirac delta, and its payoff is large: the response to a unit hammer blow, the impulse response, tells you the response to any force, which is the idea behind the Green’s function of 5.5 Variation of Parameters and the transfer function of the lesson after this one. We keep the spring-mass-damper of 5.3 Mechanical Vibrations: Spring–Mass–Damper: m=2m = 2, c=8c = 8, k=200k = 200.

▶ Video Pulses narrowing toward a delta, the sifting property, the transform e to the minus a s, a hammer blow on a spring, a train of blows, and why a grid cannot hold a deltacaptions only
Transcript
  1. A hammer blow, a bat on a ball, a gust that lasts a blink. Take a pulse of width epsilon and height one over epsilon, so its area is one, and squeeze it. It gets taller and thinner, the area stays one, and the limit is the Dirac delta. It is not a function.
  2. What converges is the integral against a smooth function. For cosine at a equals 1, the averages over width epsilon are 0.452841, 0.497364, 0.519045, 0.529728, heading for cosine of 1, 0.540302. That is sifting: delta picks out f of a.
  3. At a, the pulse transform is e to the minus a s times one minus e to the minus epsilon s, over epsilon s. The fraction tends to one, so the transform of delta is e to the minus a s. At s equals 2 the ratio climbs 0.8242, 0.906346, 0.951626, 0.975412.
  4. Now the hammer. Integrate the equation of motion across a short pulse. The damping and spring terms shrink with epsilon, but J does not. So the position stays at zero and the momentum jumps: x prime of 0 plus is J over m, 0.5 meters per second.
  5. A real pulse delivers a little less, because the spring and damper take a share. At epsilon 0.025 the three terms are 0.941746, 0.048124 and 0.010130, and they add to the impulse. The end velocities, 0.158271 up to 0.486437, close in on 0.5.
  6. After the blow the mass rings freely. In the transform domain, P of s times X equals J. Complete the square, alpha 2, beta 9.79796: h of t is 0.051031 e to the minus 2 t times sine of 9.79796 t, peaking at 0.037807. It is the Green's function.
  7. As epsilon shrinks, the pulse response collapses onto h. The largest gaps are 2.54 times 10 to the minus 2, 1.74 times 10 to the minus 2, 1.05 times 10 to the minus 2, then 5.78 and 3.01 times 10 to the minus 3. The ratios creep toward 2: first order in epsilon.
  8. The mechanism is a delay. A pulse delivers its momentum on average epsilon over 2 late, so the response is h of t minus epsilon over 2. The steepest slope of h is 0.5, so the gap is about epsilon over 4: 3.13 times 10 to the minus 3 at epsilon 0.0125, against 3.01 observed.
  9. Blows add. Three unit blows, one damped period T apart, 0.641275 seconds, ring in phase: a sum of shifted impulse responses. The peak is 0.051199, 1.354 times one blow, far below 3 because of damping. A gust is many small kicks: a convolution with h.
  10. Now the grid. A spike narrower than the step is seen only by stages that land inside it. With h 0.05 and epsilon 0.025, only the first stage does, with weight h over 6: one third of the impulse. The gap is 2.51 times 10 to the minus 2, and refining the grid does not fix it cleanly.
  11. The fix: do not integrate the spike. Start at x equals zero with velocity J over m and integrate the unforced equation. The gaps at h 0.05, 0.02, 0.01 are 5.12 times 10 to the minus 5, 1.28 times 10 to the minus 6, 7.9 times 10 to the minus 8: fourth order.
  12. The same idea in space: a point vortex is a delta of vorticity, and from memory its swirl speed falls as gamma over 2 pi r. It blows up at the center like the pulse height, so codes smooth it into a small core. An impulse is a change of state.
  13. So: the delta is the limit of unit-area pulses, defined by sifting. Its transform is e to the minus a s. A blow jumps the velocity by J over m, and h is the inverse of one over P of s. Blows add as shifted copies. Next: convolution and the transfer function.

From a pulse to a delta

Take a rectangular pulse of width ε\varepsilon and height 1/ε1/\varepsilon, so its area is 11 (the steps of the last lesson build it: [u(t)−u(t−ε)]/ε[u(t) - u(t-\varepsilon)]/\varepsilon). Squeeze ε\varepsilon and the pulse gets taller and thinner while the area stays 11. The Dirac delta δ(t)\delta(t) is what these pulses approach, but they approach it only in one sense: pointwise the heights blow up to infinity at t=0t = 0 and the limit is not a function. What converges is the integral against a test function. For smooth ff and a pulse placed at t=at = a,

∫−∞∞f(t) pε(t−a) dt=1ε∫aa+εf(t) dt=average of f over [a,a+ε]  ⟶  f(a).\int_{-\infty}^{\infty} f(t)\,p_\varepsilon(t-a)\,dt = \frac1\varepsilon\int_a^{a+\varepsilon} f(t)\,dt = \text{average of } f \text{ over } [a, a+\varepsilon] \;\longrightarrow\; f(a) .

That limit is the sifting property, and it is the entire definition: ∫f(t) δ(t−a) dt=f(a)\int f(t)\,\delta(t-a)\,dt = f(a). So the delta is a thing that only makes sense under an integral. Read δ\delta as a shorthand for “take the limit of the pulses after integrating”. Two more properties follow from the same picture. Its area is 11 (take f=1f = 1). And δ(ct)=δ(t)/∣c∣\delta(ct) = \delta(t)/|c|: a pulse in ctct of width ε\varepsilon is a pulse in tt of width ε/∣c∣\varepsilon/|c|, so for the same height it has 1/∣c∣1/|c| of the area.

How fast do the averages converge? For f=cos⁡tf = \cos t and a=1a = 1 the exact value is f(1)=0.540302f(1) = 0.540302. The averages over [1,1+ε][1, 1+\varepsilon] are 0.4528410.452841, 0.4973640.497364, 0.5190450.519045, 0.5297280.529728 for ε=0.2,0.1,0.05,0.025\varepsilon = 0.2, 0.1, 0.05, 0.025. The errors are 0.08750.0875, 0.04290.0429, 0.02130.0213, 0.01060.0106: they halve with ε\varepsilon, so the error is first order in the width. The reason is the Taylor expansion f(t)≈f(a)+f′(a)(t−a)f(t) \approx f(a) + f'(a)(t-a), whose average over the window is f(a)+f′(a)ε/2f(a) + f'(a)\varepsilon/2; here ∣f′(1)∣ε/2=0.42 ε|f'(1)|\varepsilon/2 = 0.42\,\varepsilon. Block A of the lab cell below prints these numbers, and you will meet the same first-order error again in the numerics.

Its Laplace transform

Place the pulse at t=at = a: p(t)=[u(t−a)−u(t−a−ε)]/εp(t) = [u(t-a) - u(t-a-\varepsilon)]/\varepsilon. The step lesson gives the transform at once,

L{p}=1ε e−as−e−(a+ε)ss=e−as 1−e−εsεs.\mathcal L\{p\} = \frac{1}{\varepsilon}\,\frac{e^{-as} - e^{-(a+\varepsilon)s}}{s} = e^{-as}\,\frac{1 - e^{-\varepsilon s}}{\varepsilon s} .

For small εs\varepsilon s, e−εs=1−εs+⋯e^{-\varepsilon s} = 1 - \varepsilon s + \cdots, so the fraction tends to 11 and

L{δ(t−a)}=e−as,L{δ(t)}=1.\mathcal L\{\delta(t-a)\} = e^{-as}, \qquad \mathcal L\{\delta(t)\} = 1 .

This is just the sifting property with f(t)=e−stf(t) = e^{-st}. The case a=0a = 0 is the one to be careful with: the pulse starts exactly at the lower limit of the integral, so we agree that the transform integral starts at 0−0^-, just before the blow, and counts all of the delta. Block B integrates the definition numerically for a pulse at a=0.5a = 0.5 with s=2s = 2 and divides by e−ase^{-as}; the ratio is 1−e−2ε2ε=0.824200, 0.906346, 0.951626, 0.975412\dfrac{1 - e^{-2\varepsilon}}{2\varepsilon} = 0.824200,\ 0.906346,\ 0.951626,\ 0.975412 for ε=0.2,0.1,0.05,0.025\varepsilon = 0.2, 0.1, 0.05, 0.025, climbing to 11 with a gap that halves each time, the same first-order behaviour. The sim reports the same ratio as its "L{pulse}(s=2)\mathcal L\{\text{pulse}\}(s{=}2)" line.

The delta and the step

The step u(t−a)u(t-a) jumps from 00 to 11, and the pulse is the step’s smoothed-out slope: if the jump is spread over a ramp of width ε\varepsilon, the slope on the ramp is 1/ε1/\varepsilon for ε\varepsilon seconds, which is exactly pεp_\varepsilon. So one says u′=δu' = \delta. Be honest about what this means: a true step has no derivative at the jump, and δ\delta is not a function, so the statement is shorthand for the limit above. It can be made precise in the only way that works for deltas, under an integral, by integration by parts. For a window [0,T][0, T] containing aa,

∫0Tf(t) u′(t−a) dt=[f u]0T−∫0Tf′(t) u(t−a) dt=f(T)−∫aTf′(t) dt=f(a),\int_0^T f(t)\,u'(t-a)\,dt = \bigl[f\,u\bigr]_0^T - \int_0^T f'(t)\,u(t-a)\,dt = f(T) - \int_a^T f'(t)\,dt = f(a) ,

which is the sifting property again. It also agrees with the transform: L{u′}=s⋅1s−u(0−)=1=L{δ}\mathcal L\{u'\} = s\cdot\tfrac{1}{s} - u(0^-) = 1 = \mathcal L\{\delta\}. So the derivative rule from 6.1 The Laplace Transform keeps working across jumps, provided we use the value just before t=0t = 0. This is the practical meaning of u′=δu' = \delta: differentiating a signal with a jump, in the transform world, produces a delta.

What a hammer blow does to a spring

A hammer delivers an impulse J=∫F dtJ = \int F\,dt (units N s, the momentum given to the mass). Model it as a pulse of height J/εJ/\varepsilon on [0,ε][0, \varepsilon] and integrate mx′′+cx′+kx=F(t)m x'' + c x' + k x = F(t) across it:

m[x′(ε)−x′(0)]+c[x(ε)−x(0)]+k∫0εx dt=J.m\bigl[x'(\varepsilon) - x'(0)\bigr] + c\bigl[x(\varepsilon) - x(0)\bigr] + k\int_0^\varepsilon x\,dt = J .

Over a time ε\varepsilon the position cannot move by much: xx changes by about ε x′\varepsilon\,x', so the cc term and the kk integral both shrink with ε\varepsilon while JJ does not. In the limit only the first term survives:

x(0+)=0,m x′(0+)=J⟹x′(0+)=J/m.x(0^+) = 0, \qquad m\,x'(0^+) = J \quad\Longrightarrow\quad x'(0^+) = J/m .

The blow does not move the mass, it changes its momentum instantly. For J=1J = 1 N s and m=2m = 2 kg that is x′(0+)=0.5x'(0^+) = 0.5 m/s. Block C checks the balance at ε=0.025\varepsilon = 0.025: m x′(ε)=0.941746m\,x'(\varepsilon) = 0.941746, c x(ε)=0.048124c\,x(\varepsilon) = 0.048124, k∫0εx dt=0.010130k\int_0^\varepsilon x\,dt = 0.010130, and these add to 1.0000001.000000, so x′(ε)x'(\varepsilon) is 0.47090.4709 rather than 0.50.5 because the spring and damper took a small share. As the pulse narrows, x′(ε)x'(\varepsilon) is 0.1582710.158271, 0.3469400.346940, 0.4345380.434538, 0.4708730.470873, 0.4864370.486437 for ε=0.2,0.1,0.05,0.025,0.0125\varepsilon = 0.2, 0.1, 0.05, 0.025, 0.0125, closing in on 0.50.5. The sim prints this at the end of the pulse for any width.

The impulse response

After the blow nothing pushes: the mass rings freely from the state (0,J/m)(0, J/m). In the transform domain the same story is one line. With F=Jδ(t)F = J\delta(t) and rest initial conditions, P(s)X=JP(s)X = J where P(s)=ms2+cs+kP(s) = ms^2 + cs + k, so

X(s)=Jms2+cs+k,h(t)=L−1{1P(s)}for J=1.X(s) = \frac{J}{ms^2 + cs + k}, \qquad h(t) = \mathcal L^{-1}\Bigl\{\frac{1}{P(s)}\Bigr\} \quad\text{for } J = 1 .

The function hh is the impulse response, and 1/P(s)1/P(s) is the transfer function. Complete the square: ms2+cs+k=m[(s+α)2+β2]ms^2 + cs + k = m[(s+\alpha)^2 + \beta^2] with α=c/2m=2\alpha = c/2m = 2 and β=k/m−α2=9.79796\beta = \sqrt{k/m - \alpha^2} = 9.79796, so

h(t)=1mβ e−αtsin⁡βt=0.051031  e−2tsin⁡(9.79796 t),t>0.h(t) = \frac{1}{m\beta}\,e^{-\alpha t}\sin\beta t = 0.051031\;e^{-2t}\sin(9.79796\,t), \qquad t \gt 0 .

Check the mechanism. This is the free vibration from the previous modules with x(0)=0x(0) = 0 and x′(0)=1/mx'(0) = 1/m: its sine coefficient is x′(0)/β=1/(mβ)x'(0)/\beta = 1/(m\beta), as above. It starts at 00 with slope 0.50.5, peaks at t=arctan⁡(β/α)/β=0.13977t = \arctan(\beta/\alpha)/\beta = 0.13977 s at h=0.037807h = 0.037807 m, and by t=1t = 1 has swung through to −0.002518-0.002518. The two derivations agree because they are the same statement: a delta force and an initial velocity jump are the same physics.

Why this equals the Green’s function. Variation of parameters in 5.5 Variation of Parameters built a particular solution from a kernel G(t,τ)G(t, \tau) (written there for the standard form, with the equation divided by mm). Think of the force as a stack of tiny blows, one at each time τ\tau of strength F(τ) dτF(\tau)\,d\tau. The blow at τ\tau rings as F(τ) dτ  h(t−τ)F(\tau)\,d\tau\;h(t-\tau) afterwards, so the system’s total response is the sum,

x(t)=∫0th(t−τ) F(τ) dτ,i.e. G(t,τ)=m h(t−τ).x(t) = \int_0^t h(t-\tau)\,F(\tau)\,d\tau , \qquad\text{i.e. } G(t,\tau) = m\,h(t - \tau) .

For constant coefficients the Green’s function depends only on the elapsed time t−τt - \tau. This integral is the convolution, and turning it into the product H(s)F(s)H(s)F(s) is the next lesson, 6.5 Convolution & Transfer Functions.

Narrowing the pulse

◆ Interactive Narrow a unit-area pulse and watch the spring's response collapse onto the impulse response

Loading simulation…

The sim fires a pulse of unit area, 1/ε1/\varepsilon high and ε\varepsilon wide, at the spring from rest. The thin curve is the exact response to that pulse, built from two step responses. The thick curve is h(t)h(t). Read the “halving table”. At c=8c = 8 the largest gap max⁡t∣xε−h∣\max_t |x_\varepsilon - h| over [0,4][0, 4] is 2.54×10−22.54\times10^{-2}, 1.74×10−21.74\times10^{-2}, 1.05×10−21.05\times10^{-2}, 5.78×10−35.78\times10^{-3}, 3.01×10−33.01\times10^{-3} for ε=0.2,0.1,0.05,0.025,0.0125\varepsilon = 0.2, 0.1, 0.05, 0.025, 0.0125. The observed ratios between successive rows are 1.471.47, 1.651.65, 1.821.82, 1.921.92, creeping toward 22 (the next rows give 1.961.96, 1.981.98, 1.991.99): the error is proportional to ε\varepsilon, as with the sifting averages, but only once ε\varepsilon is small against the oscillation period of 0.640.64 s.

The mechanism is a delay. A pulse of width ε\varepsilon delivers its momentum on average ε/2\varepsilon/2 after the start, so xε(t)≈h(t−ε/2)x_\varepsilon(t) \approx h(t - \varepsilon/2) and the gap is about (ε/2)max⁡∣h′∣(\varepsilon/2)\max|h'|. The steepest slope of hh is its initial slope 1/m=0.51/m = 0.5, so the prediction is ε/4\varepsilon/4: at ε=0.0125\varepsilon = 0.0125 that is 3.13×10−33.13\times10^{-3} against the observed 3.01×10−33.01\times10^{-3}.

Repeated blows and gusts

A system is linear, so blows add. If blows JiJ_i land at times tit_i, then f=∑iJi δ(t−ti)f = \sum_i J_i\,\delta(t - t_i) and

x(t)=∑iJi h(t−ti)(h=0 for negative argument).x(t) = \sum_i J_i\,h(t - t_i) \qquad (h = 0 \text{ for negative argument}) .

Each blow starts its own ringing, with no memory of the others. Try three unit blows spaced by exactly one damped period T=2π/β=0.641275T = 2\pi/\beta = 0.641275 s. After the third, every ringing is in phase and sums to h(τ) (1+q+q2)h(\tau)\,(1 + q + q^2) with q=e−αTq = e^{-\alpha T}, because sin⁡β(τ+T)=sin⁡βτ\sin\beta(\tau + T) = \sin\beta\tau and each earlier blow has had TT longer to decay. The largest displacement is then 0.0511990.051199 m, which is 1.3541.354 times the single-blow peak of 0.0378070.037807; blow-for-blow tapping in rhythm builds up, but damping caps the build-up far below the 3×3\times a lossless system would give. Block D prints these numbers and also integrates the ODE with each blow applied as a velocity jump J/mJ/m; the two agree to 4×10−124\times10^{-12}.

A delta cannot be sampled

A grid stores values, not integrals, and the area of a pulse is an integral. Hand a spike of height 1/ε1/\varepsilon to a time stepper and three things can go wrong, all visible in the sim’s “spike” mode, where the force 1/ε1/\varepsilon for t<εt \lt \varepsilon is evaluated at RK4’s stage times. If ε\varepsilon is smaller than a step, the stages that land inside the pulse carry the whole weight, and the delivered momentum depends on how many stages happened to land there, not on JJ. Take h=0.05h = 0.05 and ε=0.025\varepsilon = 0.025: only the first stage, at t=0t = 0, sees the pulse, and RK4 gives it weight h/6h/6, so the solver delivers h/(6ε)=1/3h/(6\varepsilon) = 1/3 of the intended impulse. The response is then 2/32/3 short, and since 23×0.037807=0.0252\tfrac23\times0.037807 = 0.0252, we expect a gap of about 0.02520.0252. The sim shows 2.51×10−22.51\times10^{-2}, where an exact pulse of this width would give 5.78×10−35.78\times10^{-3}. Refining the grid does not repair this smoothly: the spike gaps at h=0.1,0.05,0.025,0.01h = 0.1, 0.05, 0.025, 0.01 for ε=0.025\varepsilon = 0.025 are 1.50×10−21.50\times10^{-2}, 2.51×10−22.51\times10^{-2}, 7.32×10−37.32\times10^{-3} and 6.62×10−36.62\times10^{-3}, not a clean decreasing sequence.

The fix is the one in the step lesson, taken to its conclusion. Do not integrate the spike. Use the momentum jump we derived: start at x=0x = 0, x′=J/mx' = J/m and integrate the unforced ODE. This is exact in the delta limit, so the only error left is the solver’s own. At h=0.05,0.02,0.01h = 0.05, 0.02, 0.01 the gap to h(t)h(t) is 5.12×10−55.12\times10^{-5}, 1.28×10−61.28\times10^{-6} and 7.9×10−87.9\times10^{-8}. The ratios are 39.939.9 for a step ratio of 2.52.5 (and 2.54=39.12.5^4 = 39.1) and 16.216.2 for halving: fourth order, which is what RK4 promises. Switch the sim to “kick” and watch the dots land on hh.

Computing it

🐍 PythonLab: sifting, the pulse transform, and the momentum balance across the pulse (blocks A to C)
⌘/Ctrl + Enter

Block A prints the averages and errors quoted above, block B the ratios 0.824200…0.9754120.824200\ldots0.975412, and block C the momentum balance (and x′(ε)=0.486437x'(\varepsilon) = 0.486437 at ε=0.0125\varepsilon = 0.0125, where the balance sums to 1.0000001.000000 again).

🐍 PythonLab: a train of blows as a sum of shifted impulse responses, against the ODE with velocity jumps (block D)
⌘/Ctrl + Enter

Block D prints T=0.641275T = 0.641275, 1+q+q21 + q + q^2, the peaks 0.0378070.037807 and 0.0511990.051199 and the 4×10−124\times10^{-12} agreement. Now do the exercise: simulate a narrow pulse, find hh symbolically from its transform, and compare.

🧪 Coding exerciseExercise: impulse response from a narrow pulse against the inverse transform of 1/P(s)
⌘/Ctrl + Enter

Concept checks

✅ Concept check

What is $\mathcal L\{3\,\delta(t-2)\}$ at $s = 0.5$? (Give a number.)

✅ Concept check

A hammer gives a $0.5$ kg mass on a spring an impulse of $3$ N s while it is at rest at the equilibrium position. What is $x'(0^+)$ in m/s? (Give a number.)

✅ Concept check

You simulate a hammer blow by forcing an ODE with a spike of height $1/\varepsilon$ and width $\varepsilon$ smaller than the step $h$, sampled at the solver's stage times. What is the most reliable fix?

✅ Concept check

For the pulse of width $\varepsilon$ on the spring, the gap to the impulse response is about $\varepsilon/4$. What does that predict for the ratio of gaps when $\varepsilon$ is halved?

Problems

✏️ Pen & paperA kick on a first-order system●○○

Solve y′+2y=δ(t−1)y' + 2y = \delta(t-1) with y(0)=0y(0) = 0 and find y(2)y(2).

Show worked solution

Transform: (s+2)Y=e−s(s+2)Y = e^{-s}, so Y=e−s/(s+2)Y = e^{-s}/(s+2) and y(t)=u(t−1) e−2(t−1)y(t) = u(t-1)\,e^{-2(t-1)}. Before t=1t = 1, y=0y = 0. At t=1+t = 1^+ the delta has made yy jump from 00 to 11, the area of the delta divided by the coefficient of y′y', and after that it decays freely. y(2)=e−2=0.135335y(2) = e^{-2} = 0.135335.

✏️ Pen & paperCancelling a ring with a second blow●●○

An undamped oscillator x′′+4x=δ(t)+J2 δ(t−a)x'' + 4x = \delta(t) + J_2\,\delta(t - a) starts at rest. Find J2J_2 and aa so that the mass is again at rest after the second blow, and say why the timing works.

Show worked solution

The impulse response is h(t)=12sin⁡2th(t) = \tfrac12\sin 2t (inverse of 1/(s2+4)1/(s^2+4)), so x=12sin⁡2t+J22sin⁡2(t−a)x = \tfrac12\sin 2t + \tfrac{J_2}2\sin 2(t-a) for t>at \gt a. Choose a=π/2a = \pi/2, half the period π\pi of the oscillator. Then sin⁡2(t−π/2)=sin⁡(2t−π)=−sin⁡2t\sin 2(t - \pi/2) = \sin(2t - \pi) = -\sin 2t, and J2=1J_2 = 1 gives x≡0x \equiv 0 afterwards. Mechanism: half a period after the first blow the mass is back at x=0x = 0 moving with velocity x′=cos⁡π=−1x' = \cos\pi = -1; the second blow adds +1+1 to the velocity and stops it exactly at the origin. Any other timing leaves some ringing behind.

✏️ Pen & paperImpulse response of an underdamped system●●○

For x′′+2x′+5x=δ(t)x'' + 2x' + 5x = \delta(t) with rest initial conditions, find h(t)h(t), the time of its first peak, and the ratio of consecutive peaks.

Show worked solution

s2+2s+5=(s+1)2+4s^2 + 2s + 5 = (s+1)^2 + 4, so h(t)=12 e−tsin⁡2th(t) = \tfrac12\,e^{-t}\sin 2t. The first peak is where h′=0h' = 0: −sin⁡2t+2cos⁡2t=0-\sin 2t + 2\cos 2t = 0, i.e. tan⁡2t=2\tan 2t = 2, so t=12arctan⁡2=0.5536t = \tfrac12\arctan 2 = 0.5536 and h=0.2571h = 0.2571. Successive peaks of the same sign are one damped period T=2π/β=πT = 2\pi/\beta = \pi apart and differ by the factor e−αT=e−π=0.0432e^{-\alpha T} = e^{-\pi} = 0.0432. Successive extremes of either sign are π/2\pi/2 apart, with ratio e−π/2=0.2079e^{-\pi/2} = 0.2079.

✏️ Pen & paperSifting by hand●●●

Use a pulse of width ε\varepsilon at t=at = a to show ∫t2 δ(t−a) dt=a2\int t^2\,\delta(t-a)\,dt = a^2, and give the size of the error for finite ε\varepsilon.

Show worked solution

1ε∫aa+εt2 dt=(a+ε)3−a33ε=a2+aε+ε23\dfrac1\varepsilon\int_a^{a+\varepsilon} t^2\,dt = \dfrac{(a+\varepsilon)^3 - a^3}{3\varepsilon} = a^2 + a\varepsilon + \dfrac{\varepsilon^2}{3}. As ε→0\varepsilon \to 0 this tends to a2a^2, and the error aε+ε2/3a\varepsilon + \varepsilon^2/3 is first order in ε\varepsilon with slope 2a⋅12=a2a\cdot\tfrac12 = a: the Taylor estimate f′(a)ε/2f'(a)\varepsilon/2 is exactly the first term.

Key Takeaways

  • The delta is the limit of unit-area pulses of height 1/ε1/\varepsilon and width ε\varepsilon. It is not a function: it only has meaning under an integral, where ∫f δ(t−a) dt=f(a)\int f\,\delta(t-a)\,dt = f(a) (sifting).
  • L{δ(t−a)}=e−as\mathcal L\{\delta(t-a)\} = e^{-as}, the limit of e−as(1−e−εs)/(εs)e^{-as}(1 - e^{-\varepsilon s})/(\varepsilon s); the transform integral starts at 0−0^- so a delta at 00 counts fully.
  • u′=δu' = \delta is shorthand: a jump differentiates to a delta, made precise by integration by parts under an integral.
  • A hammer blow of impulse JJ leaves the position alone and jumps the velocity by J/mJ/m: integrate the ODE across the pulse and the cc and kk terms vanish with ε\varepsilon.
  • The impulse response is h=L−1{1/P(s)}h = \mathcal L^{-1}\{1/P(s)\}; for the spring it is e−αtsin⁡(βt)/(mβ)e^{-\alpha t}\sin(\beta t)/(m\beta), the free vibration from (0,1/m)(0, 1/m), and it is the Green’s function, G(t,τ)=m h(t−τ)G(t,\tau) = m\,h(t-\tau) in the standard form.
  • Repeated blows or a gust give x=∑Ji h(t−ti)x = \sum J_i\,h(t-t_i) (or ∫h(t−τ)F(τ) dτ\int h(t-\tau)F(\tau)\,d\tau): superposition of shifted impulse responses.
  • A width-ε\varepsilon pulse stands in for the delta with an error proportional to ε\varepsilon (observed halving ratios approach 2), but a grid cannot sample the spike reliably, so apply impulses as state jumps.
  • Next, 6.5 Convolution & Transfer Functions turns the integral ∫h(t−τ)F(τ) dτ\int h(t-\tau)F(\tau)\,d\tau into the product H(s)F(s)H(s)F(s).
Finished this lesson?

Mark it complete to track your progress (saved in this browser). Reading every section does this for you.

Where this leads

These later lessons build directly on this one: