Worked examples — Non-homogeneous — method of undetermined coefficients (annihilator method)
The scenario matrix
Every problem this topic throws at you lands in exactly one of these cells. The examples below are labelled by cell number so you can see the whole space is covered.
| # | Case class | What makes it tricky | Example |
|---|---|---|---|
| C1 | Pure polynomial | root ; need ; watch for already in | Ex 1 |
| C2 | Exponential, no overlap | copy the shape, simplest case | Ex 2 |
| C3 | Exponential, overlap (resonance) | root of already in → multiply by | Ex 3 |
| C4 | Trig, no overlap | must include both and | Ex 4 |
| C5 | Trig, overlap (pure resonance) | already in → multiply by ; grows forever | Ex 5 |
| C6 | Product (repeated-root annihilator) | need , not just | Ex 6 |
| C7 | Sum (superposition) | annihilate with , add the trial forms | Ex 7 |
| C8 | Word problem (real-world, with initial conditions) | translate physics → ODE, then apply the method | Ex 8 |
| C9 | Exam twist: double overlap ( appears) | root already doubled in → multiply by | Ex 9 |
| C10 | Polynomial × trig: | product needs ; carry -times-both-trig | Ex 10 |
| C11 | (damped oscillation drive) | complex roots ; annihilator | Ex 11 |
We also cover the degenerate / limiting corners inside these: a constant (Ex 1's simplest sub-case), the natural frequency (Ex 5, the resonance limit), and reducing back to a polynomial (noted in Ex 6).
C1 — Pure polynomial forcing
Step 1 — homogeneous part. Why this step? We always peel off first, because the trial for must avoid whatever already covers. , roots :
Step 2 — annihilate . Why this step? Here is the degree of the polynomial forcing term — for that degree is . A polynomial of degree is built from the root repeated times (the root gives ; doubled gives ; tripled gives ). The killer is ; with that is . Check: because differentiating a degree-2 polynomial three times wipes it out. ✓
Step 3 — bigger homogeneous ODE. Why this step? Applying to both sides turns the right side into , so now we only need to solve a homogeneous equation whose root list we can read off.
Step 4 — keep only the new terms. Why this step? Roots just rebuild and contribute nothing to . The new root (tripled) contributes . So: (This answers the forecast: you need all three, even though only shows — because and pull down lower powers.)
Step 5 — determine coefficients. Why this step? The form is fixed but the numbers are not; substituting into the original ODE pins them down. , . Substitute into : Group by power of : Match:
- : .
- : .
- : .
Step 6 — combine.
Recall Verify: plug
back. , . Then . ✓ matches .
C2 — Exponential, no overlap
Step 1 — homogeneous part. Why this step? Same reason as always: we must know before choosing the trial, to detect any overlap. (a pure imaginary pair, which produces oscillation), so .
Step 2 — annihilate . Why ? has root , so annihilates it: . ✓ And , so no overlap — no needed.
Step 3–4 — build and prune. Why this step? has roots ; the pair merely rebuilds , so only the new root can produce :
Step 5 — determine the coefficient. Why this step? Substituting into the original ODE fixes . , so
Step 6 — combine.
Recall Verify:
. , so . ✓
C3 — Exponential overlap (single resonance)
Step 1 — homogeneous part. Why this step? We need up front precisely because looks suspiciously like it might already live there. , root doubled:
Step 2 — annihilate . Why the collision matters: has root , already a double root of . Annihilator :
Step 3–4 — build and prune. Why this step? Root now has multiplicity three, giving . The first two are ; the genuinely new one is : (The double overlap forced — look at Ex 9 for the same effect on the exam.)
Step 5 — determine the coefficient. Why this step? Substitute the trial into the original ODE to pin down . With : , . Then So .
Step 6 — combine.
This is resonance in its exponential form: the forcing matches the system's own decaying mode, so the response is a bigger multiple of the same mode.
Recall Verify:
. Coefficient bracket , giving exactly . ✓
C4 — Trig, no overlap
Step 1 — homogeneous part. Why this step? We list first to check whether the drive frequency clashes with the natural frequency hidden in . , (natural frequency ).
Step 2 — annihilate . Why ? comes from roots ; the annihilator for is with . Since , no overlap.
Step 3–4 — build and prune. Why this step? has roots ; the pair rebuilds , so only the new pair matters:
Step 5 — determine coefficients. Why this step? Substitute into the original ODE and match the and parts separately. , so Match: ; .
Step 6 — combine.
Recall Verify:
. ; . ✓
C5 — Trig overlap (pure resonance) with a figure
Step 1 — homogeneous part. Why this step? We need 's frequency to see the collision: the drive is at frequency , exactly the natural one. , .
Step 2 — annihilate . Why the collision matters: has roots — exactly the roots of . Annihilator :
Step 3–4 — build and prune. Why this step? Roots now have multiplicity two, giving . The first two are ; the new ones carry the extra :
Step 5 — determine coefficients. Why this step? Substituting into the original ODE fixes ; watch the un-shifted -terms cancel — that cancellation is exactly why the extra was needed. Differentiate : Then : the and terms cancel, leaving Match: ; .
Step 6 — combine.
The figure below plots this particular solution . The horizontal axis is (time), the vertical axis is the displacement . In chalk blue is the wiggling solution itself; the two pink dashed lines are the envelope — the straight-line ceiling and floor the oscillation keeps bumping against. Because that envelope is a line through the origin with slope , the amplitude grows without bound as increases: each swing is a little taller than the last. That linear-growth envelope is the visual signature of resonance — the limiting behaviour when the drive frequency equals the natural frequency exactly.
Recall Verify:
. With : . ✓
C6 — Product (repeated-root annihilator)
Step 1 — homogeneous part. Why this step? We must know to detect that the drive's root already lives there. , roots :
Step 2 — annihilate . Why ? The factor needs root repeated times; here , so . (If were , this would collapse to the polynomial case C1 — that's the degenerate corner.) Because root is already in once, applying gives it total multiplicity :
Step 3–4 — build and prune. Why this step? Root triple → ; root → . Of these, and are . The new terms are :
Step 5 — determine coefficients. Why this step? Substituting into the original ODE fixes . To differentiate a product (where is any polynomial), the product rule gives because differentiates to itself. So each derivative just replaces the polynomial by . Here , so and: , . Then Match: ; .
Step 6 — combine.
Recall Verify:
. , so . ✓
C7 — Sum of forcings (superposition)
Step 1 — homogeneous part. Why this step? We list first so we can tell which of the two forcing pieces collides with it. , roots : .
Step 2 — annihilate using superposition. Why this step? Linearity lets us treat each summand's root independently, then multiply the annihilators. Split with (root , annihilator , no overlap) and (roots , annihilator , overlaps ). Combined annihilator , applied:
Step 3–4 — build and prune. Why this step? Roots: (new), doubled. New terms: from root ; and from the doubled the new pieces . So:
Step 5 — determine coefficients. Why this step? Substitute; linearity lets us solve the two pieces separately. Piece 1: : Piece 2: . Exactly as in C5 with , the surviving terms are
Step 6 — combine.
Recall Verify: substitute
symbolically. Piece 1: for , ; with this is . ✓ Piece 2: for , differentiate: , . So . ✓ Adding the two pieces gives . ✓
C8 — Word problem (mass–spring, real world)
Step 1 — homogeneous part. Why this step? encodes the spring's natural oscillation; comparing its frequency to the drive frequency tells us whether we hit resonance. , roots : . (Natural angular frequency .)
Step 2 — annihilate . Why no : drive has roots — no overlap, no resonance. Annihilator .
Step 3–4 — build and prune. Why this step? has roots (rebuild ) and the new ; only the new pair produces :
Step 5 — determine coefficients. Why this step? Substitute into the original ODE to fix . : So .
Step 6 — apply initial conditions. Why this step? A word problem has one specific motion; the two constants are fixed by the given start state. .
- :
- ;
Verify (units + sanity): m ✓; ✓. The response is a bounded beat between the two frequencies — no runaway growth because we're off resonance. Units: has , has (rad/s)m m/s, in m/s — consistent.
Recall Verify at
. . ✓
C9 — Exam twist: double overlap forces
Step 1 — homogeneous part. Why this step? We must count the multiplicity of the root that matches the drive — that count controls the power of . , root doubled: .
Step 2 — annihilate . Why is coming: has root , already multiplicity in . Annihilator pushes total multiplicity to :
Step 3–4 — build and prune. Why this step? Root triple → . First two are ; new term is :
Step 5 — determine the coefficient. Why this step? Substitute into the original ODE to fix ; use the product-rule shortcut from C6. , with so : , . Then So .
Step 6 — combine.
Notice the ladder: single overlap gave (Ex 3 for exponential / Ex 5 for trig), double overlap gives . The rule "multiplicity in + your annihilator ⟹ multiply trial by " is fully exposed here.
Recall Verify:
. Bracket , giving . ✓
C10 — Polynomial × trigonometric forcing
Step 1 — homogeneous part. Why this step? We check whether the trig frequency in (here ) matches 's frequency — it does, so overlap is coming. , roots : .
Step 2 — annihilate . Why ? The pattern (with ) comes from roots repeated times. The annihilator is . Check it: once kills ; a second application is needed to also kill the produced by the extra factor of .
Step 3–4 — build and prune. Why this step? Apply to the ODE, whose already contributes one factor : Roots now have multiplicity three, giving . The first pair is ; the four new terms form the trial. To keep the algebra clean we write it as times a full first-degree-polynomial-in- combination of both trig functions: (We use for the fourth coefficient to keep free as the derivative operator.)
Step 5 — determine coefficients. Why this step? Substitute into the original ODE and match the four independent shapes . Carrying out the differentiation (product rule twice) and simplifying gives Match against :
- : .
- : .
- : .
- : .
Step 6 — combine.
Recall Verify:
. Plugging into (see VERIFY block) returns exactly . ✓ Note both the (from the overlap raising multiplicity) and the mixing of into a -driven problem.
C11 — Damped-oscillation forcing
Step 1 — homogeneous part. Why this step? We list 's roots () to compare against the drive's root and check for overlap. , roots : .
Step 2 — annihilate . Why ? The pattern or comes from the complex roots ; the annihilator is . Here , giving . Its roots are different from (the real part ), so no overlap — no extra .
Step 3–4 — build and prune. Why this step? has roots (rebuild ) and the new pair . The new pair produces :
Step 5 — determine coefficients. Why this step? Substitute into the original ODE . Differentiating : Then Match: and . From the first, ; substitute: , .
Step 6 — combine.
Recall Verify:
. (no ) and (matches ). So . ✓
Wrap-up: the deciding questions
Recall Which cell forces an
in , and why? C9: the forcing root already has multiplicity in ; the annihilator raises it to , so the two lowest terms are and