Worked examples — Chain rule — proof, composite function derivatives
Everywhere below, "the machine" language is: outer function , inner function , composite , and the rule is — derivative of the outside (inside left frozen) times derivative of the inside.
The scenario matrix
Before working anything, let's list every class of case a chain-rule problem can be. Each worked example below is tagged with the cell it fills.
| Cell | Case class | What's special / what can go wrong | Example |
|---|---|---|---|
| A | Power-of-polynomial (positive inner) | Basic peel; sign of answer follows | Ex 1 |
| B | Trig inside polynomial | Must keep inside intact () | Ex 2 |
| C | Triple / multi-nesting | One factor per layer | Ex 3 |
| D | Inside evaluates to zero at a point | Is still defined? Check the outer at | Ex 4 |
| E | Degenerate: a gear's rate at the point | Whole product collapses to | Ex 5 |
| F | Limiting / blow-up (denominator ) | Derivative ; which sign? | Ex 6 |
| G | Negative / all-sign inner (quadrant sweep) | Answer changes sign across regions | Ex 7 |
| H | Word problem — related rates | Chain rule in time; carry units | Ex 8 |
| I | Exam twist — abstract with a table | No formula for ; use given values | Ex 9 |
| J | Trig-inside-trig (non-power nesting) | Both layers are transcendental | Ex 10 |
The rest of the page fills each cell.
Cell A — power of a polynomial
Cell B — trig inside a polynomial (keep the inside intact)
Cell C — triple nesting
Forecast: how many factors will the answer have? (Hint: count the layers.)
There are three layers, peel outside-in:
-
Layer 1 (outermost): cube, .
-
Layer 2: .
-
Layer 3 (innermost): .
- Cube's derivative: . Why? Power rule, inside () frozen.
- 's derivative: . Why? Peel the next shell, its inside () frozen.
- Inner's derivative: . Why this step? The innermost gear ratio: changes times as fast as ; this becomes the last multiplied factor.
- Multiply all three factors (DOTI, thrice): Why this step? DOTI applied at each layer boundary: one factor per shell, all multiplied together.
Verify: three layers → three factors → matches forecast. At : , , so . Correct — peaks at , so slope is there. ✓
Cell D — the inside hits zero
This is the case people fear from the proof (the "" worry). Let's see it concretely.
, evaluate where the inside is . Forecast: the inside is zero at . Is still defined there? Guess yes/no.
-
Layers. Outer , inner . Why this step? Identify the shell () and filling () before peeling — the exponential is the outer machine.
-
Outer derivative: (its own derivative) . Why this step? is the unique function equal to its own derivative; we evaluate it at the inner value .
-
Inner derivative: . Why this step? Power rule on gives the inner gear ratio .
-
Multiply (DOTI): Why this step? DOTI: compound the outer rate (step 2) with the inner rate (step 3).
-
Plug in the zero-inside point : , so Why this step? This is the whole point of the cell — testing whether "inside " breaks anything. It doesn't.
" is a non-event "Inside " only means the inner value happens to equal — the outer function is perfectly smooth at (), so nothing divides by zero. (In the parent note's rigorous proof, the only genuine danger was , the tiny change in the inside, being zero — and that proof was built to survive even that. See the chain-rule proof.) So when the inner value is , you simply substitute and go — no special care needed.
Verify: numerically . ✓ And : at the inside is increasing (), so the composite rises. Sign is right.
Cell E — degenerate: a gear's rate is zero
, evaluate at . Forecast: at the inner rate vanishes — and it happens the outer rate does too. Predict .
-
Layers. Outer , inner . Why this step? Tag shell () and filling () so we can track which gear stalls.
-
Outer derivative: . Why this step? Derivative of is , evaluated at the inner value (inside frozen).
-
Inner derivative: . Why this step? Power rule on gives the inner gear ratio — this is the factor that will vanish at .
-
Multiply (DOTI): . Why this step? DOTI: outer rate times inner rate, both packaged into one product.
-
At : . Here the inner gear is zero and the outer — both factors happen to vanish, so the product is emphatically .
any gear is stationary The chain rule is a product . If either gear has rate zero, the whole composite is momentarily flat — like a gearbox where one shaft isn't turning: the output can't turn either. At the inner shaft () is frozen; that alone forces regardless of the outer. (The outer also happens to be here, which just makes the collapse doubly certain.)
Verify: . ✓ (By symmetry is even, so its slope at must be — sanity confirmed.)
Cell F — limiting / blow-up (derivative )
The symbol means " approaches from the left" — through values smaller than (like ). The little superscript minus is a direction, not a sign. Likewise means "approaching from above" — through tiny positive values. We need the one-sided version here because only exists for , so we can only sneak up on from the left.
(upper unit semicircle), behaviour near . Forecast: as approaches from below, the semicircle plunges toward the axis. Does the slope go to , a finite number, or ? Which sign?
- Layers. Outer , inner .
- Outer derivative: . Why this step? Power rule with exponent ; the is why a blow-up is possible — as this factor explodes.
- Inner derivative: . Why this step? Power rule on ; the minus sign here is what makes the final slope negative (curve falling).
- Multiply (DOTI): Why this step? DOTI: outer rate (step 2) times inner rate (step 3), then simplify.
- Limit : numerator ; denominator (a shrinking positive number). So
Figure below (s01): the upper semicircle in violet, with three dashed tangent lines at that get steeper and steeper, tilting toward vertical as we approach . The minus sign says each tangent tilts downward (curve falling); the magnitude says it becomes vertical.

Verify: at , — large and negative, heading to . ✓
Cell G — all-sign inner (a quadrant/region sweep)
; describe the sign of across all regions of . Forecast: where is the composite increasing vs decreasing? Sketch a guess with three break-points.
- Layers. Outer , inner .
- Outer derivative: . Why this step? Power rule on the cube; crucially this is a square, so this factor is always — it can touch zero but never flip the sign.
- Inner derivative: . Why this step? Power rule on ; this is the only factor that changes sign, so it controls where rises or falls.
- Multiply (DOTI): Why this step? DOTI: outer rate times inner rate; the sign-carrying factor is the inner .
- Sign analysis — cover every region. The factor never flips sign; it is at . So the sign of is the sign of :
| Region | Behaviour | |||
|---|---|---|---|---|
| decreasing | ||||
| flat (touch) | ||||
| decreasing | ||||
| flat (min) | ||||
| increasing | ||||
| flat (touch) | ||||
| increasing |
Figure below (s02): the curve in violet with the three horizontal tangents marked at (orange dots, magenta level segments); labels flag as the true minimum and as saddle-flats.

"flat but not turning" At the inside ; because the outer exponent is , the derivative carries which touches zero without changing sign. So the curve momentarily flattens (horizontal tangent) but keeps going the same way — a saddle-flat, not a max or min. The only true turning point is (a minimum), where itself changes sign.
Verify: ✓ (increasing). ✓ (decreasing). ✓.
Cell H — word problem (related rates, carry units)
radius grows at . How fast is the volume growing when ? Forecast: volume is 3D, radius is 1D — do you expect the volume rate to be bigger or smaller than ? Guess a ballpark number.
- Write the relationship. , and depends on time . Why this step? We're asked for but only know ; we first need a formula that ties to the shared variable , otherwise there is nothing to differentiate.
- Chain rule in time. With and , Why this step? This is exactly from the parent note, with , — the topic Related rates.
- Outer rate. . Why this step? Power rule on ; the collapses to — which is exactly the sphere's surface area, a handy sanity anchor.
- Assemble with values (DOTI). At : Why this step? DOTI in time: multiply the outer rate (step 3) by the inner rate (given).
- Attach units. has units (area), has , so the product has . Therefore Why this step? Units are part of the answer in a word problem — a rate of volume change must come out in volume-per-time.
Verify: units multiply correctly ( ✓). Revisiting the forecast: yes, is far bigger than — exactly as predicted, because a large sphere gains a lot of volume for each extra cm of radius. Numeric: ✓ The takeaway: related rates is just the chain rule with time as the driving variable.
Cell I — exam twist (abstract from a table, no formula)
and given only this table. Find .
Forecast: which two numbers from this table do you actually need? Circle them before computing.
-
Write the rule at the point. . Why this step? This is the whole chain rule instantiated at ; the twist is only that have no formula, so every quantity must be read off the table.
-
Find the inner value. (row ). Why this step? The outer derivative is evaluated at , so we must first learn where to look. The answer is , not — this is the #1 exam trap.
-
Read at that inner value. (row ). Why this step? We need at (from step 2), which lives in the row — not the row.
-
Read the inner rate. (row ). Why this step? The second chain-rule factor is read directly at the original point .
-
Multiply (DOTI): Why this step? DOTI: outer rate times inner rate — the only two table entries that actually matter.
Using instead of . The outer derivative is evaluated at the inner output , never at . The decoy numbers are (a function value, not a rate) and (the outer rate at the wrong point).
Verify: the two needed numbers were and ; product . Consistent with the forecast if you circled those two. ✓
Cell J — trig-inside-trig (both layers transcendental)
Forecast: neither layer is a power — both are trig. Will the answer be times something, or ? And what sign will the extra factor bring? Guess first.
-
Layers. Outer , inner . Why this step? With two trig layers it is doubly important to name them — it is easy to accidentally differentiate the inner as if it were the outer.
-
Outer derivative, inside frozen. . Why this step? Derivative of is , evaluated at the inner value — the argument stays intact inside the outer .
-
Inner derivative. . Why this step? Derivative of is ; the minus sign is the reason the final answer carries a leading negative.
-
Multiply (DOTI): Why this step? DOTI: outer rate (step 2) times inner rate (step 3).
Do not write or . The outer keeps the whole inner as its argument, and the separate inner rate is (differentiate , not ).
Verify: at : . Sanity: has zero slope at (top of the cosine wave), so the composite is flat there. ✓ At : . ✓
Recall Self-test: name the cell, then solve
Cover the worked solutions. For each, first say which matrix cell it is, then differentiate.
Differentiate . ::: Cell A: . Differentiate (watch the trap). ::: Cell B: , not . Differentiate ; how many factors? ::: Cell C: ; three layers → three factors. Is defined where the inside of is zero? Value at ? ::: Cell D: yes; . Why is for ? ::: Cell E: the inner rate at , collapsing the product. What is ? ::: Cell F: (tangent goes vertical, falling). Where does have horizontal tangents, and which is a true min? ::: Cell G: ; only is a min (others are saddle-flats). Balloon: at , ? ::: Cell H: . From the table, where ? ::: Cell I: . Differentiate . ::: Cell J: .
Connections
- Chain rule — proof, composite function derivatives — the parent; this page is its applied stress-test.
- Related rates — Cell H is a direct related-rates use of .
- Derivative — limit definition — Cells D–F lean on the limit meaning of a derivative (blow-up, zero rate).
- Product rule and Quotient rule — often combined with the chain rule in tougher exam problems.
- Implicit differentiation — the abstract- mindset of Cell I generalises here.
- Inverse function derivative and Continuity implies differentiability fails converse — related follow-ups.