This page is a drill hall . The parent note built the machinery (secant → limit → tangent). Here we throw every kind of input at that machinery so no exam question can surprise you. We compute everything from the difference quotient
f ′ ( a ) = lim h → 0 h f ( a + h ) − f ( a )
and never skip the "why". Read the scenario matrix first, then hunt each cell.
Every derivative-at-a-point question is one of these case classes . The last column tells you which worked example below nails it.
#
Case class
What makes it tricky
Example
A
Positive slope (curve rising)
baseline sanity check
Ex 1
B
Negative slope (curve falling)
sign of the answer must come out < 0
Ex 2
C
Zero slope (flat tangent, a peak/valley)
answer is exactly 0
Ex 3
D
Denominator function 1/ x
the h hides inside a fraction
Ex 4
E
Root function x
must rationalise, not just expand
Ex 5
F
Word / rate problem with units
interpret sign + attach units
Ex 6
G
Degenerate: corner ∣ x ∣
limit fails, no derivative
Ex 7
H
Degenerate: vertical tangent x 1/3
limit is + ∞ , no finite slope
Ex 8
I
Exam twist : tangent parallel to a given line
solve f ′ ( a ) = m for a
Ex 9
Prerequisites you may want open: Average rate of change (the secant slope), Limits — formal definition (the engine), and Velocity and acceleration (for the word problem).
f ( x ) = x 2 − 2 x at x = 3
Forecast: the parabola y = x 2 − 2 x opens up and x = 3 is to the right of its lowest point, so the curve is rising . Guess: a positive number.
Step 1. Form the difference quotient.
h f ( 3 + h ) − f ( 3 ) = h [ ( 3 + h ) 2 − 2 ( 3 + h ) ] − [ 9 − 6 ]
Why this step? This is the honest slope of the secant through two real points — no division by zero yet.
Step 2. Expand the top: ( 3 + h ) 2 = 9 + 6 h + h 2 and − 2 ( 3 + h ) = − 6 − 2 h , so the bracket is 9 + 6 h + h 2 − 6 − 2 h = 3 + 4 h + h 2 . Subtract f ( 3 ) = 3 :
numerator = 4 h + h 2 .
Why this step? We need the 0 0 trouble (h in the bottom) to be matched by an h on top so we can cancel.
Step 3. Cancel and take the limit.
h 4 h + h 2 = 4 + h ( h = 0 ) , lim h → 0 ( 4 + h ) = 4.
Why this step? Cancelling kills the illegal 0/0 before we let h → 0 — "Simplify Before Limit".
So f ′ ( 3 ) = 4 .
Verify: positive, as forecast. Sanity check with a tiny secant h = 0.001 : 4 + 0.001 = 4.001 ≈ 4 . ✓ Tangent line: y − 3 = 4 ( x − 3 ) .
f ( x ) = 6 − x 2 at x = 1
Forecast: y = 6 − x 2 is an upside-down parabola; right of its peak at x = 0 it falls . Guess: a negative number.
Step 1. Difference quotient:
h [ 6 − ( 1 + h ) 2 ] − [ 6 − 1 ] .
Why this step? Same secant idea — two points, honest rise over run.
Step 2. Top = 6 − ( 1 + 2 h + h 2 ) − 5 = 6 − 1 − 2 h − h 2 − 5 = − 2 h − h 2 .
Why this step? Watch the minus sign distribute onto every term of ( 1 + h ) 2 — that is where the negative answer is born.
Step 3. h − 2 h − h 2 = − 2 − h , and h → 0 lim ( − 2 − h ) = − 2 .
So f ′ ( 1 ) = − 2 .
Verify: negative, as forecast. Slope − 2 means the curve drops 2 units of height per 1 unit right. Tiny secant h = 0.001 : − 2 − 0.001 = − 2.001 ≈ − 2 . ✓
f ( x ) = x 2 − 4 x + 5 at its vertex x = 2
Forecast: x = 2 is the bottom of this upward parabola. At the very bottom the curve is momentarily flat . Guess: exactly 0 .
Step 1. f ( 2 ) = 4 − 8 + 5 = 1 . Difference quotient:
h [ ( 2 + h ) 2 − 4 ( 2 + h ) + 5 ] − 1 .
Why this step? We compute the anchor value f ( 2 ) first so the numerator is fully numeric-plus-h .
Step 2. Top = ( 4 + 4 h + h 2 ) − ( 8 + 4 h ) + 5 − 1 = 4 + 4 h + h 2 − 8 − 4 h + 5 − 1 = h 2 .
Why this step? The 4 h from the square and the − 4 h from the linear term cancel — a hallmark of being at the vertex. Only h 2 survives.
Step 3. h h 2 = h , and h → 0 lim h = 0 . So f ′ ( 2 ) = 0 .
Verify: zero, as forecast — the tangent is horizontal, y = 1 . This is exactly why a flat tangent flags a maximum or minimum.
f ( x ) = x 1 at x = 2
Forecast: 1/ x is a decreasing curve for x > 0 (bigger x → smaller output), so we expect a negative slope, small in size.
Step 1. Difference quotient:
h 2 + h 1 − 2 1 .
Why this step? Standard secant, but the h is now buried inside a fraction , so we can't cancel yet — we must combine first.
Step 2. Combine the top fractions over the common denominator 2 ( 2 + h ) :
2 + h 1 − 2 1 = 2 ( 2 + h ) 2 − ( 2 + h ) = 2 ( 2 + h ) − h .
Why this step? This manufactures an h on top to match the h underneath. Combining fractions is the only tool that reaches a buried h .
Step 3. Divide by h (i.e. multiply by 1/ h ):
2 ( 2 + h ) − h ⋅ h 1 = 2 ( 2 + h ) − 1 , lim h → 0 2 ( 2 + h ) − 1 = 4 − 1 .
Why this step? Now the h cancels cleanly and plugging h = 0 is legal. So f ′ ( 2 ) = − 4 1 .
Verify: negative and small, as forecast. Tiny secant h = 0.001 : 0.001 1/2.001 − 1/2 ≈ − 0.2499 ≈ − 0.25 . ✓
f ( x ) = x at x = 9
Forecast: x rises but flattens as x grows. At x = 9 it should be a small positive number.
Step 1. Difference quotient:
h 9 + h − 9 = h 9 + h − 3 .
Why this step? Expanding won't help a root; the h is trapped under the radical.
Step 2. Multiply top and bottom by the conjugate 9 + h + 3 :
h 9 + h − 3 ⋅ 9 + h + 3 9 + h + 3 = h ( 9 + h + 3 ) ( 9 + h ) − 9 = h ( 9 + h + 3 ) h .
Why this step? ( A − B ) ( A + B ) = A − B turns the difference of roots into a plain difference, freeing an h on top.
Step 3. Cancel h and take the limit:
9 + h + 3 1 h → 0 9 + 3 1 = 6 1 .
So f ′ ( 9 ) = 6 1 .
Verify: small positive, as forecast. Tiny secant h = 0.001 : 0.001 9.001 − 3 ≈ 0.16666 ≈ 6 1 . ✓ (This is the Power Rule result 2 1 x − 1/2 at x = 9 in disguise.)
Worked example A ball is thrown up; height
s ( t ) = 20 t − 5 t 2 metres, t in seconds. Find its velocity at t = 1 s and at t = 3 s.
Forecast: the ball rises then falls. At t = 1 s (still climbing) velocity should be positive ; by t = 3 s (past the top) it should be negative . Units: metres per second.
Step 1 (build once). h s ( a + h ) − s ( a ) with s ( t ) = 20 t − 5 t 2 . Why? Velocity is the instantaneous rate of change of position — see Velocity and acceleration .
Step 2 (general slope). Top = 20 ( a + h ) − 5 ( a + h ) 2 − [ 20 a − 5 a 2 ] . Expand − 5 ( a + h ) 2 = − 5 a 2 − 10 ah − 5 h 2 :
top = 20 h − 10 ah − 5 h 2 .
Divide by h : 20 − 10 a − 5 h . Why? Doing it with a general a lets us reuse the work for both times.
Step 3 (limit, then substitute). h → 0 lim ( 20 − 10 a − 5 h ) = 20 − 10 a . So v ( a ) = 20 − 10 a .
At a = 1 : v = 20 − 10 = + 10 m/s.
At a = 3 : v = 20 − 30 = − 10 m/s.
Verify: signs match the forecast — + 10 climbing, − 10 descending. Units: [ m ] / [ s ] = m/s. ✓ The peak (where v = 0 ) is at a = 2 s, exactly between our two times, which is consistent with the sign flip.
f ( x ) = ∣ x − 1∣ have a slope at x = 1 ?
Forecast: ∣ x − 1∣ makes a sharp V with its point at x = 1 . A corner has no single tangent — guess: derivative does not exist .
Step 1. Difference quotient at a = 1 : since f ( 1 ) = 0 ,
h ∣1 + h − 1∣ − 0 = h ∣ h ∣ .
Why this step? We must test whether the limit even exists before claiming a number.
Step 2. Split by the sign of h — this is the "cover all cases" move:
h > 0 : ∣ h ∣ = h ⇒ h h = + 1.
h < 0 : ∣ h ∣ = − h ⇒ h − h = − 1.
Why this step? The absolute value behaves differently on each side, so the one-sided limits can disagree.
Step 3. Right limit = + 1 , left limit = − 1 . They disagree , so h → 0 lim h ∣ h ∣ does not exist . Therefore f ′ ( 1 ) does not exist .
Verify: matches forecast — a corner. This is the Differentiability and continuity lesson: continuous everywhere, yet not differentiable at the point.
f ( x ) = x 1/3 (the cube root) have a slope at x = 0 ?
Forecast: the cube-root curve passes through the origin standing almost straight up . A vertical tangent has no finite slope — guess: the limit blows up to + ∞ , so no finite derivative.
Step 1. At a = 0 , f ( 0 ) = 0 , so the difference quotient is
h ( 0 + h ) 1/3 − 0 = h h 1/3 = h 1/3 − 1 = h − 2/3 = h 2/3 1 .
Why this step? Writing it as a single power exposes the behaviour as h → 0 .
Step 2. As h → 0 (from either side), h 2/3 = ( h 1/3 ) 2 ≥ 0 shrinks toward 0 + , so h 2/3 1 grows without bound :
lim h → 0 h 2/3 1 = + ∞.
Why this step? A cube root of a negative number is negative, but squaring it makes the denominator positive on both sides — so both one-sided limits go to + ∞ together.
Step 3. The limit is not a finite number, so f ′ ( 0 ) does not exist (the tangent is vertical, slope "∞ ").
Verify: matches forecast. Check the magnitude at h = 0.001 : h − 2/3 = 0.00 1 − 2/3 = 100 , and at h = 1 0 − 6 it is 1 0 4 — clearly growing. ✓
f ( x ) = x 2 , at which point is the tangent parallel to the line y = 8 x − 3 ?
Forecast: parallel lines share a slope. The line has slope 8 , so we need f ′ ( a ) = 8 . Since y = x 2 steepens as x grows, expect a single positive a .
Step 1. Get the general slope of x 2 . Difference quotient h ( a + h ) 2 − a 2 = h 2 ah + h 2 = 2 a + h → 2 a . So f ′ ( a ) = 2 a .
Why this step? We need slope as a function of position so we can set it equal to a target.
Step 2. Set f ′ ( a ) = 8 : 2 a = 8 ⇒ a = 4.
Why this step? "Parallel" translates directly into "equal slope"; solving for a locates the point.
Step 3. The point is ( 4 , f ( 4 )) = ( 4 , 16 ) . Tangent line: y − 16 = 8 ( x − 4 ) ⇒ y = 8 x − 16 .
Why this step? Point–slope form finishes the job once we have the point and slope.
Verify: slope 8 matches the given line's slope 8 ✓, and the two lines differ (− 16 = − 3 ), so they are parallel, not the same line. ✓
Common mistake "A vertical tangent means slope
= 0 ."
Why it feels right: both "flat" and "vertical" sound extreme. Why it's wrong: flat means slope 0 (Ex 3); vertical means slope is infinite (Ex 8). Fix: flat = derivative 0 ; vertical = derivative does not exist .
Recall Rapid self-test across the matrix
Which case gives a negative answer, and why? ::: Case B — a falling curve; the minus sign distributes onto the squared term.
Which two cases have "derivative does not exist", and how do they differ? ::: Corner (Ex 7, one-sided limits + 1 vs − 1 ) and vertical tangent (Ex 8, limit → + ∞ ).
For a reciprocal or root, what algebra unblocks the buried h ? ::: Combine fractions (reciprocal) or multiply by the conjugate (root).
"Tangent parallel to line y = m x + c " becomes which equation? ::: f ′ ( a ) = m , then solve for a .
Average rate of change — every example started as a secant slope before the limit.
Limits — formal definition — Ex 7 and Ex 8 are really limit-existence questions.
Derivative as a function — Ex 6 and Ex 9 found f ′ ( a ) for general a , i.e. f ′ ( x ) .
Power Rule — Ex 5 and Ex 9 secretly confirm it.
Differentiability and continuity — Ex 7, 8 show continuity is not enough for a derivative.
Velocity and acceleration — Ex 6 is the physics face of the same limit.