5.12 · AHL · HL onlyPaper 1Paper 2

Continuity, limits and higher derivatives

In a minute

A function is where its graph has no break, and where it has no corner either. A sequence is if it settles on one value. You'll also differentiate from the limit definition, and differentiate repeatedly.

You'll be able to

  • Say when a function is or at a point.
  • Decide whether a sequence or geometric series converges, and find limits as nn or x→∞x \to \infty.
  • Differentiate a polynomial from .
  • Find higher derivatives, f(n)(x)f^{(n)}(x) or dnydxn\frac{\mathrm{d}^{n}y}{\mathrm{d}x^{n}}, and spot a formula for the nnth.
Before you start3 quick questions on what this builds on
Check yourself
On a curve y=f(x)y = f(x), the chord from P(1,3)P(1, 3) to the point with x=1+hx = 1 + h has gradient 4+h4 + h. Write down f′(1)f'(1).
Check yourself
Find d2ydx2\dfrac{\mathrm{d}^{2}y}{\mathrm{d}x^{2}} for y=2x3−5x2+xy = 2x^{3} - 5x^{2} + x.
Check yourself
Find the sum to infinity of the geometric series 9+3+1+…9 + 3 + 1 + \dots

Revise first: 1.8 Infinite geometric series, 5.1 Limits and the derivative as a rate of change and 5.7 Second derivative and concavity

Continuity and differentiability

Some graphs jump from one height to another. Some turn a sharp corner. Continuity is about jumps; differentiability is about corners.

Three graphs at a point a: a jump, a corner and a smooth join a jump not continuous a corner continuous, not differentiable a smooth join continuous and differentiable
A jump, a corner and a smooth join at x=ax = a.

So the heights from the left, from the right and f(a)f(a) itself must all agree. At a jump they don't, so ff is not there.

At a corner the gradient on the left differs from the gradient on the right. There is no single tangent, so the function is not there.

For two pieces that meet at x=ax = a, compare their heights at aa: a mismatch is a jump. If the heights match, compare their gradients at aa: a mismatch is a corner.

Worked exampleLet f(x)={x2,x≤13−2x,x>1f(x) = \begin{cases} x^{2}, & x \le 1 \\ 3 - 2x, & x > 1 \end{cases}. Decide whether ff is continuous at x=1x = 1, and whether it is differentiable there.

Try it first, then reveal one step at a time.

  1. Heights: the left piece gives f(1)=12=1f(1) = 1^{2} = 1. The right piece heads for 3−2(1)=13 - 2(1) = 1.
Worked exampleLet g(x)={4x−4,x<2x2,x≥2g(x) = \begin{cases} 4x - 4, & x < 2 \\ x^{2}, & x \ge 2 \end{cases}. Decide whether gg is continuous and differentiable at x=2x = 2.

Try it first, then reveal one step at a time.

  1. Heights: the line heads for 4(2)−4=44(2) - 4 = 4, and g(2)=22=4g(2) = 2^{2} = 4. They match, so gg is continuous at x=2x = 2.
Check yourself
Let k(x)={x+3,x<02−x2,x≥0k(x) = \begin{cases} x + 3, & x < 0 \\ 2 - x^{2}, & x \ge 0 \end{cases}. Which statement is true?
Check yourself
Let m(x)={x2+1,x≤14x−2,x>1m(x) = \begin{cases} x^{2} + 1, & x \le 1 \\ 4x - 2, & x > 1 \end{cases}. Which statement is true?

Convergence and divergence

In 1,12,14,18,…1, \tfrac{1}{2}, \tfrac{1}{4}, \tfrac{1}{8}, \dots each term halves. The terms never reach 0, but they get as close to 0 as you like.

A sequence either grows without limit, like 2,4,8,…2, 4, 8, \dots (written un→∞u_{n} \to \infty), or never settles, like 1,−1,1,−1,…1, -1, 1, -1, \dots

Powers rnr^{n} show the cases: rn→0r^{n} \to 0 if ∣r∣<1\lvert r\rvert < 1, rn→∞r^{n} \to \infty if r>1r > 1, and rnr^{n} never settles if r≤−1r \le -1. (If r=1r = 1, every power is 1.)

This explains the sum to infinity from 1.8. The partial sum is Sn=u1(1−rn)1−rS_{n} = \dfrac{u_{1}(1 - r^{n})}{1 - r}. When ∣r∣<1\lvert r\rvert < 1, rn→0r^{n} \to 0, so SnS_{n} is with limit:

In the formula booklet
S∞=u11−r,∣r∣<1S_{\infty} = \frac{u_{1}}{1 - r}, \quad \lvert r\rvert < 1

When ∣r∣≥1\lvert r\rvert \ge 1, SnS_{n} diverges, and the series has no sum to infinity.

Worked exampleConsider the geometric series 12+8+163+…12 + 8 + \tfrac{16}{3} + \dots Write SnS_{n} in terms of nn. Hence explain why the series converges, and find its limit.

Try it first, then reveal one step at a time.

  1. Find the ratio: r=812=23r = \frac{8}{12} = \frac{2}{3}, and u1=12u_{1} = 12.
Check yourself
Which geometric series has a sum to infinity?

The same idea works for a function as x→∞x \to \infty. For a fraction of two polynomials of the same degree, divide every term by the highest power of xx. Terms such as 1x\frac{1}{x} or 5x2\frac{5}{x^{2}} then tend to 0.

Worked exampleFind lim⁡x→∞6x2−x2x2+5\displaystyle\lim_{x \to \infty} \frac{6x^{2} - x}{2x^{2} + 5}.

Try it first, then reveal one step at a time.

  1. Divide top and bottom by x2x^{2}: 6x2−x2x2+5=6−1x2+5x2\dfrac{6x^{2} - x}{2x^{2} + 5} = \dfrac{6 - \frac{1}{x}}{2 + \frac{5}{x^{2}}}.
Check yourself
Find lim⁡x→∞5−2xx+3\displaystyle\lim_{x \to \infty} \frac{5 - 2x}{x + 3}.
Worked exampleThe geometric series 50+45+40.5+…50 + 45 + 40.5 + \dots converges. Use your calculator to find the least number of terms nn for which SnS_{n} is within 1 of S∞S_{\infty}.

Try it first, then reveal one step at a time.

  1. Here u1=50u_{1} = 50 and r=4550=0.9r = \frac{45}{50} = 0.9. So S∞=501−0.9=500S_{\infty} = \frac{50}{1 - 0.9} = 500 and Sn=500(1−0.9n)S_{n} = 500(1 - 0.9^{n}).

Differentiation from first principles

In 5.1 you estimated a gradient from chords. The chord from xx to x+hx + h has gradient f(x+h)−f(x)h\frac{f(x+h) - f(x)}{h}. Now you find its limit exactly, by algebra. This is differentiating from .

In the formula booklet
y=f(x)⇒dydx=f′(x)=lim⁡h→0(f(x+h)−f(x)h)y = f(x) \Rightarrow \frac{\mathrm{d}y}{\mathrm{d}x} = f'(x) = \lim_{h \to 0}\left(\frac{f(x+h) - f(x)}{h}\right)

It's in the formula booklet; you need the method. Putting h=0h = 0 straight in gives 00\frac{0}{0}, so the algebra cancels the hh first.

1. Write f(x+h)f(x + h): replace every xx with (x+h)(x + h), then expand.

2. Subtract f(x)f(x). Every term without an hh cancels.

3. Divide by hh. Every term left has a factor hh, so this cancels cleanly.

4. Let h→0h \to 0. The terms that still contain hh vanish.

From secant to tangentDrag the slider
123−112345678910111213141516PQsecant PQtangent at P

1.00

P = (1, 1.00), Q = (2.00, 8.000)

Secant gradient = 7.000

Tangent gradient f′(1) = 3

As Q slides towards P, h → 0 and the gradient of the chord PQ settles on the gradient of the tangent, f′(1).

For y=x3y = x^{3} at P(1,1)P(1, 1), the chord gradient is (1+h)3−1h=3+3h+h2\frac{(1+h)^{3} - 1}{h} = 3 + 3h + h^{2}. Slide hh towards 0: the readout heads for 3, the gradient of the tangent.

Worked exampleLet f(x)=3x2−5xf(x) = 3x^{2} - 5x. Find f′(x)f'(x) from first principles.

Try it first, then reveal one step at a time.

  1. Replace xx with (x+h)(x + h): f(x+h)=3(x+h)2−5(x+h)=3x2+6xh+3h2−5x−5hf(x+h) = 3(x+h)^{2} - 5(x+h) = 3x^{2} + 6xh + 3h^{2} - 5x - 5h.
Worked exampleLet f(x)=2x3+xf(x) = 2x^{3} + x. Use the definition of the derivative from first principles to show that f′(x)=6x2+1f'(x) = 6x^{2} + 1.

Try it first, then reveal one step at a time.

  1. Expand the cube first: (x+h)3=x3+3x2h+3xh2+h3(x+h)^{3} = x^{3} + 3x^{2}h + 3xh^{2} + h^{3}. Use the binomial theorem, or multiply (x+h)2(x+h)^{2} by (x+h)(x+h).
Check yourself
Let f(x)=4−x2f(x) = 4 - x^{2}. Simplify f(x+h)−f(x)h\dfrac{f(x+h) - f(x)}{h}.

For the gradient at one point x=ax = a, put the number in from the start: f′(a)=lim⁡h→0f(a+h)−f(a)hf'(a) = \lim_{h \to 0} \dfrac{f(a+h) - f(a)}{h}.

Check yourself
Let f(x)=x3+2x2f(x) = x^{3} + 2x^{2}. Use first principles to find f′(−1)f'(-1).

Higher derivatives

In 5.7 you differentiated f′(x)f'(x) to get f′′(x)f''(x). You can keep going.

The first three use primes: f′f', f′′f'', f′′′f'''. From the fourth on, bracket the number: f(4)(x)f^{(4)}(x). Without brackets, f4(x)f^{4}(x) reads as a power.

Learn this
f(n)(x)=ddx(f(n−1)(x)),dnydxn=ddx(dn−1ydxn−1)f^{(n)}(x) = \frac{\mathrm{d}}{\mathrm{d}x}\left(f^{(n-1)}(x)\right), \qquad \frac{\mathrm{d}^{n}y}{\mathrm{d}x^{n}} = \frac{\mathrm{d}}{\mathrm{d}x}\left(\frac{\mathrm{d}^{n-1}y}{\mathrm{d}x^{n-1}}\right)
Worked exampleFind d3ydx3\dfrac{\mathrm{d}^{3}y}{\mathrm{d}x^{3}} for y=x5−3x3+2xy = x^{5} - 3x^{3} + 2x.

Try it first, then reveal one step at a time.

  1. First: dydx=5x4−9x2+2\dfrac{\mathrm{d}y}{\mathrm{d}x} = 5x^{4} - 9x^{2} + 2.
Check yourself
Let f(x)=x4−2x3+7xf(x) = x^{4} - 2x^{3} + 7x. Find f′′′(x)f'''(x).

Sometimes a pattern gives f(n)(x)f^{(n)}(x) for every nn at once.

Worked exampleLet f(x)=xe−xf(x) = x\mathrm{e}^{-x}. Find f′(x)f'(x), f′′(x)f''(x) and f′′′(x)f'''(x). Hence suggest a formula for f(n)(x)f^{(n)}(x).

Try it first, then reveal one step at a time.

  1. Product rule: f′(x)=1⋅e−x+x⋅(−e−x)=(1−x)e−x=−(x−1)e−xf'(x) = 1 \cdot \mathrm{e}^{-x} + x \cdot \left(-\mathrm{e}^{-x}\right) = (1 - x)\mathrm{e}^{-x} = -(x - 1)\mathrm{e}^{-x}.

For xe−xx\mathrm{e}^{-x} the step is one product rule. If f(k)(x)=(−1)k(x−k)e−xf^{(k)}(x) = (-1)^{k}(x - k)\mathrm{e}^{-x}, then f(k+1)(x)=(−1)k(e−x−(x−k)e−x)=(−1)k+1(x−(k+1))e−xf^{(k+1)}(x) = (-1)^{k}\left(\mathrm{e}^{-x} - (x - k)\mathrm{e}^{-x}\right) = (-1)^{k+1}\left(x - (k + 1)\right)\mathrm{e}^{-x}.

Check yourself
Let f(x)=cos⁡(2x)f(x) = \cos(2x). Find f(4)(x)f^{(4)}(x).
Check yourself
Let y=xy = \sqrt{x}. Find the value of d3ydx3\dfrac{\mathrm{d}^{3}y}{\mathrm{d}x^{3}} when x=1x = 1, as a decimal.

Common mistakes

  • Putting h=0h = 0 before dividing. That gives 00\frac{0}{0}. Cancel the factor hh first, then let h→0h \to 0.
  • Subtracting f(x)f(x) without brackets. Put the whole of f(x)f(x) in brackets, as in f(x+h)−(3x2−5x)f(x+h) - \left(3x^{2} - 5x\right) for f(x)=3x2−5xf(x) = 3x^{2} - 5x, so every term changes sign.
  • Dropping the chain-rule factor after the first round. Every derivative of cos⁡(2x)\cos(2x) brings out another 2, so f(4)(x)f^{(4)}(x) has 242^{4}.

Exam tips

  • "From first principles": show each stage. Expand f(x+h)f(x+h), subtract f(x)f(x), divide by hh, and write lim⁡h→0\lim_{h \to 0} until the last line.
  • "Write down" a limit as t→∞t \to \infty: just the value. If asked to comment, say whether the model still holds for large tt.
  • "Hence suggest" a formula for f(n)(x)f^{(n)}(x): show the first few derivatives in full, then the pattern. "Prove" it by induction.

Key takeaways

In bookletS∞=u11−r,∣r∣<1S_{\infty} = \frac{u_{1}}{1 - r}, \quad \lvert r\rvert < 1
In booklety=f(x)⇒dydx=f′(x)=lim⁡h→0(f(x+h)−f(x)h)y = f(x) \Rightarrow \frac{\mathrm{d}y}{\mathrm{d}x} = f'(x) = \lim_{h \to 0}\left(\frac{f(x+h) - f(x)}{h}\right)
Learn thisf(n)(x)=ddx(f(n−1)(x)),dnydxn=ddx(dn−1ydxn−1)f^{(n)}(x) = \frac{\mathrm{d}}{\mathrm{d}x}\left(f^{(n-1)}(x)\right), \qquad \frac{\mathrm{d}^{n}y}{\mathrm{d}x^{n}} = \frac{\mathrm{d}}{\mathrm{d}x}\left(\frac{\mathrm{d}^{n-1}y}{\mathrm{d}x^{n-1}}\right)
Learn thisContinuous at aa: no jump, so lim⁡x→af(x)=f(a)\lim_{x \to a} f(x) = f(a). Differentiable at aa: no corner either, so there is one tangent.
Learn thisDifferentiable implies continuous, not the reverse: y=∣x∣y = \lvert x\rvert is continuous at 00 but has a corner.
Learn thisA sequence converges if it settles on one value; otherwise it diverges. rn→0r^{n} \to 0 when ∣r∣<1\lvert r\rvert < 1.
Learn thisFirst principles: expand f(x+h)f(x+h), subtract f(x)f(x), divide by hh, then let h→0h \to 0.
Learn thisf(n)(x)=dnydxnf^{(n)}(x) = \frac{\mathrm{d}^{n}y}{\mathrm{d}x^{n}}: differentiate nn times. Spot the pattern, then prove it by induction.

Practise this

1 past-paper question on 5.12 in the AA HL bank, each with a mark scheme and a worked explanation.

Practise 5.12 →