5.10 · AHL · HL onlyPaper 1Paper 2

Second derivative and concavity

In a minute

The , f′′(x)f''(x), is the of the derivative: it tells you how fast the gradient is changing. At a , if it is not 00, its sign tells a maximum from a minimum. Elsewhere its sign tells you which way the graph bends, and where the bend changes there is a , often where a quantity grows or falls fastest.

You'll be able to

  • Find f′′(x)f''(x) or d2ydx2\dfrac{\mathrm{d}^{2}y}{\mathrm{d}x^{2}} by differentiating twice, and say what it means in context.
  • Use the to decide whether a stationary point is a or a , and know when the test decides nothing.
  • Decide where a graph is or from the sign of f′′(x)f''(x).
  • Find points of inflexion, by hand or with your GDC, and interpret them in context.
Before you start3 quick questions on what this builds on
Check yourself
Find the xx-coordinates of the stationary points of f(x)=2x3−24x+7f(x) = 2x^{3} - 24x + 7.
Check yourself
Differentiate y=xe−2xy = x\mathrm{e}^{-2x}.
Check yourself
The gradient f′(x)f'(x) is negative just to the left of x=3x = 3, zero at x=3x = 3 and positive just to the right. What does the graph of ff have at x=3x = 3?

Revise first: 2.4 Key features of graphs, 5.1 Limits and the derivative as a rate of change, 5.2 Increasing and decreasing functions, 5.4 Tangents and normals, 5.6 Local maximum and minimum points and 5.9 Differentiation rules and related rates

Differentiating twice

The derivative f′(x)f'(x) is the gradient function. It is a function of xx too, so you can differentiate it again.

The result is the second derivative. It tells you how fast the gradient is changing as xx increases.

Learn this
f′′(x)=ddx(f′(x)),d2ydx2=ddx(dydx)f''(x) = \frac{\mathrm{d}}{\mathrm{d}x}\big(f'(x)\big), \qquad \frac{\mathrm{d}^{2}y}{\mathrm{d}x^{2}} = \frac{\mathrm{d}}{\mathrm{d}x}\left(\frac{\mathrm{d}y}{\mathrm{d}x}\right)

Read f′′(x)f''(x) as "f double prime of x". In d2ydx2\dfrac{\mathrm{d}^{2}y}{\mathrm{d}x^{2}} nothing is squared: the 2s mean "differentiated twice". Questions use both, so answer in the notation the question uses.

You need no new rules. Use the power rule, and the chain, product and quotient rules from 5.9, twice.

Worked exampleA curve has equation y=0.5x4−3x2+6xy = 0.5x^{4} - 3x^{2} + \dfrac{6}{x}, for x≠0x \neq 0. Find d2ydx2\dfrac{\mathrm{d}^{2}y}{\mathrm{d}x^{2}}.

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

  1. Write the fraction as a power: y=0.5x4−3x2+6x−1y = 0.5x^{4} - 3x^{2} + 6x^{-1}.

In context, f′′f'' has units of "units of ff per unit of xx, per unit of xx". It says whether the rate of change f′f' is increasing (f′′>0f'' > 0) or decreasing (f′′<0f'' < 0). In 5.13, the second derivative of displacement is acceleration.

Worked exampleThe amount of a painkiller in a patient's blood, in mg, is modelled by A(t)=12te−0.4tA(t) = 12t\mathrm{e}^{-0.4t}, where tt is the time in hours after a dose. (a) Find A′(t)A'(t). (b) Find A′′(t)A''(t). (c) Find A′′(2)A''(2), to 3 s.f., and interpret your answer together with A′(2)A'(2).

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

  1. (a) Product rule with u=12tu = 12t and v=e−0.4tv = \mathrm{e}^{-0.4t}, so u′=12u' = 12 and v′=−0.4e−0.4tv' = -0.4\mathrm{e}^{-0.4t} (chain rule): A′(t)=12e−0.4t−4.8te−0.4t=(12−4.8t)e−0.4tA'(t) = 12\mathrm{e}^{-0.4t} - 4.8t\mathrm{e}^{-0.4t} = (12 - 4.8t)\mathrm{e}^{-0.4t}.
Check yourself
Let f(x)=x4−5x3+2xf(x) = x^{4} - 5x^{3} + 2x. Find f′′(x)f''(x).
Check yourself
The number of subscribers to a podcast, in thousands, is modelled by S(t)=200+45t2−2t3S(t) = 200 + 45t^{2} - 2t^{3}, where tt is the number of months after launch. Find S′′(4)S''(4).

The second derivative test

In 5.6 you classified a stationary point by the sign of f′(x)f'(x) on each side. The second derivative gives a quicker test.

At a local minimum the gradient goes from negative, through 00, to positive. It is increasing, so the curve bends up like a cup. At a local maximum the gradient goes from positive to negative. It is decreasing, so the curve bends down like an upside-down cup.

3 6 9 12 15 18 5 10 15 20 25 t h local max (5, 22): h''(5) < 0 local min (15, 12): h''(15) > 0 h(t)
Bending down at the maximum, h′′(5)<0h''(5) < 0; bending up at the minimum, h′′(15)>0h''(15) > 0.
Worked exampleThe depth of water in a reservoir, in metres, is modelled by h(t)=0.02t3−0.6t2+4.5t+12h(t) = 0.02t^{3} - 0.6t^{2} + 4.5t + 12, where tt is the time in months after the start of a year, 0≤t≤180 \le t \le 18. (a) Find h′(t)h'(t) and h′′(t)h''(t). (b) Find the values of tt at which h′(t)=0h'(t) = 0. (c) Use the second derivative test to classify each stationary point. (d) Find the coordinates of the local maximum point.

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

  1. (a) h′(t)=0.06t2−1.2t+4.5h'(t) = 0.06t^{2} - 1.2t + 4.5 and h′′(t)=0.12t−1.2h''(t) = 0.12t - 1.2.

Sometimes f′′(x)f''(x) has the same sign for every xx in the domain. Then you need no substitution: say so, and every stationary point is the same kind.

Worked exampleThe cost, in dollars, of a 300 km delivery trip at a constant speed of vv km/h is modelled by C(v)=0.006v2+7500v+30C(v) = 0.006v^{2} + \dfrac{7500}{v} + 30, for 40≤v≤11040 \le v \le 110. (a) Find C′(v)C'(v) and C′′(v)C''(v). (b) Use your GDC to find the speed that gives the least cost. (c) Use the second derivative to justify that it is a minimum. (d) Find the least cost, to the nearest cent.

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

  1. (a) Write 7500v=7500v−1\dfrac{7500}{v} = 7500v^{-1}. Then C′(v)=0.012v−7500v−2C'(v) = 0.012v - 7500v^{-2} and C′′(v)=0.012+15000v−3C''(v) = 0.012 + 15000v^{-3}.

The test fails when f′′(a)=0f''(a) = 0. At x=0x = 0, all of y=x4y = x^{4}, y=−x4y = -x^{4} and y=x3y = x^{3} have f′(0)=0f'(0) = 0 and f′′(0)=0f''(0) = 0. Yet the first has a minimum, the second a maximum, and the third neither.

Check yourself
A bakery's weekly profit is P(x)P(x) dollars when it bakes xx hundred loaves. It finds P′(40)=0P'(40) = 0 and P′′(40)=−3.5P''(40) = -3.5. What does the graph of PP have at x=40x = 40?
Check yourself
Let k(x)=x3−6x2+12x+1k(x) = x^{3} - 6x^{2} + 12x + 1. Then k′(x)=3(x−2)2k'(x) = 3(x - 2)^{2}, so k′(2)=0k'(2) = 0, and k′′(2)=0k''(2) = 0. What does the graph of kk have at x=2x = 2?

Concavity and points of inflexion

The sign of f′f' tells you whether a graph goes up or down. The sign of f′′f'' tells you which way it bends. These are separate: a falling graph can still bend upwards.

To find one: solve f′′(x)=0f''(x) = 0 (and note any xx where f′′f'' is undefined but ff is defined, such as x=0x = 0 for y=x5/3y = x^{5/3}), then check that f′′(x)f''(x) changes sign there. Both parts matter. For y=x4y = x^{4}, f′′(x)=12x2f''(x) = 12x^{2} is 00 at the origin but positive on both sides, so the graph stays concave-up and the origin is a minimum.

The gradient at a point of inflexion can be 00, as for y=x3y = x^{3} at the origin. Often it is not. Either way, the there crosses the curve.

In context, the gradient is a rate of change. At a point of inflexion it has a local maximum or minimum. So that is where the rate of change is greatest or least, compared with nearby times.

5 10 15 20 2000 4000 6000 8000 t D point of inflexion (10, 4000) concave-up D''(t) > 0 concave-down D''(t) < 0
Downloads grow fastest at the point of inflexion, t=10t = 10.
Worked exampleThe total number of downloads of a new app, in thousands, is modelled by D(t)=60t2−2t3D(t) = 60t^{2} - 2t^{3}, where tt is the number of days after launch, 0≤t≤200 \le t \le 20. (a) Find the values of tt for which the graph of DD is concave-up. (b) Find the coordinates of the point of inflexion. (c) Interpret the point of inflexion in context.

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

  1. (a) D′(t)=120t−6t2D'(t) = 120t - 6t^{2} and D′′(t)=120−12tD''(t) = 120 - 12t. Concave-up needs D′′(t)>0D''(t) > 0: 120−12t>0120 - 12t > 0, so 0≤t<100 \le t < 10.

Often f′′(x)=0f''(x) = 0 is hard to solve by hand. Find f′′(x)f''(x) yourself, graph y=f′′(x)y = f''(x) and find its zeros. Check that the graph crosses the axis there, not just touches it.

Another GDC route: graph y=f′(x)y = f'(x) and find its local maximum and minimum points. Their xx-values are where the concavity of ff changes.

Worked exampleDuring an outbreak, the number of people in a town with flu is modelled by P(t)=50t2e−0.5tP(t) = 50t^{2}\mathrm{e}^{-0.5t}, where tt is the time in weeks, 0≤t≤120 \le t \le 12. (a) Show that P′′(t)=(12.5t2−100t+100)e−0.5tP''(t) = (12.5t^{2} - 100t + 100)\mathrm{e}^{-0.5t}. (b) Use your GDC to find the values of tt at the points of inflexion of the graph of PP. Give your answers to 3 s.f. (c) Interpret the first of these in context.

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

  1. (a) Product rule: P′(t)=100te−0.5t+50t2(−0.5)e−0.5t=(100t−25t2)e−0.5tP'(t) = 100t\mathrm{e}^{-0.5t} + 50t^{2}(-0.5)\mathrm{e}^{-0.5t} = (100t - 25t^{2})\mathrm{e}^{-0.5t}.
Check yourself
Let f(x)=x3−9x2+4xf(x) = x^{3} - 9x^{2} + 4x. For which values of xx is the graph of ff concave-down?
Check yourself
Let q(x)=(x−2)4+3xq(x) = (x - 2)^{4} + 3x, so q′′(2)=0q''(2) = 0. A student says the graph of qq has a point of inflexion at x=2x = 2. Is the student right?
Check yourself
The total number of tickets sold for a festival, N(t)N(t), is modelled as increasing for 0≤t≤140 \le t \le 14 days. The graph of NN changes from concave-up to concave-down at t=6t = 6. What happens at t=6t = 6?

Common mistakes

  • Dropping the inner factor when you differentiate again: the derivative of e−0.4t\mathrm{e}^{-0.4t} is −0.4e−0.4t-0.4\mathrm{e}^{-0.4t}. Write the product or chain rule out in full for each derivative.
  • Squaring the first derivative instead of differentiating it. f′′(x)f''(x) means differentiate f′(x)f'(x) once more.
  • Writing "minimum" with no reason. Give f′′(a)f''(a) or its sign and the conclusion, such as "C′′(v)>0C''(v) > 0 for all v>0v > 0, so the stationary point is a minimum".
  • Calling a point a point of inflexion just because f′′(x)=0f''(x) = 0 there. Check that f′′(x)f''(x) changes sign: y=x4y = x^{4} has f′′(0)=0f''(0) = 0 and a minimum.
  • Using the sign of f′(x)f'(x) for concavity. Increasing is about f′f'; concave-up is about f′′f''.
  • Leaving out units or context in the final answer, such as giving 161.58 with no dollars, or a time with no units. Answer the question in its own terms.

Exam tips

  • "Find d2ydx2\dfrac{\mathrm{d}^{2}y}{\mathrm{d}x^{2}}": show the first derivative and each rule you use. The method marks are for the product or chain rule seen in your working.
  • "Justify" or "show that" a stationary point is a minimum: substitute its xx-value into f′′f'' (or explain why f′′f'' is always positive) and state the conclusion. The reasoning mark needs both the sign and the word "minimum".
  • "Hence or otherwise find the points of inflexion": write the equation f′′(x)=0f''(x) = 0, or say you are finding the turning points of the graph of f′f', before you give the GDC values to 3 s.f.
  • "Interpret": say which rate is greatest or least at the point of inflexion, in the units of the question.

Key takeaways

Learn thisThe second derivative, f′′(x)f''(x) or d2ydx2\dfrac{\mathrm{d}^{2}y}{\mathrm{d}x^{2}}, is the derivative of f′(x)f'(x): differentiate twice. It measures how fast the gradient is changing.
Learn thisSecond derivative test, at a stationary point x=ax = a (where f′(a)=0f'(a) = 0): f′′(a)>0f''(a) > 0 gives a local minimum and f′′(a)<0f''(a) < 0 a local maximum. If f′′(a)=0f''(a) = 0, use the sign of f′(x)f'(x) on each side instead.
Learn thisThe graph of ff is concave-up on an interval where f′′(x)>0f''(x) > 0 throughout, and concave-down on an interval where f′′(x)<0f''(x) < 0 throughout.
Learn thisA point of inflexion is where the concavity changes. Find it by solving f′′(x)=0f''(x) = 0 (and checking any xx where f′′f'' is undefined but ff exists) and checking that f′′(x)f''(x) changes sign: y=x4y = x^{4} has f′′(0)=0f''(0) = 0 but no point of inflexion. The gradient there need not be 00.
Learn thisAt a point of inflexion the rate of change f′(x)f'(x) has a local maximum or minimum: in context, the quantity is changing fastest or slowest there compared with nearby values.
Learn thisWith a GDC, find f′′(x)f''(x) by hand, write down the function you graph, and find where y=f′′(x)y = f''(x) crosses the axis, or where y=f′(x)y = f'(x) has its turning points.

Practise this

4 past-paper questions on Second order derivatives and concavity in the AI HL bank, each with a mark scheme and a worked explanation.

Practise Second order derivatives and concavity →