Unit 3 · Topic 3.6 Beta

Second and Higher Derivatives

4 min read · freeNot practiced

The derivative f′f' of a function is a new function. Nothing stops us from differentiating it again. The result, the second derivative f″f'', tells how fast the rate of change is itself changing. It is the key to motion problems in the next unit and to the shape of graphs after that, so it is worth getting the mechanics and the meaning right now.

Notation

All of these mean the same second derivative:

f″(x)=y″=d2ydx2=ddx(dydx)f''(x) = y'' = \frac{d^2y}{dx^2} = \frac{d}{dx}\left(\frac{dy}{dx}\right)

Read d2ydx2\frac{d^2y}{dx^2} as "d two y, d x squared". The 2s sit where they do because the operation ddx\frac{d}{dx} is applied twice to yy. It is not (dydx)2\left(\frac{dy}{dx}\right)^2, the square of the slope, and not the derivative of y2y^2. Higher derivatives follow the same pattern: f‴(x)f'''(x) for the third, then f(4)(x)f^{(4)}(x), f(5)(x)f^{(5)}(x) and in general f(n)(x)f^{(n)}(x), with the number in brackets so it is not mistaken for a power.

Computing higher derivatives

Differentiate, then differentiate the result. Every rule still applies at every step.

Worked example 1. Find f″(x)f''(x) and f‴(x)f'''(x) for f(x)=x3−3xf(x) = x^3 - 3x.

f′(x)=3x2−3f'(x) = 3x^2 - 3, then f″(x)=6xf''(x) = 6x, then f‴(x)=6f'''(x) = 6, and f(4)(x)=0f^{(4)}(x) = 0.

A polynomial of degree nn has f(n+1)=0f^{(n+1)} = 0: every derivative lowers the degree by one.

Three stacked graphs on the same x-axis: f(x) = x³ − 3x (crimson) with level points at x = −1 and 1; f′(x) = 3x² − 3 (blue), zero at x = ±1; and f″(x) = 6x (green), the slope of f′, negative left of 0 and positive right of 0.
Figure 1. f(x)=x3−3xf(x) = x^3 - 3x, f′(x)=3x2−3f'(x) = 3x^2 - 3 and f″(x)=6xf''(x) = 6x. Each graph records the slopes of the one above it. LevlPrep original diagram.

Look at the figure from bottom to top. Where f″f'' is negative (x<0x < 0), the graph of f′f' is falling. Where f″f'' is positive, f′f' is rising. And where f′f' is 0, at x=±1x = \pm 1, the graph of ff levels off. Each function describes the slope of the one above it.

Worked example 2. Find d2ydx2\frac{d^2y}{dx^2} for y=ex2y = e^{x^2}.

First derivative, by the chain rule: dydx=2xex2\frac{dy}{dx} = 2xe^{x^2}.

This is a product of 2x2x and ex2e^{x^2}, so the second derivative needs the product rule (and the chain rule again inside it):

d2ydx2\frac{d^2y}{dx^2}=2⋅ex2= 2 \cdot e^{x^2}+2x⋅2xex2+ 2x \cdot 2xe^{x^2}=(2+4x2)ex2= (2 + 4x^2)e^{x^2}.

The most common slip is to differentiate only one factor of 2xex22xe^{x^2}.

Patterns are worth spotting. For f(x)=sin⁡(3x)f(x) = \sin(3x): f′=3cos⁡(3x)f' = 3\cos(3x), f″=−9sin⁡(3x)f'' = -9\sin(3x), f‴=−27cos⁡(3x)f''' = -27\cos(3x), and every two derivatives multiply by −9-9. For f(x)=e−2xf(x) = e^{-2x}, each derivative multiplies by −2-2, so f(n)(x)f^{(n)}(x)=(−2)ne−2x= (-2)^n e^{-2x}.

Reading f″f'' from a graph of f′f'

When the exam shows the graph of f′f', the second derivative is the slope of that graph, not its height. If the graph of f′f' is a line segment from (0,2)(0, 2) to (2,−2)(2, -2), then f″(1)f''(1)=−2−22−0= \frac{-2 - 2}{2 - 0}=−2= -2, while f′(1)=0f'(1) = 0 is the height. At a sharp corner of the graph of f′f', the slopes from the two sides differ, so f″f'' does not exist there.

On the graph of f′f'What it tells
height at x=ax = af′(a)f'(a), the slope of ff
slope at x=ax = af″(a)f''(a)
graph risingf″>0f'' > 0
graph fallingf″<0f'' < 0
sharp cornerf″f'' does not exist there

Second derivatives of implicit curves

For a curve given implicitly, differentiate the expression for dydx\frac{dy}{dx} again, remembering that yy still depends on xx. Then substitute the expression for dydx\frac{dy}{dx} wherever it appears.

Worked example 3. For x2+4y2=8x^2 + 4y^2 = 8, find d2ydx2\frac{d^2y}{dx^2} at (2,1)(2, 1).

First: 2x+8ydydx=02x + 8y\frac{dy}{dx} = 0, so dydx=−x4y\frac{dy}{dx} = -\frac{x}{4y}. At (2,1)(2, 1), dydx=−12\frac{dy}{dx} = -\frac{1}{2}.

Second, by the quotient rule: d2ydx2\frac{d^2y}{dx^2}=−(1)(4y)−x(4dydx)(4y)2= -\frac{(1)(4y) - x\left(4\frac{dy}{dx}\right)}{(4y)^2}.

At (2,1)(2, 1) with dydx=−12\frac{dy}{dx} = -\frac{1}{2}: −4−2⋅4⋅(−12)16-\frac{4 - 2 \cdot 4 \cdot \left(-\frac{1}{2}\right)}{16}=−4+416= -\frac{4 + 4}{16}=−12= -\frac{1}{2}.

In general, substituting dydx=−x4y\frac{dy}{dx} = -\frac{x}{4y} and using x2+4y2=8x^2 + 4y^2 = 8 gives d2ydx2\frac{d^2y}{dx^2}=−x2+4y216y3= -\frac{x^2 + 4y^2}{16y^3}=−12y3= -\frac{1}{2y^3}.

Slips that cost points

  • Writing dy2dx\frac{dy^2}{dx} or (dydx)2\left(\frac{dy}{dx}\right)^2 for the second derivative.
  • Forgetting that a first derivative such as 2xex22xe^{x^2} is a product.
  • Treating yy as a constant when differentiating dydx\frac{dy}{dx} for an implicit curve.
  • Reading the height of a graph of f′f' when the question asks for f″f''.

Spot a mistake on this page?