Extrema and concavity

A look at first and second derivative tests.

Contents

Higher derivatives

Derivative functions can also be differentiated. The second-derivative of ff is

d2dx2f(x)=d2fdx2=ddxf(x)=f(x)\frac{\d ^2}{\d x^2} f(x) = \frac{\d ^2f}{\d x^2} = \frac{\d }{\d x} f^\prime(x) = f^{\prime\prime}(x)

At HL, there is also the notation

dndxnf(x)=f(n)(x)\frac{\d ^n}{\d x^n} f(x) = f^{(n)}(x)

for the nnth derivative.

Purpose

In this course, the purposes of identifying extrema are primarily

  1. Curve sketching
  2. Identifying intervals of increasing and decreasing
  3. Optimization problems
  4. Kinetics problems

The purpose of identifying concavity is mainly curve sketching-related.

So extrema and concavity are not only goals, but also tools.

Extrema

If the function is not defined over all reals, then these endpoints are automatically local extrema, provided that the function is defined on these endpoints.

Or, if the function is piece-wise, then where pieces are not continuous, such endpoints are also local extrema.

Otherwise, x=x0x = x_0 is a critical point, or candidate for a local maximum or local minimum if

  1. f(x0)=0f^\prime(x_0) = 0, OR
  2. ff is continuous at x=x0x = x_0 but f(x0)f^\prime(x_0) is undefined (not differentiable).

Both tests are applicable when f(x0)=0f^\prime(x_0) = 0; only the first derivative test is applicable when f(x0)f^\prime(x_0) is undefined.

Passing either the first or second derivative test means an extremum on x=x0x = x_0. Failing either test (as long as it is applicable) means there is no extremum on x=x0x = x_0.

The global maximum (minimum) is the greatest (least) of the local maxima (minima).

Concavity

Informally, a portion of the graph looking like \cup is concave up, while ones looking like \cap is concave down. The point where a graph changes concavity is known as the point of inflexion.

First derivative test

With critical point x=x0x = x_0, pick aa and bb such that a<x0<ba < x_0 < b. There cannot be other critical points in [a,b][a, b]. Try to make the algebra easy, if applying first derivative test by hand.

caseresult
f(a)>0>f(b){f^\prime(a) > 0 > f^\prime(b)}local maximum; \cap concave down; or  concave up
f(a)<0<f(b){f^\prime(a) < 0 < f^\prime(b)}local minimum; \cup concave up; or  concave down
no sign change, f(x0)=0f^\prime(x_0) = 0horizontal (stationary) point of inflexion (eg (0,0)(0,0) on xx3x \mapsto x^3)
no sign change, f(x0)f^\prime(x_0) is undefinednot extremum or POI

Second derivative test

The second derivative test requires the first derivative to be defined (and continuous).

Find f(x)f^{\prime\prime}(x), and evaluate f(x0)f^{\prime\prime}(x_0).

caseresult
f(x0)<0{f^{\prime\prime}(x_0) < 0}local maximum; \cap concave down
f(x0)>0{f^{\prime\prime}(x_0) > 0}local minimum; \cup concave up
f(x0)=0{f^{\prime\prime}(x_0) = 0}horizontal (stationary) point of inflexion (eg (0,0)(0, 0) on xx3x \mapsto x^3)

Global extrema

Global max and min can be determined by comparing all the endpoint values and critical values, using f(x)f(x).

Points of inflexion

A point of inflexion at x=x1x = x_1 requires

  • f(x1)f^\prime(x_1) is defined
  • f(x1)=0f^{\prime\prime}(x_1) = 0
  • no sign change around f(x1)f^\prime(x_1) or sign change around f(x1)f^{\prime\prime}(x_1).

Note that a POI is only a horizontal POI if the first derivative is also 00. A POI need not satisfy either the first or second derivative tests.

Tips

  1. Does the question care about all extrema or just ones on an open or closed interval?
  2. Does question state that a max or min exist, or do you have to prove it’s a max or min?
  3. Did you check the endpoints? Do you need to check them?
  4. Does question want xx or yy or both?
  5. Is it easier to find the second derivative or evaluate at values near the critical point?