The first derivative test for increase, decrease and local extrema; concavity as the sign of the second derivative and as the direction the tangents lie; points of inflection where the concavity changes; and the second derivative test with the three functions that show why it can be inconclusive.
Subject: Calculus I · 65 slides · symbolic lesson
Open the interactive version of this deck
Title
Calculus I · Section 4.5
Reading every rise, fall and bend from two signs
Objectives
Section 4.3 found the critical points of a function but could not say what happens at them. This lesson finishes the job, and then reads a second layer of shape from the second derivative.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, pp. 339-353 — learning objectives 4.5.1 to 4.5.6
In Section 4.3 you learned to find critical points, the places where the derivative is zero or does not exist, and you learned that every local maximum or minimum must sit at one of them. What you could not yet do was look at a particular critical point and say what actually happens there. This lesson gives you that verdict.
The tool is surprisingly simple: the sign of the derivative on either side of the point. From there the lesson climbs one level higher, to the second derivative, which tells you which way the graph bends. The two signs together describe the shape of a curve completely, without plotting a single point.
By the end you will be able to take a formula, draw two number lines, and read off every peak, valley and change of bend of its graph. Section 4.6 then adds the behaviour at the far ends, and you will have everything you need to sketch a curve honestly.
Warm-up
Discussion prompt
Write the definition of a critical point of f, and say exactly what Fermat's theorem from Section 4.3 promises about critical points and what it does not.
Write your answer before you reveal it. The definition has two halves, and the second half, where the derivative fails to exist, is the one people forget. Corners and cusps are critical points too, and one of this lesson's examples turns on exactly that.
The more important part is the direction of Fermat's theorem. It says that if f has a local extremum at c, then c must be critical. It does not say that every critical point is an extremum. Critical points are the suspects, not the convicted.
Hold on to that asymmetry. The first half of this lesson is a method for interrogating each suspect: look at what the function does just before and just after, and decide whether the point is a peak, a valley, or simply a place where the curve paused.
Section
Part 1
Concept
Figure (svg): Two panels. Left: a rising curve with five short tangent segments, every one tilted upward. Right: a falling curve with five tangent segments, every one tilted downward.
\[ f'(x) > 0 \text{ on } I \;\Longrightarrow\; f \text{ increasing on } I \]
\[ f'(x) < 0 \text{ on } I \;\Longrightarrow\; f \text{ decreasing on } I \]
This is Corollary 3 of the Mean Value Theorem from Section 4.4. Everything in this lesson is built on it.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, pp. 339-340 — Figure 4.30
Look at the left panel. Every short green segment is a tangent line, and every one of them tilts upward. A positive derivative everywhere on an interval means a climbing tangent everywhere, and the graph climbs across the whole interval. The right panel is the mirror image: every tangent falls, so the graph falls.
This was proved in Section 4.4 as a corollary of the Mean Value Theorem. The proof goes through the theorem because a single positive slope at one point says nothing about far-away points; it is the positivity across the whole interval that forces the function to increase.
From now on, when you want to know where a function rises and where it falls, you will not plot it. You will find where its derivative is positive and where it is negative. That is the whole of the first half of this lesson.
Concept
The book's opening example. The cube function has a critical point at the origin:
\[ f(x) = x^3, \qquad f'(x) = 3x^2, \qquad f'(0) = 0 \]
But a square is positive whenever x is not zero, so the derivative is positive on both sides:
\[ f'(x) = 3x^2 > 0 \quad \text{for } x \ne 0 \]
The function rises into the origin and keeps rising out of it. The tangent is flat for one instant, and nothing turns.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 339 — introduction
This is the book's opening example, and it is worth remembering forever. The cube function has derivative three x squared, which is zero at the origin, so the origin is a critical point. But is it a peak or a valley? Neither.
The reason is in the sign. Three x squared is positive for every x other than zero, because a square of a nonzero number is positive. So the function is rising to the left of the origin and rising to the right of it. It flattens out for a single instant and then carries on climbing.
Picture a car that slows to a momentary stop at a junction and then continues forward in the same direction. It stopped, but it did not turn around. A flat tangent is the stop; a turn needs the direction of travel to reverse, and that means the sign of the derivative has to change.
Concept
Figure (svg): A continuous curve with four critical points marked a, b, c and d: a smooth peak at a, a smooth valley at b, a flat point at c where the curve keeps rising, and a sharp corner peak at d. Below it, the sign chart of f prime: plus, minus, plus, plus, minus.
To switch from increasing to decreasing at c, the derivative must switch from positive to negative. A continuous derivative can only do that by passing through zero; otherwise it must fail to exist at c.
\[ f'(c) = 0 \quad\text{or}\quad f'(c) \text{ undefined} \]
So the critical points are the only candidates, and the sign of the derivative on either side is what decides.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, pp. 340-341 — Figure 4.31
The curve has four critical points. At a it rises and then falls, a peak. At b it falls and then rises, a valley. At c the tangent is flat, but the curve rises on both sides, so nothing happens. At d there is a sharp corner, the derivative does not exist, and the curve goes from rising to falling: a peak with a point on it.
Now look at the sign chart underneath. Every turn in the curve sits exactly above a change of sign, and the point c, where the sign stays plus, is the one critical point with no turn. That is the whole idea of this part.
Why must a turn happen at a critical point? To change from increasing to decreasing, the derivative must change from positive to negative. A continuous derivative cannot jump from positive to negative without passing through zero, and if it is not continuous there, it may simply not exist. Either way, you are at a critical point.
Concept
First derivative test (Theorem 4.9) — Let f be continuous on an interval I containing a critical point c, and differentiable on I except possibly at c. Read the sign of the derivative just left and just right of c.
| sign of f′ left of c | sign of f′ right of c | f(c) is |
|---|---|---|
| + | − | a local maximum |
| − | + | a local minimum |
| same sign on both sides | same sign on both sides | neither |
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 341 — Theorem 4.9
Here is the theorem in its official form. The table has only three rows, and each one is a picture. Plus then minus: uphill then downhill, a peak. Minus then plus: downhill then uphill, a valley. The same sign on both sides: the curve carries on the way it was going, and there is no extremum.
Notice the hypotheses. The function must be continuous on the interval, and differentiable there except possibly at c itself. That exception is deliberate: the test only looks at the derivative on either side of c, so it works at corners and cusps where the derivative at c does not exist.
Also notice what the test does not need: it never asks for the value of the derivative at c, and it never asks for the second derivative. Just two signs. That makes it the most reliable tool in this lesson, the one you fall back on whenever anything else is silent.
Notation
Annotate
On: \( f' : + \to - \text{ across } c \;\Longrightarrow\; f(c) \text{ is a local max} \)
Step through the notes one at a time. The first is the hypothesis most people skip: continuity. It sounds like a technicality, but a few slides from now you will see a function where the signs say peak and the truth is valley, simply because the function jumps.
The second note is the generous one. The derivative is allowed to fail to exist at c itself. So the absolute value function, which has a corner at zero, is covered: its derivative is minus one on the left and plus one on the right, so its corner is a minimum.
The last two notes are the verdicts. Read the arrow as a story: on the way into c the function is climbing, on the way out it is descending, so nothing nearby can be higher than f of c. If the story does not change direction, there is no extremum to report.
Sorting
Sort into buckets
Each item describes the sign of f′ just left and just right of a critical point c of a continuous f. Sort them.
Sort each description before checking. The first four are the plain cases and should take you a moment each. The last two are there to test whether you really believe that the first derivative test does not care about the value at c.
A corner with the derivative going from minus to plus is a minimum, exactly like the bottom of the absolute value graph. A cusp with plus then minus is a maximum, like the top of the graph of minus the cube root of x squared. In neither case does the derivative exist at c, and in neither case does it matter.
The two neither cases are the ones to remember: the sign is the same on both sides, so the function simply passes through. The cube function is the example for plus then plus; its negative is the example for minus then minus.
Concept
Suppose the derivative is continuous on an interval with no critical points inside it. If it were positive at one point a and negative at another point b of that interval, the Intermediate Value Theorem would force a zero between them:
\[ f'(a) > 0, \quad f'(b) < 0 \;\Longrightarrow\; f'(d) = 0 \text{ for some } d \text{ between} \]
That d would be a critical point inside the interval, which there is not. So the derivative has one sign across the whole interval, and one test point reads it.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 341 — Problem-Solving Strategy, step 2
It can feel reckless to evaluate the derivative at a single point and declare its sign for a whole interval. Here is why it is safe. Between two consecutive critical points the derivative has no zeros and, in the usual case, is continuous.
Suppose it were positive at one spot and negative at another inside that interval. The Intermediate Value Theorem from Chapter 2 would then force it to be zero somewhere between them. That zero would be another critical point inside the interval, and you already listed every critical point. So the sign cannot change inside the interval.
The book adds one caution: if the derivative has a discontinuity, include that point as a cut too, because the Intermediate Value Theorem needs continuity. With that proviso, one test point per interval is not a shortcut but a proof.
Concept
The whole method is a sign chart: a number line, the cut points, and a plus or minus sign in each gap.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 341 — Problem-Solving Strategy: using the first derivative test
This is the book's Problem-Solving Strategy, and it turns the theorem into a routine you can run on any function. The output is a sign chart: a number line, cut at every critical point, with a plus or minus sign in each gap.
Step one is where most errors happen. Find the zeros of the derivative, but also look for places where it does not exist, typically where a denominator vanishes or a fractional power has a negative exponent. Forgetting those loses points like the corner or cusp.
Steps three and four are mechanical once the cuts are right. Pick easy test points, often whole numbers, and compute only the sign, not the full value, if you like; a factored derivative makes the sign a matter of counting negative factors. Then read the chart from left to right.
Ranking
Put in order
Order the steps of a complete first derivative test.
Why: Critical points come first because they are where the sign can change; the test points are chosen inside the gaps they create; only then can each critical point be classified.
Arrange the cards before checking. The order is not arbitrary, and it is worth being able to say why each step has to come where it does.
You need the derivative before you can find its zeros. You need the critical points before you can cut the line, because they are the only places the sign could change. The test points are chosen inside the gaps the cuts create, so they must come after. And the verdict at each critical point compares the signs on its two sides, so it comes last.
If you ever find yourself choosing test points before you have found all the critical points, stop. A test point on the wrong side of a missed critical point gives a sign that does not describe the interval you think it does.
Worked example
Use the first derivative test to find all local extrema.
\[ f(x) = x^3 - 3x^2 - 9x - 1 \]
Differentiate
Why: Power rule, term by term.
\[ f'(x) = 3x^2 - 6x - 9 \]
Take out the common factor
Why: Every coefficient is a multiple of 3.
\[ f'(x) = 3(x^2 - 2x - 3) \]
Factor the quadratic
Why: Two numbers with product minus 3 and sum minus 2.
\[ f'(x) = 3(x - 3)(x + 1) \]
Find the critical points
Why: The derivative exists everywhere, so only its zeros count.
\[ x = -1, \qquad x = 3 \]
Test the interval left of minus one
Why: Take x equal to minus 2.
\[ f'(-2) = 3(-5)(-1) = 15 > 0 \]
Test the middle interval
Why: Take x equal to 0.
\[ f'(0) = 3(-3)(1) = -9 < 0 \]
Test the interval right of three
Why: Take x equal to 4.
\[ f'(4) = 3(1)(5) = 15 > 0 \]
Classify
Why: Plus to minus at minus 1, minus to plus at 3.
\[ \text{max at } x = -1, \qquad \text{min at } x = 3 \]
Evaluate f at both
Why: Substitute into the original function.
\[ f(-1) = -1 - 3 + 9 - 1 = 4, \qquad f(3) = 27 - 27 - 27 - 1 = -28 \]
Figure (svg): The cubic f of x equals x cubed minus 3x squared minus 9x minus 1, with a peak at (−1, 4) and a valley at (3, −28). Below it, the sign chart of f prime: plus left of −1, minus between −1 and 3, plus right of 3.
Check against neighbouring values
Why: The maximum should beat the values on either side of it, and the minimum should lie below its neighbours.
\[ f(-2) = -3 < 4 > -1 = f(0), \qquad f(2) = -23 > -28 < -21 = f(4) \]
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, pp. 341-342 — Example 4.17 and Figure 4.32
Follow the maths down the slide with the words covered. The derivative factors as three times x minus three times x plus one, so the critical points are minus one and three. Nothing is undefined for a polynomial, so those are all of them.
The three test points are chosen for easy arithmetic: minus two, zero and four. At minus two both brackets are negative, so their product is positive. At zero one bracket is negative and one positive, so the product is negative. At four both are positive. The chart reads plus, minus, plus.
Now look at the figure. The sign chart sits under the graph, and each change of sign is directly below a turn: the peak at minus one with height four, the valley at three with depth minus twenty-eight. The check step confirms it the old-fashioned way, comparing the extreme values with their neighbours. That comparison is not a proof, but it catches arithmetic slips instantly.
Fill the middle
Checkpoint 4.16 uses the function below. Its derivative factors neatly; fill in its sign at the test point 0.
\[ f(x) = -x^3 + \tfrac32 x^2 + 18x \]
Fill in the blanks
f'(x) = -3(x - 3)(x + 2), \qquad f'(0) = -3(-3)(2) = 18
Why: Minus 3 times minus 3 is 9, and 9 times 2 is 18. A positive derivative between the two critical points means the function rises there: the reverse of Example 4.17's pattern.
Type the value of the derivative at zero. The derivative is already factored for you, so this is just careful arithmetic with signs.
Minus three times minus three is nine, and nine times two is eighteen. The derivative is positive between the two critical points, which means the function rises there. That is the opposite of Example 4.17, where the middle interval was the falling one.
The reason is the leading minus sign. A cubic with a negative leading coefficient falls on the far left and the far right, so its middle stretch is the one that climbs. Keeping the overall shape of the function in mind is a quick way to catch a sign error before it reaches your answer.
Worked example
\[ f(x) = -x^3 + \tfrac32 x^2 + 18x \]
Differentiate
Why: Power rule; three halves times two is three.
\[ f'(x) = -3x^2 + 3x + 18 \]
Factor
Why: Take out minus 3, then factor the quadratic.
\[ f'(x) = -3(x^2 - x - 6) = -3(x - 3)(x + 2) \]
Critical points
Why: The zeros of the derivative.
\[ x = -2, \qquad x = 3 \]
Signs on the three intervals
Why: Test points minus 3, 0 and 4.
\[ f'(-3) = -18, \quad f'(0) = 18, \quad f'(4) = -18 \]
Classify
Why: Minus to plus at minus 2; plus to minus at 3.
\[ \text{min at } x = -2, \qquad \text{max at } x = 3 \]
Evaluate
Why: Substitute each into f.
\[ f(-2) = 8 + 6 - 36 = -22, \qquad f(3) = -27 + \tfrac{27}{2} + 54 = \tfrac{81}{2} \]
Figure (svg): Two panels. Left: the parabola f prime of x equals minus 3 times (x minus 3)(x plus 2), shaded green where it is above the axis, between −2 and 3, and red outside. Right: the function f, falling to a valley at x = −2, rising to a peak at x = 3, then falling.
Check against neighbours
Why: The minimum lies below the values at minus 3 and minus 1; the maximum lies above the values at 2 and 4.
\[ f(-3) = -13.5,\; f(-1) = -15.5 > -22, \qquad f(2) = 34,\; f(4) = 32 < 40.5 \]
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 342 — Checkpoint 4.16
The derivative here is a quadratic, so its graph is a parabola, and the left panel draws it. Wherever the parabola is above the axis, shaded green, the function rises; wherever it is below, shaded red, the function falls.
Now look across to the right panel. The valley sits exactly above the root at minus two, where the parabola crosses from red to green, and the peak sits exactly above the root at three, where it crosses from green to red. That is the first derivative test drawn as a picture, and it is worth being able to read a derivative's graph this way on its own.
The values follow by substitution: minus twenty-two at the valley and eighty-one halves, or forty point five, at the peak. The check compares each with the values one unit to either side. Keep this function in mind, because it comes back twice more in this lesson.
Worked example
\[ f(x) = 5x^{1/3} - x^{5/3} \]
Differentiate
Why: Power rule with fractional exponents.
\[ f'(x) = \frac53 x^{-2/3} - \frac53 x^{2/3} \]
Write the negative power as a fraction
Why: Move it to the denominator.
\[ f'(x) = \frac{5}{3x^{2/3}} - \frac{5x^{2/3}}{3} \]
Combine over one denominator
Why: Multiply the second term top and bottom by x to the two thirds.
\[ f'(x) = \frac{5 - 5x^{4/3}}{3x^{2/3}} = \frac{5\left(1 - x^{4/3}\right)}{3x^{2/3}} \]
Zeros of the derivative
Why: The numerator vanishes when x to the four thirds is 1.
\[ x^{4/3} = 1 \iff x = \pm 1 \]
Where the derivative is undefined
Why: The denominator is zero at the origin, where f itself is defined.
\[ x = 0: \quad 3x^{2/3} = 0 \]
Signs on four intervals
Why: Test points minus 2, minus one half, one half and 2; the denominator is never negative.
\[ f'(-2) < 0, \quad f'(-\tfrac12) > 0, \quad f'(\tfrac12) > 0, \quad f'(2) < 0 \]
Classify
Why: Minus to plus, then plus to plus, then plus to minus.
\[ \text{min at } -1, \quad \text{neither at } 0, \quad \text{max at } 1 \]
Evaluate
Why: The cube root of minus 1 is minus 1.
\[ f(-1) = -5 + 1 = -4, \qquad f(1) = 5 - 1 = 4 \]
Figure (svg): The curve f of x equals 5 times the cube root of x minus x to the five thirds, with a valley at (−1, −4), a vertical tangent at the origin, and a peak at (1, 4). Below it, the sign chart of f prime with breaks at −1, 0 and 1: minus, plus, plus, minus.
Check with symmetry and a number
Why: Both terms are odd functions, so f is odd and its minimum must mirror its maximum; numerically the derivative at 2 is about minus 1.60, negative as the chart says.
\[ f(-x) = -f(x) \;\Longrightarrow\; f(-1) = -f(1) = -4, \qquad f'(2) \approx -1.60 \]
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 343 — Example 4.18 and Figure 4.33
This example is here for the critical point at zero, where the derivative does not exist. The algebra to see it is the longest in the lesson, so take it one line at a time: differentiate, move the negative power into the denominator, then combine over a common denominator of three x to the two thirds.
Once the derivative is a single fraction, the critical points can be read off. The numerator is zero at plus and minus one. The denominator is zero at the origin, and the function itself is defined there, so the origin is a critical point too. That gives four intervals and four test points.
The chart reads minus, plus, plus, minus. So minus one is a minimum, one is a maximum, and zero, where the signs are plus on both sides, is neither: in the figure the curve goes straight up through the origin with a vertical tangent. The check uses symmetry: both terms are odd powers, so the function is odd, and its minimum must be the mirror image of its maximum.
Prediction
\[ f(x) = \sqrt[3]{x - 1} \]
Predict first
Before computing anything: how many local extrema does this function have?
Correct: None
Why: It is the cube root graph shifted right by one, and the cube root rises everywhere. The derivative is undefined at 1, so 1 is a critical point, but the derivative is positive on both sides of it: no sign change, no extremum.
Make a prediction before any calculus. This function is the plain cube root graph, shifted one unit to the right. What does the cube root graph look like?
The cube root is an increasing function: bigger inputs always give bigger outputs, including for negative numbers. Shifting it sideways changes nothing about that, so there is no peak and no valley anywhere.
The interesting thing is that there is still a critical point. At x equals one the graph has a vertical tangent, and the derivative does not exist. So one is a legitimate candidate, and it fails the test. The next slide does the calculation that confirms your picture.
Worked example
\[ f(x) = (x - 1)^{1/3} \]
Differentiate
Why: Power rule and the chain rule; the inner derivative is 1.
\[ f'(x) = \frac13 (x - 1)^{-2/3} \]
Rewrite as a fraction
Why: The negative exponent moves down.
\[ f'(x) = \frac{1}{3(x - 1)^{2/3}} \]
Find the critical point
Why: The numerator is never zero; the denominator vanishes at 1.
\[ f'(1) \text{ undefined} \;\Longrightarrow\; x = 1 \text{ is critical} \]
Read the sign on each side
Why: The two-thirds power is a square of a cube root, so it is positive whenever x is not 1.
\[ (x - 1)^{2/3} = \left(\sqrt[3]{x-1}\right)^2 > 0 \;\Longrightarrow\; f'(x) > 0 \text{ for } x \ne 1 \]
Conclude
Why: Plus on both sides.
\[ \text{no local extrema} \]
Figure (svg): The curve y equals the cube root of x minus 1, rising everywhere with a vertical tangent at (1, 0), and dashed, its derivative, positive everywhere and shooting up at x equals 1.
Check with three values
Why: An increasing function keeps getting bigger straight through 1.
\[ f(0) = -1 < f(1) = 0 < f(2) = 1 \]
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 343 — Checkpoint 4.17
The derivative is one third times x minus one to the power minus two thirds. Written as a fraction, the power sits in the denominator, so the derivative is undefined exactly when x equals one. The numerator is the constant one, so the derivative is never zero.
The key step is the sign. The two-thirds power is the square of a cube root, and a square is positive unless it is zero. So the denominator is positive on both sides of one, and so is the derivative. Plus on both sides: no sign change, no extremum.
Look at the figure. The dashed curve is the derivative: always above the axis, shooting upward as you approach one, which is where the tangent becomes vertical. The solid curve passes through the critical point climbing, exactly as your prediction said.
Counterexample
Figure (svg): The graph of minus the absolute value of x drawn as two line pieces meeting at the origin, with a hollow circle at the origin and a filled dot below it at (0, −1).
Discussion prompt
Exercise 198 asks why the first derivative test needs f to be continuous. The function g equals minus the absolute value of x, except that g(0) is redefined to be minus 1. What does the test predict at 0, and what actually happens?
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 349 — Exercise 198
This is Exercise 198, and the answer is a function built to break the test. Away from zero it is minus the absolute value of x, an upside-down V with its tip at the origin. But at zero itself the value has been moved down to minus one. The hollow circle in the figure marks the hole; the filled dot marks where the value actually is.
Now run the test. On the left the slope is plus one, on the right it is minus one. Plus then minus, the signature of a maximum. But the value at zero is minus one, and every nearby value is close to zero, which is higher. So zero is actually a local minimum.
What went wrong is that the function jumped. The derivative tracks how the function changes on each side, but it knows nothing about a sudden drop at a single point. Continuity is exactly the hypothesis that rules such jumps out, which is why Theorem 4.9 insists on it.
Trap
A very common first line:
\[ f(x) = x^3, \quad f'(0) = 0 \]
\[ \Longrightarrow f \text{ has an extremum at } 0? \]
Wrong. A zero derivative only makes 0 a candidate.
Look at the signs on each side. Here the derivative is positive on both, so the function is rising straight through the origin and there is no extremum.
\[ f'(-1) = 3 > 0 \]
\[ f'(1) = 3 > 0 \]
This is the most common wrong line in the whole topic, and it looks completely reasonable. The derivative is zero, the tangent is flat, and flat tangents are what you see at peaks and valleys. So the conclusion feels automatic.
The correction is the sign check. For the cube function the derivative is positive at minus one and positive at one, and indeed positive everywhere except the origin. A function whose derivative never goes negative never goes downhill, so it cannot have a peak or a valley anywhere.
Make the habit concrete: never write the word maximum or minimum next to a critical point until you have written the two signs on either side of it, or used the second derivative test you will meet in Part 4. A zero derivative only puts a point on the list of suspects.
Section
Part 2
Concept
Figure (svg): Two panels. Left: an upward-opening curve with five tangent segments whose slopes increase from −1.5 to 1.5, each lying below the curve. Right: a downward-opening curve with five tangents whose slopes decrease from 1.5 to −1.5, each lying above the curve.
Concavity (definition) — Let f be differentiable on an open interval I. If the derivative is increasing on I, f is concave up on I; if the derivative is decreasing on I, f is concave down on I.
\[ f' \text{ increasing} \iff \text{concave up}, \qquad f' \text{ decreasing} \iff \text{concave down} \]
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 344 — Definition and Figure 4.34
Now the second question: not whether the graph rises or falls, but which way it bends. Look at the left panel. Read the tangent slopes from left to right: minus one and a half, minus three quarters, zero, three quarters, one and a half. They increase. A graph whose slopes increase bends upward like a cup, and that is what concave up means.
The right panel is the reverse: the slopes go from one and a half down to minus one and a half. The derivative is decreasing, and the graph bends downward like a cap. That is concave down.
Notice also where the tangents sit. On the concave up curve every tangent lies below the graph; on the concave down curve every tangent lies above it. You met that fact in Section 4.2, where it decided whether a linear approximation overestimates or underestimates. The book's definition uses the increasing slope; the tangent picture is a useful second reading of the same thing.
Intuition
A concave up curve can be falling: think of the left half of a bowl. The slopes there are negative, but they are becoming less negative, so the derivative is increasing.
\[ \text{slopes } -1.5,\; -0.75,\; 0: \quad \text{falling, yet concave up} \]
Increasing and decreasing describe the heights. Concavity describes the slopes. Keep those two questions separate and the four shapes of the next slides fall out on their own.
It is tempting to think concave up means going up. It does not. The left half of a bowl is falling and concave up at the same time: the slopes there are negative, but they are becoming less negative as you move right, so the derivative is increasing.
The cleanest way to keep the two ideas apart is to ask what each one is about. Increasing and decreasing describe the heights of the graph, so they are read from the first derivative. Concave up and concave down describe the slopes, how they change, so they are read from the second derivative.
Because the two questions are independent, a graph can be in any of four states at once: rising or falling, and bending up or bending down. The next few slides show all four, and then you will see how to read each one from two signs.
Concept
Concave up means the derivative is increasing. Apply Corollary 3 of the Mean Value Theorem to the derivative itself: it increases where its own derivative is positive.
\[ (f')' = f'' \]
\[ f'' > 0 \text{ on } I \;\Longrightarrow\; f' \text{ increasing} \;\Longrightarrow\; f \text{ concave up on } I \]
\[ f'' < 0 \text{ on } I \;\Longrightarrow\; f' \text{ decreasing} \;\Longrightarrow\; f \text{ concave down on } I \]
Test for concavity (Theorem 4.10) — For f twice differentiable on I: a positive second derivative throughout I means concave up; a negative one means concave down.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, pp. 344-345 — Theorem 4.10
The definition says concave up means the derivative is increasing. You already know how to decide when a function is increasing: check that its derivative is positive. So apply that rule to the derivative itself. Its derivative is the second derivative, and if the second derivative is positive on an interval, the first derivative is increasing there, and the graph is concave up.
Read the two middle lines as chains of reasoning, not as facts to memorise. Each arrow is one step: the second derivative's sign tells you whether the slope is rising or falling, and that tells you which way the graph bends.
This is Theorem 4.10. It turns concavity, which is a statement about a whole interval, into a sign you can compute. And sign questions are exactly what sign charts are for, so everything you did in Part 1 with the first derivative now works for the second.
Notation
Annotate
On: \( f''(x) > 0 \text{ for all } x \in I \;\Longrightarrow\; f \text{ concave up over } I \)
Take the notes in order. The first one is the conceptual point: the second derivative is the rate of change of the slope. It measures how quickly the tangent is turning, and it says nothing about whether the function itself is going up or down.
The second note is about the interval. The theorem concludes something about all of I, and so it needs the sign to hold at every point of I. A single positive value at one point only tells you about that point and a small neighbourhood of it.
The third note matters later. Where the second derivative does not exist, such as at the vertical tangent of the cube root, the theorem is silent, and concavity has to be judged from the signs on either side. That will matter when you look for inflection points, because a place where the second derivative is undefined can still be one.
Picture it
Figure (svg): A two by two grid of curve pieces. Top left, rising and bending upward. Top right, rising and bending downward. Bottom left, falling and bending upward. Bottom right, falling and bending downward. Each is labelled with the signs of f prime and f double prime.
The first derivative's sign says whether the path goes up or down; the second derivative's sign says which way it bends. Every smooth stretch of a graph is one of these four pieces.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, pp. 346-347 — Table 4.1 and Figure 4.37
This is Table 4.1 drawn as pictures. Each panel is a short piece of a smooth curve, and each one is fixed by two signs. Top left: rising and bending up, so it gets steeper as it climbs. Top right: rising but bending down, so it levels off. Bottom left: falling and bending up, so the descent eases. Bottom right: falling and bending down, so the descent gets steeper.
Say each one aloud with its signs. It helps to attach a real motion to each: a rocket accelerating upward, a car easing off as it reaches cruising height, a ball rolling to a stop at the bottom of a slope, a stone falling faster and faster.
Any smooth graph can be cut into pieces of these four kinds. The cuts happen at critical points, where the first sign changes, and at inflection points, where the second sign changes. That is why two sign charts are enough to describe a curve.
Comparison
Comparison matrix
| sign of f′ | sign of f″ | increasing or decreasing? | concavity |
|---|---|---|---|
| positive | positive | increasing | concave up |
| positive | negative | increasing | concave down |
| negative | positive | decreasing | concave up |
| negative | negative | decreasing | concave down |
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 346 — Table 4.1
Fill in the missing cells from the two readings you have: the sign of the first derivative gives increasing or decreasing, and the sign of the second gives concave up or concave down. Each column depends on exactly one of the two signs.
The row that tends to catch people is the third one, where you are given decreasing and concave up and must work backwards. Decreasing means the first derivative is negative, regardless of the concavity. Concave up means the second derivative is positive, regardless of the direction. The two do not constrain each other.
Once you see that the columns pair off, first derivative with direction and second derivative with bend, the table is not something to memorise at all. It is just two independent facts written side by side.
Matching
Match the pairs
Why: Growing is a statement about f′; the speed of that growth changing is a statement about f″. A peak is a flat tangent on a downward bend. A descent that eases off has a negative slope that is increasing toward zero.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 353 — Exercises 241 to 245
These sentences come from Exercises 241 to 245, where the book asks you to translate ordinary language into statements about the function and its first two derivatives. Match them before checking.
The population growing more slowly is still growing, so the first derivative is positive, but its growth rate is shrinking, so the second derivative is negative. The economy picking up speed is growing, and growing faster, so both are positive. Costs at their peak on July 1 mean a flat tangent that day on a curve that bends down: first derivative zero, second derivative negative.
The airplane is the subtle one. Its altitude is decreasing, so the first derivative is negative. But a smooth landing means the rate of descent is easing off toward zero, so the negative slope is increasing, and the second derivative is positive. Words like faster, slower, speeding up and easing off are almost always about the second derivative.
Section
Part 3
Concept
Inflection point (definition) — If f is continuous at a and f changes concavity at a, the point (a, f(a)) is an inflection point of f.
\[ f'' \text{ changes sign at } a \;\Longrightarrow\; (a, f(a)) \text{ is an inflection point} \]
It is a point on the graph, so give both coordinates. And it is defined by a change, just as an extremum was defined by a change in the first derivative's sign.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 345 — Definition and Figure 4.35
An inflection point is where the graph switches from bending one way to bending the other. The definition has two conditions: the function must be continuous there, and the concavity must actually change. Continuity rules out cases where the graph jumps and the concavity seems to change across a gap.
Two small habits. First, an inflection point is a point on the graph, so report both coordinates, not just the x value. Second, notice the structure: an extremum was defined by a change in the sign of the first derivative, and an inflection point is defined by a change in the sign of the second. The same kind of test, one level up.
The consequence is that everything you learned about sign charts transfers directly. Find the candidates, test the sign of the second derivative in each gap, and keep only the candidates where the sign changes.
Picture it
Figure (svg): The cubic x cubed minus 3x, drawn orange where it is concave down (left of 0) and blue where it is concave up (right of 0), with its tangent line y equals minus 3x at the origin crossing through the curve.
On the concave down side the curve lies below its tangent line; on the concave up side it lies above. At the inflection point the tangent line passes from one side of the curve to the other.
Look at the dashed line. It is the tangent to the curve at the origin, and it does something tangents to cups and caps never do: it crosses the curve. On the left, where the curve is concave down, the curve lies below the tangent. On the right, where it is concave up, the curve lies above it.
That crossing is a reliable visual test for an inflection point. On a concave down stretch every tangent sits above the graph, and on a concave up stretch every tangent sits below it. Where the concavity switches, the tangent has to switch sides, and it can only do that by passing through.
It also explains why inflection points matter in approximation. The tangent line approximation from Section 4.2 is unusually good near an inflection point, because the curve hugs the tangent on both sides instead of pulling away in one direction.
Concept
A continuous function can switch concavity only where the second derivative is zero or undefined. Those are the candidates:
\[ f''(a) = 0 \quad\text{or}\quad f''(a) \text{ undefined} \]
But a candidate need not be an inflection point. The fourth power is the standard warning:
\[ f(x) = x^4, \qquad f''(x) = 12x^2, \qquad f''(0) = 0 \]
\[ f''(x) = 12x^2 > 0 \quad \text{for } x \ne 0 \]
The second derivative is positive on both sides, so the graph is concave up throughout and the origin is not an inflection point.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 345 — concavity changes only where the second derivative is zero or undefined
The logic mirrors Part 1. A continuous function can switch concavity only where its second derivative is zero or undefined, because a continuous second derivative cannot change sign without passing through zero. So those points are the candidates.
And just as a zero first derivative did not guarantee a turn, a zero second derivative does not guarantee an inflection. The fourth power is the standard example. Its second derivative is twelve x squared, which is zero at the origin, but positive on both sides of it. The graph is concave up everywhere, and the origin is the bottom of a bowl.
So the candidate list is only the start. You must check the sign of the second derivative on each side of every candidate, and only a genuine change of sign earns the name inflection point.
Estimation
\[ f(x) = x^3 - 6x^2 + 9x + 30 \quad \text{is falling between } x = 1 \text{ and } x = 3 \]
Predict first
Between its peak at x = 1 and its valley at x = 3, where is this cubic steepest downhill?
Correct: At x = 2
Why: The slope is the derivative 3x² − 12x + 9, a parabola whose lowest point is at x = 2, where it equals −3. Before 2 the slope is getting more negative; after 2 it recovers. The steepest point is where the concavity switches, which is the next example's inflection point.
Figure (svg): The derivative 3x squared minus 12x plus 9 of the Example 4.19 cubic, a parabola with its lowest point at (2, −3).
Make an estimate before revealing. The cubic has a peak at one and a valley at three, so it falls between them. At both ends of that stretch the tangent is flat. Somewhere in the middle the fall is steepest. Where?
The figure answers it. It shows the derivative, the slope of the cubic, which is a parabola. The slope is zero at one and at three, and dips to its lowest value, minus three, exactly at two. So the cubic is falling fastest at x equals two.
Here is the connection to this part of the lesson. Before two, the slope is getting more negative, so the derivative is decreasing and the graph is concave down. After two, the slope recovers, so the graph is concave up. The steepest point is exactly where the concavity switches. An inflection point is where the slope itself reaches an extreme.
Worked example
Find the intervals of concavity and every inflection point.
\[ f(x) = x^3 - 6x^2 + 9x + 30 \]
First derivative
Why: Power rule.
\[ f'(x) = 3x^2 - 12x + 9 \]
Second derivative
Why: Differentiate again.
\[ f''(x) = 6x - 12 \]
Find the candidates
Why: The second derivative is defined everywhere, so solve for its zero.
\[ 6x - 12 = 0 \iff x = 2 \]
Test left of 2
Why: Take x equal to 0.
\[ f''(0) = -12 < 0 \;\Longrightarrow\; \text{concave down on } (-\infty, 2) \]
Test right of 2
Why: Take x equal to 3.
\[ f''(3) = 6 > 0 \;\Longrightarrow\; \text{concave up on } (2, \infty) \]
Find the height of the inflection point
Why: The concavity changes at 2, so evaluate f there.
\[ f(2) = 8 - 24 + 18 + 30 = 32 \]
Figure (svg): The cubic x cubed minus 6x squared plus 9x plus 30 from x = −0.5 to 5, orange and bending downward left of x = 2, blue and bending upward right of it, with the inflection point (2, 32) marked. Below, the sign chart of f double prime: minus then plus.
Check with the slope
Why: At an inflection point the slope should be at its extreme. The derivative is a parabola with its vertex at 2, and its value there is its minimum.
\[ f'(2) = 12 - 24 + 9 = -3, \qquad f'(1) = f'(3) = 0 > -3 \]
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, pp. 345-346 — Example 4.19 and Figure 4.36
Read the maths line by line. Two differentiations give a second derivative of six x minus twelve, a straight line. It is defined everywhere, so the only candidate is its zero, x equals two.
The test points zero and three give a negative and a positive value, so the graph is concave down to the left of two and concave up to the right. The concavity genuinely changes, so there is an inflection point, and its height is f of two, which is thirty-two. In the figure the orange piece frowns and the blue piece smiles, and the join is at the marked point.
The check connects back to the estimation slide. If two really is an inflection point, the slope should be at an extreme there. The derivative at two is minus three, while at the critical points one and three it is zero. So the slope is most negative at the inflection point, exactly as the picture of the parabola showed.
Worked example
\[ f(x) = -x^3 + \tfrac32 x^2 + 18x \]
First derivative
Why: From Checkpoint 4.16.
\[ f'(x) = -3x^2 + 3x + 18 \]
Second derivative
Why: Differentiate again.
\[ f''(x) = -6x + 3 \]
Find the candidate
Why: Solve for the zero.
\[ -6x + 3 = 0 \iff x = \tfrac12 \]
Test on each side
Why: Test points 0 and 1.
\[ f''(0) = 3 > 0, \qquad f''(1) = -3 < 0 \]
State the intervals
Why: Plus then minus.
\[ \text{up on } \left(-\infty, \tfrac12\right), \qquad \text{down on } \left(\tfrac12, \infty\right) \]
Locate the inflection point
Why: Evaluate f at one half.
\[ f\left(\tfrac12\right) = -\tfrac18 + \tfrac38 + 9 = \tfrac{37}{4} \]
Figure (svg): The cubic minus x cubed plus three halves x squared plus 18x, blue and concave up left of x = 1/2, orange and concave down right of it, with the inflection point (1/2, 37/4) marked. Below, the sign chart of f double prime: plus then minus.
Check with the cubic's symmetry
Why: A cubic's graph is symmetric about its inflection point, so the inflection must sit halfway between the critical points minus 2 and 3 found in Checkpoint 4.16.
\[ \frac{-2 + 3}{2} = \frac12 \;\checkmark \]
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 346 — Checkpoint 4.18
This is the function from Checkpoint 4.16 again, so the first derivative is already known. Differentiate once more and you get minus six x plus three, which is zero at one half.
The signs this time are the other way round from Example 4.19: positive at zero and negative at one. So the graph is concave up to the left of one half and concave down to the right. The negative leading coefficient flips everything. The inflection point is at one half, with height thirty-seven quarters, or nine point two five.
The check uses a fact worth knowing about cubics: every cubic's graph is symmetric about its inflection point, turned half a revolution. So the inflection point must sit exactly halfway between the two critical points, minus two and three, which you found in Checkpoint 4.16. Their average is one half. The two checkpoints agree.
Fill the middle
\[ f(x) = x^3 - 4x^2 + x + 2, \qquad f''(x) = 6x - 8 \]
Fill in the blanks
6x - 8 = 0 \iff x = \frac{4}{3}
Why: The second derivative is a line with positive slope, so it is negative left of 4/3 and positive right of it: concave down, then concave up, and the concavity really does change at x = 4/3.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 353 — Exercise 223
Type the x coordinate where the second derivative is zero. The second derivative has already been computed for you, so this is one line of algebra: add eight and divide by six.
The answer is four thirds. Before calling it an inflection point, though, you owe one more sentence: does the sign actually change there? It does, because six x minus eight is a line with positive slope, negative to the left of its zero and positive to the right.
That will be true for every cubic. Its second derivative is always a non-constant straight line, which has exactly one zero and changes sign there. So every cubic has exactly one inflection point, and you can find it in one line, as here.
Error analysis
Annotate
On: \( f(x) = x^4,\; f''(x) = 12x^2,\; f''(0) = 0 \;\Longrightarrow\; (0, 0) \text{ inflection} \)
Step through the annotations and find the line that is wrong. The derivative calculation is correct, and the value of the second derivative at zero is correct. The error is in the arrow, the step that jumps from a zero of the second derivative to the conclusion inflection point.
An inflection point requires the concavity to change. Twelve x squared is positive for every nonzero x, on the left of the origin and on the right. The graph of the fourth power is concave up on both sides, so nothing changes at zero.
In fact the origin is the minimum of the fourth power: a flat-bottomed bowl, flatter than a parabola, but a bowl all the same. Whoever wrote this found a candidate and forgot to test it. The fix is always the same: write the sign of the second derivative on each side before you use the word inflection.
Trap
The rule gets remembered as:
\[ f''(a) = 0 \;\Longrightarrow\; \text{inflection at } a \]
Wrong. It is the same mistake as reading every flat tangent as a turn.
An inflection point needs f″ to change sign. Check the sign on each side, exactly as the first derivative test does with f′.
\[ x^4: \quad f'' \text{ is } + , + \text{ around } 0 \]
\[ \text{no change, no inflection} \]
This is the second-derivative version of the first trap in this lesson, and it catches people for the same reason. A zero second derivative is what you usually see at an inflection point, so the reverse feels natural.
The correction is the same too. Inflection is defined by a change of sign, not by a zero. The fourth power has a zero second derivative at the origin and no change of sign, so no inflection. Its second derivative touches zero and bounces back, the same way three x squared touched zero without changing sign in the first trap.
If you want a single habit for this whole lesson, it is this: every verdict, max, min, or inflection, is read from a change of sign. Zeros only tell you where to look.
Prediction
\[ f(x) = ax^2 + bx + c, \quad a \ne 0 \]
Predict first
Can the graph of a degree 2 polynomial have an inflection point?
Correct: Never
Why: The second derivative is the constant 2a. It never changes sign, so a parabola is concave up everywhere (a positive) or concave down everywhere (a negative). A cubic's second derivative is a line, which changes sign exactly once: every cubic has exactly one inflection point.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 349 — Exercise 200
Predict before you reveal. A parabola is the simplest curve that bends at all. Can it switch the way it bends?
Differentiate twice. The second derivative of a quadratic is two times its leading coefficient, a constant. A constant never changes sign. So a parabola with a positive leading coefficient is concave up everywhere, and one with a negative leading coefficient is concave down everywhere. There is nowhere for an inflection to happen.
The same reasoning answers the next question too. A cubic's second derivative is a straight line with nonzero slope, which changes sign exactly once, so every cubic has exactly one inflection point. A quartic's second derivative is a quadratic, which can change sign twice, once, or not at all, as the fourth power showed.
Section
Part 4
Concept
Figure (svg): Two panels. Left: a hill-shaped curve with a peak at c, and a dashed straight line for its derivative falling through zero at c. Right: a valley-shaped curve with its bottom at c, and a dashed derivative line rising through zero at c.
Suppose the tangent is flat at c and the second derivative is negative there. Then the first derivative is decreasing near c, and it passes through zero at c:
\[ f'(c) = 0, \quad f' \text{ decreasing near } c \]
\[ \Longrightarrow\; f' > 0 \text{ just left of } c, \quad f' < 0 \text{ just right of } c \]
That is the plus to minus pattern, so by the first derivative test f has a local maximum at c. A positive second derivative gives minus to plus: a minimum.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 347 — the argument before Theorem 4.11 and Figure 4.38
The first derivative test needs signs on both sides of a critical point. This slide shows that when the tangent is flat, the second derivative can supply those signs for you.
Look at the left panel. The solid curve is a hill; the dashed line is its derivative. The derivative is zero at c, and because the second derivative is negative, the derivative is decreasing, so the dashed line slopes down. A line sloping down through zero is above zero on the left and below zero on the right: plus then minus, and the first derivative test says maximum.
The right panel is the mirror image. A positive second derivative makes the derivative increase, so it crosses zero going upward: minus then plus, a minimum. This is the argument the book gives just before Theorem 4.11, and it is worth understanding rather than memorising, because it tells you exactly when the shortcut is valid.
Concept
Second derivative test (Theorem 4.11) — Suppose the first derivative is zero at c and the second derivative is continuous on an interval containing c.
| at c | verdict | picture |
|---|---|---|
| f′(c) = 0, f″(c) > 0 | local minimum | a flat point on a cup |
| f′(c) = 0, f″(c) < 0 | local maximum | a flat point on a cap |
| f′(c) = 0, f″(c) = 0 | inconclusive | could be max, min or neither |
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 348 — Theorem 4.11
Here is Theorem 4.11. The hypotheses are that the tangent is flat at c and the second derivative is continuous near c. Then the sign of the single number f double prime of c decides.
The pictures in the table are the easiest way to remember which sign gives which verdict. A positive second derivative means concave up, a cup, and a flat point on a cup is its bottom: a minimum. A negative one means a cap, and a flat point on a cap is its top: a maximum.
The third row is the one people misread. If the second derivative at c is zero, the test is inconclusive. That does not mean there is no extremum; it means this test cannot tell you. The next slides show why, and what to do instead.
Notation
Annotate
On: \( f'(c) = 0,\; f''(c) < 0 \;\Longrightarrow\; f(c) \text{ is a local max} \)
Work through the notes in order. The first is a hypothesis that is easy to miss: the derivative must equal zero at c. Critical points where the derivative does not exist, like the corner of the absolute value or the cusp of the two-thirds power, are outside this theorem entirely.
The second note is why the test is popular. You evaluate one number at c. There is no sign chart, no choice of test points, and when the second derivative is simple the whole classification takes one line per critical point.
The last note is the warning. Inconclusive means exactly what it says: no conclusion. It is not a vote for neither. When you get a zero, you go back to the first derivative test, which never comes back empty.
Picture it
Figure (svg): Three panels on identical axes: x cubed, passing flat through the origin; x to the fourth, a flat-bottomed valley at the origin; minus x to the fourth, a flat-topped hill at the origin.
\[ f(x) = x^3,\; x^4,\; -x^4: \qquad f'(0) = 0, \quad f''(0) = 0 \]
One is neither, one a minimum, one a maximum, and the second derivative reads zero for all three. That is why case three of Theorem 4.11 says inconclusive.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 348 — the three functions after Theorem 4.11
Here are the book's three functions, drawn on identical axes. Each has a flat tangent at the origin, and each has a second derivative of zero there. If the second derivative test could decide, it would have to give the same answer for all three.
But the three graphs behave completely differently. The cube passes through, climbing: neither. The fourth power has a flat-bottomed valley: a minimum. Its negative has a flat-topped hill: a maximum. Same data for the test, three different truths.
That is the whole reason for case three of the theorem. A zero second derivative at a flat point is compatible with every possible outcome, so no rule based on it alone can be correct. For all three, the first derivative test settles it at once: the cube's derivative is positive on both sides, the fourth power's goes minus to plus, and its negative goes plus to minus.
Step zero
\[ f(x) = x^{2/3} \]
Discussion prompt
The function above has a critical point at 0. Before differentiating twice, can the second derivative test classify it? If not, classify it another way.
Answer before you reveal. The temptation is to differentiate twice straight away, because the second derivative test looks quicker. Pause and check its hypothesis first.
The derivative of x to the two thirds is two over three times the cube root of x, which does not exist at zero. The graph has a cusp there, a sharp point. The second derivative test requires the first derivative to equal zero at c, and here it does not exist at all, so the test simply does not apply.
The first derivative test has no such restriction. The cube root of x is negative for negative x and positive for positive x, so the derivative goes from minus to plus, and the cusp is a minimum. Checking the hypothesis before you start is the step zero this slide is about.
Worked example
\[ f(x) = x^5 - 5x^3 \]
First derivative
Why: Power rule.
\[ f'(x) = 5x^4 - 15x^2 \]
Factor it
Why: Take out 5x squared.
\[ f'(x) = 5x^2\left(x^2 - 3\right) \]
Critical points
Why: Either factor can be zero.
\[ x = 0, \qquad x = \pm\sqrt{3} \]
Second derivative, factored
Why: Differentiate, then take out 10x.
\[ f''(x) = 20x^3 - 30x = 10x\left(2x^2 - 3\right) \]
Evaluate at minus root three
Why: Here x squared is 3, so the bracket is 3.
\[ f''(-\sqrt3) = 10(-\sqrt3)(3) = -30\sqrt3 < 0 \;\Longrightarrow\; \text{max} \]
Evaluate at root three
Why: Same bracket, positive x.
\[ f''(\sqrt3) = 10\sqrt3(3) = 30\sqrt3 > 0 \;\Longrightarrow\; \text{min} \]
Evaluate at zero
Why: The factor 10x is zero.
\[ f''(0) = 0 \;\Longrightarrow\; \text{inconclusive} \]
Fall back on the first derivative test at zero
Why: Test points minus 1 and 1 lie between the critical points.
\[ f'(-1) = 5(1)(-2) = -10, \qquad f'(1) = 5(1)(-2) = -10 \]
Conclude at zero
Why: Minus on both sides: the function keeps falling through the origin.
\[ \text{no extremum at } 0 \]
Figure (svg): The curve x to the fifth minus 5x cubed from −2.5 to 2.5: a peak at x = −√3 of height about 10.4, a flat point at the origin where it keeps falling, and a valley at x = √3 of depth about −10.4.
Check the extreme values
Why: Root three cubed is 3 root three, and to the fifth is 9 root three.
\[ f(-\sqrt3) = -9\sqrt3 + 15\sqrt3 = 6\sqrt3 \approx 10.39, \qquad f(\sqrt3) = -6\sqrt3 \]
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, pp. 348-349 — Example 4.20 and Figure 4.39
This example shows the test at its best and at its limit, in one function. Factor both derivatives first: the first is five x squared times x squared minus three, and the second is ten x times two x squared minus three. Factored forms make every evaluation quick.
At minus root three and root three, x squared is three, so the bracket in the second derivative is three, and the value is minus or plus thirty root three. Negative gives a maximum, positive gives a minimum. Two points classified in two lines, with no sign chart.
At zero the second derivative is zero, and the test gives up. So fall back on the first derivative test with test points minus one and one. Both give minus ten, so the function is falling on both sides of the origin, and there is no extremum there. In the figure the curve flattens at the origin and carries on down. The check computes the extreme values: six root three, about ten point four, and its negative.
Worked example
\[ f(x) = x^3 - \tfrac32 x^2 - 18x, \qquad c = 3, \; -2 \]
Confirm the critical points
Why: Differentiate and factor.
\[ f'(x) = 3x^2 - 3x - 18 = 3(x - 3)(x + 2) \]
Second derivative
Why: Differentiate again.
\[ f''(x) = 6x - 3 \]
Evaluate at 3
Why: Positive means concave up.
\[ f''(3) = 15 > 0 \;\Longrightarrow\; \text{local min at } 3 \]
Evaluate at minus 2
Why: Negative means concave down.
\[ f''(-2) = -15 < 0 \;\Longrightarrow\; \text{local max at } -2 \]
Evaluate f
Why: Substitute each point.
\[ f(3) = 27 - \tfrac{27}{2} - 54 = -\tfrac{81}{2}, \qquad f(-2) = -8 - 6 + 36 = 22 \]
Check against Checkpoint 4.16
Why: This function is exactly minus the one in Checkpoint 4.16, so its graph is that one flipped upside down: the minimum at minus 2 must become a maximum and the maximum at 3 a minimum, with the values negated.
\[ f = -g: \quad g(-2) = -22 \to 22, \qquad g(3) = \tfrac{81}{2} \to -\tfrac{81}{2} \]
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 349 — Checkpoint 4.19
The book gives you the critical points, three and minus two, so first confirm them: the derivative factors as three times x minus three times x plus two. Then the second derivative is six x minus three, and evaluating it at each point takes one line.
At three it is fifteen, positive, so a minimum. At minus two it is minus fifteen, negative, so a maximum. The values of the function are minus eighty-one halves and twenty-two.
The check is a lovely coincidence that is not a coincidence. Compare this function with Checkpoint 4.16. Every coefficient has the opposite sign, so this function is exactly the negative of that one, and its graph is that graph turned upside down. Peaks become valleys and valleys become peaks, and the values change sign. Checkpoint 4.16 found a minimum of minus twenty-two at minus two and a maximum of eighty-one halves at three. Everything matches.
Trap
Having found a zero second derivative:
\[ f(x) = x^4, \quad f''(0) = 0 \]
\[ \Longrightarrow \text{no max or min at } 0? \]
Wrong. The test gave no answer, not the answer no.
The fourth power has a perfectly good minimum at 0. When the second derivative test is silent, run the first derivative test:
\[ f'(x) = 4x^3: \quad - \text{ then } + \]
\[ \text{so a local min at } 0 \]
This is the misreading of case three. The second derivative test has three outcomes: minimum, maximum and no information. The error is to treat no information as the answer neither.
The fourth power exposes it. Its second derivative at zero is zero, so the test is silent. But the fourth power obviously has a minimum at zero: every other value is positive. Writing no extremum would be wrong.
The correction is a reflex: when the second derivative test returns zero, switch to the first derivative test. For the fourth power the derivative is four x cubed, negative on the left and positive on the right, which settles it as a minimum. The first derivative test is the one that never comes back empty.
Tweak it
Parameter explorer
The graph shows f(x) = x³ + ax. Slide a from negative to positive. When are there two critical points, when one flat point, and when none? Where is the inflection point in every case?
\[ f(x) = x^3 + ({a})x \]
Start with a at minus three. The graph has a peak and a valley, because the derivative, three x squared plus a, has two zeros. Now slide a toward zero and watch the peak and valley move toward each other and flatten.
At a equal to zero they merge into the flat point of the plain cube function, the critical point with no turn from the start of this lesson. Push a positive and the flat point disappears: the derivative is positive everywhere, and the curve rises the whole way with no critical points at all.
Now watch the origin the whole time. The second derivative is six x no matter what a is, so the inflection point never moves. The peak and valley, when they exist, sit symmetrically on either side of it, just as the cubic symmetry check in Checkpoint 4.18 predicted.
Intuition
| first derivative test | second derivative test | |
|---|---|---|
| what you compute | the sign of f′ on every interval | one number, f″(c), at each point |
| works at corners and cusps | yes | no: needs f′(c) = 0 |
| can come back empty | never | yes, when f″(c) = 0 |
| also tells you | where f rises and falls | the concavity at c |
Use the second derivative test when the second derivative is easy and the critical points are smooth; use the first derivative test whenever it is silent, and whenever you also need the intervals of increase.
You now have two ways to classify a critical point, and each has its strengths. Read the table across each row.
The second derivative test is fast when the second derivative is easy to compute: one evaluation per critical point. But it needs the first derivative to be zero at c, so it cannot handle corners and cusps, and it goes silent whenever the second derivative is zero there. The first derivative test is slower, since it needs a full sign chart, but it works everywhere the function is continuous and it never fails to give a verdict.
There is also a bonus to the first derivative test: the sign chart you build to classify the critical points also gives you the intervals of increase and decrease, which many questions ask for anyway. So in practice, if you need the intervals, use the first derivative test; if you only need to classify smooth flat points, try the second.
Section
Part 5
Pattern
Figure (svg): The Example 4.17 cubic again, now with the inflection point (1, −12) marked between the peak and the valley, and two sign charts: f prime plus, minus, plus with breaks at −1 and 3; f double prime minus, plus with a break at 1.
Here is Example 4.17 again, now with its second chart. The first chart is the one you built earlier: plus, minus, plus, giving the peak at minus one and the valley at three. The second derivative is six x minus six, so the second chart has one cut, at one, with minus then plus.
Look at how the two charts carve the curve into pieces. Left of minus one it rises and bends down. Between minus one and one it falls and bends down. Between one and three it falls and bends up. Right of three it rises and bends up. Those are four of the shapes from Table 4.1, in order, and the inflection point at one, height minus twelve, is exactly halfway between the peak and the valley.
That is the routine in the list. Two derivatives, two charts, keep only real sign changes, and evaluate the function at every special point. The next three examples run it on functions with a little more going on.
Worked example
\[ f(x) = x^4 - 6x^3 \]
First derivative
Why: Power rule.
\[ f'(x) = 4x^3 - 18x^2 \]
Factor
Why: Take out 2x squared.
\[ f'(x) = 2x^2(2x - 9) \]
Critical points
Why: Each factor gives one.
\[ x = 0, \qquad x = \tfrac92 \]
Signs on three intervals
Why: Test points minus 1, 1 and 5.
\[ f'(-1) = -22, \quad f'(1) = -14, \quad f'(5) = 50 \]
Read the chart
Why: Minus, minus, plus.
\[ \text{decreasing on } \left(-\infty, \tfrac92\right), \quad \text{increasing on } \left(\tfrac92, \infty\right) \]
Classify
Why: No change at 0; minus to plus at nine halves.
\[ \text{neither at } 0, \qquad \text{min at } x = \tfrac92 \]
Evaluate the minimum
Why: Write both powers over 16.
\[ f\left(\tfrac92\right) = \frac{6561}{16} - \frac{8748}{16} = -\frac{2187}{16} \]
Check why 0 is not an extremum
Why: The factor 2x squared is never negative, so near 0 the sign of the derivative is the sign of 2x minus 9, which is negative on both sides.
\[ 2x^2 \ge 0, \quad 2x - 9 < 0 \text{ near } 0 \;\Longrightarrow\; f' \le 0 \text{ on both sides} \]
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 353 — Exercise 226
This is the kind of full analysis the exercise set asks for, split over two slides. Part one is the first derivative. It factors as two x squared times two x minus nine, so the critical points are zero and nine halves.
The test points minus one, one and five give minus twenty-two, minus fourteen and fifty. So the function falls all the way to nine halves and rises after it. Nine halves is a minimum, with value minus two thousand one hundred eighty-seven sixteenths, about minus one hundred thirty-seven. Zero is a critical point with no sign change, so it is not an extremum.
The check explains why zero is not an extremum without any test points. The factor two x squared is never negative, so near zero the derivative has the sign of the other factor, two x minus nine, which is negative on both sides. A squared factor in a derivative is a flag: that critical point will not be a turn.
Worked example
\[ f'(x) = 4x^3 - 18x^2 \]
Second derivative
Why: Differentiate again.
\[ f''(x) = 12x^2 - 36x \]
Factor
Why: Take out 12x.
\[ f''(x) = 12x(x - 3) \]
Candidates
Why: Both factors give one.
\[ x = 0, \qquad x = 3 \]
Signs on three intervals
Why: Test points minus 1, 1 and 4.
\[ f''(-1) = 48, \quad f''(1) = -24, \quad f''(4) = 48 \]
Read the chart
Why: Plus, minus, plus: both candidates are real changes.
\[ \text{up on } (-\infty, 0), \quad \text{down on } (0, 3), \quad \text{up on } (3, \infty) \]
Inflection points
Why: Evaluate f at both.
\[ f(0) = 0, \qquad f(3) = 81 - 162 = -81 \]
Figure (svg): The quartic x to the fourth minus 6x cubed from −1.5 to 6.3: flat but still falling at the origin, bending over at (3, −81), bottoming out at (9/2, −2187/16). Below, two sign charts: f prime is minus, minus, plus with breaks at 0 and 9/2; f double prime is plus, minus, plus with breaks at 0 and 3.
Check with Exercise 197
Why: The origin is a flat critical point AND an inflection point, which is allowed; what a twice-differentiable function cannot have is an inflection point that is also an extremum, and 0 is not an extremum here.
\[ (0, 0): \; f'(0) = 0, \; f'' \text{ changes sign}, \; f' \text{ does not} \]
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, pp. 349, 353 — Exercises 197 and 226
Now the second derivative, which factors as twelve x times x minus three. The candidates are zero and three, and the test points minus one, one and four give plus, minus, plus. Both candidates are genuine sign changes, so both are inflection points: the origin and the point three, minus eighty-one.
Look at the figure with both charts under it. The origin is special: it is a critical point, where the tangent is flat, and it is also an inflection point, where the bend changes. The curve comes down, flattens for an instant while switching from a cup to a cap, and carries on down.
The check links to Exercise 197, which asks whether a point can be both an inflection point and a local extremum of a twice-differentiable function. It cannot. At an inflection point the slope reaches an extreme value; if that value is zero, the slope is on the same side of zero just before and just after, so the first derivative does not change sign and the function does not turn. Here the origin is an inflection point and a critical point, but not an extremum, so there is no conflict.
Worked example
\[ f(x) = x^2 + \cos x \]
First derivative
Why: The derivative of cosine is minus sine.
\[ f'(x) = 2x - \sin x \]
Find the critical points
Why: The sine of x is smaller in size than x itself whenever x is not zero, so it can never cancel 2x.
\[ |\sin x| < |x| < 2|x| \quad \text{for } x \ne 0 \;\Longrightarrow\; x = 0 \text{ only} \]
Second derivative
Why: The derivative of minus sine is minus cosine.
\[ f''(x) = 2 - \cos x \]
Bound the second derivative
Why: Cosine is at most 1.
\[ f''(x) = 2 - \cos x \ge 2 - 1 = 1 > 0 \]
Concavity
Why: Positive everywhere.
\[ \text{concave up on } (-\infty, \infty), \quad \text{no inflection points} \]
Classify the critical point
Why: Second derivative test.
\[ f'(0) = 0, \quad f''(0) = 1 > 0 \;\Longrightarrow\; \text{min } f(0) = 1 \]
Figure (svg): Two panels. Left: the curve x squared plus cos x, a single smooth valley with its bottom at (0, 1). Right: its second derivative 2 minus cos x, a wave that never drops below the dashed line at height 1.
Check with nearby values
Why: A minimum of 1 at 0 means nearby values are larger.
\[ f(\pm 1) = 1 + \cos 1 \approx 1.540 > 1 \]
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 353 — Exercise 222
This function mixes a polynomial and a trigonometric term, so the critical points cannot be found by factoring. The derivative is two x minus the sine of x, and the argument is a size comparison: for every nonzero x, the sine of x is smaller in size than x, and so certainly smaller than two x. The two terms can only cancel at zero.
The second derivative is two minus the cosine of x. Since the cosine is never more than one, this is never less than one. So the graph is concave up everywhere and has no inflection points at all; the right panel of the figure shows the second derivative waving but never dipping below the dashed line.
A curve that bends upward everywhere can have at most one flat point, and it must be the bottom. The second derivative test confirms it: at zero the second derivative is one, positive, so the only critical point is a minimum with value one. The check compares with the values at plus and minus one, about one point five four.
Worked example
\[ f(x) = e^{x}\sin x, \qquad -\pi \le x \le \pi \]
First derivative
Why: Product rule.
\[ f'(x) = e^{x}\sin x + e^{x}\cos x = e^{x}(\sin x + \cos x) \]
Critical points
Why: The exponential is never zero, so the bracket must be.
\[ \sin x = -\cos x \iff \tan x = -1 \iff x = -\tfrac{\pi}{4}, \; \tfrac{3\pi}{4} \]
Signs of the first derivative
Why: Test points minus pi over 2, 0 and pi.
\[ f'\left(-\tfrac{\pi}{2}\right) = -e^{-\pi/2} < 0, \quad f'(0) = 1 > 0, \quad f'(\pi) = -e^{\pi} < 0 \]
Classify
Why: Minus to plus, then plus to minus.
\[ \text{min at } -\tfrac{\pi}{4}, \qquad \text{max at } \tfrac{3\pi}{4} \]
Second derivative
Why: Product rule again, then collect terms.
\[ f''(x) = e^{x}(\sin x + \cos x) + e^{x}(\cos x - \sin x) = 2e^{x}\cos x \]
Inflection candidates
Why: Cosine is zero at plus or minus pi over 2.
\[ x = \pm\tfrac{\pi}{2} \]
Signs of the second derivative
Why: Test points minus 3 pi over 4, 0 and 3 pi over 4.
\[ f''\left(-\tfrac{3\pi}{4}\right) < 0, \quad f''(0) = 2 > 0, \quad f''\left(\tfrac{3\pi}{4}\right) < 0 \]
Read the concavity
Why: Minus, plus, minus: both candidates are inflection points.
\[ \text{down, up, down; inflections at } x = \pm\tfrac{\pi}{2} \]
Figure (svg): The curve e to the x times sin x from −π to π: nearly flat and slightly negative on the left with a shallow minimum at −π/4, then climbing to a peak near 7.46 at 3π/4 before dropping to 0 at π. Below, sign charts: f prime is minus, plus, minus with breaks at −π/4 and 3π/4; f double prime is minus, plus, minus with breaks at −π/2 and π/2.
Check the values numerically
Why: The minimum is negative and tiny, the maximum large, and the inflection heights are minus and plus e to the power of plus or minus pi over 2.
\[ f\left(-\tfrac{\pi}{4}\right) \approx -0.322, \quad f\left(\tfrac{3\pi}{4}\right) \approx 7.460, \quad f\left(\pm\tfrac{\pi}{2}\right) \approx -0.208, \; 4.810 \]
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 353 — Exercise 237
This is the longest example, so take it in two halves. The first derivative, by the product rule, is e to the x times the sum of sine and cosine. The exponential is never zero, so the critical points are where sine equals minus cosine, which on this interval means x is minus a quarter of pi or three quarters of pi.
The test points give minus, plus, minus, so there is a minimum at minus a quarter of pi and a maximum at three quarters of pi. The second derivative collects neatly to two e to the x times the cosine of x, which is zero at plus and minus half of pi and changes sign at both: down, up, down.
Now look at the figure. On the left the exponential factor squashes everything so close to zero that the minimum and the left inflection point are invisible at this scale. This is exactly why the analysis matters: the charts find features the picture hides. The check lists the numerical values, a minimum of about minus zero point three two and a maximum of about seven point four six.
Discrimination
\[ f \text{ is a cubic}, \qquad f'(1) = 0, \quad f'(3) = 0 \]
Sort into buckets
Exercises 246 to 248. Sort each statement about this cubic.
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 353 — Exercises 246 to 248
These are Exercises 246 to 248. You are told only that a cubic has flat tangents at one and three. Sort each statement by whether it must be true or can fail.
The derivative is a quadratic with zeros at one and three, so it is a multiple of x minus one times x minus three. Its own derivative, the second derivative of f, is zero at the midpoint two, so the second statement must hold. Both zeros are simple, so the first derivative changes sign at each, and one is a maximum and the other a minimum. And any cubic runs off to infinity in one direction, so no point can be an absolute maximum.
The first statement is the one that fails. The flat points say nothing about height, because adding a constant to f changes no derivative. Lift the graph high enough and it never touches the axis between one and three.
Real world
Figure (svg): Two panels over 45 days. Left: an S-shaped cumulative total climbing from near 0 to near 1000, with its inflection point at day 20, height 500. Right: the daily rate, a single hump peaking at 62.5 per day on day 20.
Discussion prompt
In an outbreak, C(t) is the running total of cases after t days. News reports say 'the curve is bending'. In derivative language, what happens at the inflection point of C, and why is that the day health officials watch for, even though the total is still rising?
OpenStax Calculus Volume 1, §4.5 Derivatives and the Shape of a Graph §4.5, p. 353 — Exercise 241, the population growing more slowly
The left panel is a running total, such as the cumulative number of cases in an outbreak. It is always rising, so its first derivative is positive throughout. But it bends upward at first, as the daily count grows, and then bends over, as the daily count falls. The switch happens at day twenty, the inflection point.
The right panel is the daily rate, the derivative of the total. It peaks exactly at day twenty, at sixty-two and a half cases per day. That is the connection you saw in Example 4.19: an inflection point of a function is an extreme point of its derivative.
This is why officials watch for the curve to bend. The total is still rising on day twenty-one, so it is easy to miss, but the rate of new cases has begun to fall. The inflection point is the first mathematical evidence that things have turned, long before the total levels off.
Check
Check your understanding
The derivative of a continuous function is f′(x) = (x + 2)²(x − 1). What does the first derivative test say at x = −2 and at x = 1?
Answer: B
Why: The squared factor is never negative, so near −2 the sign of f′ is the sign of x − 1, which is negative on both sides: no sign change, no extremum. At 1 the factor x − 1 goes from negative to positive while the square stays positive: minus to plus, a local minimum.
Choose an answer before revealing. The derivative is given already factored, so you do not need to differentiate anything; you only need to read signs.
The factor x plus two is squared, so it is never negative. Near minus two, the sign of the whole derivative is the sign of x minus one, which is negative on both sides of minus two. No change, so no extremum. At one, the squared factor is positive and x minus one goes from negative to positive. Minus then plus, so a local minimum.
If you picked an answer with an extremum at minus two, look again at the lesson from Exercise 226: a squared factor in the derivative marks a critical point where the sign does not change. It is the same situation as the cube function at the origin.
Check
Check your understanding
For f(x) = x⁴ − 4x³, which list gives every inflection point?
Answer: A
Why: f″(x) = 12x² − 24x = 12x(x − 2) is positive left of 0, negative between 0 and 2, positive right of 2. Both zeros are real sign changes, so both are inflection points: f(0) = 0 and f(2) = 16 − 32 = −16. The point x = 3 is the minimum, a zero of f′, not of f″.
Work it out on paper before choosing. Differentiate twice, factor, find the candidates and test the sign on each side.
The second derivative is twelve x times x minus two. It is positive left of zero, negative between zero and two, and positive right of two. Both zeros are real sign changes, so both are inflection points. The heights are zero and minus sixteen.
The distractors test two different confusions. One mixes up the zeros of the first derivative with the zeros of the second: three is where the first derivative changes sign, so it is the minimum, not an inflection point. The other drops the origin because it is also a flat critical point, but as in Exercise 226, a point can be both.
Check
Check your understanding
At a critical point c, f′(c) = 0 and f″(c) = 0. What can you conclude?
Answer: C
Why: Case three of Theorem 4.11: the test is inconclusive. x⁴, −x⁴ and x³ all have f′(0) = f″(0) = 0 and show a minimum, a maximum and neither. Only the signs of f′ around c decide.
Choose before revealing. This is the one situation the second derivative test cannot handle, and the question is what you are entitled to say.
Nothing yet. Case three of Theorem 4.11 is inconclusive, and the three functions from the visual slide prove it: the cube, the fourth power and its negative all have both derivatives zero at the origin, and they give neither, a minimum and a maximum.
Each wrong answer draws a conclusion that one of those three functions breaks. Neither is broken by the fourth power. A minimum is broken by its negative. And an inflection point is broken by the fourth power again, which bends upward on both sides. The right move is always the first derivative test.
Explain it to yourself
Discussion prompt
Explain, in two or three sentences and without formulas, why a positive second derivative at a flat critical point forces the first derivative to go from negative to positive there.
Write your explanation in words, without formulas. The aim is to see that the second derivative test is not a separate piece of magic but a shortcut for the first derivative test.
The chain goes like this. A positive second derivative means the slope is increasing. An increasing quantity that is exactly zero at c must be negative just before c and positive just after. So the function falls into c and rises out of it, and that is a minimum.
If you can say that chain aloud, you also know exactly when the shortcut fails. If the second derivative at c is zero, you no longer know that the slope is increasing or decreasing at c, and the chain breaks at its first link. Then you go back to the signs themselves.
Exit ticket
\[ f(x) = x^3 - 12x \]
Discussion prompt
Find every local extremum using the second derivative test, and find the inflection point.
Try this on paper in under five minutes. It uses every idea in the lesson: factor the first derivative, classify with the second derivative test, and find where the second derivative changes sign.
The derivative factors as three times x minus two times x plus two, so the critical points are plus and minus two. The second derivative is six x: twelve at two, a minimum of minus sixteen, and minus twelve at minus two, a maximum of sixteen. Six x changes sign at zero, so the origin is the inflection point.
Notice again that the inflection point sits exactly halfway between the maximum and the minimum. For a cubic that is always true, and it makes a quick check on any cubic you analyse.
Recap
| tool | what it reads | what it concludes |
|---|---|---|
| sign of f′ | f′ on an interval | increasing (+) or decreasing (−) |
| first derivative test | the sign of f′ on each side of c | max (+ to −), min (− to +), neither (no change) |
| test for concavity | f″ on an interval | concave up (+) or down (−) |
| inflection point | a change of sign of f″ at a | (a, f(a)) where the bend reverses |
| second derivative test | f″(c) where f′(c) = 0 | min (+), max (−), inconclusive (0) |
Every conclusion comes from a sign change or a sign, never from a zero alone. Next, Section 4.6 looks at the ends of the graph, and the two sign charts become a full curve sketch.
Stewart, Calculus: Early Transcendentals 8e, §4.3 How Derivatives Affect the Shape of a Graph §4.3, pp. 293-303 — the same material in Stewart
Read the table from top to bottom and notice the pattern in the last column. Every conclusion comes from a sign or a change of sign. The first derivative's sign gives direction; its change of sign gives extrema. The second derivative's sign gives concavity; its change of sign gives inflection points.
Zeros only ever tell you where to look. A zero first derivative made the cube function's origin a suspect that turned out innocent, and a zero second derivative did the same for the fourth power. Every time you are tempted to conclude from a zero, write the two signs on either side first.
Section 4.6 adds limits at infinity and asymptotes, the behaviour at the far ends of the graph. Put that together with the two sign charts from this lesson, and you can sketch almost any function you meet in this course from its formula alone. Stewart's Section 4.3 covers the same material if you want a second explanation.
Want this taught 1-on-1? Alexander tutors Calculus I — $55/session, free consultation.