NOTE: You may view the original PDFs of the notes written in class by clicking on the link for the 1:30 Section or 2:30 Section, respectively.
Lecture 18: Concavity and The 2nd Derivative Test
Goal: Use the 2nd derivative to determine properties of the original function.
Concave Up [CU] vs Concave Down [CD]
Figure 1 shows an example of a graph that is concave up. Formally, a function is concave up if for any two points on the graph, the line segment connecting them is above the graph. However, informally, a function is concave up if it is bowed upward to resemble a cup or a bowl. If we draw a bunch of tangent lines, we see their slope are increasing as we read the graph from left to right. We saw last lecture is a function is increasing, its derivative is positive. Hence, if a function \(f\) is concave up, then \([f']' = f''>0\).
Figure 1
Figure 2
We want a way to determine the concavity of a function on an interval. Ideally, without having to fiddle around with inequalities.
Theorem (C Test)
Let \( f \) be a function where \( f'' \) exists on an open interval \( I \)
- If \( f''(x) > 0 \) for \( x \) in \( I \), \( f \) is CU on \( I \).
- If \( f''(x) < 0 \) for \( x \) in \( I \), \( f \) is CD on \( I \).
There is a diagram to help remember the results of the C Test.
Figure
Example 1
Discuss the shape of \( f(x) = x^3 - 12x + 1 \)
Step 1 Find when \( f' = 0 \) and \( f'' = 0 \).
\begin{align*} f'(x) =&~ 3x^2 - 12\\[0.5em] =&~ 3(x-2)(x+2) \overset{\text{set}}{=} 0 \\[0.5em] \Rightarrow x =&~ \pm 2 \end{align*} \begin{align*} f''(x) = 6x \overset{\text{set}}{=} 0\\[0.5em] \Rightarrow x =&~ \pm 0 \end{align*}Step 2 Make sign chart for both \( f' \) and \( f'' \)
| \( (-\infty,-2) \) | \( (-2,0) \) | \( (0,2) \) | \( (2,\infty) \) | |
|---|---|---|---|---|
| Test Points | \(-100\) | \(-1\) | \(1\) | \(100\) |
| Sign of \( 3(x-2)(x+2) = f' \) | \(+\) | \(-\) | \(-\) | \(+\) |
| Sign of \( 6x = f'' \) | \(-\) | \(-\) | \(+\) | \(+\) |
| Results of the ID Test | Inc | Dec | Dec | Inc |
| Results of C Test | CD | CD | CU | CU |
Conclusion:
- \( f \) is increasing and CD on \( (-\infty, -2) \)
- \( f \) is decreasing and CD on \( (-2, 0) \)
- \( f \) is decreasing and CU on \( (0, 2) \)
- \( f \) is increasing and CU on \( (2, \infty) \)
Figure 3
We saw last lecture that when a function changed from increasing to decreasing, interesting things can happen (mainly there is a relative extrema there). Where the concavity changes is also a point of interest.
Definition: If \( f \) changes concavity at a point \( P \), then \( P \) is called an inflection point (IP).
The function in Example 1 has one inflection point, labeled as \("P"\) in Figure 3. There, it is the location of where the derivative reaches a minimum and starts increasing again. This causes, well, an inflection in the graph. As with relative extrema, we want a way to locate inflection points. To do so, we have an analog of Fermat's Theorem.
Theorem:
If \( (c, f(c)) \) is an inflection point, then either \( f''(c) = 0 \) or undefined.
Why? \( c \) is the location of a relative extrema for \(f'\). Then just use Fermat's Theorem.
NOTE:
As with Fermat's Theorem, the reverse is not true. For example, take \(x^4\) and \(\sqrt{|x|}\); their second derivative at \(0\) is \(0\) and undefined respectively. However, they do not have an inflection point at \(0\) since concavity does not change at \(0\).
Figure 4
Figure 5
Example 2
Determine the locations of all inflection points for \( f(x) = 3x^4 - 4x^3 - 6x^2 + 12x + 1 \)
Step 1 Find when \( f'' \) is 0 or undefined
\[ f'(x) = 12x^3 - 12x^2 + 12x + 12 \]\[ \begin{aligned} f''(x) &= 12\cdot 3x^2 - 12\cdot 2x + 12\cdot 1 \\ &= 12(3x^2 - 2x + 1) \\ &= 12(x-1)(3x+1) \overset{\text{set}}{=\!=} 0 \end{aligned} \]Thus, \( x = -\frac{1}{3}, 1 \) are the only possible locations.
Step 2 Make sign chart for \( f'' \)
| \( \left(-\infty, -\frac{1}{3}\right) \) | \( \left(-\frac{1}{3}, 1\right) \) | \( (1,\infty) \) | |
|---|---|---|---|
| Test Points | \( -100 \) | \( 0 \) | \( 100 \) |
| Sign of 12 | \(+\) | \(+\) | \(+\) |
| Sign of \( x-1 \) | \(-\) | \(-\) | + |
| Sign of \( 3x+1 \) | \(-\) | \(+\) | \(+\) |
| Sign of \( f'' \) | \(+\) | \(-\) | \(+\) |
| Results of C Test | CU | CD | CU |
Conclusion: \( f \) has inflection points at both \( x = -\frac{1}{3} \) and \( x = 1 \)
Example 3
Find all inflection points of
\[ f(x) = \frac{1}{20}x^5 - \frac{1}{6}x^4 + \frac{1}{6}x^3 - x + 42 \]Step 1
\[ f'(x) = \frac{1}{4}x^4 - \frac{2}{3}x^3 + \frac{1}{2}x^2 - 1 \] \[ \begin{aligned} f''(x) &= x^3 - 2x^2 + x \\ &= x(x^2 - 2x + 1) \\ &= x(x-1)^2 \overset{\text{set}}{=\!=} 0\\ \Rightarrow x &= 0, 1 \end{aligned} \]Step 2
| \( (-\infty, 0) \) | \( (0,1)\) | \( (1, \infty) \) | |
|---|---|---|---|
| Test Point | \( -100 \) | \( \frac{1}{2} \) | \( 100 \) |
| Sign of \( x \) | − | + | + |
| Sign of \( (x-1)^2 \) | + | + | + |
| Sign of \( f'' \) | − | + | + |
| Results of C Test | CD | CU | CU |
Conclusion: \( f \) has only one IP at \( (0, f(0)) = (0, 42) \)
Locating and Classifying Relative Extrema
Concavity can be used to detect relative extrema. Let \( c \) be a critical number of \( f \). If \(f''(c) > 0\), then \(f\) is concave down near \(c\) implying there is a relative minimum at \(c\) (See Figure 6).
Figure 6
Likewise, if \(f''(c) < 0\), then \(f\) is concave down near \(c\) implying there is a relative minimum at \(c\) (See Figure 7).
Figure 7
Theorem (The 2nd Derivative Test)
Let \( f'' \) be continuous near \( c \) and \( f'(c) = 0 \).
- If \( f''(c) > 0 \), then \( f \) has a rel. min. at \( c \)
- If \( f''(c) < 0 \), then \( f \) has a rel. max at \( c \)
- If \( f''(c) = 0 \), the test is inconclusive. [I.e., the test failed and you'll need to pick another test. This is not the same thing as saying \(f\) has neither a relative max nor min at \(c\)]
Why (3)?
Take for example, \(x^4,~-x^4\), and \(x^3\). In each case, their first and second derivatives at \(0\) are \(0\). However, \(x^4\) has a relative minimum at \(0\), \(-x^4\) has a relative maximum at \(0\), and \(x^3\) has neither at \(0\) (See Figures 8-10). Hence, if \(f''(c) = 0\), then we gain no information about relative extrema at that point.
Figure 8
Figure 9
Figure 10
Example 4
Locate and classify all relative extrema for \( f(x) = x^3 - 3x^2 \)
Step 1 Solve \( f'(x) = 0 \)
\[ f'(x) = 3x^2 - 6x = 3x(x-2) \overset{\text{set}}{=\!=} 0 \]So relative extrema could occur at \( x = 0, 2 \).
Step 2 Determine the sign of \( f''(0) \) and \( f''(2) \)
\[ \begin{gathered} f''(x) = 6x - 6 \\[0.5em] f''(0) = -6 < 0 \\[0.5em] f''(2) = 6 > 0 \end{gathered} \]Conclusion: \( f \) has a relative maximum at \( x = 0 \) and a relative minimum at \( x = 2 \).
Example 5
Locate and classify all relative extrema for \( f(x) = e^x(x^2 - 5x - 13) \)
-
Find when \(f'=0\)
\[ \begin{aligned} f'(x) &= e^x(x^2 - 5x - 13) + e^x(2x - 5) \\ &= e^x(x^2 - 3x - 18) \\ &= e^x(x - 6)(x + 3) \overset{\text{set}}{=\!=} 0 \Rightarrow x = -3, 6 \end{aligned} \] -
Compute the sign of \( f''(-3) \) and \( f''(6) \)
\[ \begin{aligned} f''(x) &= e^x(x^2 - 3x - 18) + e^x(2x - 3) \\ &= e^x(x^2 - x - 21) \end{aligned} \] \[ \begin{gathered} f''(-3) = e^{-3}(9 + 3 - 21) = -9e^{-3} < 0 \\[0.5em] f''(6) = e^6(36 - 6 - 21) > 0 \end{gathered} \]
Conclusion: \(f\) has a relative maximum at \( x = -3 \) and relative minimum at \( x = 6 \).