Why Derivative Tables Require Care
A derivative table can summarize how a function behaves, but what it proves depends on what the entries mean. A sign recorded throughout an interval is stronger evidence than a derivative value listed at one point. Confusing those types of information can lead to unsupported claims about concavity or inflection points.
For example, knowing that \(f''(2)>0\) tells us the second derivative is positive at \(x=2\). It does not, by itself, tell us that \(f''\) is positive on an interval around 2, or that \(f\) is concave up on an entire interval. Similarly, two values of \(f'\) can show a net increase between their input values without showing that \(f'\) increased at every point between them.
What Concavity Means in a Derivative Table
On an interval, a differentiable function is concave up when its tangent slopes increase as \(x\) increases; it is concave down when those slopes decrease. Since the tangent slope is \(f'(x)\), interval-wide information that \(f'\) increases or decreases directly describes concavity.
If \(f''\) exists on an interval, the Second Derivative Concavity Theorem says that \(f''(x)>0\) throughout the interval implies \(f\) is concave up there, while \(f''(x)<0\) throughout implies \(f\) is concave down. A table may also give \(f''=0\) or \(f''\) undefined at particular inputs; neither fact alone establishes an inflection point.
A Justification Using First-Derivative Information
Why does an increasing first derivative indicate concavity up? The Mean Value Theorem gives a direct explanation. Suppose \(f\) is continuous on \([a,c]\), differentiable on \((a,c)\), and \(f'\) is increasing there. Choose any \(a<b<c\). Applying the Mean Value Theorem on \([a,b]\) gives a number \(u\in(a,b)\) such that
Applying it again on \([b,c]\) gives a number \(v\in(b,c)\) such that
Because \(u<b<v\) and \(f'\) is increasing, \(f'(u)\leq f'(v)\). Therefore the average slope from \(a\) to \(b\) is no greater than the average slope from \(b\) to \(c\). This ordered-slope behavior is the defining geometric pattern of concavity up. If \(f'\) is decreasing, the inequalities reverse and the function is concave down. The same reasoning applies on any interval where the stated conditions hold.
Worked Example: Reading an Interval Sign Table for the Second Derivative
Worked Example: Reading an Interval Sign Table for the Second Derivative
Suppose \(f\) is continuous on \([-3,4]\), and a table reports the following interval signs and point values:
| Location | Information |
|---|---|
| \((-3,-1)\) | \(f''(x)<0\) throughout |
| \(x=-1\) | \(f''(-1)=0\) |
| \((-1,2)\) | \(f''(x)>0\) throughout |
| \(x=2\) | \(f''(2)=0\) |
| \((2,4)\) | \(f''(x)>0\) throughout |
On \((-3,-1)\), \(f''<0\), so \(f\) is concave down. On \((-1,2)\), \(f''>0\), so \(f\) is concave up. The concavity changes at \(x=-1\). Since \(f\) is continuous there, \((-1,f(-1))\) is an inflection point.
At \(x=2\), the table gives \(f''(2)=0\), but \(f''\) is positive on both adjacent intervals. The function is concave up on both sides, so the table does not support an inflection point at \(x=2\).
Check: The sign comparison is negative on the left of \(-1\) and positive on the right, so concavity changes there. At 2 the comparison is positive on both sides, so concavity does not change there. The conclusions follow from the interval signs, not just from the two zero values.
Worked Example: What Two First-Derivative Values Establish
Worked Example: What Two First-Derivative Values Establish
Assume \(f'\) is continuous on \([0,2]\) and differentiable on \((0,2)\). A table gives \(f'(0)=2\), \(f'(1)=5\), and \(f'(2)=1\). What can be concluded about concavity?
From \(0\) to \(1\), the first derivative changes by \(5-2=3\), over an interval of length \(1-0=1\). The Mean Value Theorem applied to \(f'\) guarantees some \(c\in(0,1)\) such that
From \(1\) to \(2\), the change is \(1-5=-4\), over a length of \(2-1=1\). There is some \(d\in(1,2)\) such that
Thus \(f''\) is positive at at least one point in \((0,1)\), and negative at at least one point in \((1,2)\). These results do not show that \(f''\) is positive throughout \((0,1)\) or negative throughout \((1,2)\). Nor do they locate an inflection point. The calculations check: the slopes of the secant lines for \(f'\) are exactly \(3\) and \(-4\), as given by the differences in the table.
Worked Example: A Table of First-Derivative Trends
Worked Example: A Table of First-Derivative Trends
Suppose \(f\) is continuous and differentiable on an interval containing \([-2,3]\), and the table describes how \(f'\) behaves on each interval:
| Interval | Behavior of \(f'\) |
|---|---|
| \((-2,0)\) | Decreasing throughout |
| \((0,1)\) | Increasing throughout |
| \((1,3)\) | Decreasing throughout |
On \((-2,0)\), tangent slopes decrease, so \(f\) is concave down. On \((0,1)\), tangent slopes increase, so \(f\) is concave up. On \((1,3)\), tangent slopes decrease again, so \(f\) is concave down.
The reported behavior therefore shows a concavity change at \(x=0\), from down to up, and another at \(x=1\), from up to down. Provided \(f\) is defined and continuous at each boundary, the corresponding points on the graph are inflection points. No particular value of \(f'(0)\), \(f'(1)\), or \(f''\) at those inputs is needed for this conclusion: the interval-wide trends are the evidence.
Check: The trend sequence is decreasing, increasing, decreasing. Translating those trends into concavity gives down, up, down, so each of the two boundaries separates intervals with opposite concavity.
Isolated Second-Derivative Samples and Sparse Data
A table can instead list \(f''\) only at selected points. For instance, suppose \(f''(0)=-2\), \(f''(1)=0\), and \(f''(2)=3\). These entries establish the stated values at \(0\), \(1\), and \(2\), but do not determine the sign of \(f''\) throughout \((0,1)\) or \((1,2)\). Without more information, the entries do not establish concavity on either interval or identify an exact inflection point.
Additional assumptions can strengthen what follows. If \(f''\) is continuous on \([0,2]\), its negative value at 0 implies it remains negative in some neighborhood of 0, and its positive value at 2 implies it remains positive in some neighborhood of 2. The change between those signs guarantees at least one zero of \(f''\) between 0 and 2. But the table still does not specify that zero’s location, or prove there is only one. A zero at the listed point \(x=1\) is not automatically the sign-changing zero.
When a table is sparse, report the strongest conclusion justified by the entries and assumptions. If it only gives isolated values, say that interval-wide concavity is inconclusive. Do not silently treat a list of samples as if it gave a continuous sign chart.
A Reliable Reading Process
Check whether the entries describe \(f''\) signs, \(f'\) trends, or derivative values at isolated inputs.
A value at \(x=a\) is point information. A sign or monotonicity statement “throughout” an interval is interval information.
Positive \(f''\) or increasing \(f'\) indicates concave up; negative \(f''\) or decreasing \(f'\) indicates concave down.
A candidate boundary is an inflection point only if the concavity actually changes there and the point belongs to the graph.
If the table gives too little information to establish an interval sign or a change in concavity, say the conclusion is inconclusive.
Common Mistake: Treating a Zero as an Inflection Point
Suppose a table says \(f''(4)=0\), with \(f''(x)>0\) both just to the left and just to the right of 4. A tempting but incorrect calculation is: “\(f''(4)=0\), so \(x=4\) is an inflection point.” This applies a candidate test as though it were a sufficient condition.
The correction is to compare the concavity on both sides. Since \(f''\) is positive on both sides, the function is concave up on both sides. There is no concavity change, so \(x=4\) is not an inflection-point location on this evidence. The zero identifies a place to check; the sign behavior decides whether concavity changes.
Check Your Understanding
For each question, distinguish what the table establishes from what would require additional information.
- A table reports \(f''(x)<0\) throughout \((1,3)\). What can you conclude about concavity on that interval?
- A table gives \(f''(2)=0\), and \(f''\) is positive on both sides of 2. Is 2 an inflection-point location?
- If \(f'(0)=1\) and \(f'(2)=7\), what does the Mean Value Theorem guarantee, assuming \(f'\) is continuous on \([0,2]\) and differentiable on \((0,2)\)?
- A table says \(f'\) increases on \((-1,0)\) and decreases on \((0,2)\). What concavity change is indicated at 0?
- Why do isolated values \(f''(0)<0\) and \(f''(1)>0\), without further assumptions, fail to establish concavity throughout \((0,1)\)?
