AP Calculus AB/BC · Unit 2 · Topic 2.3
Estimating Derivatives of a Function at a Point
Estimate a derivative from a tangent line, symmetric difference, table, or nearby secant slopes.
1. Topic Focus
Define derivatives as limits, estimate slopes from representations, and establish the fundamental derivative rules.
This topic: Estimate a derivative from a tangent line, symmetric difference, table, or nearby secant slopes.
2. Key Relationship
Read every symbol with its domain, direction, units, and hypotheses before applying the relationship.
3. Visual Connection
4. Worked Example
Use table values equally spaced around a to estimate the tangent slope.
Write the governing relationship first, carry out the algebra cleanly, and finish with a sentence that answers the mathematical question.
5. Concept Development
A derivative is a limit of secant slopes, so a nearby secant slope can estimate the tangent slope even when no formula for the function is available. The estimate should use data close to the target input and, when possible, information from both sides.
The estimate is not found by averaging function values. It is a quotient of output change and input change, so it has output-units per input-unit.
| Available information | Estimate | When to use it |
|---|---|---|
| Value at \(a\) and a nearby value to the left | \(\dfrac{f(a)-f(a-h)}h\) | Only left-side data are available |
| Value at \(a\) and a nearby value to the right | \(\dfrac{f(a+h)-f(a)}h\) | Only right-side data are available |
| Equally spaced values around \(a\) | \(\dfrac{f(a+h)-f(a-h)}{2h}\) | Preferred centered estimate for smooth data |
| Unequally spaced values bracketing \(a\) | \(\dfrac{f(x_R)-f(x_L)}{x_R-x_L}\) | Use the closest reliable pair with \(x_L<a<x_R\) |
Why centered data help. For a smooth curve, a secant through points on opposite sides often balances some of the curvature error. A centered difference also works when the table omits \(f(a)\). It is still an approximation: symmetry does not make it exact for every function.
A practical selection routine.
- Locate the target input and inspect the spacing of nearby data.
- Choose the closest reliable values that bracket the target whenever possible.
- Compute the slope using the actual input difference, not an assumed step size.
- Compare left and right secant slopes if both are available.
- Report an appropriate approximation sign, units, and contextual meaning.
Estimating from a graph. If a tangent line is drawn, choose two clear points on that line and use rise over run. The points do not have to lie on the original curve. If only the curve is shown, draw or imagine the local tangent and use nearby curve points on opposite sides as a secant approximation. Read both axis scales first; one grid square may represent different amounts horizontally and vertically.
Reliability and existence. Smaller intervals usually track a smooth tangent more closely, but rounded measurements can make extremely small differences unstable. Widely different left and right slopes can signal a corner or other failure of differentiability. A short table can suggest this behavior but cannot prove a limit does not exist without additional information.
6. Detailed Worked Example and Error Check
Example 1: Compare three table estimates. Suppose the table contains:
| \(x\) | \(1.8\) | \(1.9\) | \(2.0\) | \(2.1\) | \(2.2\) |
|---|---|---|---|---|---|
| \(f(x)\) | \(3.24\) | \(3.61\) | \(4.00\) | \(4.41\) | \(4.84\) |
The closest centered estimate is
The left and right estimates surround 4 and agree closely, which supports the centered estimate. The wider centered pair at 1.8 and 2.2 also gives 4, but the closer pair generally provides stronger local evidence.
Example 2: Unequal spacing and contextual units. A population table gives \(P(4.8)=116.2\), \(P(5.0)=120.0\), and \(P(5.3)=126.3\), with population measured in thousands and time in years. Using the closest points that bracket 5,
At year 5, the population is increasing at approximately 20.2 thousand people per year. The denominator is 0.5, not twice one of the unequal distances from 5.
Example 3: Read a drawn tangent line. A tangent line at \(x=1\) passes through the convenient grid points \((-1,4)\) and \((3,-2)\). Therefore
The derivative is negative because the tangent falls from left to right. The points used for the slope calculation belong to the tangent line; they need not be points where the tangent meets the curve.
Example 4: Do not average incompatible one-sided slopes. Suppose secant slopes approaching \(a\) from the left are \(-2.1,-2.01,-2.001\), while slopes from the right are \(2.9,2.99,2.999\). The evidence suggests one-sided derivatives near \(-2\) and \(3\). Averaging them to obtain \(0.5\) would hide the disagreement; the data instead suggest that \(f'(a)\) does not exist.
Example 5: Match precision to the data. If table values are rounded to the nearest tenth, reporting \(f'(a)\approx2.738416\) implies unsupported accuracy. Carry enough digits during the calculation, then round the derivative consistently with the quality of the given measurements.
AP error check. Do not divide by the number of table rows, confuse \(f(a)\) with \(f'(a)\), use two points on the curve when a tangent line is explicitly supplied, ignore unequal axis scales, or omit quotient units in context.
7. AP Reasoning Routine
Identify the function structure, state the applicable rule, preserve notation and units, and check differentiability before interpreting a derivative.
- Identify the representation and requested quantity.
- State the rule or theorem and verify its conditions.
- Keep exact values until the final requested approximation.
- Interpret sign, units, interval, and context.
Estimate carefully and explain the data choice.
(a) A table gives \(f(2.9)=8.41\), \(f(3.0)=9.00\), and \(f(3.1)=9.61\). Estimate \(f'(3)\) with a centered difference and compare it with the two one-sided secant slopes.
(b) Values \(G(1.7)=4.8\) and \(G(2.2)=6.7\) are the closest data bracketing \(x=2\). Estimate \(G'(2)\).
(c) A table does not list \(H(5)\), but gives \(H(4.9)=12.4\) and \(H(5.1)=13.0\). Explain why a derivative estimate is still possible and find it.
(d) A tangent line at \(x=4\) passes through \((2,7)\) and \((6,-1)\). Estimate \(f'(4)\).
(e) Water volume \(V(t)\), measured in liters, satisfies \(V(9.8)=51.6\) and \(V(10.2)=50.0\), where \(t\) is minutes. Estimate and interpret \(V'(10)\).
(f) Left secant slopes approach 1.5 while right secant slopes approach 1.5. What derivative estimate is supported? How would the conclusion change if the right slopes approached 4?
(g) Explain why the closest pair is not automatically best when measurements have been heavily rounded.
(h) On a graph, one horizontal grid square represents 2 seconds and one vertical grid square represents 5 meters. A tangent rises 3 vertical squares while running 4 horizontal squares. Estimate the derivative with units.
Check the solution
(a) The centered estimate is \((9.61-8.41)/(3.1-2.9)=6\). The left and right slopes are 5.9 and 6.1, so both support \(f'(3)\approx6\).
(b) \(G'(2)\approx(6.7-4.8)/(2.2-1.7)=3.8\).
(c) A centered secant does not require \(H(5)\): \(H'(5)\approx(13.0-12.4)/0.2=3\).
(d) The tangent slope is \((-1-7)/(6-2)=-2\), so \(f'(4)\approx-2\).
(e) \(V'(10)\approx(50.0-51.6)/(10.2-9.8)=-4\) liters per minute. At 10 minutes, volume is decreasing at approximately 4 liters per minute.
(f) Matching one-sided trends support \(f'(a)\approx1.5\). If the right side approached 4, the mismatch would suggest that the two-sided derivative does not exist.
(g) Subtracting nearly equal rounded outputs can magnify measurement and rounding error; a slightly wider interval may produce a more stable estimate.
(h) The rise is \(3(5)=15\) meters and the run is \(4(2)=8\) seconds, so the derivative is approximately \(15/8=1.875\) meters per second.