Local linearity and linearization — AP Calculus AB Study Guide

For: AP Calculus AB candidates sitting AP Calculus AB.

Covers: Geometric interpretation of local linearity, the tangent line approximation (linearization) formula, differential notation for linear changes, approximating function values, and error estimation for approximations, aligned to AP CED requirements.

You should already know: How to compute derivatives of elementary functions using differentiation rules. How to find the equation of a tangent line at a point. How to evaluate functions at given input values.

A note on the practice questions: All worked questions in the "Practice Questions" section below are original problems written by us in the AP Calculus AB style for educational use. They are not reproductions of past College Board / Cambridge / IB papers and may differ in wording, numerical values, or context. Use them to practise the technique; cross-check with official mark schemes for grading conventions.

1. What Is Local linearity and linearization?

Local linearity is a core geometric property of differentiable functions that states when you zoom in far enough on any point on the graph of a differentiable function, the graph becomes almost identical to its tangent line at that point. This topic is part of AP Calculus AB Unit 4: Contextual Applications of Differentiation, which accounts for 18-20% of the total AP exam score, and linearization questions appear in both multiple-choice (MCQ) and free-response (FRQ) sections every year. Linearization (also called tangent line approximation) is the practical application of local linearity: we use the simple, easy-to-evaluate tangent line to approximate the value of a complicated nonlinear function near the point of tangency. This works because for differentiable functions, the difference between the function and its tangent line becomes negligible for inputs very close to the point of tangency. On the AP exam, you will be expected to find linearizations, use them to approximate function values, and interpret approximations in real-world contexts, often paired with other differentiation skills like concavity or related rates.

2. Local Linearity: The Geometric Foundation

Local linearity is the reason differentiation works as a tool for approximating nonlinear behavior. For any differentiable function $y=f(x)$ at a given point $x=a$, the derivative $f^{\prime }(a)$ gives the slope of the tangent line at that point. Because the tangent line and the original function agree in both value and slope at $x=a$, the function behaves almost exactly like the straight tangent line for $x$-values very close to $a$. For example, take $y=\sin x$ at $x=0$: if you zoom in from the interval $[-2,2]$ to $[-0.1,0.1]$, the sine curve becomes indistinguishable from the line $y=x$, which is its tangent line at $x=0$. This is not a coincidence: it is a direct consequence of the definition of the derivative as the limit of secant slopes. The difference between $f(x)$ and its tangent line $f(a)+f^{\prime }(a)(x-a)$ approaches 0 faster than $(x-a)$ approaches 0, so the difference is negligible for small changes in $x$ near $a$. This is the core insight: near the point of tangency, the linear tangent is the best possible approximation to the nonlinear function.

Worked Example

Problem: Confirm that $f(x)=\sqrt{x}$ is locally linear at $x=4$ by comparing the function value to its tangent line approximation at $x=4.1$ and $x=6$. Solution:

1. First calculate $f(4)=\sqrt{4}=2$, then find the derivative: $f^{\prime }(x)=\frac{1}{2\sqrt{x}}$, so $f^{\prime }(4)=\frac{1}{4}=0.25$.
2. Write the tangent line equation at $x=4$ using point-slope form: $y=f(4)+f^{\prime }(4)(x-4)=2+0.25(x-4)=0.25x+1$.
3. Compare at $x=4.1$: The actual function value is $f(4.1)=\sqrt{4.1}\approx 2.0248$, while the tangent line gives $y=2+0.25(0.1)=2.025$. The difference is only ~0.0002, which is negligible for most purposes.
4. Compare at $x=6$: The actual function value is $f(6)=\sqrt{6}\approx 2.4495$, while the tangent line gives $y=2+0.25(2)=2.5$. The difference is ~0.05, much larger. Conclusion: The function is almost identical to its tangent line near $x=4$, so it is locally linear at $x=4$, but the approximation breaks down far from the point of tangency.

Exam tip: When the AP exam asks you to explain why a function is locally linear at a point, always mention that the function is differentiable at that point, and that the tangent line matches the function's value and slope for inputs near the point — that is the explicit reasoning exam graders look for.

3. Linearization (Tangent Line Approximation): Formula and Computation

The linearization of a function $f(x)$ at $x=a$ is simply the tangent line to $f$ at $x=a$, written as a function for approximation. The standard formula for linearization is: $$L(x)=f(a)+f^{\prime }(a)(x-a)$$ To use this formula, you need two key values: $f(a)$ (the exact value of the function at $a$) and $f^{\prime }(a)$ (the exact slope at $a$). We always choose $a$ to be a nice, round value near the input we want to approximate, where we can calculate $f(a)$ exactly. For example, if we want to approximate $\sqrt{15}$, we choose $a=16$, since we know $\sqrt{16}=4$ exactly, and 15 is very close to 16. Finding and using linearization is the most heavily tested skill for this topic on the AP exam.

Worked Example

Problem: Find the linearization of $f(x)=x^3-2x+1$ at $a=2$, then use it to approximate $f(1.98)$. Solution:

1. Calculate $f(a)=f(2)=(2{)}^3-2(2)+1=8-4+1=5$.
2. Calculate the derivative: $f^{\prime }(x)=3x^2-2$, so $f^{\prime }(a)=f^{\prime }(2)=3(4)-2=10$.
3. Substitute into the linearization formula: $L(x)=f(2)+f^{\prime }(2)(x-2)=5+10(x-2)=10x-15$.
4. Evaluate $L(1.98)$ to get the approximation: $L(1.98)=10(1.98)-15=4.8$. The actual value of $f(1.98)\approx 4.801$, so the approximation is extremely accurate.

Exam tip: Always simplify your linearization before evaluating the approximation to reduce arithmetic errors. If you leave the linearization in factored form, double-check your distribution of the slope term.

4. Differentials and Approximating Change

Differentials are an alternative notation for linear approximation that focuses on changes in quantities, rather than absolute values. This notation is especially useful for real-world problems where we want to approximate how much a quantity changes, or to estimate error from measurement. For a function $y=f(x)$, we let $dx=\Delta x=x-a$, which is a small change in the input $x$. The differential $dy$, the linear approximation of the change in output $y$, is given by: $$dy=f^{\prime }(x)dx$$ The actual change in $y$ is $\Delta y=f(a+dx)-f(a)$, and for small $dx$, $\Delta y\approx dy$. This notation is directly equivalent to linearization: rearranging the linearization formula gives $L(x)-f(a)=f^{\prime }(a)(x-a)$, which is exactly $\Delta y\approx dy$.

Worked Example

Problem: The radius of a spherical birthday balloon is measured as 10 cm with a maximum measurement error of $\pm 0.1$ cm. Use differentials to approximate the maximum possible error in the calculated surface area of the balloon. Solution:

1. The formula for the surface area of a sphere is $S(r)=4\pi r^2$. We know the measured radius is $r=10$ cm, and the maximum error in $r$ is $|dr|=0.1$ cm.
2. Calculate the derivative of $S(r)$: $S^{\prime }(r)=8\pi r$.
3. The maximum error in surface area is approximated by $|dS|=|S^{\prime }(r)dr|$.
4. Substitute values: $|dS|=|8\pi (10)(0.1)|=8\pi \approx 25.13$ cm². The maximum possible error in the calculated surface area is approximately $8\pi$ cm².

Exam tip: Always read the question carefully: if it asks for absolute error, you just need $|dy|$; if it asks for relative error, you need $\frac{|dy|}{y}$, and percentage error is that value multiplied by 100.

5. Common Pitfalls (and how to avoid them)

• Wrong move: Writing the linearization as $L(x)=f(a)+f^{\prime }(a)x$ instead of $L(x)=f(a)+f^{\prime }(a)(x-a)$, omitting the $(x-a)$ term. Why: Students misremember the formula and forget it is derived from point-slope form of a line. Correct move: Always start from point-slope form $(y-f(a))=f^{\prime }(a)(x-a)$, then rearrange to get $L(x)$.
• Wrong move: Choosing $a$ as the input you want to approximate instead of a nearby known point (e.g., approximating $\sqrt{15}$ by choosing $a=15$ instead of $a=16$). Why: Students misread the problem and forget the purpose of $a$ is to have an exactly known value of $f(a)$. Correct move: Always pick the closest round number to your unknown input where you can compute $f(a)$ exactly.
• Wrong move: Forgetting the chain rule when differentiating composite functions for linearization (e.g., writing the derivative of $e^{3x}$ as $e^{3x}$ instead of $3e^{3x}$). Why: Students rush through differentiation after setting up the problem and skip the chain rule step. Correct move: After calculating any derivative for linearization, double-check that you applied the chain rule for composite inner functions.
• Wrong move: Claiming a linear approximation is accurate for a point far from $a$. Why: Students assume linearization works everywhere, not just locally. Correct move: After calculating an approximation, check how far your input is from $a$, and explicitly note that accuracy decreases as distance from $a$ increases.
• Wrong move: Writing $\Delta y=dy$ instead of $\Delta y\approx dy$, treating the approximation as exact. Why: Students forget that $dy$ is a linear approximation, not the actual change. Correct move: Always use an approximation symbol when relating $\Delta y$ to $dy$, and explicitly state the relationship is approximate for small $dx$.
• Wrong move: Confusing the measured value of a quantity with its error in differential error problems (e.g., using $dr=10$ cm instead of $dr=0.1$ cm in the worked example above). Why: Students mix up which value corresponds to the input and which corresponds to the error. Correct move: Label all variables before substituting: $r$ = measured input, $dr$ = error in $r$.

6. Practice Questions (AP Calculus AB Style)

Question 1 (Multiple Choice)

What is the linearization of $f(x)=\ln (3x+1)$ at $a=0$? A) $L(x)=3x$ B) $L(x)=1+3x$ C) $L(x)=3x+\ln 2$ D) $L(x)=x$

Worked Solution: We use the linearization formula $L(x)=f(a)+f^{\prime }(a)(x-a)$ with $a=0$. First, $f(0)=\ln (3(0)+1)=\ln 1=0$. Next, use the chain rule to find the derivative: $f^{\prime }(x)=\frac{3}{3x+1}$, so $f^{\prime }(0)=\frac{3}{1}=3$. Substitute into the formula: $L(x)=0+3(x-0)=3x$. The wrong answers come from forgetting $f(0)=0$ (option B), choosing $a=1$ instead of $a=0$ (option C), and forgetting the 3 from the chain rule (option D). The correct answer is A.

Question 2 (Free Response)

Let $f(x)=\sqrt{x^2+9}$. (a) Find the linearization $L(x)$ of $f(x)$ at $a=4$. (b) Use $L(x)$ to approximate $f(4.2)$. Is this approximation an overestimate or an underestimate of the actual value of $f(4.2)$? Justify your answer. (c) Use differentials to approximate the change in $f(x)$ when $x$ increases from 4 to 4.2.

Worked Solution: (a) First calculate $f(4)=\sqrt{4^2+9}=\sqrt{25}=5$. Next find the derivative using the chain rule: $f^{\prime }(x)=\frac{1}{2}(x^2+9{)}^{-1/2}(2x)=\frac{x}{\sqrt{x^2+9}}$. Evaluate at $a=4$: $f^{\prime }(4)=\frac{4}{5}=0.8$. Substitute into the linearization formula: $L(x)=f(4)+f^{\prime }(4)(x-4)=5+0.8(x-4)=0.8x+1.8$.

(b) Evaluate $L(4.2)=0.8(4.2)+1.8=3.36+1.8=5.16$, so the approximation of $f(4.2)$ is 5.16. To check over/underestimate, calculate the second derivative: $f^{\prime \prime }(x)=\frac{9}{(x^2+9{)}^{3/2}}>0$ for all $x$, so $f(x)$ is concave up near $x=4$. A tangent line to a concave up function lies below the function, so $L(4.2)<f(4.2)$, meaning the approximation is an underestimate.

(c) The change in $x$ is $dx=4.2-4=0.2$. The approximate change in $f(x)$ is $dy=f^{\prime }(4)dx=0.8(0.2)=0.16$, which matches the result from (b): the approximate change is $f(4.2)-f(4)\approx 0.16$.

Question 3 (Application / Real-World Style)

The monthly profit of a small coffee shop is modeled by $P(x)=12x-0.005x^2-2000$ dollars, where $x$ is the number of lattes sold per month. The shop currently sells 1500 lattes per month, and they project sales will increase by 60 lattes next month. Use linearization to approximate the change in profit for next month, and find the approximate total profit for next month.

Worked Solution: We need the linear approximation of $P(x)$ at $x=1500$, for a small change of $dx=60$. First calculate $P^{\prime }(x)=12-0.01x$. Evaluate at $x=1500$: $P^{\prime }(1500)=12-0.01(1500)=-3$ dollars per latte. The approximate change in profit is $dP=P^{\prime }(1500)dx=-3(60)=-180$ dollars. The current profit at $x=1500$ is $P(1500)=12(1500)-0.005(1500{)}^2-2000=18000-11250-2000=4750$ dollars. The approximate total profit for next month is $4750-180=4570$ dollars. In context: A 60-latte increase in monthly sales from 1500 lattes will decrease total monthly profit by approximately $180,toanewtotalof$4570.

7. Quick Reference Cheatsheet

8. What's Next

Local linearity and linearization is the foundational prerequisite for the next core topics in Unit 4. The immediate next topic is related rates, which relies on the idea of differentials (small linear changes) to relate the rates of change of connected quantities; without understanding that $\Delta y\approx dy=f^{\prime }(x)dx$, you will struggle to set up related rate equations correctly. This topic also introduces the core idea of approximating complicated nonlinear functions with simple linear models, which underpins all applied calculus in science, economics, and engineering, and prepares you for curve sketching and optimization. The next topics you will study after mastering this chapter are:

• Related Rates
• Mean Value Theorem
• Interval Concavity and Points of Inflection
• Applied Optimization