# Local linearity and linearization

> AP Calculus BC · Contextual Applications of Differentiation
> Source: https://www.owlsprep.com/study/ap-calculus-bc-u4-local-linearity-and-linearization/

This module covers the geometric interpretation of local linearity, the linearization formula, approximation of function values and small changes, and error bounds for linear approximations for AP Calculus BC.

**Prerequisites:** Derivative definition and tangent line slope calculation; Chain rule for composite functions; Point-slope form of a line

## Learning objectives

- Explain the geometric principle of local linearity for differentiable functions
- Find the linearization of a function at a given point and approximate function values near the center
- Use differentials to approximate small changes and measurement errors
- Calculate error bounds for linear approximations of twice-differentiable functions

## Core Concept of Local Linearity

Local linearity is a fundamental geometric property of differentiable functions. If a function $f(x)$ is differentiable at $x=a$, zooming in sufficiently close to $(a, f(a))$ makes the graph of $f(x)$ nearly indistinguishable from its tangent line at that point.

This principle gives rise to linearization (also called tangent line approximation), a technique that uses the tangent line to approximate $f(x)$ for $x$ near $a$. It simplifies calculations with complicated non-linear functions and is the foundation for advanced approximation methods like Euler's Method and Taylor polynomials, which you will encounter later in the course. This topic makes up 3-6% of the total AP Calculus BC exam score, appearing in both multiple-choice and free-response sections.

**Local Linearity** — A property of differentiable functions where the graph of the function becomes approximately straight and indistinguishable from its tangent line when viewed at a sufficiently small zoom near a point $x=a$.

*Example:* The function $f(x)=\sqrt{x}$ looks almost straight near $x=16$ when zoomed in to a small window from $x=15.9$ to $x=16.1$.

## Linearization Formula and Function Approximation

The tangent line to $f(x)$ at $x=a$ passes through $(a, f(a))$ with slope equal to $f'(a)$, the derivative of $f$ at $a$. Starting from point-slope form of a line: $y - f(a) = f'(a)(x - a)$. Rearranging gives the standard linearization formula:

$$L(x) = f(a) + f'(a)(x - a)$$

Intuition: $f(a)$ is the exact known function value at the center point $a$, and $f'(a)(x-a)$ approximates the change in $f$ from $a$ to $x$ using the instantaneous rate of change. The closer $x$ is to $a$, the more accurate the approximation. Always choose a center $a$ close to your target $x$ where $f(a)$ and $f'(a)$ are known exactly.

**Worked example:** Find the linearization of $f(x) = \sqrt{x}$ at $a=16$, then use it to approximate $\sqrt{15.9}$.

1. Calculate the exact value of $f$ at the center $a=16$:
2. $$f(16) = \sqrt{16} = 4$$
3. Compute the derivative of $f(x)$ and evaluate it at the center:
4. $$f'(x) = \frac{1}{2\sqrt{x}}, \quad f'(16) = \frac{1}{2(4)} = \frac{1}{8} = 0.125$$
5. Substitute into the linearization formula:
6. $$L(x) = 4 + 0.125(x - 16)$$
7. Evaluate $L(x)$ at the target $x=15.9$:
8. $$L(15.9) = 4 + 0.125(15.9 - 16) = 4 - 0.0125 = 3.9875$$

> **Exam tip:** Always pick a center point $a$ that is extremely close to the target $x$, and where $f(a)$ is an exact, easy-to-compute value. MCQ distractors are almost always designed to reward students who pick the wrong center, so double-check this step first.

## Differentials and Approximation of Change

Differentials are an alternative notation for linear approximation that explicitly describes approximate change in $f(x)$ when $x$ changes by a small amount $\Delta x$. By definition, $dx = \Delta x$ (the actual change in the independent variable $x$). The differential of the dependent variable $y = f(x)$, written $dy$, is the approximate change in $y$ along the tangent line, given by the formula:

$$dy = f'(x) dx$$

The actual change in $y$ is $\Delta y = f(x + \Delta x) - f(x)$. For small $\Delta x$, $dy \approx \Delta y$. This notation is especially useful for estimating measurement error and the effect of small input changes, which is common in real-world AP FRQ problems.

**Worked example:** The side length of a square is measured to be 10 cm, with a maximum measurement error of 0.05 cm. Use differentials to approximate the maximum error in the calculated area of the square.

1. Define the function: Let $s$ = side length, $A(s) = s^2$ = area. The measured side is $s=10$, maximum error in $s$ is $|\Delta s| = 0.05$, so set $ds = 0.05$.
2. Compute the derivative of area with respect to side length:
3. $$A'(s) = 2s$$
4. Calculate the differential for area, which is the approximate maximum error:
5. $$dA = A'(s) ds = 2s ds$$
6. Substitute the given values:
7. $$dA = 2(10)(0.05) = 1 \text{ cm}^2$$

> **Exam tip:** When asked for relative or percentage error, remember it is $\frac{dA}{A}$ (relative) or $100 \cdot \frac{dA}{A}$ (percentage), not just $dA$. This is one of the most commonly missed points on linear approximation FRQs.

## Error Bounds for Linear Approximation

Linear approximation is never exact. For twice-differentiable functions, we can bound the absolute error using the second derivative, which measures how much the function curves away from the tangent line. If $|f''(t)| \leq M$ for all $t$ between $a$ and $x$, the absolute error satisfies:

$$|f(x) - L(x)| \leq \frac{M}{2} |x - a|^2$$

This is the remainder term for first-order Taylor approximation, which is what linearization is. Larger $M$ (more curvature) means larger error, and error decreases quadratically as $x$ gets closer to $a$.

**Worked example:** For the approximation $\sqrt{15.9} \approx 3.9875$ centered at $a=16$, find an upper bound for the absolute error.

1. Compute the second derivative of $f(x) = x^{1/2}$:
2. $$f'(x) = \frac{1}{2}x^{-1/2}, \quad f''(x) = -\frac{1}{4}x^{-3/2} = -\frac{1}{4x^{3/2}}$$
3. Find $M$, the maximum value of $|f''(t)|$ for $t$ between 15.9 and 16. $|f''(t)| = \frac{1}{4t^{3/2}}$ decreases as $t$ increases, so the maximum occurs at the smallest $t=15.9$.
4. Find a safe upper bound for $M$: since $15.9 < 16$, $(15.9)^{3/2} < 16^{3/2} = 64$, so:
5. $$\frac{1}{4(15.9)^{3/2}} < \frac{1}{4(64)} = \frac{1}{256} \approx 0.00391$$
6. We can safely take $M = \frac{1}{256}$. Substitute into the error bound formula:
7. $$|f(15.9) - L(15.9)| \leq \frac{M}{2} |15.9 - 16|^2 = \frac{1/(256)}{2} (0.1)^2 \approx 0.00002$$

> **Exam tip:** When bounding $M$, always round up to a safe value, never round down. A larger $M$ that is still an upper bound is still correct, but a smaller $M$ that underestimates maximum curvature is wrong and will lose points.

## Concept Check

**Check your understanding**

Test your understanding with these AP-style questions:

1. Which of the following is the correct linear approximation of $(2.01)^6$ centered at $a=2$?

   - A) $64 + 192(0.01) = 65.92$
   - B) $64 + 6(0.01) = 64.06$
   - C) $12 + 64(0.01) = 12.64$
   - D) $64 + 30(0.01) = 64.3$

   *Answer:* A) $64 + 192(0.01) = 65.92$

   *Why:* Correct! You correctly calculated $f(2) = 64$ and $f'(2) = 6(2^5) = 192$, then applied the linearization formula.

## Common pitfalls

- **Wrong:** Picking the target $x$ as the center $a$, and using the unknown point as the center to approximate the known point.
  - Why it fails: Students reverse which point is the known center, because they confuse the goal of the problem.
  - Correct: Always set $a$ equal to the point near your target where you know $f(a)$ exactly, not the other way around.
- **Wrong:** Evaluating the derivative $f'(x)$ at the target $x$ instead of the center $a$.
  - Why it fails: Students compute the general derivative, then accidentally plug in the wrong point before applying the formula.
  - Correct: After computing $f'(x)$, immediately plug in $a$ to get $f'(a)$ before working with the target $x$.
- **Wrong:** Using the second derivative itself instead of its absolute value when calculating $M$ for error bounds.
  - Why it fails: Concave down functions have negative second derivatives, so students carry the negative sign into $M$, resulting in a negative error bound that is meaningless.
  - Correct: Always compute the maximum of $|f''(t)|$, not the maximum of $f''(t)$, when finding $M$.
- **Wrong:** Writing the differential change as $dy = f'(x) \Delta y$ instead of $dy = f'(x) dx$.
  - Why it fails: Students mix up which variable's change is multiplied by the derivative.
  - Correct: Remember $dy = \frac{dy}{dx} dx$, so the derivative is always multiplied by the change in the independent variable $x$.
- **Wrong:** Claiming the linear approximation is accurate for any $x$, even points far from $a$.
  - Why it fails: Students forget that local linearity is a local property, only valid near the center.
  - Correct: When interpreting an approximation, always note that it is only valid for $x$ very close to $a$, and error grows as $x$ moves away from $a$.
- **Wrong:** Calculating percentage error as $100 \cdot dQ$ instead of $100 \cdot \frac{dQ}{Q}$.
  - Why it fails: Students confuse absolute error with percentage error, which is a common AP distractor.
  - Correct: When asked for percentage error, always divide the absolute error by the total value of the quantity before multiplying by 100.

## Cheatsheet

| Category | Formula | Notes |
| --- | --- | --- |
| Linearization of $f(x)$ at $a$ | $L(x) = f(a) + f'(a)(x - a)$ | Only valid for $x$ near $a$; $a$ is a point with known $f(a), f'(a)$ |
| Differential of $y = f(x)$ | $dy = f'(x) dx$ | $dx = \Delta x$ (actual change in $x$); $dy \approx \Delta y$ (approximate change in $y$) |
| Absolute change approximation | $\Delta y = f(x + \Delta x) - f(x) \approx dy$ | For small changes in $x$ only |
| Absolute error bound | $\|f(x) - L(x)\| \leq \frac{M}{2} \|x - a\|^2$ | $M = \max \|f''(t)\|$ for $t$ between $a$ and $x$ |
| Absolute error | $\approx dQ$ | $Q$ = quantity of interest |
| Relative error | $\frac{dQ}{Q}$ | Error relative to the total value of $Q$ |
| Percentage error | $100 \cdot \frac{dQ}{Q}$ | Relative error expressed as a percentage |
| Local Linearity | $f(x) \approx L(x)$ for $x$ near $a$ | Only holds if $f$ is differentiable at $a$ |

## What's next

Local linearity and linear approximation are the foundational prerequisite for Euler's Method, the next core topic in Unit 4 for AP Calculus BC, where we extend linear approximation step-by-step to approximate solutions to first-order differential equations. Without mastering how linear approximation works near a point, you will not be able to correctly set up or interpret Euler's Method steps, and you will lose easy points on exam questions. Beyond Unit 4, linear approximation is the first-order special case of Taylor polynomials and Taylor series, which are high-weight topics in Unit 10 of AP Calculus BC. All of the core ideas here (center points, approximation error, error bounding) carry directly over to higher-order approximation, so mastering this topic is critical for later BC content.

- [L'Hopital's Rule for Indeterminate Forms](https://www.owlsprep.com/study/ap-calculus-bc-u4-l-hopital-s-rule-for/)
- [Analytical Applications of Differentiation Overview](https://www.owlsprep.com/study/ap-calculus-bc-u5-overview/)
- [Mean Value Theorem (MVT)](https://www.owlsprep.com/study/ap-calculus-bc-u5-mean-value-theorem/)

---

From [OwlsPrep](https://www.owlsprep.com) — free study guides for A-Level, IB, AP and IGCSE, written against the official syllabus. Canonical page: https://www.owlsprep.com/study/ap-calculus-bc-u4-local-linearity-and-linearization/