AP Calculus AB: Linearization and Differentials

At the heart of calculus is a powerful, practical idea: complex, curving functions can behave very simply if you look at a sufficiently small scale. Linearization is the process of using a tangent line to approximate the value of a function near a specific, known point. This technique connects the abstract concept of the derivative—instantaneous rate of change—to concrete estimation problems in physics, engineering, and economics, where exact solutions are often difficult or impossible to calculate directly. Mastering linearization and differentials provides you with a crucial tool for simplification, error analysis, and informed approximation.

The Tangent Line as a Local Proxy

The foundational insight of linearization is geometric. For a function $f$ that is differentiable at a point $x=a$, the tangent line to the graph at the point $(a,f(a))$ closely "hugs" the curve very near that point. The key question is: how can we find the equation of this special line and use it as a stand-in for the function itself?

Recall that the slope of this tangent line is given by the derivative, $f^{\prime }(a)$. Using the point-slope form of a line, $y-y_1=m(x-x_1)$, we can write the equation of the tangent line. Substituting the point $(a,f(a))$ and slope $f^{\prime }(a)$ gives us: $$y-f(a)=f^{\prime }(a)(x-a)$$ We then solve for $y$ to define our linear approximation function, $L(x)$: $$y=f(a)+f^{\prime }(a)(x-a)$$ This function $L(x)$ is our linear approximation, or linearization, of $f$ at $x=a$. The point $x=a$ is called the center of the linearization. For values of $x$ close to $a$, we confidently state that $f(x)\approx L(x)$.

Example: Let's find the linearization of $f(x)=\sqrt{x}$ centered at $a=9$.

1. $f(a)=f(9)=\sqrt{9}=3$.
2. $f^{\prime }(x)=\frac{1}{2\sqrt{x}}$, so $f^{\prime }(a)=f^{\prime }(9)=\frac{1}{2\sqrt{9}}=\frac{1}{6}$.
3. The linearization is: $L(x)=3+\frac{1}{6}(x-9)$.

This linearization allows us to quickly estimate square roots near 9. For instance, to approximate $\sqrt{9.1}$, we simply calculate: $$L(9.1)=3+\frac{1}{6}(9.1-9)=3+\frac{1}{6}(0.1)\approx \mathrm{3.01667.}$$ The actual value of $\sqrt{9.1}\approx 3.01662$, demonstrating the remarkable accuracy of the tangent line approximation for inputs close to the center.

The Notation of Differentials: $dy$ and $dx$

While linearization gives us a function $L(x)$ for approximation, the concept of differentials provides a closely related notation for estimating changes. Let $y=f(x)$ be a differentiable function. We define two new variables:

• The differential $dx$ represents an arbitrary (small) change in the independent variable $x$.
• The differential $dy$ represents the corresponding change along the tangent line for the given change $dx$.

The relationship is defined by the derivative: $dy=f^{\prime }(x)dx$. This is not a new formula, but a new interpretation of the derivative notation $\frac{dy}{dx}=f^{\prime }(x)$. We can now think of "multiplying" both sides by $dx$ to get the differential form.

What is the practical meaning? The differential $dy$ estimates the change in the function's output as predicted by the tangent line. The actual change in the function is denoted by $\Delta y=f(x+dx)-f(x)$. For small values of $dx$, we expect $dy\approx \Delta y$.

Example: Consider a circle whose radius increases from $r=10$ cm to $r=10.1$ cm. Let's use differentials to estimate the change in its area, $A=\pi r^2$.

1. Find $dA$: $A^{\prime }(r)=2\pi r$, so $dA=A^{\prime }(r)dr=(2\pi r)dr$.
2. Here, $r=10$ cm and $dr=0.1$ cm.
3. The estimated change is $dA=2\pi (10)(0.1)=2\pi \approx 6.283$ cm².

This differential $dA$ estimates how much the area would increase if it grew linearly with radius (following the tangent line). The actual change $\Delta A=\pi (10.1{)}^2-\pi (10{)}^2=\pi (102.01-100)=2.01\pi \approx 6.314$ cm² is very close to our estimate of $2\pi$, confirming the utility of the method.

Quantifying the Approximation Error

A responsible approximation always considers its potential error. The error (or remainder) in our linear approximation is the vertical gap between the curve $y=f(x)$ and the tangent line $y=L(x)$ at a given $x$. We denote it as: $$\text{Error}=f(x)-L(x)$$ For the approximation to be trustworthy, this error must be very small when $(x-a)$ is small. In advanced calculus, theorems like Taylor's Theorem provide formal bounds for this error. For AP Calculus, a visual and conceptual understanding is key.

The error is closely related to the curvature of the function. A function with a large second derivative $|f^{\prime \prime }(x)|$ (one that bends sharply) will generally have a larger linearization error for a given step away from the center than a function that is almost straight. This is why linearization works best very close to the point of tangency—the function has less "room" to curve away from the line.

You can investigate error by calculating the percent error: $$\text{Percent Error}=\left|\frac{\text{Actual Value}-\text{Approximate Value}}{\text{Actual Value}}\right|\times 100\%$$ In our $\sqrt{9.1}$ example, the percent error is approximately $0.0017\%$, which is negligible for most purposes. However, if we rashly used the same linearization $L(x)=3+\frac{1}{6}(x-9)$ to estimate $\sqrt{16}$, we would get $L(16)=4.1667$, while the true value is 4. This is a $4.2\%$ error—a clear sign that we moved too far from the center $a=9$ for the linear model to remain accurate.

Common Pitfalls

1. Linearizing at the Wrong Point: Always ensure your center $a$ is a point where you know the exact function value $f(a)$ and derivative $f^{\prime }(a)$, and that it is close to the $x$-value you want to approximate. Choosing $a=0$ to approximate $f(100)$ will almost certainly yield a useless result.

• Correction: Select a center strategically. To approximate $\sqrt{26}$, choose a perfect square near 26, like $a=25$ or $a=36$, as your center.

1. Confusing $dy$ (the estimate) with $\Delta y$ (the actual change): It is a common conceptual error to treat the differential $dy$ as the exact change in the function. Remember, $dy$ estimates the change along the tangent line, while $\Delta y$ is the actual change along the curve.

• Correction: When a problem asks "use differentials to estimate the change," you are solving for $dy$. When it asks for "the actual change" or "compute the change," you must calculate $\Delta y=f(x+dx)-f(x)$.

1. Misapplying the Formula Through Poor Notation: Students sometimes incorrectly write $L(x)=f^{\prime }(a)+f(a)(x-a)$, reversing the roles of the function value and the derivative.

• Correction: Use a structured approach. First, compute and box $f(a)$. Second, compute and box $f^{\prime }(a)$. Then plug them directly into the formula: $L(x)=[f(a)]+[f^{\prime }(a)](x-a)$.

1. Forgetting the $(x-a)$ Factor: The formula is not $f(a)+f^{\prime }(a)\cdot x$. The derivative's effect must be multiplied by the change in the input, $(x-a)$.

• Correction: Verbally read the formula as: "The value at the center, plus the rate of change at the center, multiplied by how far you've moved from the center."

Summary

• Linearization uses the tangent line $L(x)=f(a)+f^{\prime }(a)(x-a)$ at a center $x=a$ to approximate function values $f(x)$ for $x$ near $a$. It is the practical application of differentiability.
• Differentials provide complementary notation, where $dy=f^{\prime }(x)dx$ estimates the change in the function's output ($\Delta y$) for a small change $dx$ in the input. Here, $dy$ represents the change along the tangent line.
• The approximation is only reliable when $x$ is sufficiently close to the center $a$. The error $f(x)-L(x)$ grows as you move farther away, particularly for functions with high curvature (large second derivative).
• This technique is indispensable for making quick, reasonable estimates in applied fields when exact computation is complex, and it forms the foundational idea for more advanced approximation methods like Taylor polynomials.