5.1 Tangent Lines, Normal Lines & Linear Approximation

5.1 Tangent Lines, Normal Lines & Linear Approximation

Core CLEP Concept: A differentiable curve is locally linear—when viewed under high magnification near a point of tangency $(a, f(a))$, the smooth curve becomes virtually indistinguishable from its tangent line. We exploit this property to formulate exact tangent and normal line equations, construct local linear approximations $L(x) = f(a) + f'(a)(x - a)$, analyze over- and under-estimation errors via concavity $f''(x)$, and quantify propagating measurement errors using differentials $dy = f'(x),dx$.

---

1. Tangent Lines to Differentiable Curves

Geometrically, the derivative $f'(a)$ represents the instantaneous rate of change of $y = f(x)$ and the exact slope of the line tangent to the curve at the point $(a, f(a))$.

The Point-Slope Tangent Equation

Using the fundamental point-slope equation of a line $y - y_1 = m(x - x_1)$ with slope $m = f'(a)$ and coordinate point $(x_1, y_1) = (a, f(a))$, the equation of the tangent line is:

$y - f(a) = f'(a)(x - a)$

Expressed in slope-intercept form ($y = mx + b$):

$y = f'(a)x + [f(a) - a\cdot f'(a)]$

Horizontal and Vertical Tangent Lines

• Horizontal Tangents ($f'(a) = 0$): Occur at stationary points where the instantaneous rate of change vanishes. The tangent line equation reduces to the horizontal constant: $y = f(a)$
• Vertical Tangents ($f'(a) = \pm\infty$): Occur where the difference quotient limit diverges to infinity while the function remains continuous (such as $f(x) = x^{1/3}$ at $x = 0$). The tangent line equation is vertical: $x = a$

Step-by-Step Worked Example: Explicit Polynomial Curve

Problem: Find the equation of the tangent line to $f(x) = 2x^3 - 5x + 3$ at $x = 2$.

1. Evaluate the coordinate point: $f(2) = 2(2)^3 - 5(2) + 3 = 2(8) - 10 + 3 = 16 - 10 + 3 = 9 \implies (x_1, y_1) = (2, 9)$
2. Differentiate to find the general slope function: $f'(x) = 6x^2 - 5$
3. Evaluate the slope at $x = 2$: $m_{\text{tan}} = f'(2) = 6(2)^2 - 5 = 6(4) - 5 = 24 - 5 = 19$
4. Substitute into the point-slope formula: $y - 9 = 19(x - 2) \implies y - 9 = 19x - 38 \implies y = 19x - 29$

Worked Example: Implicit Algebraic Curve

Problem: Find the equation of the tangent line to the circle $x^2 + y^2 = 25$ at the point $(3, -4)$.

1. Differentiate implicitly with respect to $x$: $\frac{d}{dx}[x^2 + y^2] = \frac{d}{dx}[25] \implies 2x + 2y\frac{dy}{dx} = 0 \implies \frac{dy}{dx} = -\frac{x}{y}$
2. Evaluate the tangent slope at $(3, -4)$: $m_{\text{tan}} = -\frac{3}{-4} = \frac{3}{4}$
3. Formulate the tangent equation: $y - (-4) = \frac{3}{4}(x - 3) \implies y + 4 = \frac{3}{4}x - \frac{9}{4} \implies y = \frac{3}{4}x - \frac{25}{4}$

---

2. Normal Lines to Curves

A normal line to a curve at a specified point $(a, f(a))$ is defined as the unique line that is perpendicular to the tangent line at that exact point of tangency.

          y ^                     Tangent Line (slope = f'(a))
            |                         / 
            |       Curve y=f(x)     / 
            |          .---.        /  
            |         /     \      /   
            |        /       \    /    
            |       |         \  /     
            |--------\----------*------------> Normal Line (slope = -1/f'(a))
            |         \        / (a, f(a))
            |          '---'  /        
            +----------------+--------------> x
                             a

Perpendicular Slope Mechanics

Two non-vertical lines with slopes $m_1$ and $m_2$ are perpendicular if and only if their product equals $-1$ ($m_1 \cdot m_2 = -1$). Therefore, the slope of the normal line $m_{\text{norm}}$ is the negative reciprocal of the tangent slope $f'(a)$:

$m_{\text{norm}} = -\frac{1}{f'(a)}, \quad \text{provided } f'(a) \neq 0$

The Point-Slope Normal Equation

$y - f(a) = -\frac{1}{f'(a)}(x - a)$

Special Perpendicular Cases

• If $f'(a) = 0$ (Horizontal Tangent $y = f(a)$): The normal line is perpendicular to a horizontal line, meaning it is strictly vertical: $x = a$
• If $f'(a) = \pm\infty$ (Vertical Tangent $x = a$): The normal line is perpendicular to a vertical line, meaning it is strictly horizontal: $y = f(a)$

Step-by-Step Worked Example: Rational Function Normal Line

Problem: Find the equation of the normal line to $f(x) = \frac{x + 1}{x - 1}$ at $x = 3$.

1. Find the point: $f(3) = \frac{3 + 1}{3 - 1} = \frac{4}{2} = 2 \implies (3, 2)$.
2. Compute derivative using the Quotient Rule: $f'(x) = \frac{(1)(x - 1) - (x + 1)(1)}{(x - 1)^2} = \frac{x - 1 - x - 1}{(x - 1)^2} = \frac{-2}{(x - 1)^2}$
3. Compute the tangent slope at $x = 3$: $m_{\text{tan}} = f'(3) = \frac{-2}{(3 - 1)^2} = \frac{-2}{4} = -\frac{1}{2}$
4. Compute the perpendicular normal slope: $m_{\text{norm}} = -\frac{1}{-1/2} = +2$
5. Formulate the normal line equation: $y - 2 = 2(x - 3) \implies y - 2 = 2x - 6 \implies y = 2x - 4$

---

3. Local Linear Approximation (Linearization)

Because differentiable curves exhibit local linearity, we can approximate complicated non-linear functions near a known reference point $x = a$ using the tangent line equation. The tangent line function is designated as the linearization $L(x)$.

Formal Definition of Linearization

Local Linear Approximation: If $f$ is differentiable at $x = a$, the approximating function: $L(x) = f(a) + f'(a)(x - a)$ is called the linearization of $f$ at $a$. For values of $x$ sufficiently close to $a$: $f(x) \approx L(x) = f(a) + f'(a)(x - a)$

4-Step Linear Approximation Protocol

1. Select an anchor point $a$: Choose a nearby base point $a$ where both $f(a)$ and $f'(a)$ can be evaluated easily and exactly without a calculator.
2. Compute $f(a)$ and $f'(a)$: Evaluate the baseline height and slope.
3. Construct the linearization formula: Write $L(x) = f(a) + f'(a)(x - a)$.
4. Evaluate at target $x$: Substitute the desired value $x$ to calculate the linear estimate.

Worked Example: Estimating Radicals Without a Calculator

Problem: Use a local linear approximation to estimate $\sqrt{25.6}$.

1. Define function and anchor point: Let $f(x) = \sqrt{x} = x^{1/2}$. The nearest perfect square to $25.6$ is $a = 25$.
2. Evaluate baseline values:
   • $f(a) = f(25) = \sqrt{25} = 5$
   • $f'(x) = \frac{1}{2\sqrt{x}} \implies f'(25) = \frac{1}{2\sqrt{25}} = \frac{1}{10} = 0.1$
3. Build the linearization equation: $L(x) = 5 + 0.1(x - 25)$
4. Evaluate at $x = 25.6$: $L(25.6) = 5 + 0.1(25.6 - 25) = 5 + 0.1(0.6) = 5 + 0.06 = 5.06$ (Comparison: The true value of $\sqrt{25.6} \approx 5.059644...$. The linear estimate $5.06$ is accurate to 3 decimal places!)

---

4. Concavity and Over/Under-Estimation Analysis

A critical conceptual question on the CLEP Calculus exam asks whether a tangent line approximation produces an overestimate or an underestimate of the true function value. This behavior is governed entirely by the second derivative $f''(x)$ (concavity) on the interval connecting $a$ and $x$.

       CONCAVE DOWN: f''(x) < 0                     CONCAVE UP: f''(x) > 0
    y ^                                          y ^
      |         Tangent Line (Overestimate)        |                    Tangent Line
      |           /                                |            Curve   (Underestimate)
      |   L(x) ->*                                 |            .---.         /
      |         / \   Curve y = f(x)               |           /     \       / 
      |        /   * <- True f(x)                  |   f(x) ->*       \     /
      |       /     \                              |           \   * <- L(x)
      |      /       \                             |            '-/-------'
      +-----+---------+---------> x                +-------------+---------+---------> x
            a         x                                          a         x

Analytical Verification for $f(x) = \sqrt{x}$

• $f'(x) = \frac{1}{2}x^{-1/2}$
• $f''(x) = -\frac{1}{4}x^{-3/2} = -\frac{1}{4\sqrt{x^3}}$
• For all $x > 0$, $f''(x) < 0$, which proves the graph of $y = \sqrt{x}$ is concave down everywhere on its domain.
• Because the curve is concave down, the tangent line lies strictly above the curve, guaranteeing that our estimate $L(25.6) = 5.06$ is an overestimate ($5.06 > 5.059644...$).

---

5. Differentials and Propagating Error Analysis

While linear approximation estimates the total value $f(x)$, differentials estimate the change in output $\Delta y$ resulting from a small change in input $\Delta x$.

Definition of Differentials

Let $y = f(x)$ be a differentiable function. The differential $dx$ is an independent variable representing an increment in $x$ ($dx = \Delta x$). The differential $dy$ is defined as:

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

• Actual Change: $\Delta y = f(x + \Delta x) - f(x)$
• Differential (Linear) Approximation: $\Delta y \approx dy = f'(x),dx$
• Approximation Error: $|\Delta y - dy| \to 0$ faster than $\Delta x \to 0$.

Error Propagation Hierarchy

When measuring a physical quantity $x$ with potential measurement error $dx$, the error propagates into any calculated dependent variable $y = f(x)$ according to three standard metrics:

\text{\textbf{Absolute Error:}} & \quad |\Delta y| \approx |dy| = |f'(x)\,dx| \\[6pt] \text{\textbf{Relative Error:}} & \quad \frac{\Delta y}{y} \approx \frac{dy}{y} = \frac{f'(x)\,dx}{f(x)} \\[6pt] \text{\textbf{Percentage Error:}} & \quad \frac{dy}{y} \times 100\% \end{aligned}$$ ### The Power Rule Error Scaling Law If $y = c\cdot x^n$, then $dy = c\cdot n x^{n-1}\,dx$. Dividing by $y = c\cdot x^n$ yields: $$\frac{dy}{y} = \frac{c\cdot n x^{n-1}\,dx}{c\cdot x^n} = n\cdot \left(\frac{dx}{x}\right)$$ > **The Power Scaling Law:** The relative (and percentage) error in $y = x^n$ is exactly **$n$ times** the relative (and percentage) error in $x$. ### Step-by-Step Worked Example: Spherical Measurement Propagation **Problem:** The radius of a solid sphere is measured as $r = 10\text{ cm}$ with a maximum possible measurement error of $dr = \pm 0.04\text{ cm}$. Use differentials to estimate: 1. The maximum absolute error in the calculated volume $V = \frac{4}{3}\pi r^3$. 2. The relative error and percentage error in the volume. **Solution:** 1. **Calculate the differential $dV$:** $$V(r) = \frac{4}{3}\pi r^3 \implies dV = V'(r)\,dr = 4\pi r^2\,dr$$ 2. **Compute the absolute error at $r = 10$ with $dr = 0.04$:** $$|dV| = 4\pi (10)^2 (0.04) = 4\pi (100) (0.04) = 16\pi \approx 50.27\text{ cm}^3$$ 3. **Compute the relative error:** $$\frac{dV}{V} = \frac{4\pi r^2\,dr}{\frac{4}{3}\pi r^3} = 3\left(\frac{dr}{r}\right) = 3\left(\frac{0.04}{10}\right) = 3(0.004) = 0.012$$ 4. **Compute the percentage error:** $$\text{Percentage Error} = 0.012 \times 100\% = 1.2\%$$ *(Notice: A $0.4\%$ error in radius produces a $3 \times 0.4\% = 1.2\%$ error in volume!)* --- ## 6. Common CLEP Traps & Strategic Summary - **Trap 1: Forgetting the negative sign in normal lines.** The normal slope is $-1/f'(a)$, NOT $+1/f'(a)$. If $f'(a) = -3$, the normal slope is $+1/3$. - **Trap 2: Checking $f'(x)$ instead of $f''(x)$ for over/under-estimates.** A common student error is thinking that because $f'(x) > 0$ (increasing), the tangent is an overestimate. The slope $f'(x)$ has zero influence on over/under-estimation; **only concavity $f''(x)$** determines whether the tangent line lies above or below the curve. - **Trap 3: Using degree angle inputs in trigonometric linearizations.** Linearizations of $\sin x, \cos x, \tan x$ require all increments $\Delta x = (x - a)$ to be converted strictly to **radians** (e.g., $1^\circ = \frac{\pi}{180}\text{ rad}$).