AP Calc 4.6Unit 4: Contextual Applications of Differentiation · 6 min video

Linearization and Tangent Line Approximation Explained

Why the tangent line is a good estimate near the point of tangency, how to build L(x), and how concavity tells you if the estimate is too high or too low.

What this lesson covers

This lesson explains why the tangent line is a good estimate near the point of tangency and how to build L(x) = f(a) + f'(a)(x - a). It starts with a square root estimated without a calculator and a table showing how the error grows as x moves away from a.

Concavity tells you which way the estimate misses. On a concave down graph the tangent line overestimates, and on a concave up graph it underestimates. The lesson includes an AP-style example from given values, the most common mistake, and two check questions.

  • 4.6Approximating Values of a Function Using Local Linearity and Linearization

Practice it on CalcLearn

Transcript

0:00The square root of 25.4

No calculator. What's the square root of 25.4? You know the square root of 25 is 5, so it's a bit more than 5. But how much more? Watch the graph of the square root while we zoom in on the point (25, 5).

The closer we get, the less it curves. Zoom in far enough and it's a straight line. And a line, we can work with. This is linearization, AP Calculus topic 4.6.

0:27Local linearity

Here's the idea. That straight line is the tangent line at x = 25. Near the point of tangency, a smooth curve and its tangent line are almost the same, so the line's height is a good estimate of the curve's height.

This is called local linearity: zoomed in close, a differentiable function looks linear. It's the same reason the ground looks flat where you're standing, even though the Earth is round.

0:51Building L(x)

Build the line. Start at the point (a, f(a)). Move to the right by x − a. The line rises by its slope times that run: f′(a) times (x − a). So the height of the tangent line is L(x) = f(a) + f′(a)(x − a). That's the linearization of f at a.

At x = a the run is zero, so L(a) = f(a). The line and the curve meet at the point, with the same slope. No other line hugs the curve that closely there. Start height, plus slope times how far you moved. Every tangent-line estimate is just that.

1:28Worked example

Back to the square root of 25.4. Take f(x) = √x and a = 25, because 25 is close and easy. f(25) = 5. The derivative is f′(x) = 1/(2√x), so f′(25) = 1/10. The tangent line is L(x) = 5 + (1/10)(x − 25).

Now plug in 25.4. You move 0.4 to the right, and the line rises one tenth of that, 0.04. So the estimate is 5.04. A calculator says 5.0398. We're off by about two ten-thousandths.

2:07Over or underestimate?

But the further you go from a, the worse it gets. At 26, the line says 5.1 and the truth is about 5.099. At 30, 5.5 against about 5.477. At 36, the line says 6.1, but the square root of 36 is exactly 6.

And notice every estimate came out too high. That's not luck. Look at the shape. The square-root curve is concave down: it bends downward, away from its tangent line. So the curve sits below the line, and the line's estimate is an overestimate.

Flip the shape. The curve y = e^x is concave up: it bends upward, so it sits above its tangent lines. Its tangent line at 0 is y = 1 + x, which estimates e^0.2 as 1.2. The truth is about 1.2214, so the line gives an underestimate. So: if f″ is negative near a, the tangent line overestimates. If f″ is positive, it underestimates.

3:09Example from given values

Example two, the way the AP exam often asks it, with no formula. A function f has f(2) = 7 and f′(2) = −3, and f″(x) > 0 for all x. Estimate f(2.1), and say whether it's too high or too low.

The tangent line is L(x) = 7 − 3(x − 2). At 2.1 you move 0.1, so L(2.1) = 7 − 0.3, which is 6.7. And f″ is positive, so f is concave up, the curve sits above the line, and 6.7 is an underestimate.

3:48The usual mistake

Here's the slip students make. They look at whether f is increasing or decreasing to decide over or under. That has nothing to do with it. The square root is increasing, and its estimate was too high. Example two was decreasing, and its estimate was too low. Only the concavity decides: the sign of f″.

4:09Check yourself

Your turn. Use the tangent line to f(x) = x³ at x = 2 to estimate (2.1)³. Pause and work it out. f(2) = 8, and f′(x) = 3x², so f′(2) = 12. L(2.1) = 8 + 12 times 0.1, which is 9.2. The true value is 9.261, so 9.2 is an underestimate: x³ is concave up for x > 0.

One more. A function g has g(4) = 10 and g′(4) = −2, and g″(x) < 0 for all x. Estimate g(4.3). Is it an overestimate or an underestimate? Pause. L(4.3) = 10 − 2 times 0.3, which is 9.4. And g″ is negative, so g is concave down, the curve is below the line, and 9.4 is an overestimate.

5:16Recap

Recap. Near a point, a smooth curve looks like its tangent line, so the tangent line gives good estimates nearby: L(x) = f(a) + f′(a)(x − a). Pick an a that's close and easy. The estimate gets worse as you move away. Concave down means an overestimate, and concave up means an underestimate. For more practice, go to calclearn.org, glossary, linearization.