Skip to content
Family Table Math

Linearization and Tangent Line Approximation

If you zoom in far enough on a smooth curve, it starts to look like a straight line: its tangent line. That idea, called local linearity, lets you estimate hard-to-compute values like 4.1\sqrt{4.1} or ln⁡1.05\ln 1.05 using nothing more than a line. On the AP exam, tangent line approximations appear often, usually with a follow-up question: is your estimate too big or too small?

If ff is differentiable at x=ax = a, then near aa the graph of ff is very close to its tangent line. The closer xx is to aa, the better the match.

The tangent line at x=ax = a passes through (a,f(a))(a, f(a)) with slope f′(a)f'(a). Written as a function, it is the linearization of ff at aa:

L(x)=f(a)+f′(a)(x−a)L(x) = f(a) + f'(a)(x - a)

For xx near aa:

f(x)≈L(x)f(x) \approx L(x)

This is the same line as the point-slope form y−f(a)=f′(a)(x−a)y - f(a) = f'(a)(x - a), just solved for yy.

Pick aa so that f(a)f(a) and f′(a)f'(a) are easy to find exactly, and aa is close to the value you want. To estimate 4.1\sqrt{4.1}, use a=4a = 4. To estimate e0.1e^{0.1}, use a=0a = 0.

The concavity of ff between aa and xx tells you which side of the curve the tangent line is on:

ConcavityTangent line isApproximation L(x)L(x) is
Concave up, f′′(x)>0f''(x) \gt 0below the curvean underestimate
Concave down, f′′(x)<0f''(x) \lt 0above the curvean overestimate

On the AP exam, justify with the second derivative: ”f′′(x)<0f''(x) \lt 0 on the interval, so the graph is concave down there, and the tangent line lies above it. The approximation is an overestimate.”

The curve y = square root of x and its tangent line L(x) = 2 + one quarter (x - 4) at the point (4, 2). Near x = 4 the line and curve almost match; the line stays above the curve because the curve is concave down. 1 2 3 4 5 6 7 8 1 2 3 (4, 2) y = √x L(x) = 2 + ¼(x − 4)
Near x=4x = 4, the tangent line is almost the same as y=xy = \sqrt{x}. The curve is concave down, so the line sits above it.

In a word problem, the linearization reads: “new value ≈\approx value now ++ rate now ×\times time elapsed”. If a tank holds 500500 L at t=10t = 10 min and is filling at 1212 L/min, then about 0.50.5 min later it holds about 500+12(0.5)=506500 + 12(0.5) = 506 L.

Use a tangent line approximation to estimate 4.1\sqrt{4.1}. Is your estimate too big or too small?

Solution. Let f(x)=xf(x) = \sqrt{x} and a=4a = 4. Then f(4)=2f(4) = 2 and

f′(x)=12x⇒f′(4)=14f'(x) = \frac{1}{2\sqrt{x}} \quad\Rightarrow\quad f'(4) = \frac{1}{4} L(x)=2+14(x−4)⇒4.1≈L(4.1)=2+14(0.1)=2.025L(x) = 2 + \frac{1}{4}(x - 4) \quad\Rightarrow\quad \sqrt{4.1} \approx L(4.1) = 2 + \frac{1}{4}(0.1) = 2.025

Check the concavity: f′′(x)=−14x−3/2<0f''(x) = -\dfrac{1}{4}x^{-3/2} \lt 0 for x>0x \gt 0, so ff is concave down and the tangent line lies above the curve. The estimate 2.0252.025 is an overestimate. (Check: a calculator gives 4.1≈2.024846\sqrt{4.1} \approx 2.024846, a little less.)

Find the linearization of f(x)=ln⁡xf(x) = \ln x at x=1x = 1, and use it to estimate ln⁡1.05\ln 1.05.

Solution. f(1)=ln⁡1=0f(1) = \ln 1 = 0 and f′(x)=1xf'(x) = \dfrac{1}{x}, so f′(1)=1f'(1) = 1:

L(x)=0+1(x−1)=x−1L(x) = 0 + 1(x - 1) = x - 1 ln⁡1.05≈L(1.05)=0.05\ln 1.05 \approx L(1.05) = 0.05

Since f′′(x)=−1x2<0f''(x) = -\dfrac{1}{x^2} \lt 0, the graph is concave down, so 0.050.05 is an overestimate. (The true value is about 0.0487900.048790.)

A function ff has f(3)=7f(3) = 7 and f′(3)=−2f'(3) = -2, and f′′(x)>0f''(x) \gt 0 for all xx. Estimate f(3.2)f(3.2) and say whether the estimate is too big or too small.

Solution.

L(x)=7−2(x−3)⇒f(3.2)≈7−2(0.2)=6.6L(x) = 7 - 2(x - 3) \quad\Rightarrow\quad f(3.2) \approx 7 - 2(0.2) = 6.6

f′′(x)>0f''(x) \gt 0 means ff is concave up, so the tangent line lies below the graph. The estimate 6.66.6 is an underestimate.

V(t)V(t) is the volume of water in a tank, in litres, tt minutes after a pump is turned on. V(10)=500V(10) = 500 and V′(10)=12V'(10) = 12. Use the tangent line at t=10t = 10 to estimate V(10.5)V(10.5). If V′′(t)<0V''(t) \lt 0 for 10≤t≤10.510 \le t \le 10.5, is the estimate too big or too small?

Solution.

V(10.5)≈V(10)+V′(10)(10.5−10)=500+12(0.5)=506V(10.5) \approx V(10) + V'(10)(10.5 - 10) = 500 + 12(0.5) = 506

About 506506 litres. Since V′′<0V'' \lt 0, the graph is concave down: the rate of filling is slowing, so the tangent line (which assumes the rate stays at 1212 L/min) lies above the graph. The estimate is an overestimate.

Using f(x)f(x) in place of f(a)f(a) in the formula. The point of the linearization is that you only need values at aa. Write L(x)=f(a)+f′(a)(x−a)L(x) = f(a) + f'(a)(x - a) and plug in numbers for f(a)f(a) and f′(a)f'(a).

Choosing an aa that’s far away or awkward. For 4.1\sqrt{4.1}, a=4a = 4 is ideal. Using a=1a = 1 would work in principle but gives a poor estimate.

Getting the concavity rule backwards. Concave up means the curve bends upward away from the tangent line, so the line is below: underestimate. Picture y=x2y = x^2 and its tangent at the origin.

Justifying with the first derivative. Whether ff is increasing or decreasing has nothing to do with over- or underestimating. Use f′′f'' (concavity).

Forgetting what the tangent line means in context. In Example 4, the estimate assumes the rate stays at 1212 L/min. If the rate is really slowing down, the estimate is too big.

1. (Warm-up) Find the linearization of f(x)=x2f(x) = x^2 at a=3a = 3 and use it to estimate 3.123.1^2. Is the estimate too big or too small?

Solution

f(3)=9f(3) = 9 and f′(x)=2xf'(x) = 2x, so f′(3)=6f'(3) = 6.

L(x)=9+6(x−3)⇒3.12≈9+6(0.1)=9.6L(x) = 9 + 6(x - 3) \quad\Rightarrow\quad 3.1^2 \approx 9 + 6(0.1) = 9.6

f′′(x)=2>0f''(x) = 2 \gt 0, so the graph is concave up and 9.69.6 is an underestimate. (Indeed 3.12=9.613.1^2 = 9.61.)

2. (Warm-up) f(2)=5f(2) = 5 and f′(2)=3f'(2) = 3. Estimate f(2.1)f(2.1).

Solutionf(2.1)≈5+3(0.1)=5.3f(2.1) \approx 5 + 3(0.1) = 5.3

3. (Warm-up) Use a tangent line approximation to estimate 8.123\sqrt[3]{8.12}.

Solution

Let f(x)=x1/3f(x) = x^{1/3} and a=8a = 8. Then f(8)=2f(8) = 2 and

f′(x)=13x−2/3⇒f′(8)=13⋅14=112f'(x) = \frac{1}{3}x^{-2/3} \quad\Rightarrow\quad f'(8) = \frac{1}{3} \cdot \frac{1}{4} = \frac{1}{12}8.123≈2+112(0.12)=2.01\sqrt[3]{8.12} \approx 2 + \frac{1}{12}(0.12) = 2.01

4. (Core) Use the tangent line to f(x)=exf(x) = e^x at x=0x = 0 to estimate e0.1e^{0.1}. Is your estimate an overestimate or an underestimate? Justify.

Solution

f(0)=1f(0) = 1 and f′(0)=e0=1f'(0) = e^0 = 1, so L(x)=1+xL(x) = 1 + x and e0.1≈1.1e^{0.1} \approx 1.1.

f′′(x)=ex>0f''(x) = e^x \gt 0 for all xx, so the graph is concave up and the tangent line lies below it. The estimate is an underestimate. (The true value is about 1.1051711.105171.)

5. (Core) Let f(x)=x3−2xf(x) = x^3 - 2x. Use the linearization at a=2a = 2 to estimate f(1.95)f(1.95), and decide whether it’s too big or too small.

Solution

f(2)=8−4=4f(2) = 8 - 4 = 4 and f′(x)=3x2−2f'(x) = 3x^2 - 2, so f′(2)=10f'(2) = 10.

f(1.95)≈4+10(1.95−2)=4−0.5=3.5f(1.95) \approx 4 + 10(1.95 - 2) = 4 - 0.5 = 3.5

f′′(x)=6x>0f''(x) = 6x \gt 0 near x=2x = 2, so ff is concave up there and 3.53.5 is an underestimate. (The true value is 3.5148753.514875.)

6. (Core) Find the linearization of f(x)=tan⁡xf(x) = \tan x at a=π4a = \dfrac{\pi}{4} (radians), and use it to estimate tan⁡0.8\tan 0.8 to 3 decimal places.

Solution

f(π4)=1f\left(\dfrac{\pi}{4}\right) = 1 and f′(x)=sec⁡2xf'(x) = \sec^2 x, so f′(π4)=(2)2=2f'\left(\dfrac{\pi}{4}\right) = (\sqrt{2})^2 = 2.

L(x)=1+2(x−π4)L(x) = 1 + 2\left(x - \frac{\pi}{4}\right)tan⁡0.8≈1+2(0.8−π4)≈1+2(0.014602)≈1.029\tan 0.8 \approx 1 + 2\left(0.8 - \frac{\pi}{4}\right) \approx 1 + 2(0.014602) \approx 1.029

(A calculator gives tan⁡0.8≈1.030\tan 0.8 \approx 1.030; the estimate is a slight underestimate, since tan⁡x\tan x is concave up on (0,π2)\left(0, \dfrac{\pi}{2}\right).)

7. (Core) D(t)D(t) is the depth of snow on the ground, in centimetres, tt hours after midnight. D(4)=30D(4) = 30 and D′(4)=2.5D'(4) = 2.5.

  • (a) Estimate the depth at 4:24 a.m.
  • (b) If D′′(t)<0D''(t) \lt 0 for 4≤t≤54 \le t \le 5, is your estimate too big or too small? Explain.
Solution

(a) 4:24 a.m. is t=4.4t = 4.4 hours.

D(4.4)≈30+2.5(0.4)=31D(4.4) \approx 30 + 2.5(0.4) = 31

About 3131 cm.

(b) D′′<0D'' \lt 0, so the graph of DD is concave down and the tangent line lies above it. The estimate is too big (an overestimate).

8. (Challenge) A function ff has f(3)=2f(3) = 2 and f′(x)=x2+7f'(x) = \sqrt{x^2 + 7}. Estimate f(3.1)f(3.1) and decide whether it’s an overestimate or an underestimate.

Solution

f′(3)=9+7=4f'(3) = \sqrt{9 + 7} = 4, so

f(3.1)≈2+4(0.1)=2.4f(3.1) \approx 2 + 4(0.1) = 2.4

Differentiate f′f' to find the concavity:

f′′(x)=xx2+7f''(x) = \frac{x}{\sqrt{x^2 + 7}}

which is positive for x>0x \gt 0. So ff is concave up near x=3x = 3, and 2.42.4 is an underestimate.

9. (Challenge) The curve x2+xy+y2=7x^2 + xy + y^2 = 7 passes through (1,2)(1, 2). Use the tangent line at that point to estimate the yy-value on the curve near (1,2)(1, 2) when x=1.1x = 1.1.

Solution

Differentiate implicitly:

2x+y+xdydx+2ydydx=0⇒dydx=−2x+yx+2y2x + y + x\frac{dy}{dx} + 2y\frac{dy}{dx} = 0 \quad\Rightarrow\quad \frac{dy}{dx} = -\frac{2x + y}{x + 2y}

At (1,2)(1, 2): dydx=−45\dfrac{dy}{dx} = -\dfrac{4}{5}.

y≈2−45(1.1−1)=2−0.08=1.92y \approx 2 - \frac{4}{5}(1.1 - 1) = 2 - 0.08 = 1.92

(Check: solving y2+1.1y−5.79=0y^2 + 1.1y - 5.79 = 0 gives y≈1.918y \approx 1.918, very close.)