2026 Fall Section 3.8 Differentials & Linear Approximation
Overview
The lesson develops differentials and linear approximation by distinguishing the actual change Δy from the tangent-line change dy = f′(x) dx, which estimates Δy when dx is small. It applies the method to polynomial changes, a blood vessel’s circular area, bacterial population, estimates of 51² and √27, and ball-bearing volume, then connects differentials to tangent-line linearization.
Key takeaways
- For a differentiable function, dy = f′(x) dx estimates the actual change Δy = f(x + dx) − f(x); a smaller dx generally makes the estimate more accurate.
- The linearization at x = a is L(x) = f(a) + f′(a)(x − a), the tangent-line equation used to estimate nearby function values.
- For a vessel radius shrinking from 1.4 mm to 1.3 mm, differentials estimate the area decrease as 0.28π mm², close to the exact decrease of 0.27π mm².
- For P(t) = 20t / (10 + t²) million, the differential estimate from 2 to 2.4 hours is about 0.245 million bacteria, while the exact increase is about 0.189 million, showing the effect of a relatively large dt.
- Linearization at 50 estimates 51² as 2,600, compared with the exact 2,601; linearization of √x at 25 estimates √27 as 5.2, compared with approximately 5.19615.
- For a sphere with radius 0.6 mm and radius tolerance ±0.015 mm, dV estimates a volume tolerance of ±0.0216π mm³.
Chapters
- Reviews the derivative limit definition, the power rule, trigonometric derivatives, and product, quotient, and chain rules.
- Recaps exponential and logarithmic relationships, including ln x having derivative 1/x and logarithmic functions being inverses of exponentials.
- States the change-of-base formula: log base b of x can be computed as ln x / ln b, with b positive and not equal to 1.
- Defines dx as the input change Δx and dy as f′(x) dx, the corresponding change along the tangent line.
- Distinguishes Δy = f(x + Δx) − f(x), the actual change along the curve, from dy; the two are close when Δx is small.
- Rearranges the approximation as f(x + Δx) ≈ f(x) + dy, providing the basis for linear approximation.
- Differentiates f(x) = 4x³ − 3x to obtain f′(x) = 12x² − 3, then writes dy = (12x² − 3) dx.
- At x = 3 and dx = 0.2, calculates dy = 21 as the tangent-line estimate of the function’s change.
- Computes the actual change using f(3.2) − f(3), illustrating why dy is a simpler estimate and why reducing dx improves closeness.
- Uses the product rule to differentiate y = x(4 − x)³, giving y′ = (4 − x)²(4 − 4x).
- At x = 5 and dx = 0.1, evaluates dy = −1.6, indicating an estimated decrease.
- Compares the estimate with the actual change f(5.1) − f(5) = −1.7881 and notes that the estimate improves for smaller input changes.
- Models the circular cross-sectional area as A = πr² and differentiates to get dA = 2πr dr.
- For a radius change from 1.4 mm to 1.3 mm, uses dr = −0.1 mm to estimate dA = −0.28π mm².
- Interprets the negative sign as a decrease and compares the estimate with the exact area change, −0.27π mm².
- Uses P(t) = 20t / (10 + t²), measured in millions of bacteria, and differentiates with the quotient rule.
- Obtains P′(t) = (200 − 20t²) / (10 + t²)² and estimates the change from t = 2 to t = 2.4 using dt = 0.4.
- Calculates dP ≈ 0.245 million, or about 245,000 bacteria; the exact increase is about 189,000, and the lesson notes that dt = 0.4 is relatively large.
- Chooses f(x) = x² and the nearby convenient input x = 50, where f(50) = 2,500.
- With dx = 51 − 50 = 1 and f′(x) = 2x, calculates dy = 2(50)(1) = 100.
- Estimates 51² as 2,500 + 100 = 2,600, one less than the exact value 2,601.
- Finds the point (50, 2,500) and tangent slope f′(50) = 100 for f(x) = x².
- Uses point-slope form and rearranges fully to the slope-intercept equation L(x) = 100x − 2,500.
- Evaluates L(51) = 2,600, matching the differential estimate and approximating the exact square 2,601.
- Models the operation with f(x) = √x and chooses x = 25 because √25 = 5 is easy to calculate.
- Uses dx = 27 − 25 = 2 and f′(x) = 1 / (2√x), giving dy = 1/5 at x = 25.
- Estimates √27 as 5 + 1/5 = 5.2, close to the calculator value of approximately 5.19615.
- Uses the tangent point (25, 5) and slope f′(25) = 1/10 for f(x) = √x.
- Rearranges the tangent-line equation into L(x) = x/10 + 5/2.
- Evaluates L(27) = 5.2, reproducing the differential estimate for √27.
- Models a spherical ball bearing with V = (4/3)πr³, so dV = 4πr² dr.
- For nominal radius 0.6 mm and tolerance dr = ±0.015 mm, estimates the maximum volume error as ±0.0216π mm³.
- Compares the differential estimate with exact volume changes at radii 0.615 mm and 0.585 mm, which are close but not perfectly symmetric.
Transcript excerpt
0:18 Good morning, class. This is chapter three. Derivative section 3.8. Differentials and linear approximation. So, in this chapter we've covered derivatives and rates of change, derivative as a function, basic differentiation formulas, product and quotient rules, chain rule, implicit differentiation, related rates, derivatives of logs, and now differentials and linear approximation. Reminding you of differentiation formulas such as its definition f prime of x is limit of f of x plus h minus f of x over h as h approaches zero or dy dx is limit of delta y over delta x as delta x approaches zero. And the meaning of e, there are two ways
1:06 to approach it, is e is equal to the limit as n approaches infinity of 1 plus 1 over n to the power of n, or limit as h approaches zero of e to the power of h minus 1 over h equals 1. Differentiation starting with the the power rule, the derivative of x to the power of n is n x to the power of n minus 1, which gives rise to derivative of to derivative of a constant to be zero, kx k k is a constant, and we continue with more such as uh trigonometric functions, the derivative of sine is cosine, cosine is negative sine, tangent is secant squared, cotangent is minus cosecant squared, secant results in secant x tan x, and cosecant x minus cosecant x cotangent x. We had other rules such as the product,
1:57 the quotient, the power rule, and chain rule, and of course the extended power rule. Exponentials, in which case the derivative is always starts with itself, and then we look at more, for example, the chain rule or the logarithmic properties. And we looked at logs, starting with natural log and other logs based on
This is the start of the transcript. Save this video free for the whole transcript, searchable and time-coded.
In this course
Related notes
Summary, takeaways, and chapters were generated by AI from the video's transcript and may contain errors. The video belongs to its creator, Professor Ara Araeipour.