Linear approximation: use the tangent line as a local calculator

Near a known input, a smooth function behaves like its tangent line—and the derivative measures the approximation’s sensitivity.

LaTeX article Updated July 11, 2026

L(x)=f(a)+f(a)(xa)L(x)=f(a)+f'(a)(x-a)

Why the tangent line works

Differentiability means the error between the function and its best linear model becomes small compared with the input change. The tangent line matches both the value and first-order rate at a.

The approximation is local. Moving far from a lets curvature accumulate and weakens the estimate.

Choose the center strategically

Pick a close value with simple arithmetic: a perfect square for square roots, a familiar angle for trig, or zero for exponentials and logarithms when allowed.

A closer center usually improves accuracy, but simplicity and distance should be balanced.

Differentials and error estimates

Writing dy = f′(a) dx expresses the predicted output change. It is useful for sensitivity and measurement error: a small input uncertainty is scaled by the local derivative.

Worked example

Common mistakes

  • Using x instead of x−a in the tangent formula.
  • Choosing a center that is easy but not close.
  • Treating a local approximation as exact.

Keep these ideas

  • Match value and slope at a convenient nearby point.
  • The derivative predicts small output changes.
  • Curvature controls how quickly the approximation degrades.
Vocab
Mean value theorem
Math glossaryMean value theorem
f(c)=f(b)f(a)baf'(c)=\frac{f(b)-f(a)}{b-a}

Guarantees an instantaneous rate equal to an interval's average rate.

Learn more
Critical number
Math glossaryCritical number
f(c)=0 or undefinedf'(c)=0\ \text{or undefined}

A domain input where the derivative is zero or does not exist.

Learn more
Local maximum
Math glossaryLocal maximum
f(c)f(x)f(c)\ge f(x)

A function value at least as large as nearby values.

Learn more
Concavity
Math glossaryConcavity
f(x)>0concave upf''(x)>0\Rightarrow\text{concave up}

Describes whether graph slopes are increasing or decreasing.

Learn more
Linearization
Math glossaryLinearization
L(x)=f(a)+f(a)(xa)L(x)=f(a)+f'(a)(x-a)

A tangent-line approximation near a chosen input.

Learn more
Math glossary