Source

Adapted from my own handwritten and hand-typed notes from Year 1 to Year 3 of my EEE degree

Near a point , a smooth curve looks almost like its tangent line. Linearization uses that tangent to approximate , and differentials use it to estimate small changes and errors. The same tangent-line idea powers Newton's method.

Linearization

The linearization of at is

and for close to . The graph is exactly the tangent line at .

Steps:

  1. Choose and a nearby point where and are easy
  2. Compute and
  3. Write and evaluate it at the required

Worked Examples

with , :

  • and , so
  • , so (true value )

with , :

  • and
  • , so

with , :

  • and

Higher accuracy uses the Taylor polynomial ; linearization is the case .

Differentials

If changes from to , the true change is . The differential is the change along the tangent:

Type of changeActualEstimated
Absolute
Relative
Percentage

Note: is a single symbol for the derivative, not literally " divided by ".

Worked Example: Measurement Error

A disc has radius cm, measured with maximum error cm. Area :

  • cm², the estimated maximum error in area
  • relative error
  • percentage error