Source

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

The directional derivative is the rate of change of at a point in the direction of a unit vector . Partial derivatives are the special cases (gives ) and (gives ).

Definition

Formula

If is differentiable,

The same holds in 3 or variables: .

Proof idea: let . Then , and the multivariable chain rule gives .

Method

  1. Find and evaluate it at the point
  2. Turn the given direction into a unit vector: , or for an angle, or for "from to "
  3. Take the dot product

Forgetting to normalise is the most common mistake.

Maximum and minimum rate of change

Since :

ValueDirection
Maximum rate (along the gradient, )
Minimum rate (against the gradient, )
Zero rateperpendicular to (along the level curve)

Worked example

Find for at in the direction of .

Worked example: max rate

at , direction from to .

Working:

  • , so
  • with length , so
  • Maximum rate is , in the direction

The gradient also gives the normal vector for a tangent plane.