Source

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

On a general domain, a function may have no global max or min. The extreme value theorem gives a condition that guarantees them.

Extreme value theorem

If is continuous on a set that is closed and bounded, then has a global maximum and a global minimum on .

Meaning of the conditions:

  • Bounded: the region does not run off to infinity (fits inside some big disk)
  • Closed: the region includes its boundary (the boundary uses or , not )
  • Continuous: true for the usual functions (polynomials, , , etc.) on their domains

Method (closed and bounded region)

  1. Interior: find the critical points of that lie inside the region. Ignore any outside it
  2. Boundary: find candidate points on the boundary, either by substituting the boundary equation into (simple boundaries such as line segments) or by Lagrange multipliers. For a boundary made of several pieces, also include the corner points
  3. Compare: evaluate at all candidates. Largest is the global max, smallest is the global min

No second derivative test is needed here, since we just compare values.

Point vs value

The extreme value is a number . The point is where it happens. Answer both when asked.

Worked example

Find the extreme values of on the disk . The disk is closed and bounded and is continuous, so the extrema exist.

Step 1 (interior): gives , which is inside. .

Step 2 (boundary): on , Lagrange multipliers give and .

Step 3 (compare):

Point

Global maximum at , global minimum at .

Boundary by substitution

On the same circle, gives for . Then at , plus the end points . This gives the same candidates as Lagrange.