Optimizing a Function of Several Variables
Using Partial Derivatives

Local maximums and local minimums of a function of one variable, f(x), are found at critical points (points where f'(x) = 0 or f'(x) is undefined) and at endpoints, as shown in the figure below. Thus, one way to find local maximums and minimums is by looking at critical points and endpoints.

Missing figure

The same idea works for a function of two or more variables except that a critical point is a point at which all the partial derivatives are zero or undefined.

Example:

Suppose that a manufacturer wants to build a box that will hold one cubic foot of sand. The box will be in the shape of a rectangular prism as shown in the figure below. The manufacturer wants to minimize the cost of materials for the box by minimizing the total surface area of all six sides.

Missing figure

Notice that the total surface area is

S(x, y) = 2xy + 2xh + 2yh

Since the total volume is one cubic foot

xyh = 1

and

h = 1/xy

we can write

Missing equation

The two partial derivatives are

Missing equation

So, to find the critical points we must solve the two equations

Missing equation

or

Missing equation

Thus,

x = y = h = 1

and the box should be a cube.


  1. Suppose that a manufacturer wants to build a box with an open top that will hold one cubic foot of sand. She wants to minimize the cost of materials by minimizing the total surface area of the box. What dimensions should the box have? answer

  2. Suppose that a manufacturer wants to build a box with an open top that will hold one cubic foot of sand. She wants to minimize the cost of materials. Because sand is so heavy the bottom of the box will cost four times as much per square foot as the sides. What dimensions should the box have? answer

  3. Find the largest box (in volume) whose total surface area is six cubic feet. answer

  4. Find all the critical points of the function

    Missing equation

    Draw a graph of this function showing all the critical points and describe each one in prose. answer


Copyright c 1997 by Frank Wattenberg, Department of Mathematics, Montana State University, Bozeman, MT 59717