2.2. THE EXTREME VALUE THEOREM REDUX 53
Solution:
In this problem we are cutting a parabolic segment out of the plane s.x; y/ D 2x C y 4. We
get that the ends of the domain of optimization are .2; 1/ and .2; 1/ by plugging in the ends of
the interval in x to the formula for the parabolic segment. Substituting the parabolic segment
into the plane yields
s.x; x
2
3/ D 2x C x
2
3 4 D x
2
C 2x 7
is means our critical point appears at 2x C 2 D 0 or x D 1, making the candidate point
.1; 2/. Plugging the candidate points into s.x; y/ yields:
s.2; 1/ D 7
s.1; 2/ D 8
s.2; 1/ D 1
is means that the maximum value is 1 at .2; 1/, and the minimum is 8 at .1; 2/.
˙
Some sets of boundaries are simple enough that we can use geometric reasoning to avoid needing
to use calculus on the boundaries.
Example 2.26 Find the maximum value of
h.x; y/ D x
2
C y
2
for 2 x; y 3.
Solution:
First of all, we know that this surface has a single critical point at .0; 0/ – this surface is
an old friend (see graph in Figure 1.2).
e domain of optimization is a square with corners .2; 2/ and .3; 3/. Along each side
of the square, we see that the boundary is a line – and so has extreme values at its ends. is
means that we need only check the corners of the square; the interior of the edges – viewed as
line segments – cannot attain maximum or minimum values by the extreme value theorem. is
means we need only add the points .2; 2/, .2; 3/, .3; 2/, and .3; 3/ to our candidates.