Skip to main content

Unit 8.1 Optimization in geometry

We’ll start with a few geometric problems, where the objective function is a little more obvious.

Example 8.1.

We’re going to build a window in the shape of a rectangle topped by an isosceles triangle whose height is the same as the width of the rectangular portion. We want to make a window which lets in the most light -- that is, with the greatest possible area. In order to build the window, we have to use wood trim. We have 16 feet of wood trim to build the window with.
Such a window has two dimensions: the width \(w\) and the height \(h\) of the rectangle. The rectangular portion has area \(wh\) and the triangular portion has area \(\frac{1}{2}w^2\text{.}\) So the total area is
\begin{equation*} A(w,h)=wh+\frac{1}{2}w^2\ . \end{equation*}
We also need to record the fact that our supplies are limited. A little geometry shows that to build the window requires two pieces of trim with length \(h\text{,}\) two of length \(w\text{,}\) and two of length \(\frac{\sqrt{3}}{2}w\text{.}\)
Figure 8.2. A window made from a rectangle and a triangle.
Technically we don’t have to use all the trim, but if we had some left over, we could have used it to build a bigger window. So let’s assume we use all 16 feet; that is, we assume
\begin{equation*} 16=2h+2\left(1+\frac{\sqrt{3}}{2}\right)w\ . \end{equation*}
We can solve this equation for either \(w\) or \(h\text{.}\) Let’s solve for \(h\text{:}\)
\begin{equation*} h=8-\left(1+\frac{\sqrt{3}}{2}\right)w \end{equation*}
and substitute that into the formula for area:
\begin{equation*} A(w,h(w))=w\left(8-\left(1+\frac{\sqrt{3}}{2}\right)w\right)+\frac{1}{2}w^2 \end{equation*}
Now we’ve got a function which we can optimize. We want to have a sensible result, so we know that \(w\) can’t be less than 0, and can be at most \(\frac{8}{1+\frac{\sqrt{3}}{2}}\text{.}\) So we want to optimize on the interval \(\left[0,\frac{8}{1+\frac{\sqrt{3}}{2}}\right]\text{.}\)
Differentiating, we get \(\frac{dA}{dw}=8-\left(1+\sqrt{3}\right)w\text{.}\) So there is a single critical point at \(w=\frac{8}{1+\sqrt{3}}\text{.}\) We have
\begin{gather*} A(0,h(0))=0\\ A\left(\frac{8}{1+\frac{\sqrt{3}}{2}},h\left(\frac{8}{1+\frac{\sqrt{3}}{2}}\right)\right)=A\left(\frac{8}{1+\frac{\sqrt{3}}{2}},0\right)=\frac{1}{2}\left(\frac{8}{1+\frac{\sqrt{3}}{2}}\right)^2=\frac{128}{\left(2+\sqrt{3}\right)^2}\\ A\left(\frac{8}{1+\sqrt{3}},h\left(8+\sqrt{3}\right)\right)=\frac{32}{1+\sqrt{3}} \end{gather*}
Since \(\frac{32}{1+\sqrt{3}}\approx 11.71\) is greater than either 0 or \(\frac{128}{\left(2+\sqrt{3}\right)^2}\approx 9.12\text{,}\) we see that the maximal area occurs when we choose \(w=\frac{8}{1+\sqrt{3}}\) and \(h=h(\frac{8}{1+\sqrt{3}})=\frac{8}{1+\sqrt{3}}\frac{\sqrt{3}}{2}\text{.}\)
This example shows a few things. First, notice that many optimization problems come equipped with a constraint. Here that was the fact that we only had so much trim. If you think about it, constrained optimization is the kind you usually deal with -- our world is full of scarcity.
Second, we can interpret each of the values we compared in the context of the problem. At \(w=0\text{,}\) the window has width zero. At \(w=\frac{8}{1+\frac{\sqrt{3}}{2}}\text{,}\) \(h=0\) so we only have a triangular section of window. \(w=\frac{8}{1+\sqrt{3}}\) is somewhere in between.
Third, consider the optimal dimensions: we have \(h=w\frac{\sqrt{3}}{2}\text{.}\) In other words, the height of the rectangular portion is equal to the side length of the triangular section. This is typical -- optimizers are often symmetric in some sense.
Here are some other geometric optimization problems.

Checkpoint 123.

Given a total length of 10 meters of rope, what’s the greatest area we can enclose in a rectangle? Answer to three decimal places.
Using the same length of rope, what base length should we use to obtain the greatest area in an isosceles triangle? Answer to three decimal places.
Answer 1.
\(6.25\)
Answer 2.
\(3.33333\)
Figure 8.3. Windows made from a rectangle and a triangle.

Checkpoint 124.

In the window example, let’s say we remove the “middle” piece of trim (which had length \(w\)). First, a gut check: does this increase or decrease the optimal area?
What if we double up the middle trim? Does this increase or decrease the optimal area?
Now verify your intuition by computing the optimal area in each scenario. Answer to three decimal places.
area with no crossbar:
area with double crossbar:
Answer 1.
\(\text{increase}\)
Answer 2.
\(\text{decrease}\)
Answer 3.
\(18.4752\)
Answer 4.
\(7.16829\)