1
$\begingroup$

This is a question from Brilliant.org

The triangle $ABC$ has $AB = 9$ and $AC:BC = 40:41$. What is the maximum possible area of $ABC$?

For this question, I considered the equation $A=\frac 12ab\sin\theta$.

Since $\sin\theta\le 1$, then $A$ is maximised when $\sin\theta = 1$.

This meant $ABC$ was a right-angled triangle, after some working I got the answer as 180. However, it was wrong, Brilliant said it is not necessarily a right-angled triangle and then used Heron's formula to find the maximised area, 820.

I checked other posts which were similar to my question such as, How to maximize the area of a triangle, given two sides?. However, they followed the same method I did

I am interested in why a right-angled triangle would not maximise the area in this case and what is wrong with my logic?

$\endgroup$
  • $\begingroup$ You are not given two sides of the triangle, but rather different data. Somehow, with base $9$ and the constraint given, you need to maximise the height $h$ of the triangle. $\endgroup$ – Mark Bennet Jun 7 '18 at 22:15
  • $\begingroup$ Well, if $a,b$ would have been fixed or independent, yes, but we have an expression with two constraints. I will try a similar question. The triangle $ABC$ has $AB=9$ and $AC:BC=1$. What is the maximum possible area of $ABC$? Please solve this problem first... $\endgroup$ – dan_fulea Jun 7 '18 at 22:15
  • $\begingroup$ You aren't given a triangle with two sides. You are given a triangle with one side and a proportion to the other two. $\endgroup$ – fleablood Jun 7 '18 at 22:21
  • $\begingroup$ Ohh, that makes much more sense. Thank you! $\endgroup$ – Confused Bob Jun 7 '18 at 23:00
1
$\begingroup$

Your confusion stems from the fact that the area is maximized at $\theta=90$, if you keep the side lengths $BC,AC$ fixed. However in this problem $BC,AC$ can vary in length.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ AB cannot vary. $\endgroup$ – Szeto Jun 7 '18 at 22:19
  • $\begingroup$ @Szeto: whoops wrong side. Fixed. $\endgroup$ – Alex R. Jun 7 '18 at 22:19
1
$\begingroup$

Let the sides $AC=40x, BC=41x$. Using the Heron's formula: $$S=\sqrt{\frac{81x+9}{2}\cdot \frac{81x-9}{2}\cdot \frac{9+x}{2}\cdot \frac{9-x}{2}}=\frac{81}{4}\sqrt{\left(x^2-\frac 1{81}\right)\left(81-x^2\right)}\overbrace{\le}^{GM-AM} \\ \frac{81}{4}\cdot \frac{\left(x^2-\frac 1{81}\right)+\left(81-x^2\right)}{2}=820,$$ equality occurs when $x^2-\frac1{81}=81-x^2 \Rightarrow x\approx 3$.

Note: $GM-AM$ is the Geometric Mean-Arithmetic Mean inequality.

| cite | improve this answer | |
$\endgroup$
0
$\begingroup$

As mentioned in a comment, the best way to realize where is the error is to consider the similar problem where $AB=9$ is given and fixed, and where $AC:BC=1$. Then in this case the triangle is isosceles, and we have the option to choose its height as big as we want. The bigger this heigth $h$, (the smaller the angle in $C$, and) the bigger the area. There is no limit for the area, we can let $h$ tend to infinity, the area goes to infinity!

As seen also in this example, in the given situation the sides $a=BC$, $b=AC$ and the angle in $C$ variate together in a "complicated way", and contribute more or less to the area. A maximum for $\sin \hat C$ does not lead to a maximal area, since the sides $a,b$, and the height $h$ are dramatically constrained to be "small".


Now let us also give a simple solution to the "brilliant" problem.

First of all, let us consider the following example of a triangle satisfying the given conditions, the one with sides $9,40,41$. Yes, it has a right angle, but not there where i supposed first that the question supposed it to be... In a (bad) picture:

Triangle ABC''

(Sorry, i cannot get it smaller. But maybe it is psychologically good so.)

A possible choice of $C$ is as above, in the position which makes the angle in $B$ a right angle. An other position is the symmetrical position, which makes the angle in $A$ a right angle. With the one or the other choice we get a first weak attempt to maximize the area, it is $\frac 12\cdot 9\cdot 40=180$.

Now let us try to get a better guess. Consider a point on $AB$ which divides $AB$ in two parts, $s$ and $9-s$.

In fact we allow $$s\in\Bbb R\ ,$$ for a negative value of $s$ we consider the point on $AB$ to be on the half-line $(AB$ "after" $B$.

We draw a long perpendicular in this point to $AB$, choose on it that one point (in the "upper" half-plane) $C''$, so that the sides $AC$ and $BC$ are in the given proportion: $$ \frac{AC''}{BC''}=\frac{41}{40}\ . $$ This is maybe possible, maybe impossible. (For instance, for $s=9/2$ this is impossible.) If it is possible, we want to get the height $h=h(s)$ as a function of $s$, that we will maximize.

The above figure is maybe wrong, maybe $C''$ is so that the corresponding height is $\le 40$, and maybe it is there where $?C''$ is placed, well, we have to compute. (Yes, it is wrong.)

Note: This is somehow related to the posted question. It turns out, that the maximal height is obtained for a negative $s$, (which is relatively big in absolute value,) the point on the line $AB$ which realizes the maximal height is not on the segment $AB$.

Computations. Note that from $$ \frac{41}{40} =\frac{AC''}{BC''}= \frac {\sin \hat B} {\sin \hat A} = \frac {h/\sqrt{h^2+s^2}} {h/\sqrt{h^2+(9-s)^2}} = \frac {\sqrt{h^2+(9-s)^2}} {\sqrt{h^2+s^2}} $$ we get a simple relation determining $h$, $$ \frac{41^2}{40^2} = \frac {h^2+(9-s)^2} {h^2+s^2} \ , $$ so $$ \begin{aligned} \underbrace{(41^2-40^2)}_{=9^2}h^2 &= 40^2(9-s)^2-41^2s^2 \\ &= 9(40-9s)(s+360)\ . \end{aligned} $$ The roots of the last expression are $40/9$ and $-360$. To have a positive quantity, we need $40-9s\ge 0$, so $s\le 40/9$.

The maximal value of the expression is taken exactly in the middle of the segment determined by the roots, which is $s^*=-1600/9$. For this value of the $s$-variable we get a corresponding maximal $h^*$, which is $$ h^* =\frac 19\sqrt{9\cdot (40+1600)\left(-\frac {1600}9+360\right)} = \frac {1640}9\ . $$ So the maximal area is $\frac 12 \cdot AB\cdot h^*=\frac{1640}2=820$.

| cite | improve this answer | |
$\endgroup$
0
$\begingroup$

Since we have expressions for the side lengths as $a,b=px,c=qx$ ($a=9,p=40,q=41$), we can ignore all the angles and use Heron’s formula for the area \begin{align} S&=\tfrac14\sqrt{4a^2b^2-(a^2+b^2-c^2)^2} \tag{1}\label{1} ,\\ S(x)&=\tfrac14\sqrt{2x^2a^2(q^2+p^2)-x^4(p-q)^2(q+p)^2-a^4} \tag{2}\label{2} . \end{align}

The $x$ that maximizes the area $S(x)$, also maximizes $S(x)^2$, so we can consider \begin{align} S_2(x)&=S(x)^2=\tfrac1{16}(2x^2a^2(q^2+p^2)-x^4(p-q)^2(q+p)^2-a^4) ,\\ S_2'(x)&= \tfrac14x(a^2(q^2+p^2)-(p-q)^2(q+p)^2x^2) \end{align}

And the sought maximum can be found at

\begin{align} x_{\max}&= \frac{a\sqrt{q^2+p^2}}{q^2-p^2} =\tfrac19\sqrt{3281} \approx 6.36 ,\\ S_{\max}&= S(x_{\max})=820 . \end{align}

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.