# The area of an inscribed triangle

3 min read
Table of Contents

Some days ago, a friend presented me a problem: If you have a circle of radius rr, Which is the inscribed quadrilateral with the biggest area.

Currently I don’t have an answer for that question, but I decided to try something more easy. Which is the inscribed triangle with the biggest area?

Even if it seems more easy, I don’t have a simple solution. The intuition tells me that probably is the quadrilateral triangle, but in math intuition is not enough, it’s necessary a proof.

To attack the problem, I decided to take some suppositions. First, the circle will be of radius $$. Second the triangle will have three vertices, AA, BB and CC, that lies in the circle.Third CC will be the point (1,0)(-1,0) and last I will express AA and BB with polar coordinates, as follows:

A=(1;α),B=(1;β),π<β<α<π.A = (1; \alpha), \qquad B = (1; \beta), \quad -\pi < \beta < \alpha < \pi.

The first thought I had was: How I can get the area of the triangle? Well, this is an interesting question, what I will do is to know how to get the area of the parallelogram that defines two vectors. If I know this area, then I can just divide by two and get the area of the triangle that defines two vector. And how this can help me? Easy, if I know how to calculate the area of this triangle I have two options to calculate the area of the triangle ABC\triangle ABC:

  1. Use CAC-A and CBC-B as the vectors, since the triangle defined by these two vectors is congruent of the triangle ABC\triangle ABC.
  2. If OO is the origin (0,0)(0,0), then calculate the area of the triangles AOB\triangle AOB, BOC\triangle BOC and AOC\triangle AOC.

The area of the parallelogram defined by two vectors

Suppose that we have two vectors u\gdef\u{\bec u} \u and v\gdef\v{\bec v} \v, together with the origin and the vector u+v\u + \v form a parallelogram. If θ\theta is the angle defined by u\u and v\v a easy way to calculate the area AuvA_{\u\v} of the parallelogram defined by u\u and v\v is

Auv=uvsinθ.A_{\u\v}= \norm{\u} \norm{\v} \sin\theta.

The proof of this property is simple, the formula for the area of a parallelogram it’s just base times height. We can use any vector as the base, and for the height, we can just use trigonometry and it will be the other vector times the sine of the angle between the vectors.

It’s possible to calculate sinθ\sin\theta in terms of u\u and v\v. To do this, first, it’s easies to calculate Auv2A_{\u\v}^2, because we can convert sin2θ=1cos2θ\sin^2\theta = 1 - \cos^2\theta. This way we got the next formula:

Auv=u2v2(1cos2θ) A_{\u\v} = \norm{\u}^2 \norm{\v^2} (1 - \cos^2\theta)

It might looks worse than the last formula, the change of sine to cosine makes the formula more easiest, just because if we use the cosine law we got the next formula

cosθ=uvuv.\cos\theta = \frac{\u\cdot\v}{\norm{\u}\norm{\v}}.

This way, if we join both formulas, we finally got a really easy way of calculate the area of the parallelogram defined by u\u and v\v:

Auv2=u2v2(uv)2.A_{\u\v}^2 = \norm{\u}^2 \norm{v}^2 - (u\cdot v)^2.

Some people may look this formula familiar. And that’s because is related to the Cauchy-Swartz inequality. Indeed this is a nice proof of the Cauchy-Swartz inequality, if we asume that we are in a Euclidean space (since we use terms like area or angle between two vectors).

Profile pic

Thanks for reading :3


More Posts

Comments