7
$\begingroup$

I am given the lengths of 4 sides of a convex quadrilateral as $a$, $b$, $c$, $d$.

I am also given the sequence of the sides (i.e. it is $abcd$, not $abdc$, or $acbd$, or $acdb$,... etc).

I know that Bretschneider's_formula http://en.wikipedia.org/wiki/Bretschneider%27s_formula

Bretschneider's formula

(where $ s = (a+b+c+d)/2 $ is the semi-perimeter)

can give the area if the sum of either pair of opposite angles are known.

But I do not know the angles of any of the vertices (interior or exterior).

I can generate two quadrilateral equations by generating diagonal lines $e$, $f$ and using the cosine rule of the triangle...

$$ a^2 + b^2 - 2.a.b.\cos(\alpha) (= e^2) = c^2 + d^2 - 2.c.d.\cos(\gamma) $$

$$ b^2 + c^2 - 2.b.c.\cos(\beta) (= f^2) = a^2 + d^2 - 2.a.d.\cos(\delta) $$

These two equations have four unknowns (i.e. the interior angles: $\alpha$, $\beta$, $\gamma$, $\delta$). $\delta$ is given by:

$$ \delta = 360 - \alpha-\beta-\gamma $$

so this leave three unknowns but only two equations.

Where do I go from here?

ANSWER

I have accepted Ted Shifrin's answer, if opposite angles sum to $\pi$ then we can use a simplified form of Bretschneider's formula to get the (maximum) area K.

$$ K^2 = [(s-a)(s-b)(s-c)(s-d)] $$

For finding the maximum area we dont need to know what the angles actually are.

NOTES

I dont think it matters either that the maximum area quadrilateral is concyclic although Ted indicates that it can be proved that a quadrilateral with opposite angles summing to $\pi$ must be concyclic.

Also I deduce that any angle of a cyclic quadrilateral can be obtained from:- $$ cos(\alpha) = (a^2 + b^2 - c^2 - d^2)/(2(a.b +c.d)) $$ where $\alpha$ is the interior angle between sides a and b. Which is nice. :)

$\endgroup$
3
$\begingroup$

You will find that the optimal solution is for the opposite angles to add to $\pi$ and, thus, for the quadrilateral to be concyclic (inscribed in a circle).

$\endgroup$
  • $\begingroup$ Hah, so much simpler than my solution. $\endgroup$ – Lee Mosher Jul 21 '14 at 17:02
  • $\begingroup$ I read that a quadrilateral can only be concyclic if opposite angles sum to $\pi$. But is the reverse proven? i.e. "All quadrilaterals whose opposite angles sum to $\pi$ can be inscribed in a circle". $\endgroup$ – steveOw Jul 22 '14 at 13:19
  • $\begingroup$ Yes, @steveOw. There is a unique circle through 3 of the 4 points. If the 4th point lies inside or outside the circle, then geometry tells us that the opposite angles will not sum to $\pi$. $\endgroup$ – Ted Shifrin Jul 22 '14 at 15:08
  • $\begingroup$ Thanks @Ted Shifrin. I can see how a proof might be made using the relationships between a chord and peripheral angles and central angles in a circle. $\endgroup$ – steveOw Jul 22 '14 at 15:38
1
$\begingroup$

Notice that on the right hand side of Bretschneider's formula, everything is constant except $\alpha$ and $\gamma$. So the area is maximized if and only if $\cos^2 (\frac{\alpha+\gamma}{2})$ is minimized. But $\alpha$ and $\gamma$ are not independent. From the figure one can see that $\alpha$ and $\gamma$ are constrained by applying the law of cosines to two of the triangles in the picture: $$(*) \qquad b^2 + c^2 - 2bc \cos(\gamma) = a^2 + d^2 - 2ad \cos(\alpha) $$ So now you have a constrained optimization problem, and you can try to solve it using Lagrange's method: find the minimum of $\cos^2 (\frac{\alpha+\gamma}{2})$ subject to the constraint $(*)$ (and restricted to $\alpha,\gamma \in [0,\pi]$).

You might be able to simplify the problem even further using some symmetry. For example: if you restrict to $\alpha,\gamma \in [0,\pi/2]$, so that $\cos(\alpha+\gamma) \ge 0$, then $\cos^2(\frac{\alpha+\gamma}{2})$ is minimized if and only if $\cos(\frac{\alpha+\gamma}{2})$ is also minimized.

$\endgroup$
  • $\begingroup$ thanks. Like you say:- minimizing the RHS of the Bretschneider's formula gives the maximum area which is all I need and so I just need to assume $\alpha + \gamma = \pi$...I dont need to know the actual values of the angles. I should have seen that myself. $\endgroup$ – steveOw Jul 22 '14 at 18:52

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.