This is a paragraph.

Condition for bisecting a diagonal

Convex quadrilaterals here.

Consequently, the bisection condition to be established in today’s post applies only to convex quadrilaterals — quadrilaterals where none of the interior angles exceed 180^{\circ}.

Concave quadrilaterals also exist, but we won’t cover those.

Check how many attacks we faced this week! Cause for concern?


Let ABCD be a convex quadrilateral whose diagonals AC and BD intersect at E. If AC bisects BD, PROVE that \frac{AE}{EC}=\frac{d^2-a^2}{b^2-c^2}.

As usual, a,b,c,d are the side-lengths AB,BC,CD,DA. Assume for the time being that none of these sides are equal.

Rendered by QuickLaTeX.com

Let the length of diagonal BD be q. Since AC bisects BD, both AE and CE are medians in triangles ABD and BCD, respectively. Thus:

    \[AE^2=\frac{2a^2+2d^2-q^2}{4},~CE^2=\frac{2b^2+2c^2-q^2}{4}\]

The angles AEB and CED are equal, so:

    \begin{align*} \cos(\angle AEB)&=\cos(\angle CED)\\ \frac{AE^2+BE^2-a^2}{2\times AE\times BE}&=\frac{CE^2+ED^2-c^2}{2\times CE\times ED}\\ \frac{AE^2+BE^2-a^2}{AE}&=\frac{CE^2+ED^2-c^2}{CE}\\ \frac{AE}{EC}&=\frac{AE^2+BE^2-a^2}{CE^2+ED^2-c^2}\\ &=\frac{\frac{2a^2+2d^2-q^2}{4}+\left(\frac{q}{2}\right)^2-a^2}{\frac{2b^2+2c^2-q^2}{4}+\left(\frac{q}{2}\right)^2-c^2}\\ &=\frac{d^2-a^2}{b^2-c^2} \end{align*}

We later realized that applying Stewart’s theorem would have made the proof simpler. Same with the converse in example 2 below.

Let ABCD be a convex quadrilateral whose diagonals AC and BD intersect at E. If \frac{AE}{EC}=\frac{d^2-a^2}{b^2-c^2}, PROVE that AC bisects BD.

To this end, consider the four triangles AEB, BEC, AED, and CED. For simplicity let’s set

    \[AE=p_1,~EC=p_2,~BE=q_1,~ED=q_2\]

Thus the assumption \frac{AE}{EC}=\frac{d^2-a^2}{b^2-c^2} can then be re-written as \frac{p_1}{p_2}=\frac{d^2-a^2}{b^2-c^2}, or p_1=\left(\frac{d^2-a^2}{b^2-c^2}\right)p_2. The angles \angle AEB and \angle BEC are supplementary, so:

    \begin{align*} \cos(\angle AEB)&=-\cos(\angle(BEC)\\ \implies\frac{AE^2+BE^2-AB^2}{2\times AE\times BE}&=-\left(\frac{BE^2+EC^2-BC^2}{2\times BE\times EC}\right)\\ \implies \frac{p_1^2+q_1^2-a^2}{2p_1q_1}&=-\left(\frac{q_1^2+p_2^2-b^2}{2q_1p_2}\right)\\ p_2(p_1^2+q_1^2-a^2)&=-p_1(q_1^2+p_2^2-b^2)\\ (p_1+p_2)q_1^2&=b^2p_1+a^2p_2-p_1p_2^2-p_1^2p_2\\ &=b^2\left[\left(\frac{d^2-a^2}{b^2-c^2}\right)p_2\right]+a^2p_2-p_1p_2(p_1+p_2)\\ &=\left(\frac{b^2d^2-a^2c^2}{b^2-c^2}\right)p_2-p_1p_2(p_1+p_2)\\ \therefore q_1^2&=\frac{p_2}{p_1+p_2}\left(\frac{b^2d^2-a^2c^2}{b^2-c^2}\right)-p_1p_2 \end{align*}

Similarly, the angles \angle CED and \angle AED are supplementary; so:

    \begin{align*} \cos(\angle CED)&=-\cos(\angle AED)\\ \implies \frac{ED^2+EC^2-CD^2}{2\times ED\times EC}&=-\left(\frac{AE^2+ED^2-DA^2}{2\times AE\times ED}\right)\\ \frac{q_2^2+p_2^2-c^2}{2q_2p_2}&=-\left(\frac{p_1^2+q_2^2-d^2}{2p_1q_2}\right)\\ p_1(q_2^2+p_2^2-c^2)&=-p_2(p_1^2+q_2^2-d^2)\\ (p_1+p_2)q_2^2&=c^2p_1+d^2p_2-p_1p_2^2-p_2p_1^2\\ &=c^2\left[\left(\frac{d^2-a^2}{b^2-c^2}\right)p_2\right]+d^2p_2-p_1p_2(p_1+p_2)\\ &=\left(\frac{b^2d^2-a^2c^2}{b^2-c^2}\right)p_2-p_1p_2(p_1+p_2)\\ \therefore q_2^2&=\frac{p_2}{p_1+p_2}\left(\frac{b^2d^2-a^2c^2}{b^2-c^2}\right)-p_1p_2 \end{align*}

The preceding calculation shows that q_1^2=q_2^2, or q_1=q_2. Thus BE=ED and the diagonal BD is bisected as desired.

Let ABCD be a convex quadrilateral whose diagonals intersect at E. Explain how the condition \frac{AE}{EC}=\frac{d^2-a^2}{b^2-c^2} works in the case of parallelograms.

For parallelograms, opposite sides are equal, so we can set b=d and a=c and obtain \frac{AE}{EC}=1. For squares and rhombuses, we need to take care: pretend that a and d are different, and that b and c are also different; at the same time, use the fact that opposite sides are equal. Doing this gives \frac{AE}{EC}=1.

Rendered by QuickLaTeX.com

Let ABCD be a convex quadrilateral whose diagonals intersect at E. Explain how the condition \frac{AE}{EC}=\frac{d^2-a^2}{b^2-c^2} works in the case of kites.

This is a special case because a=d and b=c and opposite sides are not equal in general kites. So we’re confronted with the indeterminate form \frac{0}{0}. However, the longer diagonal in a kite can be as long as possible, and so the ratio \frac{AE}{EC} evaluates differently depending on the kite in question. Does this somehow explain why \frac{0}{0} is said to be undefined?

Rendered by QuickLaTeX.com

If quadrilateral ABCD is such that diagonal AC bisects diagonal BD and b=c, PROVE that ABCD is a kite.

Let E be the point of intersection of AC and BD. Since b=c, triangle BCD is isosceles, and so CE is an altitude. Thus AE is also an altitude. Since BE=ED, this forces a=d. So we obtain a kite.

Takeaway

Let ABCD be a convex cyclic quadrilateral whose diagonals AC and BD intersect at E. If the side-lengths are AB=a, BC=b, CD=c, and DA=d, then the following statements are equivalent:

  1. ab=cd
  2. \frac{AE}{EC}=\frac{d^2-a^2}{b^2-c^2}
  3. \frac{AE}{EC}=\frac{d^2+a^2}{b^2+c^2}
  4. AC bisects BD.

Notice the second and third statements.

Task

  • (Bisection consequence) Let ABCD be a convex quadrilateral with side-lengths AB=a, BC=b, CD=c, and DA=d. If AC bisects BD and ab=cd (quasi-harmonic), PROVE that ABCD is cyclic.
  • (Basic characteristics) Let ABCD be a convex cyclic quadrilateral whose diagonals AC=p and BD=q intersect at E. If ab=cd (quasi-harmonic), PROVE that:
    1. p^2=AE^2+BE^2+CE^2+DE^2
    2. pq\geq 2ab.