2
$\begingroup$

In $\triangle ABC$, let $E$ be a point on $AC$, $F$ be a point on $AB$, and let $D$ be the intersection of $CF$ and $BD$. We are given that $A,E,D,F$ are concyclic, let this circle be $\Omega$. Let $\omega$ be the circle with diameter $BC$. Also, let $\omega$ and $\Omega$ intersect at $X$ and $Y$. If $M$ is the midpoint of $BC$, prove that $MX$ and $MY$ are tangent to $\Omega$.

I am not sure how to prove this. It may be worth noting that it is easy to see the case where $CF$ and $BE$ are altitudes and $D$ is the orthocenter.

Also, applying Pascal's Theorem on cyclic hexagon $FAEEDF$ implies the intersection of the tangents to $\Omega$ at $E$ and $F$ intersect on $BC$. Not sure if this helps though.

enter image description here

1 Answers 1

2

I have an algorithmic solution, but it is very far from being elegant (its style is "I have to solve this tough problem of a math competition in a very limited amount of time. Unleash brutality").

Let $\widehat{FCB}=\theta$. Then $\widehat{CFA}=\widehat{B}+\theta$ and since $AFDE$ is cyclic, $\widehat{AEB}=\pi-(\widehat{B}+\theta)$ and $\widehat{CBE}=\pi-(\widehat{B}+\widehat{C}+\theta)=\widehat{A}-\theta$, also because $\widehat{BDC}=\pi-\widehat{A}$.

Now we may use trigonometry to compute the lengths of $AF$ and $AE$ in terms of $a,b,c,\theta$, the cosine theorem to compute the length of $EF$ and $EF=2R_{\Omega}\sin\widehat{A}$ to compute $R_\Omega$.

In a skew reference system with $AE$ and $AF$ being the directions of the axis, we may also easily compute $MO_\Omega$. The orthogonality of $\omega$ and $\Omega$ then follows from checking that: $$ MO_\Omega^2 = \frac{a^2}{4}+R_{\Omega}^2.$$

Interesting (and probably useful for an elementary solution) facts:

  1. $BY$ and $CX$ meet on $\Omega$;
  2. $EX$ and $FY$ meet on $BC$;
  3. by changing $\theta$, $O_\Omega$ travels on a line.
  • 0
    Thanks for the solution! It would be cool if someone had a purely synthetic approach to this problem though.2017-01-05
  • 0
    @Math: have you attempted a circular inversion with respect to a circle centered at $A$ with radius $\sqrt{AB\cdot AC}$?2017-01-05
  • 0
    I should look in that. I did try an inversion about $\Omega$ which would leave $X$ and $Y$ fixed and map $MX$ and $MY$ to $(OM'X)$ and $(OM'Y)$ (where $O$ is the circumcenter of $\Omega$) and it would remain to show that these two circles are tangent to $\Omega$. Getting $M'$ is quite difficult though.2017-01-05