2
$\begingroup$

Let $BB',CC'$ be altitudes in $\triangle ABC$, and assume $AB\neq AC$. Let $M$ be the midpoint of $BC$, $H$ the orthocenter of $\triangle ABC$, and define $D$ as the intersection of lines $BC$ and $B'C'$. Show that $DH$ is perpendicular to $AM$

I am trying to apply the converse of the radical axis theorem to prove the desired prependicularity (since $AC'HQ$ cyclic implies $DQ'\perp AH$), but am having trouble on finding what circles to use.

The theorem states: Suppose that $ABCD$ and $CDEF$ are cyclic quadrilaterals and that line $AB,CD,EF$ are concurrent. Then $EFAB$ is also cyclic. More generally, if $\omega_1,\omega_2$ are two cricles with radical axis $l$, and $A,B$ are points on $\omega_1$, and $AB$ and $EF$ meet at a point on $l$, then $ABEF$ are concylic.

Anyone have any ideas?

$\endgroup$
1
  • $\begingroup$ This is known as the Humpty Point in modern geometry vocabulary; invert at H about the polar circle $\endgroup$ Commented Sep 30, 2024 at 5:23

1 Answer 1

1
$\begingroup$

Not using the radical axis, and not even purely geometrical, but here is a relatively straightforward proof.

Assume WLOG that $A, B, C\,$ lie in a complex plane with the origin set to the circumcenter $O$ of $\triangle ABC$, and scaled such as $\lvert OA \rvert = \lvert OB \rvert = \lvert OC \rvert = 1$.

With the usual notation where $z$ is the complex number representing point $Z$, let $z = a + b + c$. Then $\langle z - a, b - c \rangle = \langle b + c, b - c \rangle = \langle b, b \rangle + \langle b, c \rangle - \langle b, c \rangle - \langle c, c \rangle = {\lvert b \rvert}^2 - {\lvert c \rvert}^2 = 1 - 1 = 0$ thus $AZ \perp BC$ so $Z$ is on the altitude through $A$. By symmetry, Z must be on the other 2 altitudes of the triangle as well, thus $Z \equiv H$ the orthocenter of the triangle, and therefore:

$$ h = a + b + c $$

(As a side note, the relation is well known in the context of the Euler line, though it is more commonly stated in vector notation $\overrightarrow{OH} = \overrightarrow{OA} + \overrightarrow{OB} + \overrightarrow{OC}$, see for example paragraph "9. Vector Algebra II" at Existence of the Orthocenter.)

Let $A'$ be the foot of the altitude through $A$. Since $A'$ is on $BC$:

$$ a' = \lambda b + \mu c \;\;\;\; \text{for some real}\;\; \lambda, \mu \;\; \text{with} \;\; \lambda + \mu = 1 $$

Furthermore, since $A'H \perp BC$:

$$ \begin{align} 0 = \langle h - a', b - c \rangle & = \langle a + (1 - \lambda) b + (1 - \mu) c, b - c \rangle \\ & = \langle a + \mu b + \lambda c, b - c \rangle \\ & = \langle a, b \rangle - \langle a, c \rangle + \mu \langle b, b \rangle - \mu \langle b, c \rangle + \lambda \langle b, c \rangle - \lambda \langle c, c\rangle \\ & = -\lambda + \mu + (\lambda - \mu) \langle b, c \rangle + \langle a, b \rangle - \langle a, c \rangle \end{align} $$

By construction, point $D$ is the harmonic conjugate of $A'$ with respect to $BC$ (easy to show using the Ceva/Menelaus theorems, for example). Since $AB \ne AC$ it follows that $\lambda \ne \mu$ and $D$ is a proper point (not at infinity) given by:

$$ d = \frac{\lambda b - \mu c}{\lambda - \mu} $$

Finally, the midpoint $M$ of $BC$ is:

$$ m = \frac{b + c}{2} $$

Now, it remains to be shown that $\langle h - d, m - a \rangle = 0$ in order to complete the proof.

$$ \begin{align} \begin{split} 2 (\lambda - \mu) \langle h - d, m - a \rangle =& \langle (\lambda - \mu) a - \mu b + \lambda c, b + c - 2 a \rangle \\ =& (\lambda - \mu) \langle a, b \rangle + (\lambda - \mu) \langle a, c \rangle - 2 (\lambda - \mu) {\lvert a \rvert}^2 \\ &{} - \mu {\lvert b \rvert}^2 - \mu \langle b, c \rangle + 2 \mu \langle a, b \rangle \\ &{} + \lambda \langle b, c \rangle + \lambda {\lvert c \rvert}^2 - 2 \lambda \langle a, c \rangle \\ =& -\lambda + \mu + (\lambda + \mu) \langle a, b \rangle - (\lambda + \mu) \langle a, c \rangle + (\lambda - \mu) \langle b, c \rangle \\ =& -\lambda + \mu + (\lambda - \mu) \langle b, c \rangle + \langle a, b \rangle - \langle a, c \rangle \\ \end{split} \end{align} $$

The last line matches the expression for $0 = \langle h - a', b - c \rangle$ calculated earlier. It follows that $\langle h - d, m - a \rangle = 0$, thus $DH \perp AM$.


[ EDIT ]  The following elaborates on a couple of steps in the proof, as asked in a comment.

By construction, point $D$ is the harmonic conjugate of $A'$ with respect to $BC$

This is in fact the canonical construction for the projective harmonic conjugate point, which can be proven as follows. Since cevians $AA'$, $BB'$, $CC'$ are concurrent, it follows from Ceva's theorem that (using signed segments):

$$ \frac{BA'}{A'C} \cdot \frac{CB'}{B'A} \cdot \frac{AC'}{C'B} = 1 $$

Since points $B'$, $C'$, $D$ are collinear, it follows from Menelaus' theorem that (again using signed segments):

$$ \frac{BD}{DC} \cdot \frac{CB'}{B'A} \cdot \frac{AC'}{C'B} = -1 $$

Since the last 2 fractions are identical between the 2 products, it follows that:

$$ \frac{BD}{DC} = - \frac{BA'}{A'C} $$

therefore $D$ is by definition the harmonic conjugate of $A'$ with respect to $BC$.

$D$ is a proper point (not at infinity) given by $d = \frac{\lambda b - \mu c}{\lambda - \mu}$

Let $Z$ be an arbitrary point on the line through $BC$ written as a linear combination $z = (\lambda b + \mu c) / (\lambda + \mu)$ with real $\lambda$, $\mu$. Then the signed segments ratio:

$$ \frac{BZ}{ZC} = \frac {z - b}{c - z} = \frac{\lambda b + \mu c - \lambda b - \mu b}{\lambda c + \mu c - \lambda b - \mu c} = \frac{\mu (c - b)}{\lambda (c - b)} = \frac{\mu}{\lambda} $$

Therefore to obtain the harmonic conjugate $Z'$ for which $\frac{BZ'}{Z'C} = - \frac{BZ}{ZC}$ it is enough to replace $\mu$ with $-\mu$ which gives $z' = \frac{\lambda b - \mu c}{\lambda - \mu}$.

$\endgroup$
2
  • $\begingroup$ Sorry for the late response, nice solution though! I have a question about the harmonic conjugate property you used, how do you prove D is harmonic conjugate to A with respect to BC, and why is $d=\frac{\lambda b-\mu c}{\lambda-\mu}$ (I am not familiar with projective geometry). Thanks $\endgroup$ Commented Jan 19, 2016 at 14:53
  • 1
    $\begingroup$ Edited the answer to address both questions. Also, to clarify one other point, the entire proof could have been carried out by explicitly calculating $\lambda$ and $\mu$ (and $d$ from there) but it would have been a more tedious calculation. Keeping $\lambda$ and $\mu$ as such exploited the "symmetry" in the harmonic conjugation more directly. $\endgroup$ Commented Jan 20, 2016 at 1:57

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.