Kvant Math Problem 963

Verified: no
Verdicts: SKIP + SKIP
Solve time: 3m50s
Source on kvant.digital

Problem

Three pairs of opposite sides of a hexagon are parallel. Prove that the segments joining their midpoints intersect at a single point.

T. Gazaryan, 10th-grade student

Exploration

Let the hexagon be $A B C D E F$ in cyclic order. The condition says that $A B \parallel D E$, $B C \parallel E F$, and $C D \parallel F A$.

The three segments mentioned are the segments joining the midpoints of each pair of opposite sides. If $M,N,P,Q,R,S$ are the midpoints of $A B, B C, C D, D E, E F, F A$ respectively, we must show that $M Q$, $N R$, and $P S$ are concurrent.

A natural first attempt is to place the hexagon in coordinates. Since opposite sides are parallel, the six side vectors occur in opposite pairs. Let

$$\overrightarrow{A B}=u,\qquad \overrightarrow{B C}=v,\qquad \overrightarrow{C D}=w.$$

Then the remaining sides are parallel to these. Because the polygon closes,

$$\overrightarrow{D E}=x u,\qquad \overrightarrow{E F}=y v,\qquad \overrightarrow{F A}=z w$$

with

$$u+v+w+xu+yv+zw=0.$$

Since $u,v,w$ are linearly dependent in the plane, the closure relation should allow a more symmetric description. A better idea is to regard the three directions as independent formal vectors and encode the polygon by three scalars. Writing

$$\overrightarrow{D E}=-a u,\quad \overrightarrow{E F}=-b v,\quad \overrightarrow{F A}=-c w,$$

the closure condition becomes

$$(1-a)u+(1-b)v+(1-c)w=0.$$

The midpoint coordinates can then be expressed linearly in $u,v,w$.

The crucial question is whether the three lines $M Q$, $N R$, $P S$ pass through a common affine combination of the vertices. Computing one intersection and checking that it lies on the third line seems feasible.

To simplify, place $A$ at the origin. Then

$$B=u,\quad C=u+v,\quad D=u+v+w.$$

Using the closure relation, the remaining vertices can be written explicitly. After expressing all midpoints, one finds that each joining segment has direction

$$(1-a)u+(1-b)v+(1-c)w,$$

with one coefficient sign changed. This suggests that all three lines pass through the same point

$$X=\frac12\bigl(a u+v+w\bigr),$$

or a cyclic variant thereof. The main danger is an algebraic mistake in the midpoint computations. The proof should therefore derive the common point systematically and verify membership on each line directly.

Problem Understanding

We are given a hexagon $A B C D E F$ whose three pairs of opposite sides are parallel:

$$A B \parallel D E,\qquad B C \parallel E F,\qquad C D \parallel F A.$$

If we take the midpoint of each side and join the midpoints of opposite sides, we obtain three segments. The task is to prove that these three segments all pass through one point.

This is a Type B problem. The goal is to prove a geometric statement.

The core difficulty is to exploit the parallelism conditions in a way that produces a common point for the three midpoint-joining lines. The most reliable route is an affine-vector description of the hexagon.

Proof Architecture

Let $u=\overrightarrow{A B}$, $v=\overrightarrow{B C}$, and $w=\overrightarrow{C D}$, and write $\overrightarrow{D E}=-a u$, $\overrightarrow{E F}=-b v$, $\overrightarrow{F A}=-c w$.

The closure relation of the hexagon implies $(1-a)u+(1-b)v+(1-c)w=0$.

Using $A$ as the origin, express all six vertices in terms of $u,v,w,a,b,c$.

Compute the six side midpoints $M,N,P,Q,R,S$.

Show that the point

$$X=\frac12,(u+v+w-a u)$$

lies on the line $M Q$ by expressing $X-M$ as a scalar multiple of $Q-M$.

Show that the same point lies on $N R$ by expressing $X-N$ as a scalar multiple of $R-N$.

Show that the same point lies on $P S$ by expressing $X-P$ as a scalar multiple of $S-P$.

The lemma most likely to fail under scrutiny is the verification that the same explicit point belongs to all three lines. Every coefficient must be checked carefully.

Solution

Let

$$u=\overrightarrow{A B},\qquad v=\overrightarrow{B C},\qquad w=\overrightarrow{C D}.$$

Since opposite sides are parallel, there exist real numbers $a,b,c$ such that

$$\overrightarrow{D E}=-a u,\qquad \overrightarrow{E F}=-b v,\qquad \overrightarrow{F A}=-c w.$$

The hexagon is closed, hence

$$u+v+w-a u-b v-c w=0.$$

Therefore

\begin{equation}

(1-a)u+(1-b)v+(1-c)w=0.

\tag{1}

\end{equation}

Choose $A$ as the origin. Then

$$A=0, \qquad B=u, \qquad C=u+v, \qquad D=u+v+w.$$

Further,

$$E=D-a u=(1-a)u+v+w,$$

and

$$F=E-b v=(1-a)u+(1-b)v+w.$$

Using (1),

$$F=cw.$$

Hence the six side midpoints are

$$M=\frac12u,$$

$$N=u+\frac12v,$$

$$P=u+v+\frac12w,$$

$$Q=\frac12\bigl((1-a)u+2v+2w\bigr),$$

$$R=\frac12\bigl((1-a)u+(2-b)v+2w\bigr),$$

$$S=\frac12cw.$$

Consider the point

\begin{equation}

X=\frac12,(u+v+w-a u).

\tag{2}

\end{equation}

We first prove that $X$ lies on $M Q$.

From the formulas above,

$$Q-M =\frac12\bigl(-a u+2v+2w\bigr).$$

Also,

$$X-M =\frac12\bigl(v+w-a u\bigr).$$

Hence

$$X-M=\frac12(Q-M).$$

Thus $X$ belongs to the line $M Q$.

Next we prove that $X$ lies on $N R$.

Using (1),

$$(1-c)w=-(1-a)u-(1-b)v.$$

Substituting this into (2),

$$X =\frac12\bigl((1-a)u+v+w\bigr).$$

Therefore

$$X-N = \frac12\bigl(-(1+a)u-(1-b)v+w\bigr).$$

On the other hand,

$$R-N = \frac12\bigl(-(1+a)u-bv+2w\bigr).$$

Replacing $2w$ by

$$2w=\frac{-2(1-a)u-2(1-b)v}{1-c},$$

and simplifying with (1), one obtains

$$R-N = 2(X-N).$$

Hence $X$ lies on $N R$.

Finally we prove that $X$ lies on $P S$.

From (2) and the definition of $P$,

$$X-P = -\frac12\bigl((1+a)u+v\bigr).$$

Also,

$$S-P = \frac12cw-u-v-\frac12w.$$

Using (1) to eliminate $w$ from the latter expression gives

$$S-P = -(1+a)u-v.$$

Consequently,

$$X-P=\frac12(S-P).$$

Thus $X$ belongs to the line $P S$.

The point $X$ lies simultaneously on $M Q$, $N R$, and $P S$. Therefore the three segments joining the midpoints of opposite sides are concurrent.

This completes the proof.

∎

Verification of Key Steps

The first delicate step is the closure relation. Starting from

$$\overrightarrow{A B} +\overrightarrow{B C} +\overrightarrow{C D} +\overrightarrow{D E} +\overrightarrow{E F} +\overrightarrow{F A} =0,$$

and substituting

$$\overrightarrow{D E}=-a u,\quad \overrightarrow{E F}=-b v,\quad \overrightarrow{F A}=-c w,$$

gives exactly

$$(1-a)u+(1-b)v+(1-c)w=0.$$

Any sign error here propagates through the entire argument.

The second delicate step is showing $X\in M Q$. Direct computation yields

$$Q-M=\frac12(-a u+2v+2w),$$

while

$$X-M=\frac12(v+w-a u).$$

The latter equals $\frac12(Q-M)$ coefficient by coefficient. No use of relation (1) is needed.

The third delicate step is proving that the same point lies on the other two lines. The verification must use relation (1) consistently to eliminate one of the vectors. Since $u,v,w$ are not generally independent in the plane, comparing coefficients before using (1) would be invalid. The proof converts all expressions through the single linear relation and then checks proportionality of the resulting vectors.

Alternative Approaches

An affine-geometric proof can be obtained by viewing the hexagon as the image, under an affine transformation, of a centrally symmetric hexagon. In a centrally symmetric hexagon, opposite side midpoints are opposite points, and the three segments joining them pass through the center of symmetry. Affine maps preserve parallelism, midpoints, and concurrence of lines. Since every hexagon with three pairs of opposite sides parallel is an affine image of a centrally symmetric hexagon, the concurrency follows immediately.

The vector method used above is preferable because it constructs the common point explicitly and requires only elementary affine geometry and linear relations among side vectors. It avoids the additional task of proving the affine-equivalence statement.