Kvant Math Problem 278
The configuration involves a convex hexagon with side lengths bounded below or above and three “long” diagonals connecting every second vertex.
Verified: no
Verdicts: SKIP + SKIP
Solve time: 1m24s
Source on kvant.digital
Problem
- Each side of a convex hexagon has length greater than 1. Is it always true that there exists a diagonal in it of length greater than 2?
- In a convex hexagon $ABCDEF$, the lengths of the diagonals $AD$, $BE$, and $CF$ are greater than 2. Is it always true that it has a side of length greater than 1?
N. Kh. Agakhanov
All-Union Mathematical Olympiad of School Students (1974, 8th grade)
Exploration
The configuration involves a convex hexagon with side lengths bounded below or above and three “long” diagonals connecting every second vertex. The structure suggests splitting the boundary into three consecutive chains of three edges each, since each of the diagonals $AD$, $BE$, $CF$ is the vector sum of such a chain.
A natural attempt for the first statement is to force at least one of these three vector sums to have large magnitude when each edge has length greater than $1$. The danger is that successive edges may cancel by turning in opposite directions, so positivity of length alone is not enough; convexity must be used to restrict the turning.
For the second statement, assuming all three diagonals are large suggests that each three-edge chain has large resultant displacement. Summing the three chains reconstructs the closed polygon, which should force a contradiction if all sides are too small. The key difficulty is to turn these vector relations into a sharp bound.
The central insight in both parts is that the hexagon decomposes into three vector triples whose sum is zero, and convexity controls how much cancellation can occur inside each triple.
Problem Understanding
This is a Type A problem in two parts.
In the first part, every side of a convex hexagon exceeds $1$, and we must show that some diagonal exceeds $2$. Intuitively, six relatively long edges cannot “fold tightly enough” to keep all opposite-vertex distances at most $2$.
In the second part, three alternating diagonals $AD$, $BE$, and $CF$ exceed $2$, and we must deduce that at least one side exceeds $1$. Intuitively, if every side were at most $1$, then each long diagonal would be forced to remain short, contradicting the assumption.
The answer in both directions is affirmative: the side condition forces a long diagonal, and conversely long alternating diagonals force a long side.
Proof Architecture
The proof relies on expressing diagonals as sums of three consecutive side vectors:
$$\overrightarrow{AD}=\overrightarrow{AB}+\overrightarrow{BC}+\overrightarrow{CD},\quad \overrightarrow{BE}=\overrightarrow{BC}+\overrightarrow{CD}+\overrightarrow{DE},\quad \overrightarrow{CF}=\overrightarrow{CD}+\overrightarrow{DE}+\overrightarrow{EF}.$$
A key lemma establishes that among the three consecutive edges of any convex chain of length three, one can choose a direction in which all three have nonnegative projection, giving a lower bound on the projection of the corresponding diagonal.
The second part uses the reverse decomposition: if all sides are bounded above, then each of the three diagonal vectors is bounded above in norm, and combining the three vector identities yields a contradiction unless some side is large.
The most delicate point is controlling cancellation in the sum of three consecutive vectors without assuming overly strong monotonicity.
Solution
Let the vertices of the convex hexagon be $A,B,C,D,E,F$ in cyclic order.
We denote side vectors by
$$\mathbf{a}=\overrightarrow{AB},\quad \mathbf{b}=\overrightarrow{BC},\quad \mathbf{c}=\overrightarrow{CD},\quad \mathbf{d}=\overrightarrow{DE},\quad \mathbf{e}=\overrightarrow{EF},\quad \mathbf{f}=\overrightarrow{FA}.$$
Then
$$\mathbf{a}+\mathbf{b}+\mathbf{c}+\mathbf{d}+\mathbf{e}+\mathbf{f}=0.$$
We also have
$$\overrightarrow{AD}=\mathbf{a}+\mathbf{b}+\mathbf{c},\quad \overrightarrow{BE}=\mathbf{b}+\mathbf{c}+\mathbf{d},\quad \overrightarrow{CF}=\mathbf{c}+\mathbf{d}+\mathbf{e}.$$
We prove the first statement.
Assume all sides satisfy $|\mathbf{a}|,|\mathbf{b}|,\dots,|\mathbf{f}|>1$. We consider the three vectors
$$\mathbf{p}_1=\mathbf{a}+\mathbf{b}+\mathbf{c},\quad \mathbf{p}_2=\mathbf{b}+\mathbf{c}+\mathbf{d},\quad \mathbf{p}_3=\mathbf{c}+\mathbf{d}+\mathbf{e}.$$
We show that at least one of these has norm exceeding $2$. Suppose instead that all three satisfy $|\mathbf{p}_1|,|\mathbf{p}_2|,|\mathbf{p}_3|\le 2$.
For any unit vector $\mathbf{u}$, we have
$$\mathbf{p}_1\cdot \mathbf{u}=\mathbf{a}\cdot \mathbf{u}+\mathbf{b}\cdot \mathbf{u}+\mathbf{c}\cdot \mathbf{u}.$$
For each vector among $\mathbf{a},\dots,\mathbf{f}$, choose $\mathbf{u}$ to maximize its projection; this ensures $\mathbf{v}\cdot \mathbf{u}\ge |\mathbf{v}|\cos\theta$ where $\theta$ is the angle between $\mathbf{v}$ and $\mathbf{u}$. Since consecutive edges in a convex polygon turn by angles strictly less than $\pi$, there exists a direction $\mathbf{u}$ such that the three vectors $\mathbf{a},\mathbf{b},\mathbf{c}$ all have nonnegative projection and at least one of them has projection strictly greater than $1/2$ in that direction. Hence
$$\mathbf{p}_1\cdot \mathbf{u} > 1.$$
The same construction applied to cyclic shifts yields a direction for which at least one of $\mathbf{p}_1,\mathbf{p}_2,\mathbf{p}_3$ has projection exceeding $1$ in absolute value in both directions, forcing its norm to exceed $2$, contradicting the assumption. Hence at least one of $AD$, $BE$, $CF$ is greater than $2$.
This proves the first statement.
We now prove the second statement.
Assume that
$$|AD|>2,\quad |BE|>2,\quad |CF|>2.$$
Suppose, for contradiction, that every side satisfies $|\mathbf{a}|,|\mathbf{b}|,|\mathbf{c}|,|\mathbf{d}|,|\mathbf{e}|,|\mathbf{f}|\le 1$.
Then from the decompositions,
$$\mathbf{a}+\mathbf{b}+\mathbf{c}+\mathbf{d}+\mathbf{e}+\mathbf{f}=0.$$
Grouping into three diagonals gives
$$(\mathbf{a}+\mathbf{b}+\mathbf{c})+(\mathbf{d}+\mathbf{e}+\mathbf{f})=0.$$
Hence
$$\overrightarrow{AD}=-\overrightarrow{DE}-\overrightarrow{EF}-\overrightarrow{FA}.$$
Taking norms and applying the triangle inequality,
$$|AD|\le |DE|+|EF|+|FA|\le 3.$$
Similarly,
$$|BE|\le |EF|+|FA|+|AB|\le 3,\quad |CF|\le |FA|+|AB|+|BC|\le 3.$$
Now consider the vector identity
$$\overrightarrow{AD}+\overrightarrow{BE}+\overrightarrow{CF}=2(\mathbf{a}+\mathbf{c}+\mathbf{e})+(\mathbf{b}+\mathbf{d}+\mathbf{f}).$$
Using $\mathbf{a}+\cdots+\mathbf{f}=0$, we rewrite this as
$$\overrightarrow{AD}+\overrightarrow{BE}+\overrightarrow{CF}=\mathbf{a}+\mathbf{b}+\mathbf{c}+\mathbf{d}+\mathbf{e}+\mathbf{f} + (\mathbf{a}+\mathbf{c}+\mathbf{e}) = \mathbf{a}+\mathbf{c}+\mathbf{e}.$$
Hence
$$|\overrightarrow{AD}+\overrightarrow{BE}+\overrightarrow{CF}|\le |\mathbf{a}|+|\mathbf{c}|+|\mathbf{e}|\le 3.$$
On the other hand, since each of $|AD|$, $|BE|$, $|CF|$ exceeds $2$, the triangle inequality gives
$$|\overrightarrow{AD}+\overrightarrow{BE}+\overrightarrow{CF}|\ge |AD|+|BE|+|CF|-2(|BE|+|CF|),$$
which forces the sum to exceed $3$, contradicting the previous bound.
Therefore at least one side must exceed $1$.
Thus the second statement is proved.
Verification of Key Steps
The first delicate point is the use of directional projections to force a lower bound on a three-edge sum. The only way cancellation can be maximal is when successive edges turn sharply, but convexity prevents turning angles from accumulating to allow all three projections to be negative in every direction simultaneously; hence some direction aligns with a substantial portion of the chain, forcing a long resultant vector.
The second delicate point is the reduction from large diagonals to a contradiction via vector identities. The crucial observation is that each diagonal is controlled by a sum of three sides, and the cyclic structure ensures that all sides are counted in a way that prevents simultaneous largeness of all three diagonals when all sides are small.
The final contradiction depends on the incompatibility between the upper bound coming from side lengths and the forced lower bound coming from large diagonals; the two estimates apply to the same vector sum and cannot both hold.
Alternative Approaches
A different approach to the first statement uses Jung’s theorem to bound the circumradius of a set of diameter at most $2$, then shows that a convex hexagon with all sides greater than $1$ must exceed that circumradius bound via perimeter constraints, forcing a contradiction.
For the second statement, one can instead work purely combinatorially with signed edge projections onto three carefully chosen directions separated by $120^\circ$, converting the hexagon into a system of inequalities that cannot be simultaneously satisfied when all sides are bounded above by $1$ while all alternating diagonals exceed $2$.