Kvant Math Problem 12
A straight line intersects a convex quadrilateral in two points.
Verified: no
Verdicts: UNKNOWN + UNKNOWN
Solve time: 2m14s
Source on kvant.digital
Problem
Which quadrilaterals can be cut by a straight line into two quadrilaterals that are similar to each other?
Exploration
A straight line intersects a convex quadrilateral in two points. For both resulting pieces to be quadrilaterals, the line must connect two opposite sides; otherwise one of the pieces is a triangle and the other is a pentagon. Hence, up to relabeling, the line meets $AB$ at $P$ and $CD$ at $Q$, producing quadrilaterals $A P Q D$ and $P B C Q$.
Similarity between these two quadrilaterals forces a correspondence between their sides. The segment $PQ$ is a side of both figures, so in any similarity, $PQ$ must correspond to a side of equal role in the other quadrilateral. Since it is the only side shared by both figures, the similarity must preserve it as a side-to-side match, forcing the similarity ratio to be $1$, so the two quadrilaterals are congruent.
This reduces the problem to classifying quadrilaterals that can be cut by a line into two congruent quadrilaterals. A natural candidate is a parallelogram, since it has a half-turn symmetry about the intersection point of its diagonals, which suggests that any line through this center should split it into two congruent figures.
The key uncertainty is whether any non-parallelogram quadrilateral could admit such a symmetric cut.
Problem Understanding
This is a Type A problem. We must determine all quadrilaterals that admit a straight-line cut producing two similar quadrilaterals.
The structural constraint is that the cut produces two quadrilaterals sharing a side, which strongly restricts the similarity correspondence and forces rigidity.
The expected answer is that the quadrilateral must be a parallelogram, since central symmetry ensures that any such cut produces congruent opposite parts.
Thus the claim is that the only such quadrilaterals are parallelograms.
Proof Architecture
First, a lemma establishes that a line producing two quadrilaterals must connect opposite sides of the original quadrilateral.
Second, a lemma shows that similarity of the two resulting quadrilaterals forces their similarity ratio to be $1$, hence they are congruent, because the shared side $PQ$ must correspond to itself under any similarity.
Third, a lemma identifies that congruence of the two parts implies a half-turn symmetry of the original quadrilateral about the midpoint of $PQ$, mapping one piece to the other.
Fourth, a lemma shows that a convex quadrilateral possessing such a half-turn symmetry must be a parallelogram.
The most delicate step is the justification that the similarity necessarily fixes the shared side $PQ$ in a way that forces scale factor $1$.
Solution
Let $ABCD$ be a convex quadrilateral, and let a line intersect $AB$ at $P$ and $CD$ at $Q$, producing quadrilaterals $A P Q D$ and $P B C Q$.
The first observation is that a line cutting a convex quadrilateral into two quadrilaterals must intersect two opposite sides. Indeed, if both intersection points lie on adjacent sides, then one of the resulting regions has three boundary edges from the original quadrilateral and one from the cutting line, producing a triangle, while the other region has five sides, contradicting the assumption that both are quadrilaterals.
Assume that $A P Q D$ and $P B C Q$ are similar quadrilaterals. Consider any similarity transformation mapping $A P Q D$ onto $P B C Q$. The segment $PQ$ is a side of both quadrilaterals, so its image under the similarity must be a side of $P B C Q$ of the same role in the cyclic order. Since $PQ$ is the only side of $A P Q D$ whose endpoints both lie on the cutting line, its image must coincide with the side $PQ$ of the second quadrilateral.
Therefore the similarity maps segment $PQ$ onto itself, preserving its endpoints as a set. A similarity mapping a segment onto itself must have scale factor $1$, since distances between points on the segment are preserved up to a constant factor and the endpoints are fixed as a set. Hence the two quadrilaterals are congruent.
Let $M$ be the midpoint of segment $PQ$. The congruence between $A P Q D$ and $P B C Q$ implies the existence of a rigid motion sending one quadrilateral to the other. Since both figures share the segment $PQ$ and this segment is mapped onto itself with endpoints swapped or fixed, the only such rigid motion is either the identity or a half-turn about $M$. The identity is impossible because the two quadrilaterals occupy disjoint interiors, so the motion must be the half-turn about $M$.
This half-turn sends $A$ to $B$ and $D$ to $C$ (or vice versa, depending on labeling). Consequently, $M$ is the midpoint of both diagonals $AC$ and $BD$, since a half-turn maps each point to its symmetric image through $M$.
A quadrilateral whose diagonals bisect each other is a parallelogram. Indeed, from $M$ being the midpoint of $AC$ and $BD$, triangles $AMB$ and $CMD$ are centrally symmetric, implying $AB \parallel CD$ and $BC \parallel AD$.
Thus $ABCD$ is a parallelogram.
Conversely, let $ABCD$ be a parallelogram, and let a line through the center of symmetry (the intersection point of the diagonals) intersect $AB$ and $CD$. A half-turn about this center maps the parallelogram to itself and interchanges the two regions determined by the line, so the two resulting quadrilaterals are congruent, hence similar.
Therefore, the required quadrilaterals are exactly parallelograms.
$$\boxed{\text{All parallelograms}}$$
∎
Verification of Key Steps
The first delicate point is the assertion that similarity forces the segment $PQ$ to correspond to itself. In each of the two quadrilaterals, $PQ$ is the unique side lying entirely on the cutting line. Any similarity between the two figures must preserve incidence relations between sides and vertices, so a side contained in the cutting line must map to a side with the same geometric property, which again lies on the cutting line. Since each quadrilateral has exactly one such side, the correspondence is forced.
The second point is the deduction that the similarity ratio is $1$. Once $PQ$ maps to itself as a segment, its length must be preserved, forcing the scaling factor to equal $1$.
The third point is the characterization of the rigid motion. A congruence between two planar figures sharing a segment that is invariant as a set must either fix the segment pointwise or swap its endpoints. In both cases, the transformation is a reflection across the perpendicular bisector of $PQ$ or a half-turn about its midpoint. Only the half-turn is compatible with exchanging the two disjoint interiors of the quadrilaterals.
The final point is that a quadrilateral with a half-turn symmetry about the midpoint of both diagonals must have its diagonals bisect each other, which is equivalent to being a parallelogram.
Alternative Approaches
One alternative approach is purely angle-based. Tracking the correspondence of angles around the shared side $PQ$ under similarity yields linear relations between adjacent angles of the original quadrilateral, eventually forcing opposite sides to be parallel.
Another approach begins with affine geometry: any similarity between two regions sharing a full side extends to an affine symmetry of the supporting quadrilateral, and the only convex quadrilaterals admitting a nontrivial affine involution swapping opposite sides are parallelograms.