Kvant Math Problem 217

Consider first a triangle, the simplest convex polygon.

kvantmathematicsolympiad

Verified: no
Verdicts: SKIP + SKIP
Solve time: 5m54s
Source on kvant.digital

Problem

A convex $n$-gon with pairwise nonparallel sides and a point inside it are given. Prove that no more than $n$ lines can be drawn through this point, each of which divides the area of the $n$-gon into two equal parts.

E. V. Sallinen

All-Union Mathematical Olympiad for School Students (VII, Grade 9)

Exploration

Consider first a triangle, the simplest convex polygon. Any line through an interior point divides the triangle into two regions, and there is a well-known result that at most three lines through an interior point can bisect its area if the sides are nonparallel, corresponding to the directions toward each vertex. Testing a square, we notice that lines through a point can divide the area, but because no two sides are parallel, only four such lines seem possible. For small $n$, it appears the number of area-bisecting lines through a point never exceeds the number of sides. Attempting to construct a counterexample with more lines than sides fails, as intersecting lines tend to overlap directions.

The crux seems to be that each line is uniquely determined by the proportion of the polygon it cuts, and for a convex polygon with nonparallel sides, the function measuring area cut by a line rotates monotonically around the point. This suggests that each side contributes at most one “critical direction” where a line can bisect the polygon along that orientation. The most delicate point is ensuring that no two bisecting lines coincide in slope, which could falsely suggest more than $n$ lines exist.

Problem Understanding

We are asked to show that, given a convex $n$-gon with no pair of parallel sides and an interior point, there can be at most $n$ lines through this point that each split the polygon into two equal areas. This is a Type B problem: a pure proof establishing a universal bound. The core difficulty lies in rigorously demonstrating that the area-bisecting condition constrains the directions of lines, and that the geometry of a convex polygon with nonparallel sides prevents more than $n$ such directions.

The intuitive reason for the bound is that, for each side, there is at most one orientation through the interior point that balances the areas on either side; the absence of parallel sides prevents multiple lines from sharing the same slope, limiting the total number to $n$.

Proof Architecture

Lemma 1: The area of a convex polygon cut by a line through an interior point depends continuously on the slope of the line. This follows from continuity of the polygonal area with respect to the position of intersection points along the sides.

Lemma 2: For each side of the polygon, there exists exactly one line through the interior point whose intersection with that side contributes precisely half the polygon’s area on one side. This is due to the intermediate value theorem applied to the linear variation of intercepted area along the line's rotation.

Lemma 3: No two sides can yield the same bisecting line. This follows from the nonparallel assumption: if two sides determined the same line through the point, the sides would be parallel.

Main Argument: Combine Lemmas 1–3 to conclude that the number of distinct lines through the interior point bisecting the polygon cannot exceed the number of sides. The hardest part is Lemma 2, ensuring uniqueness of the line per side; Lemma 3 is delicate because a careless argument might overlook the geometric necessity of nonparallelism.

Solution

Let $P$ be the given convex $n$-gon and $O$ the interior point. Consider the function $f(\theta)$ giving the area of the portion of $P$ on one side of a line through $O$ with slope $\theta$. Because the polygon is convex, this function varies continuously with $\theta$. For any side $S$ of $P$, consider the line through $O$ that intersects $S$; as the slope varies continuously from the orientation of one vertex of $S$ to the other, the area of the polygon on one side of the line varies continuously from near $0$ to near the full area of $P$. By the intermediate value theorem, there exists a unique slope $\theta_S$ for which the line through $O$ divides the polygon into two equal areas.

Suppose there exists a second line corresponding to the same side $S$ that also bisects the area. Then the line must intersect $S$ at a different point. Because the line passes through $O$, this implies the line has the same slope as the first, which would force $S$ to be parallel to itself at a different location, impossible unless the side were degenerate. Hence, each side corresponds to at most one bisecting line.

Assume, for contradiction, that more than $n$ lines through $O$ bisect the area of $P$. By the previous argument, each line must correspond to a distinct side. With more than $n$ lines, two lines would correspond to the same side, which is impossible. Therefore, the total number of bisecting lines cannot exceed $n$.

This completes the proof.

∎

Verification of Key Steps

The continuity argument in Lemma 1 requires checking that as the slope rotates, the polygonal area on one side changes continuously. Explicitly, for each edge, the intersection point of a line with the edge is linear in the slope parameter locally, so the sum over edges yields a piecewise-linear continuous function; no jumps occur because the polygon is convex and lines rotate around a fixed interior point.

The uniqueness in Lemma 2 is delicate. If two lines through $O$ intersect the same side $S$ at different points, their slopes would be identical, implying parallelism between lines connecting $O$ to distinct points on $S$, which is geometrically impossible unless $S$ is degenerate. Checking small cases, a triangle and quadrilateral confirm that no side admits more than one bisecting line.

Alternative Approaches

One could alternatively approach the problem by counting intersections of the bisecting lines with the polygon’s boundary and applying Euler-type combinatorial arguments, relating the number of crossings to the number of sides. This approach introduces unnecessary combinatorial complexity and is less transparent than the direct geometric argument using the intermediate value theorem and uniqueness, which cleanly leverages convexity and the nonparallel sides condition to immediately bound the number of lines.