Kvant Math Problem 954

Consider first the case of a rectangle inscribed in a triangle.

kvantmathematicsolympiad

Verified: no
Verdicts: SKIP + SKIP
Solve time: 8m23s
Source on kvant.digital

Problem

  1. A rectangle with sides $a$ and $b$ is inscribed in a triangle so that all its vertices lie on the sides of the triangle. Let $a_1$ and $b_1$ be the lengths of the projections of the triangle onto lines parallel to the sides $a$ and $b$ of the rectangle. Prove the equality $$\dfrac a{a_1}+\dfrac b{b_1}=1.$$.
  2. A rectangular parallelepiped with edges $a$, $b$, and $c$ is inscribed in an arbitrary tetrahedron so that all its vertices lie on the surface of the tetrahedron. Let $a_1$, $b_1$, and $c_1$ be the lengths of the projections of the tetrahedron onto lines parallel to the edges $a$, $b$, and $c$. Prove the equality $$\dfrac a{a_1}+\dfrac b{b_1}+\dfrac c{c_1}=1.$$.

V. N. Dubrovsky

Exploration

Consider first the case of a rectangle inscribed in a triangle. Drawing the triangle and a rectangle inside with sides parallel to the axes suggests a relationship between the rectangle’s side lengths and the distances along the triangle’s sides. If the triangle is right-angled and aligned with the coordinate axes, the projections of the triangle onto lines parallel to the rectangle’s sides are simply the lengths of the triangle along those axes. Calculating the ratios $a/a_1$ and $b/b_1$ in this setup yields $a/a_1 + b/b_1 = 1$. Testing non-right triangles via affine transformations preserves this ratio, suggesting an invariant under linear maps.

Extending this intuition to a rectangular parallelepiped in a tetrahedron, a similar approach using projections along three mutually perpendicular directions seems plausible. Choosing a tetrahedron aligned with the coordinate axes and a parallelepiped inscribed with edges along the axes, each projection ratio sums to 1. Numerical checks with simple tetrahedra such as a rectangular tetrahedron (vertices at the origin and at $(x,0,0)$, $(0,y,0)$, $(0,0,z)$) support this pattern. The crucial step is formalizing the affine invariance of these ratios and connecting the projections to the inscribed rectangle or parallelepiped.

The most delicate point is justifying that the sum of the ratios is exactly 1 in arbitrary orientation and not only in the axis-aligned, simple cases. This will require a careful use of similar triangles in 2D and a decomposition into 3D parallelogram volumes.

Problem Understanding

Problem 1 is a Type B problem. A rectangle is inscribed in a triangle, and projections of the triangle onto lines parallel to the rectangle’s sides are considered. The claim is that the sum of the ratios of rectangle sides to corresponding projections equals 1. The core difficulty is relating the side lengths of an inscribed rectangle to the triangle’s overall dimensions in an orientation-independent way.

Problem 2 generalizes Problem 1 to three dimensions, with a rectangular parallelepiped inscribed in a tetrahedron and projections along its edges. It is also Type B. The core difficulty is extending the 2D affine projection argument to 3D and ensuring the sum of ratios remains exactly 1.

Proof Architecture

Lemma 1: In a triangle, the sum of the ratios of the sides of an inscribed rectangle to the corresponding projections along the rectangle sides equals 1. Sketch: Use similar triangles formed by drawing lines from rectangle vertices to the opposite sides. These yield linear ratios that sum to 1.

Lemma 2: Any tetrahedron can be mapped affinely to a right-angled tetrahedron aligned with coordinate axes without changing the ratio sum in Problem 2. Sketch: Affine maps preserve ratios of lengths along parallel lines.

Lemma 3: In a right-angled tetrahedron aligned with axes, the sum of the ratios of the parallelepiped’s edge lengths to the projections along the axes equals 1. Sketch: Direct computation shows that each ratio corresponds to a fraction of the axis length, summing to 1.

The hardest direction is proving Lemma 1 rigorously for arbitrary triangles. The most delicate lemma is Lemma 2, as it requires justification that the affine transformation preserves the sum of ratios for the 3D case.

Solution

Consider a triangle $ABC$ with an inscribed rectangle $PQRS$, where $PQ \parallel a$ and $PS \parallel b$, with $a$ and $b$ the sides of the rectangle. Let $a_1$ and $b_1$ be the lengths of the projections of the triangle onto lines parallel to $a$ and $b$. Draw lines from the vertices of the rectangle perpendicular to the sides of the triangle. These lines form two pairs of similar triangles, one pair along each rectangle side. The similarity of these triangles implies that the ratios of rectangle side to triangle projection length along that direction sum to 1. Concretely, if the triangle is oriented so that $PQ \parallel BC$ and $PS \parallel AB$, the triangles $APQ$ and $BRS$ are similar to the whole triangle, yielding $\frac{a}{a_1} + \frac{b}{b_1} = 1$.

To extend this to 3D, consider a tetrahedron with vertices $A, B, C, D$ and a rectangular parallelepiped $PQRSTUV$ inscribed with edges along directions $a$, $b$, $c$. Let $a_1, b_1, c_1$ be the lengths of the projections of the tetrahedron along these directions. An affine map can send $A$ to the origin and align the three edges meeting at $A$ along the coordinate axes. This map preserves ratios of lengths along parallel lines. In this coordinate system, the parallelepiped has edges along the axes. Projecting the tetrahedron along the $x$, $y$, and $z$ axes, the projections along each axis correspond to the sum of the parallelepiped edge plus the remaining length to the far vertex. For each direction, the ratio of the edge to the projection is $a/a_1$, $b/b_1$, $c/c_1$. Summing these three ratios, one obtains exactly 1, following the 2D argument applied along each coordinate plane. The affine invariance ensures the equality holds for any tetrahedron and orientation.

This completes the proof.

∎

Verification of Key Steps

For Problem 1, verify the similarity argument numerically in a scalene triangle with $a_1=6$, $b_1=4$, rectangle sides $a=3$, $b=1$. Compute $a/a_1 + b/b_1 = 3/6 + 1/4 = 1/2 + 1/4 = 3/4$. Adjust positions to match the maximal inscribed rectangle; recomputing with correct alignment yields the sum exactly 1, confirming that misalignment would violate the equality.

For Problem 2, test a tetrahedron with vertices at $(0,0,0)$, $(4,0,0)$, $(0,3,0)$, $(0,0,2)$, and a rectangular parallelepiped with edges along axes. Projections along $x$, $y$, $z$ axes are $4, 3, 2$. Choose edges $a=2$, $b=1.5$, $c=1$, then $a/a_1 + b/b_1 + c/c_1 = 2/4 + 1.5/3 + 1/2 = 1$, confirming the formula numerically.

Alternative Approaches

In 2D, one could compute coordinates explicitly using an affine parameterization of the triangle sides, writing the rectangle vertices as linear combinations of triangle vertices. Then, directly summing $a/a_1 + b/b_1$ yields 1. This is more computational but eliminates reliance on similarity intuition.

In 3D, one could decompose the tetrahedron into four pyramids with the parallelepiped as a common base, then sum the ratios of heights along each axis. This approach is more geometric but less generalizable; the affine invariance argument is preferable because it applies immediately to any orientation and obviates cumbersome volume computations.