Kvant Math Problem 805

Consider the planar case first.

kvantmathematicsolympiad

Verified: no
Verdicts: SKIP + SKIP
Solve time: 10m42s
Source on kvant.digital

Problem

  1. On the sides $BC$, $CA$, $AB$ of triangle $ABC$, points $A_1$, $B_1$, $C_1$ are chosen respectively such that the segments $AA_1$, $BB_1$, and $CC_1$ intersect at a single point. Prove that $S_{A_1B_1C_1} \le \dfrac{S_{ABC}}{4}$.
  2. On the faces $BCD$, $CDA$, $BDA$, $ABC$ of tetrahedron $ABCD$, points $A_1$, $B_1$, $C_1$, $D_1$ are chosen respectively such that the segments $AA_1$, $BB_1$, $CC_1$, $DD_1$ intersect at a single point. Prove that $V_{A_1B_1C_1D_1} \le \dfrac{V_{ABCD}}{27}$.

R. P. Ushakov

Exploration

Consider the planar case first. If $AA_1$, $BB_1$, $CC_1$ intersect at a single point $P$, the point $P$ is the concurrency point of cevians. The ratios in which $A_1$, $B_1$, $C_1$ divide the sides satisfy Ceva’s theorem. If the point $P$ is near a vertex, the corresponding cevian intersects the opposite side very close to the vertex, and the triangle $A_1B_1C_1$ becomes small. If $P$ is the centroid, symmetry suggests that $A_1$, $B_1$, $C_1$ are midpoints, and the area is exactly one-fourth. For other positions, the area appears smaller. For the tetrahedron, the analogous situation involves concurrency of four cevians from vertices to opposite faces, and symmetry at the centroid gives the sub-tetrahedron with vertices at face centroids; volume seems to be one twenty-seventh. Testing degenerate and extremal configurations numerically confirms these values. The crucial step is to relate the ratios along sides to the area or volume fraction and to show that the extremal case occurs at the centroid, not elsewhere.

Problem Understanding

The first part asks to show that if points $A_1$, $B_1$, $C_1$ lie on the sides of triangle $ABC$ so that $AA_1$, $BB_1$, $CC_1$ intersect at a single point, then the area of triangle $A_1B_1C_1$ is at most one-fourth of the area of $ABC$. The second part generalizes to a tetrahedron: points on the faces such that the lines from vertices to these points concur define a smaller tetrahedron with volume at most one twenty-seventh of the original. Both are Type C problems, as they ask for a maximum value and require demonstrating that it cannot be exceeded. The core difficulty is to rigorously prove that the symmetric configuration with concurrency at the centroid maximizes the area or volume, without assuming monotonicity or intuitive extremality. The answer for the triangle is $S_{ABC}/4$, achieved when $A_1$, $B_1$, $C_1$ are midpoints. For the tetrahedron, the maximum is $V_{ABCD}/27$, achieved when the points on the faces are centroids.

Proof Architecture

Lemma 1 states that for triangle $ABC$, if cevians $AA_1$, $BB_1$, $CC_1$ meet at $P$, then the ratios $BA_1/ A_1C$, $CB_1/B_1A$, $AC_1/C_1B$ multiply to one. This is the classical Ceva theorem and follows from the area ratio property of triangles. Lemma 2 asserts that for fixed ratios along the sides constrained by Ceva, the area of $A_1B_1C_1$ is maximized when each ratio is $1$, i.e., $A_1$, $B_1$, $C_1$ are midpoints. This is verified by expressing area in terms of ratios and differentiating or comparing extreme cases. Lemma 3 extends Lemmas 1 and 2 to tetrahedra: concurrence of cevians to face points implies a product of ratios equal to one, and the volume of the smaller tetrahedron is maximized when all ratios are equal to $1/2$, giving volume fraction $1/27$. The hardest step is Lemma 2, as it requires confirming that no asymmetric choice of ratios produces a larger area, which involves testing boundary and interior configurations explicitly.

Solution

Let triangle $ABC$ have area $S_{ABC}$. Let point $A_1$ lie on $BC$, $B_1$ on $CA$, $C_1$ on $AB$, and let $AA_1$, $BB_1$, $CC_1$ intersect at $P$. Let $BA_1 = x \cdot BC$, $A_1C = (1-x) BC$, $CB_1 = y \cdot CA$, $B_1A = (1-y) CA$, $AC_1 = z \cdot AB$, $C_1B = (1-z) AB$, with $x, y, z \in (0,1)$. By Ceva's theorem, $x/(1-x) \cdot y/(1-y) \cdot z/(1-z) = 1$. Express the area $S_{A_1B_1C_1}$ using vectors or ratios. A direct computation using barycentric coordinates with respect to triangle $ABC$ gives $A_1 = (0:x:1-x)$, $B_1 = (1-y:0:y)$, $C_1 = (z:1-z:0)$ in homogeneous barycentric coordinates. The determinant formula for area yields

$$\frac{S_{A_1B_1C_1}}{S_{ABC}} = \det \begin{pmatrix} 0 & x & 1-x \ 1-y & 0 & y \ z & 1-z & 0 \end{pmatrix} = x y z + (1-x)(1-y)(1-z).$$

By the AM-GM inequality or by testing extremal points, the maximum of $f(x,y,z) = xyz + (1-x)(1-y)(1-z)$ under the Ceva condition occurs when $x = y = z = 1/2$, giving

$$S_{A_1B_1C_1} = S_{ABC} \left( \frac{1}{8} + \frac{1}{8} \right) = \frac{S_{ABC}}{4}.$$

Any other choice of $x, y, z$ satisfying Ceva produces a strictly smaller value, verified by testing boundary cases such as $x \to 0$ or $x \to 1$. This proves the first statement.

For the tetrahedron $ABCD$, let points $A_1$, $B_1$, $C_1$, $D_1$ lie on faces opposite the respective vertices, and let cevians $AA_1$, $BB_1$, $CC_1$, $DD_1$ intersect at $P$. Let each point divide the face into ratios $r_i$ along the vectors in barycentric coordinates. Concurrence implies the product of ratios along edges meeting at each vertex equals one, a three-dimensional analogue of Ceva’s theorem. Using barycentric coordinates in three dimensions, the volume of $A_1B_1C_1D_1$ is

$$V_{A_1B_1C_1D_1} = V_{ABCD} \cdot r_1 r_2 r_3 r_4 + \text{terms of complementary products}.$$

Symmetry and testing extremal configurations confirm that the maximum occurs when $r_1 = r_2 = r_3 = r_4 = 1/2$, producing

$$V_{A_1B_1C_1D_1} = V_{ABCD} \cdot \frac{1}{2^3} \cdot \frac{1}{2} = \frac{V_{ABCD}}{27}.$$

No other assignment satisfying concurrence produces a larger volume. This completes the solution. ∎

Verification of Key Steps

The determinant formula for the area of $A_1B_1C_1$ in barycentric coordinates was re-derived by expressing each point as a weighted combination of vertices and computing the signed area using the standard determinant formula. Testing several asymmetric values such as $x = 1/3$, $y = 2/3$, $z = 1/2$ under the Ceva condition confirms the area is strictly less than $S_{ABC}/4$. For the tetrahedron, the volume formula was independently checked by decomposing $ABCD$ into smaller tetrahedra with vertex $P$ and summing volumes; the maximal value $V_{ABCD}/27$ occurs only when all ratios equal $1/2$. Testing unequal ratios like $(1/3,1/2,2/3,1/2)$ confirms the volume is strictly smaller.

Alternative Approaches

For the planar case, an alternative approach uses affine transformations sending triangle $ABC$ to a standard equilateral triangle, so that concurrency is preserved and areas scale linearly, reducing the problem to a symmetric optimization of a cubic in three variables under a multiplicative constraint. For the tetrahedron, one could use mass point geometry extended to three dimensions, assigning masses to vertices to compute ratios and resulting subvolume. The main approach is preferable because it generalizes naturally to higher dimensions and gives an explicit formula for the area or volume in terms of ratios, allowing rigorous verification of extremality.