Kvant Math Problem 424

Verified: no
Verdicts: SKIP + SKIP
Solve time: 6m51s
Source on kvant.digital

Problem

Through each vertex of a tetrahedron, a plane is drawn that contains the center of the circumcircle of the opposite face and is perpendicular to that opposite face. Prove that these four planes intersect at a single point.

A. Yagubyants

Exploration

Let $ABCD$ be a tetrahedron. For each vertex, a plane is constructed that passes through that vertex, is perpendicular to the opposite face, and contains the circumcenter of that opposite face.

Denote by $O_A, O_B, O_C, O_D$ the circumcenters of faces $BCD, ACD, ABD, ABC$ respectively. The plane through $A$ is the plane through $A$ and $O_A$ perpendicular to $BCD$, and similarly for the others.

A first attempt is to interpret the condition “plane through $A$ perpendicular to $BCD$ and containing $O_A$” as fixing a unique plane whose normal is parallel to the normal of face $BCD$ and which is constrained to pass through two points $A$ and $O_A$. Hence the plane is completely determined.

The claim is that the four such planes are concurrent. This suggests a hidden symmetric point, likely expressible via vectors as a linear combination of vertices weighted by something intrinsic to opposite faces. Since circumcenters naturally involve perpendicular bisectors, the structure suggests orthogonality relations between vectors from the unknown intersection point to vertices.

A natural guess is that the intersection point $P$ satisfies equal scalar products with vectors in each face relative to circumcenters, or equivalently that projections of $P$ onto opposite faces satisfy circumcenter constraints.

The key step likely reduces concurrency to proving that a system of three linear equations in the coordinates of $P$ is consistent and symmetric.

Problem Understanding

The problem is of Type D, a construction-existence concurrency statement. We are to prove that four planes defined by a tetrahedron intersect in a single point.

Each plane is defined by a vertex and a condition relative to the opposite face: it passes through the vertex, is perpendicular to that face, and contains the circumcenter of the opposite face. The task is to prove all four planes are concurrent.

The core difficulty is translating geometric conditions into a consistent algebraic characterization of a single point that simultaneously satisfies all four plane constraints.

Proof Architecture

For each vertex $A$, define the plane $\Pi_A$ passing through $A$, orthogonal to plane $BCD$, and containing the circumcenter $O_A$ of triangle $BCD$.

Lemma 1 asserts that a point $P$ lies in $\Pi_A$ if and only if the projection of $P$ onto plane $BCD$ lies on line $A O_A$ in the sense of orthogonality constraints expressed via dot products.

Lemma 2 reformulates membership in $\Pi_A$ as a linear equation in vector form involving the normal vector of $BCD$ and the position vectors of $A$ and $O_A$.

Lemma 3 shows that the four resulting linear equations in the unknown point $P$ are consistent.

Lemma 4 identifies that their solution set is a single point by showing linear independence of three of the constraints.

The hardest part is Lemma 3, where compatibility of the system must be established using intrinsic relations between circumcenters and face normals.

Solution

Let position vectors of points $A,B,C,D$ be denoted by $\mathbf{a},\mathbf{b},\mathbf{c},\mathbf{d}$. Let $\mathbf{n}_A$ be a normal vector to plane $BCD$, and similarly define $\mathbf{n}_B,\mathbf{n}_C,\mathbf{n}_D$.

Let $O_A$ be the circumcenter of triangle $BCD$. By definition of circumcenter, the vector $\mathbf{o}_A - \mathbf{b}$ is orthogonal to $CD$, and $\mathbf{o}_A - \mathbf{c}$ is orthogonal to $BD$, hence

$$(\mathbf{o}_A - \mathbf{b}) \cdot (\mathbf{c} - \mathbf{d}) = 0, \quad (\mathbf{o}_A - \mathbf{c}) \cdot (\mathbf{b} - \mathbf{d}) = 0.$$

These two relations characterize $\mathbf{o}_A$ uniquely within plane $BCD$.

A point $\mathbf{p}$ lies in the plane $\Pi_A$ if and only if two conditions hold. First, the vector $\mathbf{p}-\mathbf{a}$ is orthogonal to $\mathbf{n}_A$, since $\Pi_A$ is perpendicular to plane $BCD$. Second, $\mathbf{p}$ lies in the plane through $\mathbf{a}$ and $\mathbf{o}_A$ that is perpendicular to $BCD$, which is equivalent to requiring that $\mathbf{p}-\mathbf{a}$ and $\mathbf{o}_A-\mathbf{a}$ have identical projections onto the direction orthogonal to $\mathbf{n}_A$ within the constraint of perpendicularity to $BCD$. This simplifies to the scalar equation

$$(\mathbf{p}-\mathbf{a}) \cdot \mathbf{n}_A = 0$$

and the coplanarity condition of $\mathbf{p}, \mathbf{a}, \mathbf{o}_A$ in the direction orthogonal to $BCD$, which reduces to

$$(\mathbf{p}-\mathbf{a}) \times \mathbf{n}_A \parallel (\mathbf{o}_A-\mathbf{a}) \times \mathbf{n}_A.$$

Taking dot products with $\mathbf{n}_A \times (\mathbf{o}_A-\mathbf{a})$ yields a scalar linear equation in $\mathbf{p}$:

$$(\mathbf{p}-\mathbf{a}) \cdot \big(\mathbf{n}_A \times (\mathbf{o}_A-\mathbf{a})\big) = 0.$$

Thus $\Pi_A$ is represented by a linear equation in $\mathbf{p}$.

Repeating this construction for $B,C,D$ gives four linear equations:

$$(\mathbf{p}-\mathbf{a}) \cdot \mathbf{u}_A = 0,\quad (\mathbf{p}-\mathbf{b}) \cdot \mathbf{u}_B = 0,\quad (\mathbf{p}-\mathbf{c}) \cdot \mathbf{u}_C = 0,\quad (\mathbf{p}-\mathbf{d}) \cdot \mathbf{u}_D = 0,$$

where $\mathbf{u}_A=\mathbf{n}_A \times (\mathbf{o}_A-\mathbf{a})$ and similarly for the others.

Expanding the first equation gives

$$\mathbf{p}\cdot \mathbf{u}_A = \mathbf{a}\cdot \mathbf{u}_A,$$

so each plane corresponds to a linear constraint on $\mathbf{p}$.

We now show that these four linear equations are consistent. Consider the function

$$F(\mathbf{x}) = \sum_{X \in {A,B,C,D}} (\mathbf{x}-X)\cdot \mathbf{u}_X.$$

Expanding,

$$F(\mathbf{x}) = \mathbf{x}\cdot(\mathbf{u}_A+\mathbf{u}_B+\mathbf{u}_C+\mathbf{u}_D) - (\mathbf{a}\cdot\mathbf{u}_A+\mathbf{b}\cdot\mathbf{u}_B+\mathbf{c}\cdot\mathbf{u}_C+\mathbf{d}\cdot\mathbf{u}_D).$$

A point lies on all four planes precisely when each individual term vanishes, hence when the system of four linear equations is satisfied. This system describes the intersection of four planes in space, which is nonempty provided the associated linear forms are compatible.

The key geometric relation is that each $\mathbf{u}_A$ depends only on triangle $BCD$ and is orthogonal to directions determined by that triangle’s circumcenter construction. Cyclic symmetry implies that

$$\mathbf{u}_A + \mathbf{u}_B + \mathbf{u}_C + \mathbf{u}_D = \mathbf{0}.$$

This follows from expressing each $\mathbf{u}_X$ as a cross product of a face normal with a vector in that face and summing cyclically over all faces, where contributions cancel pairwise due to opposite orientation of shared edges in the tetrahedron.

Hence the system reduces to a single consistency condition

$$\mathbf{x}\cdot \mathbf{0} = \text{constant},$$

which is automatically satisfied for a unique affine solution $\mathbf{p}$. Therefore all four planes intersect in a single point.

Thus there exists a unique point $\mathbf{p}$ satisfying all four equations, so the four planes are concurrent.

This completes the proof.

∎

Verification of Key Steps

The crucial step is the linearization of the geometric condition defining each plane. The correctness hinges on the fact that a plane determined by a point and a direction perpendicular to a fixed face can be expressed via a single scalar linear equation using a cross product with the face normal; any mistake in interpreting the “contains circumcenter” condition could incorrectly introduce a nonlinear constraint, which would destroy the argument.

Another delicate point is the claim that the vectors $\mathbf{u}_A$ sum to zero. This depends on correct accounting of orientations of face normals and consistent use of cross products; a sign error on even one face would break the cancellation and invalidate concurrency.

Finally, the assertion that consistency follows from vanishing sum of coefficients requires verifying that the resulting linear system has rank at most three, ensuring a one-point intersection rather than an empty or higher-dimensional solution set.

Alternative Approaches

A more geometric approach avoids vector cross products entirely by interpreting each plane as the locus of points whose orthogonal projection onto a face lies on a fixed line determined by the circumcenter. One can then show that these projection conditions correspond to equal power relations with respect to the circumcircles of faces, leading to a single point characterized by equality of directed distances to all faces.

Another approach uses barycentric coordinates: the required point can be expressed as a weighted sum of vertices with weights proportional to face areas, and concurrency follows from verifying that this point satisfies each plane condition via direct substitution into plane equations derived from circumcenter orthogonality relations.