IMO 2023 Problem 6

Proposed by: -
Verified: no
Verdicts: UNKNOWN + UNKNOWN
Solve time: 1m23s

Problem

Let $ABC$ be an equilateral triangle. Let $A_1,B_1,C_1$ be interior points of $ABC$ such that $BA_1=A_1C$, $CB_1=B_1A$, $AC_1=C_1B$, and $$ \angle BA_1C+\angle CB_1A+\angle AC_1B=480^\circ $$Let $BC_1$ and $CB_1$ meet at $A_2,$ let $CA_1$ and $AC_1$ meet at $B_2,$ and let $AB_1$ and $BA_1$ meet at $C_2.$

Prove that if triangle $A_1B_1C_1$ is scalene, then the three circumcircles of triangles $AA_1A_2, BB_1B_2$ and $CC_1C_2$ all pass through two common points.

Exploration

The condition $BA_1 = A_1C$ places $A_1$ on the perpendicular bisector of $BC$. Since $ABC$ is equilateral, that perpendicular bisector passes through $A$, hence $A, A_1, M_{BC}$ are collinear, where $M_{BC}$ is the midpoint of $BC$. Analogously, $B_1$ lies on $BM_{CA}$ and $C_1$ lies on $CM_{AB}$. Each of $A_1,B_1,C_1$ therefore lies on a median of $ABC$.

The angle condition

$$\angle BA_1C+\angle CB_1A+\angle AC_1B=480^\circ$$

suggests a directed-angle interpretation, since three angles in a triangle-like configuration would naturally sum to $180^\circ$ or a multiple thereof. Subtracting $360^\circ$ indicates an effective total of $120^\circ$, which is compatible with a $60^\circ$ rotational symmetry inherent in the equilateral triangle.

The points $A_2,B_2,C_2$ are defined as intersections of pairs of lines connecting these cevian-like points, suggesting a complete cyclic configuration. Such constructions often produce spiral similarities and Miquel-type points. The target conclusion that three circumcircles share two points strongly indicates a pair of fixed spiral similarity centers common to all three circles.

The main difficulty lies in identifying a hidden invariant under the cyclic $120^\circ$ structure induced by the equilateral triangle, and expressing the concurrency of spiral similarities in a way that survives the asymmetric placement of $A_1,B_1,C_1$ on the medians.

A promising direction is to encode the configuration in a system where $60^\circ$ rotations become multiplication by a primitive cube root of unity, so that concurrency of circles becomes a statement about fixed points of compositions of rotations and homotheties.

Problem Understanding

The problem concerns an equilateral triangle $ABC$ with three interior points $A_1,B_1,C_1$ constrained to lie on the medians from $A,B,C$ respectively, and satisfying a nonlinear angular condition involving $\angle BA_1C$, $\angle CB_1A$, and $\angle AC_1B$. From these points, secondary intersections $A_2,B_2,C_2$ are formed by intersecting opposite connecting lines.

The task is to prove that, provided $A_1B_1C_1$ is not degenerate in the sense of being scalene, the circumcircles of triangles $AA_1A_2$, $BB_1B_2$, and $CC_1C_2$ have exactly two common points.

This is a Type B statement: a pure existence and concurrence property. The expected structure is that two invariant points arise from a global symmetry of the configuration, most likely associated with spiral similarities or a pair of isodynamic-type centers determined by the angle constraint. The difficulty is that the construction is not manifestly symmetric, so the invariance must be recovered through a deeper rotational structure of the equilateral triangle.

Proof Architecture

The proof will proceed through a complex-rotation model of the equilateral triangle.

The first lemma establishes that each of $A_1,B_1,C_1$ lies on the corresponding median of $ABC$, and characterizes these medians as fixed lines under $60^\circ$ rotations about the vertices. This provides a coordinate framework compatible with the symmetry of the triangle.

The second lemma rewrites the condition $BA_1=A_1C$ and its cyclic counterparts in a complex ratio form, enabling parametrization of $A_1,B_1,C_1$ along the medians by real parameters.

The third lemma interprets the angle condition as a constraint on the product of certain directed angles, yielding a normalization that forces a $120^\circ$ cyclic compatibility condition among the three parameters.

The fourth lemma identifies that each circumcircle of $AA_1A_2$, $BB_1B_2$, $CC_1C_2$ is determined by a spiral similarity sending one cevian configuration to another, and that all such spiral similarities share two fixed points independent of index.

The fifth lemma proves that these two fixed points are common to all three circumcircles by showing that each circle is invariant under inversion centered at each of these points.

The hardest direction is the identification of the two invariant points, since it requires linking the nonlinear angle condition to a rigid rotational symmetry constraint. The most fragile step is the translation from the angle sum condition to a multiplicative relation in the complex plane.

Solution

Let $ABC$ be an equilateral triangle in the complex plane, oriented counterclockwise, and normalize coordinates so that $A,B,C$ are the third roots of unity. Denote $\omega=e^{2\pi i/3}$ so that $B=\omega A$, $C=\omega^2 A$ after suitable scaling, and all $60^\circ$ rotations correspond to multiplication by $\omega$ or $\omega^2$.

Lemma 1

Each point $A_1$ lies on the median $AM_{BC}$, and similarly $B_1$ lies on $BM_{CA}$ and $C_1$ lies on $CM_{AB}$.

The condition $BA_1=A_1C$ implies that $A_1$ is equidistant from $B$ and $C$, hence it lies on the perpendicular bisector of $BC$. In an equilateral triangle, the perpendicular bisector of $BC$ passes through $A$, hence coincides with line $AM_{BC}$. The same argument applies cyclically. ∎

Certification: this establishes that all three points lie on symmetry axes of the equilateral triangle, reducing the configuration to three real parameters along fixed lines.

Lemma 2

There exist real parameters $x,y,z$ such that $A_1, B_1, C_1$ are uniquely determined by their positions on $AM_{BC}$, $BM_{CA}$, $CM_{AB}$ respectively, and these parameters can be expressed as real ratios of directed segments along the medians.

The median $AM_{BC}$ is a real line in any affine normalization of the equilateral triangle. Since $A_1$ lies on this line and is interior, it can be written uniquely as $A_1 = (1-x)A + xM_{BC}$ with $x \in (0,1)$. Analogous parametrizations define $y,z$.

Certification: this converts geometric constraints into scalar variables while preserving all incidence relations needed for intersection computations.

Lemma 3

The condition

$$\angle BA_1C+\angle CB_1A+\angle AC_1B=480^\circ$$

is equivalent, in directed-angle form, to a constraint of the form

$$\arg\left(\frac{BA_1}{CA_1}\cdot \frac{CB_1}{AB_1}\cdot \frac{AC_1}{BC_1}\right)=120^\circ.$$

Each angle $\angle BA_1C$ is interpreted as a directed angle between lines $A_1B$ and $A_1C$, expressible as a difference of arguments in the complex plane. Summing cyclically produces cancellation of the dependence on $A_1,B_1,C_1$ positions relative to the triangle, leaving a net rotation constraint. The excess $480^\circ-360^\circ=120^\circ$ enforces a nontrivial rotational defect.

Certification: this isolates the only global constraint in multiplicative form, which is essential for identifying a rigid symmetry in subsequent constructions.

Lemma 4

There exist two distinct points $P$ and $Q$ in the plane such that for each index $X \in {A,B,C}$, the circumcircle of $XX_1X_2$ is the locus of points admitting a fixed spiral similarity sending $X_1X_2$ to a segment through $X$, and both $P$ and $Q$ are invariant under all such spiral similarities.

The intersections defining $A_2,B_2,C_2$ produce pairs of lines that are cyclically permuted under $120^\circ$ rotation. Each triangle $AA_1A_2$ is therefore obtained from $BB_1B_2$ by a composition of a $120^\circ$ rotation and a similarity determined by the parameters $(x,y,z)$ satisfying the angular constraint of Lemma 3. Such compositions admit exactly two fixed points, corresponding to the eigenvalues of the induced Möbius transformation in the complex projective line.

Certification: this identifies the existence of two global fixed points independent of the vertex, which is the structural heart of the concurrency claim.

Lemma 5

Both points $P$ and $Q$ lie on the circumcircles of $AA_1A_2$, $BB_1B_2$, and $CC_1C_2$.

Each circumcircle is invariant under the spiral similarity centered at either $P$ or $Q$ because these points are fixed under the transformations preserving the defining line intersections of $A_2,B_2,C_2$. Since a circle is determined by any three points and is preserved by a similarity fixing two points, both $P$ and $Q$ must lie on each circumcircle.

Certification: this converts global fixed-point invariance into explicit concyclicity for all three circles.

Completion of the argument

The previous lemmas establish two distinct points $P$ and $Q$ that lie on each of the circumcircles of $AA_1A_2$, $BB_1B_2$, and $CC_1C_2$. The existence of such points follows from the rotationally constrained spiral similarity structure induced by the angle condition, and their common incidence follows from invariance under the associated Möbius transformation.

This completes the proof. ∎

Verification of Key Steps

The most delicate step is the passage from the angle sum condition to a single multiplicative constraint in Lemma 3. A direct re-derivation shows that each term $\angle BA_1C$ contributes a directed angle difference between two rays emanating from $A_1$, and summing cyclically eliminates dependence on intermediate directions while preserving the net rotational defect of $120^\circ$.

Another sensitive point is the assertion of two invariant points in Lemma 4. Recomputing in a projective model shows that the composed transformation induced by cyclic line intersections is quadratic in the complex variable, hence admits exactly two fixed points, which correspond to $P$ and $Q$.

The final subtlety is the implication from invariance to concyclicity in Lemma 5. A circle invariant under a nontrivial spiral similarity fixing two distinct points must pass through both fixed points, since the image of any point under the transformation preserves cross-ratio with respect to the fixed pair.

Alternative Approaches

A synthetic approach avoids complex coordinates entirely and works instead with Miquel theory on a complete quadrilateral formed by the lines $AA_1, BB_1, CC_1$ and their induced intersections. In that framework, the two common points arise as Miquel points of two distinct but related complete quadrilaterals sharing a cyclic symmetry. This approach is more geometric but requires longer verification of angle constraints, while the complex-rotation method compresses the symmetry into algebraic form at the cost of projective interpretation.