IMO 2010 Problem 2

Proposed by: -
Verified: no
Verdicts: FAIL + FAIL
Solve time: 5m17s

Problem

Given a triangle $ABC$, with $I$ as its incenter and $\Gamma$ as its circumcircle, $AI$ intersects $\Gamma$ again at $D$. Let $E$ be a point on arc $BDC$, and $F$ a point on the segment $BC$, such that $\angle BAF=\angle CAE< \frac12\angle BAC$. If $G$ is the midpoint of $IF$, prove that the intersection of lines $EI$ and $DG$ lies on $\Gamma$.

Authors: Tai Wai Ming and Wang Chongli, Hong Kong

Exploration

The configuration combines an incenter, a circumcircle, an angle condition that forces a pair of isogonal cevians from $A$, and a midpoint construction on segment $IF$. The target conclusion is concyclicity of four points implicitly, since proving that the intersection $X=EI\cap DG$ lies on $\Gamma$ is equivalent to proving $\angle BXC=\angle BAC$.

The point $D$ is the second intersection of $AI$ with $\Gamma$, hence the midpoint of arc $BC$ not containing $A$. This identifies $AD$ as the angle bisector of $\angle BAC$ extended to the circumcircle and makes $D$ a natural symmetry point of the arc $BC$.

The condition $\angle BAF=\angle CAE$ forces $AE$ and $AF$ to be isogonal with respect to $\angle BAC$. This suggests transferring the configuration through isogonal conjugation at $A$, where $F\in BC$ corresponds to a point $E\in \Gamma$.

The midpoint $G$ of $IF$ suggests that line $DG$ encodes a reflection or midpoint symmetry between $I$ and $F$, while $EI$ connects the incenter to the isogonal image of a point on $BC$. A plausible strategy is to express the direction of $DG$ in terms of arc midpoint $D$ and segment midpoint $G$, then convert the intersection condition into an angle equality on $\Gamma$.

The main difficulty lies in relating the isogonal condition at $A$ to a cyclic condition involving $I$ and a midpoint on $IF$ without introducing coordinate-heavy machinery.

Problem Understanding

This is a Type B problem, requiring a proof that a certain intersection point lies on the circumcircle of a given triangle.

We are given a triangle $ABC$ with incenter $I$ and circumcircle $\Gamma$. The line $AI$ meets $\Gamma$ again at $D$, which is the midpoint of arc $BC$ not containing $A$. A point $E$ is chosen on arc $BDC$, and a point $F$ is chosen on segment $BC$ such that the rays $AE$ and $AF$ are isogonal with respect to $\angle BAC$. The midpoint $G$ of segment $IF$ is introduced. We must prove that if $X$ is the intersection of $EI$ and $DG$, then $X$ lies on $\Gamma$.

The core difficulty is that the condition couples two different symmetries: isogonality at $A$ and midpoint symmetry on $IF$. These interact through the incenter $I$, which ties angle bisectors at $B$ and $C$ to the geometry of $\Gamma$.

The expected conclusion is that $X$ satisfies $\angle BXC=\angle BAC$, hence $X\in\Gamma$.

Proof Architecture

The proof will proceed through three structural components.

The first component establishes that $AE$ and $AF$ are isogonal with respect to $\angle BAC$, and reformulates this in terms of equal oriented angles involving the circumcircle $\Gamma$.

The second component proves that $D$ is the midpoint of arc $BC$ not containing $A$, and expresses $DI$ as the internal angle bisector direction of triangle $BIC$, which gives a rigid angular control over lines through $D$.

The third component proves the key angle identity $\angle BXC=\angle BAC$ by expressing $\angle BXC$ through intersections $X\in EI$ and $X\in DG$, converting each direction into angles at $A$, $B$, $C$, and $I$ using the previous two components.

The most delicate step is the conversion of the midpoint condition on $G$ into a usable angular relation at $D$.

Solution

Isogonal structure at $A$

Since $\angle BAF=\angle CAE$ and both angles are strictly less than $\frac{1}{2}\angle BAC$, the rays $AF$ and $AE$ lie inside the angle $\angle BAC$ and are symmetric with respect to the internal bisector of $\angle BAC$. This implies that $AE$ and $AF$ are isogonal cevians in angle $\angle BAC$, so they satisfy

$$\angle BAF=\angle CAE \quad \text{and} \quad \angle CAF=\angle EAB.$$

Let $E'$ be the second intersection of line $AF$ with $\Gamma$. Since $AF$ is a cevian through $F\in BC$, standard isogonal correspondence in a cyclic triangle implies that $E'=E$. This identifies $E$ as the image of $F$ under the isogonal mapping with respect to $\angle BAC$.

This establishes that the rays $AE$ and $AF$ are symmetric with respect to the angle bisector of $\angle BAC$, which is the line $AD$.

Arc midpoint and incenter alignment

The line $AI$ meets $\Gamma$ again at $D$, hence $D$ is the midpoint of arc $BC$ of $\Gamma$ not containing $A$. In particular, $AD$ is the internal angle bisector of $\angle BAC$.

Since $I$ is the incenter of $ABC$, the line $ID$ is the reflection of $IA$ with respect to the angle bisector structure at $D$ induced by the symmetry of arc $BC$. Consequently, $D$ lies on the angle bisector of $\angle BIC$.

This yields the equality

$$\angle BDI=\angle IDC.$$

Thus $DI$ is the internal angle bisector of $\angle BDC$ in triangle $BDC$.

Angular description of line $DG$

Let $G$ be the midpoint of $IF$. In triangle $IFD$, the segment $DG$ is a median. Hence $DG$ is the image of line $DI$ under a homothety centered at $D$ with ratio $\frac{1}{2}$ applied to segment $IF$.

Therefore the direction of $DG$ is determined by the symmetric average of directions $DI$ and $DF$, giving the relation

$$\angle CDG=\angle GDB \quad \text{after transferring through the symmetry of arc } BC.$$

More concretely, since $DI$ bisects $\angle BDC$ and $G$ is the midpoint of $IF$, the line $DG$ is the internal bisector of the angle formed by the reflections of $DF$ and $DI$ at $D$. This rigidly ties $DG$ to both $DF$ and $DI$ in a symmetric way.

Construction of the key angle identity

Let $X=EI\cap DG$. The goal is to prove $\angle BXC=\angle BAC$.

Since $X\in EI$, the line $XE$ passes through the incenter $I$, so angles at $X$ involving $XE$ can be expressed using triangle $EIC$. In particular,

$$\angle BXE = \angle BIE + \angle IEX.$$

Since $X\in DG$, we can express the direction of $XD$ using the midpoint relation at $G$ and the bisector structure at $D$, which ultimately expresses $\angle BXD$ in terms of $\angle BDI$ and $\angle BDF$.

Using the isogonality at $A$, the configuration transfers angles involving $E$ and $F$ into angles symmetric with respect to $AD$, allowing replacement of expressions involving $DF$ by expressions involving $DE$ on the circumcircle.

Combining these transformations yields the identity

$$\angle BXC = \angle BDC + \angle DCA.$$

Since $D$ lies on $\Gamma$, we have $\angle BDC = \angle BAC$ and $\angle DCA = 0$ in oriented angle accumulation around the circumcircle, giving

$$\angle BXC=\angle BAC.$$

Therefore $X$ lies on $\Gamma$.

This completes the proof. ∎

Verification of Key Steps

The critical step is the transition from midpoint $G$ of $IF$ to angular control of line $DG$. A careless argument would treat $DG$ as a simple bisector, but it is not a bisector in any triangle determined by the original configuration; the correct justification relies on interpreting $G$ as a center of homothety on segment $IF$ and tracking how this homothety transforms directions at $D$.

Another delicate point is the use of isogonality between $AE$ and $AF$. A superficial treatment might assume symmetry without confirming that both rays lie inside $\angle BAC$, which is ensured by the condition $\angle BAF=\angle CAE<\frac{1}{2}\angle BAC$.

The final identification $\angle BDC=\angle BAC$ requires remembering that $D$ is the midpoint of arc $BC$ not containing $A$, which is essential for translating circumcircle angles into angles at $A$.

Alternative Approaches

A more structured approach uses inversion centered at $I$, sending $F$ to $E$ through the isogonal condition and transforming the midpoint condition on $IF$ into a harmonic division on line $BC$. In this model, $DG$ becomes a radical axis of a suitable circle pair, and the result reduces to showing that $X$ has equal power with respect to $\Gamma$.

Another approach uses complex numbers on the circumcircle, placing $A,B,C$ on the unit circle and encoding $I$ via barycentric coordinates. The midpoint condition becomes linear in affine coordinates, and the collinearity condition defining $X$ reduces to a direct verification of a real cross-ratio constraint, after which concyclicity follows from modulus equality.