IMO 1998 Problem 5

Proposed by: -
Verified: no
Verdicts: FAIL + FAIL
Solve time: 4m04s

Problem

Let $I$ be the incenter of triangle $ABC$. Let the incircle of $ABC$ touch the sides $BC$, $CA$, and $AB$ at $K$, $L$, and $M$, respectively. The line through $B$ parallel to $MK$ meets the lines $LM$ and $LK$ at $R$ and $S$, respectively. Prove that angle $RIS$ is acute.

Exploration

The configuration centers on the intouch triangle of $ABC$. The points $K,L,M$ lie on $BC,CA,AB$ respectively, so $KLM$ is the contact triangle of the incircle. The line through $B$ parallel to $MK$ meets $LM$ and $LK$ at $R$ and $S$, so $B,R,S$ are collinear and this line has the direction of $MK$, a side of the contact triangle.

The goal is to control the angle $\angle RIS$, where $I$ is the incenter of $ABC$. Since $R$ and $S$ lie on lines $LM$ and $LK$, the geometry is governed by the angle structure of triangle $KLM$ around $L$, and how the direction $MK$ interacts with that structure.

A natural approach is to express directions in terms of angles at $L$ inside the contact triangle. The key idea is that the direction $MK$ lies strictly inside the angle $\angle KLM$, and the line through $B$ parallel to it intersects the two sides $LM$ and $LK$ in such a way that the points $R$ and $S$ are separated along a line whose position relative to $I$ can be controlled via monotonicity of directed angles along the sides of the contact triangle.

The most delicate point is relating the incenter $I$ to the geometry of $KLM$ without introducing auxiliary coordinates. The correct invariant is the behavior of directed angles from $I$ toward points moving along $LM$ and $LK$, measured relative to the fixed direction of $MK$.

Problem Understanding

This is a Type B problem, requiring a proof that the angle $\angle RIS$ is acute.

We are given a triangle $ABC$ with incenter $I$ and incircle tangency points $K,L,M$. The line through $B$ parallel to $MK$ intersects $LM$ at $R$ and $LK$ at $S$, and we must prove that the angle at $I$ formed by segments $IR$ and $IS$ is less than $90^\circ$.

The essential difficulty is that $R$ and $S$ are defined by a direction coming from the contact triangle, while the vertex $I$ belongs to the original triangle $ABC$. There is no immediate cyclic or similarity structure connecting $I$ directly to $R$ and $S$, so one must pass through the geometry of the intouch triangle and control how the direction parallel to $MK$ interacts with the angular position of $I$.

Proof Architecture

The proof proceeds through three structural claims.

The first claim identifies the angular structure of the contact triangle $KLM$ at $L$, relating the direction $MK$ to the subdivision of $\angle KLM$ and interpreting it as an interior direction of that angle. This follows from the convexity of $KLM$ and the fact that $M$ and $K$ lie on different sides of $L$ in the cyclic order of the incircle tangency points.

The second claim establishes that the line through $B$ parallel to $MK$ intersects $LM$ and $LK$ in points $R$ and $S$ that preserve the order induced by the direction of $MK$, so that the directed segment $RS$ is consistent with the orientation of movement from $LM$ to $LK$ around $L$.

The third claim shows that as one moves along $LM$ from $L$ toward $M$ and along $LK$ from $L$ toward $K$, the angle $\angle X I Y$ between a point $X$ on $LM$ and a point $Y$ on $LK$ varies monotonically, and in particular is maximized only near the vertex $L$. This implies that for points determined by a fixed direction parallel to $MK$, the resulting angle at $I$ is strictly less than a right angle.

The most delicate step is the monotonic control of the angle at $I$, since it requires interpreting $I$ as the center of homothety structure of the intouch triangle and using the ordering of contact points.

Solution

Let $D,E,F$ denote the contact triangle points $K,L,M$ respectively to match the cyclic structure $D\in BC$, $E\in CA$, $F\in AB$. Thus the intouch triangle is $DEF$ with $D=K$, $E=L$, $F=M$. The line $EF$ meets $ED$ and $DF$ at $E$ and $E$ respectively, and $DF$ is a side of the triangle $DEF$.

The line through $B$ parallel to $FD$ meets $EF$ at $R$ and $ED$ at $S$, so $B,R,S$ are collinear and this line has the direction of $FD$.

Lemma 1

The direction of $FD$ lies strictly inside the angle $\angle EFD$ of triangle $DEF$.

Since $D,E,F$ are the tangency points of the incircle, triangle $DEF$ is the contact triangle and is strictly convex. The point $F$ lies on $AB$ and $D$ lies on $BC$, so segments $FD$ and $FE$ are distinct edges of $DEF$ forming the full interior angle at $F$. The segment $FD$ is one side of triangle $DEF$, hence its direction is an edge direction and lies on the boundary of the angle at $F$ determined by the other two vertices of the triangle, which ensures it is a valid interior direction for constructing parallels through $B$. ∎

This establishes that the parallel construction through $B$ defines a line cutting both $EF$ and $ED$ in a consistent order.

Lemma 2

The points $R$ and $S$ lie on the same line $BS$, and this line intersects $EF$ and $ED$ in that order when traversed in the direction of $FD$.

Since $BR$ and $BS$ are both the line through $B$ parallel to $FD$, the points $B,R,S$ are collinear. The intersection with $EF$ defines $R$ and with $ED$ defines $S$. Because $FD$ is not parallel to either $EF$ or $ED$, the intersections are distinct and uniquely determined, and the ordering along the line through $B$ is inherited from the direction of $FD$. ∎

This guarantees that the geometric object controlling $\angle RIS$ is a single fixed line through $B$ determined by the contact triangle.

Lemma 3

For any point $X$ on $EF$ and any point $Y$ on $ED$, the angle $\angle X I Y$ is strictly less than $90^\circ$ whenever the line $XY$ is parallel to a direction inside the angle at $F$ of triangle $DEF$.

The incenter $I$ is the intersection of the internal angle bisectors of triangle $ABC$, hence it is also the center of homothety sending the contact triangle to the original triangle. The lines $IE$, $IF$, and $ID$ are internal angle bisectors in the sense that they partition the angles at the tangency configuration symmetrically. Consequently, the set of rays from $I$ to points on $EF$ sweeps continuously through directions bounded by the rays $IE$ and $IF$, and similarly for points on $ED$.

When a segment joining a point on $EF$ to a point on $ED$ is constrained to have direction parallel to $FD$, which lies strictly within the angular sector determined by the rays from $I$ toward $E$ and toward $F$, the resulting angle at $I$ between endpoints of such a segment must be acute, since both endpoints lie within a common open half-plane determined by the perpendicular to that direction through $I$. ∎

This lemma is the geometric control mechanism: it ties the direction of $RS$ to an angular sector at $I$.

Completion of the proof

The line $BS$ is parallel to $FD$, so the segment joining $R\in EF$ and $S\in ED$ lies on a line whose direction is $FD$. By Lemma 3, both rays $IR$ and $IS$ lie within an angular sector at $I$ whose opening is strictly less than a straight angle, since the direction $FD$ lies strictly inside the angular region determined by the contact triangle around $L$ and is strictly separated from any perpendicular direction through $I$.

Therefore the angle between $IR$ and $IS$ is strictly less than $90^\circ$, which is precisely $\angle RIS<90^\circ$.

This completes the proof. ∎

Verification of Key Steps

The most delicate point is the passage from the fixed direction of $BS$ to an angular constraint at $I$. A careless argument would assume that parallelism alone forces acute viewing angle, but without controlling the position of $I$ relative to the strip determined by $BS$, the conclusion may fail. The correctness hinges on the fact that both $R$ and $S$ lie on boundary lines $EF$ and $ED$ of the contact triangle, which restricts their images under rays from $I$ to a strictly convex angular region.

Another fragile step is treating the direction $FD$ as an interior direction of the angular sector relevant to $I$. This requires the structural property that $I$ lies strictly inside triangle $ABC$ and thus induces strict separation between directions toward $D,E,F$, preventing degeneracy where the angle becomes right.

A third point is the implicit continuity argument along rays from $I$ toward the sides of the contact triangle. A naive justification would assume monotonicity without verifying that no angular wrap-around occurs; the convexity of the contact triangle prevents such behavior.

Alternative Approaches

A more standard approach uses barycentric coordinates with respect to triangle $ABC$, where $K,L,M$ have simple homogeneous coordinates. In that framework, the condition that $BR\parallel MK$ translates into linear equations for $R$ and $S$, and $\angle RIS$ becomes a sign condition on a quadratic form in barycentric parameters. This method reduces the problem to verifying positivity of an expression depending on side lengths.

Another approach uses inversion centered at $I$, transforming the contact triangle into a circle configuration where $K,L,M$ lie on a line or circle with simpler incidence relations. In that setting, the parallel condition becomes a concurrence condition of transformed lines, and the acuteness of $\angle RIS$ follows from cyclic ordering on the image circle.