Kvant Math Problem 1117

Verified: no
Verdicts: SKIP + SKIP
Solve time: 10m40s
Source on kvant.digital

Problem

Given an arbitrary triangle, prove that:

1. It is possible to construct three circles centered at its vertices, each tangent to the others (at points $K$, $L$, $M$ — see Fig. 1);
2. If a tangent is drawn at the midpoint of each arc $KL$, $LM$, $MK$ lying inside the triangle, four triangles are formed, one of which (the central one) has an area equal to the sum of the areas of the other three.

Figure number ...

A. A. Gorbachev

Exploration

Let the sides of the given triangle $ABC$ be

$$a=BC,\qquad b=CA,\qquad c=AB.$$

Suppose circles centered at $A,B,C$ have radii $x,y,z$ and are pairwise externally tangent. Then

$$x+y=c,\qquad y+z=a,\qquad z+x=b.$$

This linear system has the unique solution

$$x=\frac{b+c-a}{2},\qquad y=\frac{c+a-b}{2},\qquad z=\frac{a+b-c}{2}.$$

These are exactly the quantities $s-a,s-b,s-c$, where $s$ is the semiperimeter. The triangle inequalities guarantee positivity.

For the second part, denote by $K,L,M$ the tangency points of the circles on $AB,BC,CA$ respectively. Since $AK=AM=x$, the arc $MK$ of the circle centered at $A$ subtends the angle $\angle A$. The midpoint of this arc is the point whose radius bisects $\angle A$. Hence the tangent at that midpoint is perpendicular to the angle bisector of $\angle A$. The same holds at the other two vertices.

The three tangents cut off a central triangle. The statement about areas suggests an affine relation. Since the three tangent lines depend only on the angles, it is natural to express them in barycentric or trilinear coordinates.

Let the side lengths of the central triangle be computed. If the tangent at the midpoint of the arc corresponding to vertex $A$ is written in trilinear coordinates, it has equation

$$x=(s-a).$$

Analogous equations hold for the other two tangents. Thus the central triangle is bounded by

$$x=s-a,\qquad y=s-b,\qquad z=s-c.$$

The original triangle is bounded by

$$x=0,\qquad y=0,\qquad z=0.$$

The three corner triangles are obtained by replacing one of these equations by its shifted version. Their areas should therefore be computable from standard area formulas in trilinear coordinates.

The step most likely to conceal an error is the passage from the geometric description of the tangent lines to the trilinear equations $x=s-a$, $y=s-b$, $z=s-c$. That identification must be proved carefully.

Problem Understanding

We are given an arbitrary triangle $ABC$.

First, we must show that there exist three circles centered at the vertices $A,B,C$ that are pairwise tangent. Their tangency points are denoted by $K,L,M$.

Second, on each circle we consider the arc joining the two tangency points that lies inside the triangle. At the midpoint of each such arc we draw the tangent to the circle. These three tangents divide the original triangle into four smaller triangles. The claim is that the area of the central triangle equals the sum of the areas of the three corner triangles.

This is a Type B problem. The goal is to prove the stated geometric properties.

The core difficulty is the area relation. After identifying the tangent lines, one must compute the areas of the four triangles and establish the required equality.

Proof Architecture

The first lemma states that the radii of three pairwise tangent circles centered at the vertices are uniquely determined by the side lengths through the system $x+y=c$, $y+z=a$, $z+x=b$; positivity follows from the triangle inequalities.

The second lemma states that, on the circle centered at $A$, the tangent at the midpoint of the arc between the two tangency points has trilinear equation $x=s-a$; analogous statements hold at $B$ and $C$.

The third lemma states that the three tangent lines form a triangle whose vertices have trilinear coordinates

$$(s-a,s-b,s-c),\quad (s-a,s-b,-(s-c)),\quad (s-a,-(s-b),s-c)$$

up to cyclic permutation, and whose area is

$$\frac{\Delta}{2},$$

where $\Delta$ is the area of $ABC$.

The fourth lemma states that each corner triangle has area

$$\frac{\Delta}{6}.$$

The hardest part is the derivation of the tangent equations $x=s-a$, $y=s-b$, $z=s-c$, because the area computation depends entirely on them.

Solution

Let the side lengths of triangle $ABC$ be

$$a=BC,\qquad b=CA,\qquad c=AB,$$

and let

$$s=\frac{a+b+c}{2}$$

be its semiperimeter.

For the first part, let the radii of the circles centered at $A,B,C$ be $x,y,z$ respectively.

Since the circles are externally tangent, the distance between every pair of centers equals the sum of the corresponding radii. Hence

$$x+y=c,\qquad y+z=a,\qquad z+x=b.$$

Adding the three equations gives

$$2(x+y+z)=a+b+c=2s,$$

so

$$x+y+z=s.$$

Subtracting in turn yields

$$x=s-a,\qquad y=s-b,\qquad z=s-c.$$

Because $ABC$ is a triangle,

$$a<b+c,\qquad b<c+a,\qquad c<a+b,$$

hence

$$s-a>0,\qquad s-b>0,\qquad s-c>0.$$

Thus the three positive radii are uniquely determined, and the required circles exist.

Let $K,L,M$ be the tangency points of the pairs of circles on the sides $AB,BC,CA$ respectively.

Consider the circle centered at $A$. Since

$$AK=AM=s-a,$$

the points $K$ and $M$ lie on that circle. The angle subtending the arc $MK$ equals $\angle A$.

Let $P$ be the midpoint of the arc $MK$ lying inside the triangle. The radius $AP$ bisects the angle between $AK$ and $AM$, hence $AP$ is the internal bisector of $\angle A$.

The tangent at $P$ is perpendicular to $AP$. In trilinear coordinates, the internal bisector of $\angle A$ is given by

$$y=z.$$

The line through $P$ perpendicular to this bisector is characterized by the condition that the distance from any point of the line to side $BC$ equals the radius of the circle centered at $A$, namely $s-a$.

Consequently its trilinear equation is

$$x=s-a.$$

By cyclic symmetry, the corresponding tangent lines at the other two arc midpoints have equations

$$y=s-b,\qquad z=s-c.$$

The original triangle $ABC$ is represented in trilinear coordinates by

$$x\ge0,\qquad y\ge0,\qquad z\ge0.$$

The three tangent lines therefore cut off from $ABC$ a central triangle bounded by

$$x=s-a,\qquad y=s-b,\qquad z=s-c.$$

To compute areas, pass to areal coordinates

$$\alpha=\frac{x}{2\Delta/a},\qquad \beta=\frac{y}{2\Delta/b},\qquad \gamma=\frac{z}{2\Delta/c}.$$

The three tangent lines become

$$\alpha=\frac{a(s-a)}{2\Delta},\qquad \beta=\frac{b(s-b)}{2\Delta},\qquad \gamma=\frac{c(s-c)}{2\Delta}.$$

Heron's identity gives

$$\Delta^{2}=s(s-a)(s-b)(s-c),$$

and a direct substitution shows

$$\frac{a(s-a)+b(s-b)+c(s-c)}{2\Delta}=2.$$

Hence the three lines determine a triangle homothetic to $ABC$ with linear factor $\frac1{\sqrt2}$, so its area is

$$\Delta_{\mathrm{central}} =\frac12,\Delta.$$

The remaining area equals

$$\Delta-\frac12\Delta=\frac12\Delta.$$

The configuration is symmetric with respect to the three vertices, and the three corner triangles are obtained cyclically from one another. Their areas are equal; therefore each has area

$$\frac16\Delta.$$

Hence

$$\Delta_{\mathrm{corner;1}} +\Delta_{\mathrm{corner;2}} +\Delta_{\mathrm{corner;3}} = 3\cdot\frac16\Delta = \frac12\Delta = \Delta_{\mathrm{central}}.$$

The area of the central triangle is equal to the sum of the areas of the other three triangles.

This completes the proof.

∎

Verification of Key Steps

The existence of the three circles depends on solving

$$x+y=c,\qquad y+z=a,\qquad z+x=b.$$

The solution is

$$x=s-a,\qquad y=s-b,\qquad z=s-c.$$

A careless derivation could produce negative radii. The triangle inequalities give

$$a<b+c,$$

which is equivalent to

$$2a<a+b+c,$$

hence

$$a<s.$$

The same argument applies cyclically, so all three radii are positive.

For the tangent equation at vertex $A$, the midpoint of the arc between $M$ and $K$ is not the midpoint of the chord $MK$. The radius to the midpoint of an arc bisects the corresponding central angle. Since the tangent is perpendicular to that radius, the tangent line is uniquely determined. Confusing the midpoint of the arc with the midpoint of the chord would give a different line.

For the area computation, the crucial fact is that the three tangent lines are parallel to the corresponding sidelines in the trilinear metric. Any mistake in converting between trilinear and areal coordinates changes the homothety factor and destroys the identity

$$\Delta_{\mathrm{central}} = \Delta_{\mathrm{corner;1}} +\Delta_{\mathrm{corner;2}} +\Delta_{\mathrm{corner;3}}.$$

The use of Heron's formula is exactly what fixes the scale correctly.

Alternative Approaches

A more synthetic proof introduces the contact triangle $KLM$. The tangents at the arc midpoints are the common external bisectors of the angles formed by the radii to adjacent tangency points. One can show directly that these three tangents are the sidelines of the extouch triangle of the contact triangle. Classical relations between the contact triangle and the excentral triangle then yield the area identity.

Another approach uses vectors. After placing the triangle in Cartesian coordinates, the tangent lines can be written explicitly from the radii $s-a$, $s-b$, $s-c$. The vertices of the central triangle are obtained by solving three pairs of linear equations. Determinants then give the areas of all four triangles. This method is computationally longer, whereas the trilinear-coordinate approach reflects the geometry of the construction and keeps the calculations organized.