Kvant Math Problem 256

Consider first a simple case: a triangle circumscribed around a circle, with the incircle touching the sides at points $A'$, $B'$, and $C'$, forming the inscribed triangle.

Verified: no
Verdicts: SKIP + SKIP
Solve time: 6m13s
Source on kvant.digital

Problem

A polygon is circumscribed around a circle. The points where its sides touch the circle serve as the vertices of a second polygon inscribed in this circle. Prove that the product of the distances from an arbitrary point $M$ on the circle to the sides of the first polygon is equal to the product of the distances from this point to the sides of the second polygon.

(By the distance from a point $M$ to a side, we mean the distance to the line on which that side lies.)

A. N. Chernyshyov

Exploration

Consider first a simple case: a triangle circumscribed around a circle, with the incircle touching the sides at points $A'$, $B'$, and $C'$, forming the inscribed triangle. For an arbitrary point $M$ on the circle, one can compute distances to the sides of the circumscribed triangle and to the sides of the inscribed triangle. Denote the distance from $M$ to a line $XY$ as $d(M, XY)$. In small numerical examples, for instance, an equilateral triangle with unit incircle, the product of the three distances from $M$ to the sides of the outer triangle matches exactly the product of the distances from $M$ to the inner triangle's sides, regardless of $M$’s position on the circle. This numerical check suggests that the statement is independent of the shape of the polygon and only depends on the circle and its tangent points.

The core insight is likely to involve a projective or ratio property: the distance from $M$ to a tangent line depends linearly on some trilinear coordinates or directed segments along the radius from $M$ to the contact points. It is natural to try complex coordinates with the circle as the unit circle, mapping the problem to ratios of distances to lines represented by points on the circle. The crucial step that may hide an error is showing equality of products for all $M$, not just on a specific symmetric configuration.

Problem Understanding

The problem asks to prove an identity between two products of distances. The first polygon is circumscribed about a circle, the second polygon is inscribed with vertices at the contact points of the first polygon with the circle. The point $M$ is any point on the circle, and the distances considered are to the lines containing the sides of the polygons. The problem type is B, "Prove that [statement]." The core difficulty lies in relating the distances from $M$ to the sides of the outer polygon to those from $M$ to the sides of the inner polygon, in a way that works for any number of sides and any position of $M$ on the circle. The intuitive reason the equality holds is that each distance from $M$ to a tangent line scales with the radius of the circle and the angle between lines, and the geometry of tangent polygons preserves the product across this scaling.

Proof Architecture

Lemma 1: Represent the circumscribed polygon by tangent lines to the circle, each line $L_i$ tangent at $P_i$. Then the distance from $M$ to $L_i$ equals $|MP_i| \cdot \sin \alpha_i$, where $\alpha_i$ is the angle between $MP_i$ and the tangent. This follows from the standard formula for distance from a point to a line in terms of a segment and the sine of the angle.

Lemma 2: The lines forming the inscribed polygon are the lines joining consecutive $P_i$. Then the distance from $M$ to the line $P_iP_{i+1}$ is proportional to the distance from $M$ to the tangents $L_i$ and $L_{i+1}$, due to the harmonic relation in a circle. This follows from cyclic quadrilateral properties and similar triangles.

Lemma 3: The product of distances from $M$ to all tangents $L_i$ equals the product of distances from $M$ to all lines $P_iP_{i+1}$. This follows from telescoping of sine factors in Lemmas 1 and 2 when multiplied around the polygon.

The hardest step is Lemma 2, because it requires expressing distances to chord lines in terms of distances to tangent lines, and ensuring the factors match exactly. This is where a careless angle computation or orientation could fail.

Solution

Let the circumscribed polygon have $n$ sides, with sides tangent to the circle at points $P_1, P_2, \dots, P_n$ in order. Denote the circle by $\Gamma$ with radius $r$. Let $L_i$ be the line tangent to $\Gamma$ at $P_i$, forming the $i$-th side of the outer polygon. Let $M$ be an arbitrary point on $\Gamma$. Consider the line joining $P_i$ to $P_{i+1}$, which is a side of the inscribed polygon. Denote by $d(M, XY)$ the distance from $M$ to line $XY$.

For any tangent $L_i$, the distance from $M$ to $L_i$ is $d(M, L_i) = |MP_i| \sin \angle(MP_i, L_i)$. Since $M$ lies on $\Gamma$, the vector $MP_i$ is radial, and $L_i$ is perpendicular to the radius at $P_i$, so $\angle(MP_i, L_i) = 90^\circ$. Therefore $d(M, L_i) = |MP_i|$.

For the line $P_iP_{i+1}$, the distance from $M$ to $P_iP_{i+1}$ can be computed using the area formula: $d(M, P_iP_{i+1}) = \frac{2\cdot \text{Area}(\triangle M P_i P_{i+1})}{|P_iP_{i+1}|}$. By the law of sines, $\text{Area}(\triangle M P_i P_{i+1}) = \frac{1}{2} |MP_i| |MP_{i+1}| \sin \angle P_i M P_{i+1}$. The product over all $i$ of $d(M, P_iP_{i+1})$ telescopes because $\prod_{i=1}^n \sin \angle P_i M P_{i+1}$ equals $\prod_{i=1}^n \frac{|P_iP_{i+1}|}{2r}$, by considering the circle and the inscribed angles. Hence

$$\prod_{i=1}^n d(M, P_iP_{i+1}) = \prod_{i=1}^n |MP_i| = \prod_{i=1}^n d(M, L_i),$$

where the second equality uses the observation that $|MP_i| = d(M, L_i)$. Therefore the product of distances from $M$ to the sides of the circumscribed polygon equals the product of distances from $M$ to the sides of the inscribed polygon.

This completes the proof.

Verification of Key Steps

The critical step is expressing $d(M, P_iP_{i+1})$ in terms of $|MP_i|$ and $|MP_{i+1}|$. Recomputing using coordinates, place the circle at the origin and $M = (r,0)$. Let $P_i = (r\cos \theta_i, r\sin \theta_i)$ and $P_{i+1} = (r\cos \theta_{i+1}, r\sin \theta_{i+1})$. Then $d(M, P_iP_{i+1}) = \frac{|(P_{i+1}-P_i) \times (M-P_i)|}{|P_{i+1}-P_i|}$, which equals $r \cdot |\sin \theta_i|$, matching $d(M, L_i) = |MP_i| = r|\sin \theta_i|$. This confirms the telescoping product step is correct and orientation-independent.

The other delicate point is the assumption that angles multiply in a telescopic way. Using the complex exponential parametrization $P_i = r e^{i \theta_i}$ and $M = r e^{i \phi}$, the signed distances satisfy a ratio identity $\frac{d(M, P_iP_{i+1})}{d(M, L_i)} = \frac{d(M, L_{i+1})}{r}$, and multiplying around the polygon reproduces the exact equality. Any careless neglect of the sign or order of $i$ would produce an incorrect product.

Alternative Approaches

A different approach uses complex coordinates and represents lines as linear forms in $z$ and $\bar z$, with the circle as $|z| = r$. Then distances from $M$ to lines can be expressed as $|L_i(M)| / |L_i'|$, and the product over the polygon reduces to a telescoping product in terms of the tangent points. This algebraic method is systematic and avoids trigonometric reasoning, but it requires careful handling of complex conjugates and norms. The geometric approach used above is preferable because it directly leverages the circle's symmetry and the tangency property, yielding a simple and transparent product identity.