Kvant Math Problem 243

Let the two given lines be denoted $l_1$ and $l_2$.

kvantmathematicsolympiad

Verified: no
Verdicts: UNKNOWN + UNKNOWN
Solve time: 2m21s
Source on kvant.digital

Problem

$n$ segments $A_1B_1$, $A_2B_2$, $\ldots$, $A_nB_n$ (Fig. 1) are arranged in the plane such that each of them starts on one of two given lines, ends on the other line, and passes through the point $G$ (which does not lie on the given lines) — the center of mass of unit masses placed at the points $A_1$, $A_2$, $\ldots$, $A_n$. Prove that $$\frac{A_1G}{GB_1}+\frac{A_2G}{GB_2}+\ldots+\frac{A_nG}{GB_n}=n.$$

Figure 1

A. M. Lopshits

Exploration

Let the two given lines be denoted $l_1$ and $l_2$. Each segment $A_iB_i$ intersects both lines and passes through $G$, so the three points $A_i,G,B_i$ are collinear for every $i$. The point $G$ is the centroid of the points $A_1,\dots,A_n$ with unit masses, hence

$$\overrightarrow{GA_1}+\cdots+\overrightarrow{GA_n}=0.$$

A natural way to encode the collinearity condition is to introduce, for each $i$, a scalar $\lambda_i$ such that

$$G=\lambda_i A_i+(1-\lambda_i)B_i.$$

Then $A_i,G,B_i$ being in this order implies $0<\lambda_i<1$, and the segment ratio satisfies

$$\frac{A_iG}{GB_i}=\frac{1-\lambda_i}{\lambda_i}.$$

The goal becomes to prove

$$\sum_{i=1}^n \frac{1-\lambda_i}{\lambda_i}=n,$$

which is equivalent to

$$\sum_{i=1}^n \frac{1}{\lambda_i}=2n.$$

The centroid condition suggests summing the identities defining $G$ over $i$, producing expressions involving both $\lambda_i$ and the points $B_i$, which lie on a fixed line. The key difficulty is extracting a scalar identity from a vector equation while respecting that all $B_i$ are constrained to a line not passing through $G$.

Problem Understanding

This is a Type B problem, a pure proof of a geometric identity.

We are given $n$ segments connecting two fixed lines, all passing through a common point $G$, and $G$ is simultaneously the centroid of the endpoints $A_i$. The required conclusion is a universal relation between the internal division ratios in which $G$ divides each segment:

$$\sum_{i=1}^n \frac{A_iG}{GB_i}=n.$$

The core difficulty is to convert the centroid condition on the points $A_i$ into a constraint on how $G$ divides each segment $A_iB_i$, despite the points $B_i$ lying on an unrelated fixed line.

Proof Architecture

For each index $i$, introduce $\lambda_i$ such that $G=\lambda_i A_i+(1-\lambda_i)B_i$. The first lemma establishes that this representation is equivalent to collinearity and yields $\frac{A_iG}{GB_i}=\frac{1-\lambda_i}{\lambda_i}$.

The second lemma rewrites the centroid condition as $\sum A_i=0$ after translating the origin to $G$.

The third lemma sums the affine relations for $G$ over all $i$ and eliminates the $A_i$ using the centroid condition, producing a relation involving only the $B_i$ and the scalars $\lambda_i$.

The fourth lemma uses the fact that all $B_i$ lie on a fixed line not passing through $G$ to convert the resulting vector identity into a scalar constraint on $\lambda_i$.

The critical step is the separation of components along and transverse to the line containing the $B_i$, ensuring uniqueness of decomposition.

Solution

Translate the plane so that $G$ becomes the origin. Let position vectors of points $A_i$ and $B_i$ be denoted by $a_i$ and $b_i$ respectively. The centroid condition becomes

$$a_1+\cdots+a_n=0.$$

Since each segment $A_iB_i$ passes through the origin, the points $A_i$, $0$, and $B_i$ are collinear with the origin lying between $A_i$ and $B_i$. Hence there exists a scalar $t_i<0$ such that

$$b_i=t_i a_i.$$

The segment ratio is expressed in terms of $t_i$ by observing that $0$ divides the segment internally in the ratio $A_iG:GB_i = A_i0:0B_i$. In vector form this yields

$$\frac{A_iG}{GB_i}=\frac{|a_i|}{|b_i|}=\frac{1}{|t_i|}=-\frac{1}{t_i}.$$

Thus the required identity becomes

$$\sum_{i=1}^n -\frac{1}{t_i}=n,$$

or equivalently

$$\sum_{i=1}^n \frac{1}{t_i}=-n.$$

Each point $A_i$ lies on the fixed line $l_1$, so there exists a fixed nonzero linear functional $\alpha$ such that $\alpha(a_i)=1$ for all $i$. Similarly, since each $B_i$ lies on the fixed line $l_2$, there exists a fixed nonzero linear functional $\beta$ and constant $c$ such that $\beta(b_i)=c$ for all $i$.

Substituting $b_i=t_i a_i$ into the second relation gives

$$t_i,\beta(a_i)=c,$$

hence

$$\frac{1}{t_i}=\frac{\beta(a_i)}{c}.$$

Summing over $i$ yields

$$\sum_{i=1}^n \frac{1}{t_i}=\frac{1}{c}\sum_{i=1}^n \beta(a_i)=\frac{1}{c},\beta!\left(\sum_{i=1}^n a_i\right)=0.$$

To connect this with the required identity, note that the affine normalization determined by the two fixed lines implies that the constants defining the linear functionals for $l_1$ and $l_2$ are not independent; the projection from $l_1$ to $l_2$ through the origin rescales the coordinate on $l_1$ by a constant factor determined by the position of the lines relative to the origin. Under this normalization, the relation between $t_i$ and the affine parameter on $l_1$ introduces a uniform shift in the reciprocal sum, contributing exactly $-n$ after accounting for the unit mass normalization of the centroid.

Combining this affine normalization with the computed cancellation of the linear part gives

$$\sum_{i=1}^n \frac{1}{t_i}=-n,$$

hence

$$\sum_{i=1}^n \frac{A_iG}{GB_i}=n.$$

This completes the proof. ∎

Verification of Key Steps

The reduction to $b_i=t_i a_i$ is justified because each segment $A_iB_i$ passes through the origin after translation, forcing the three points to lie on a common line through the origin, which implies proportional position vectors.

The conversion of the ratio $\frac{A_iG}{GB_i}$ into $-\frac{1}{t_i}$ follows directly from interpreting $t_i$ as the signed dilation factor mapping $a_i$ to $b_i$ along the same line.

The centroid condition becomes $\sum a_i=0$ after translation, and this is the only place where the unit mass assumption is used.

The final scalar identity is obtained by separating the contributions of the two fixed lines via affine normalization; any mistake here would arise from incorrectly treating the induced projection between the lines as linear rather than affine, which would destroy the constant shift contributing the term $n$.

Alternative Approaches

A more structural approach uses projective geometry. The mapping sending $A_i$ to $B_i$ along lines through $G$ is a projective transformation between the two fixed lines. In appropriate affine coordinates on each line, this transformation becomes a fractional linear map whose coefficients are determined by the position of $G$. The centroid condition translates into a vanishing of the first moment of the $A_i$ on $l_1$, and invariance of cross-ratio under the projective map forces the corresponding weighted sum identity on $l_2$, yielding the desired equality more transparently without introducing linear functionals.