Kvant Math Problem 408

Let the original rectangles have side lengths $a$ and $b$, and let their ratio be $r = \frac{a}{b} > 0$.

kvantmathematicsolympiad

Verified: no
Verdicts: UNKNOWN + UNKNOWN
Solve time: 9m21s
Source on kvant.digital

Problem

From 30 congruent rectangles, a rectangle similar to the original ones is constructed. What can be the ratio of the side lengths of this rectangle?

P. Pankov

Exploration

Let the original rectangles have side lengths $a$ and $b$, and let their ratio be $r = \frac{a}{b} > 0$. The large rectangle is similar to the small ones, so its sides are proportional to $a$ and $b$, hence can be written as $ka$ and $kb$ for some $k>0$.

Since the large rectangle is composed of $30$ congruent copies of the small rectangle, its area is $30ab$. On the other hand, its area is also $k^2 ab$, so $k^2 = 30$ and $k = \sqrt{30}$.

Thus the large rectangle must have side lengths $\sqrt{30}a$ and $\sqrt{30}b$. The tiling therefore partitions a $\sqrt{30}a \times \sqrt{30}b$ rectangle into $30$ copies of $a \times b$, possibly rotated.

The key structural restriction is that each small rectangle contributes either an $a$-segment or a $b$-segment along each side of the large rectangle depending on its orientation. This forces a linear relation between $a$ and $b$ coming from counting how many rectangles are placed in each orientation.

Trying small configurations suggests splitting the $30$ rectangles into those aligned in the same direction and those rotated. The main difficulty is to ensure that both side lengths of the big rectangle are consistent with the same mixture.

The central suspicion is that any imbalance between orientations destroys similarity unless $a=b$.

Problem Understanding

This is a Type A problem: determine all possible ratios of side lengths of the original rectangles such that $30$ congruent copies tile a rectangle similar to them.

We are asked for all possible values of $r=\frac{a}{b}$. The geometric constraint of similarity combined with tiling forces strong arithmetic restrictions on how side lengths decompose along the boundary of the large rectangle. The expectation is that only the square case survives consistency conditions.

Proof Architecture

Let $a,b>0$ be the side lengths of the small rectangle and $r=\frac{a}{b}$.

Let $n$ be the number of rectangles placed without rotation and $30-n$ the number rotated.

First lemma establishes expressions for the side lengths of the large rectangle in terms of $n,a,b$.

Second lemma enforces similarity between the large rectangle and the small one, producing an equation relating $r$ and $n$.

Third lemma solves the resulting equation and checks all admissible cases $n\in{0,1,\dots,30}$.

The most delicate step is verifying that boundary decomposition forces linear sums rather than more complicated arrangements, and ensuring both similarity directions are treated.

Solution

Let the small rectangles have sides $a$ and $b$, and define $r=\frac{a}{b}>0$. The large rectangle is composed of $30$ congruent copies of these rectangles.

Each small rectangle is placed either with side $a$ parallel to a fixed side of the large rectangle or rotated so that side $b$ is parallel to that direction. Let $n$ denote the number of rectangles in the first orientation and $30-n$ the number in the rotated orientation.

Along one side of the large rectangle, the contributions of rectangles are additive. In the first orientation, the projection contributes length $a$ in one direction and $b$ in the perpendicular direction. In the rotated orientation, these contributions are swapped. Therefore the side lengths $X$ and $Y$ of the large rectangle satisfy

$$X = na + (30-n)b, \quad Y = nb + (30-n)a.$$

Since the large rectangle is similar to the small one, its side ratio equals either $r$ or $\frac{1}{r}$. Thus

$$\frac{X}{Y} = r \quad \text{or} \quad \frac{X}{Y} = \frac{1}{r}.$$

First consider

$$\frac{na + (30-n)b}{nb + (30-n)a} = \frac{a}{b}.$$

Dividing numerator and denominator by $b$ yields

$$\frac{nr + (30-n)}{n + (30-n)r} = r.$$

Cross-multiplication gives

$$nr + 30 - n = rn + r(30-n)r = rn + r^2(30-n).$$

Subtracting $rn$ from both sides yields

$$30 - n = r^2(30-n).$$

If $n \neq 30$, division by $30-n$ gives $r^2=1$, hence $r=1$ since $r>0$.

If $n=30$, all rectangles are unrotated. Then $X=30a$ and $Y=30b$, so $\frac{X}{Y}=\frac{a}{b}=r$. However similarity would require the scale factor $k$ to satisfy simultaneously $X=ka$, $Y=kb$, giving $k=30$. The area of the large rectangle would then be $900ab$, contradicting the fact that it must equal $30ab$. Hence $n=30$ is impossible.

Now consider

$$\frac{nr + (30-n)}{n + (30-n)r} = \frac{1}{r}.$$

Cross-multiplying gives

$$r(nr + 30 - n) = n + (30-n)r.$$

Expanding,

$$nr^2 + r(30-n) = n + r(30-n).$$

Subtracting $r(30-n)$ from both sides yields

$$nr^2 = n.$$

If $n\neq 0$, dividing by $n$ gives $r^2=1$, hence $r=1$.

If $n=0$, all rectangles are rotated, and the same area contradiction arises as before, since the scaling factor would again be $30$ instead of $\sqrt{30}$.

Thus in all admissible cases the only possible ratio is $r=1$.

Therefore the original rectangles must be squares, and the ratio of their side lengths equals

$$\boxed{1}.$$

Verification of Key Steps

The expression for $X$ and $Y$ follows directly from decomposing the boundary into disjoint contributions of rectangles classified by orientation; no overlapping or missing segments occur because the tiling partitions the boundary into full side-length segments of each rectangle.

In the first equation, the cancellation leading to $30-n = r^2(30-n)$ does not discard solutions because the exceptional case $n=30$ was checked separately and shown inconsistent with area constraints.

In the second equation, the cancellation of $r(30-n)$ is valid since both sides contain identical linear contributions from rotated rectangles, and the remaining equation reduces cleanly to $nr^2=n$, with boundary cases $n=0$ excluded by the same area mismatch.

The possibility $r=-1$ is excluded by positivity of side lengths.

Alternative Approaches

A different approach uses a lattice model: interpreting each rectangle as a vector contribution in a decomposition of a $\sqrt{30}\times \sqrt{30}$ scaled grid. One then shows that consistency of integer tilings forces equal side lengths, reducing the problem to a Diophantine constraint on the aspect ratio. This method is more geometric but ultimately reduces to the same linear system on orientation counts, while the present argument is more direct and avoids coordinate embedding.