Maths Olympiad Prep

Library / /49 of 115

Geometry Difficulty 7.2 National olympiad, round 2 Find the answer

( Ricky Liu ) For what values of k>0k > 0 is it possible to dissect a 1×k1 \times k rectangle into two similar, but incongruent, polygons?

A number or a short expression. Spacing and $ signs are ignored.

Solution

We will show that a dissection satisfying the requirements of the problem is possible if and only if k1k\neq 1 .
We first show by contradiction that such a dissection is not possible when k=1k = 1 . Assume that we have such a dissection. The common boundary of the two dissecting polygons must be a single broken line connecting two points on the boundary of the square (otherwise either the square is subdivided in more than two pieces or one of the polygons is inside the other). The two dissecting polygons must have the same number of vertices. They share all the vertices on the common boundary, so they have to use the same number of corners of the square as their own vertices. Therefore, the common boundary must connect two opposite sides of the square (otherwise one of the polygons will contain at least three corners of the square, while the other at most two). However, this means that each of the dissecting polygons must use an entire side of the square as one of its sides, and thus each polygon has a side of length 1. A side of longest length in one of the polygons is either a side on the common boundary or, if all those sides have length less than 1, it is a side of the square. But this is also true of the polygon, which means that the longest side length in the two polygons is the same. This is impossible since they are similar but not congruent, so we have a contradiction.
We now construct a dissection satisfying the requirements of the problem when k1k\neq 1 . Notice that we may assume that k>1k > 1 , because a 1×k1\times k rectangle is similar to a 1×1k1\times\frac{1}{k} rectangle.
We first construct a dissection of an appropriate chosen rectangle (denoted by ABCDABCD below) into two similar noncongruent polygons. The construct depends on two parameters ( nn and rr below). By appropriate choice of these parameters we show that the constructed rectangle can be made similar to a 1×k1\times k rectangle, for any k>1k > 1 . The construction follows.
Let r>1r > 1 be a real number. For any positive integer nn , consider the following sequence of 2n+22n + 2 points: A0=(0,0),A1=(1,0),A2=(1,r),A3=(1+r2,r),A4=(1+r2,r+r3),A5=(1+r2+r4,r+r3),A_0 = (0,0), A_1 = (1,0), A_2 = (1,r), A_3 = (1 + r^2, r), \\ A_4 = (1 + r^2, r + r^3), A_5 = (1 + r^2 + r^4, r + r^3), and so on, until A2n+1=(1+r2+r4++r2n,r+r3+r5++r2n1).A_{2n+1} = (1 + r^2 + r^4 + \cdots + r^{2n}, r + r^3 + r^5 + \cdots + r^{2n - 1}). Define a rectangle ABCDABCD by A=A0,B=(1+r2++r2n,0),C=A2n+1, and D=(0,r+r3++r2n1).A = A_0, B = (1 + r^2 + \cdots + r^{2n}, 0), C = A_{2n + 1}, \text{ and }D = (0, r + r^3 + \cdots + r^{2n - 1}). The sides of the (2n+2)(2n+2) -gon A1A2A2n+1BA_1A_2\ldots A_{2n+1}B have lengths r,r2,r3,,r2n,r+r3+r5++r2n1,r2+r4+r6++r2n,r, r^2, r^3, \ldots, r^{2n}, r + r^3 + r^5 + \cdots + r^{2n-1}, r^2 + r^4 + r^6 + \cdots + r^{2n}, and the sides of the (2n+2)(2n+2) -gon A0A1A2A2nDA_0A_1A_2\ldots A_{2n}D have lengths 1,r,r2,,r2n1,1+r2+r4++r2n2,r+r3+r5++r2n1,1, r, r^2, \ldots, r^{2n-1}, 1 + r^2 + r^4 + \cdots + r^{2n-2}, r + r^3 + r^5 + \cdots + r^{2n-1}, respectively. These two polygons dissect the rectangle ABCDABCD and, apart from orientation, it is clear that they are similar but noncongruent, with coefficient of similarity r>1r > 1 . The rectangle ABCDABCD and its dissection are thus constructed.
The rectangle ABCDABCD is similar to a rectangle of size 1×fn(r)1\times f_n(r) , where fn(r)=1+r2++r2nr+r3++r2n1.f_n(r) = \frac{1 + r^2 + \cdots + r^{2n}}{r + r^3 + \cdots + r^{2n-1}}. It remains to show that fn(r)f_n(r) can have any value k>1k > 1 for appropriate choices of nn and rr . Choose nn sufficiently large so that 1+1n<k1 + \frac{1}{n} < k . Since fn(1)=1+1n<k<k1+k2++k2nk2+k4++k2n=fn(k)f_n(1) = 1 + \frac{1}{n} < k < k\frac{1 + k^2 + \cdots + k^{2n}}{k^2 + k^4 + \cdots + k^{2n}} = f_n(k) and fn(r)f_n(r) is a continuous function for positive rr , there exists an rr such that 1<r<k1 < r < k and fn(r)=kf_n(r) = k , so we are done.
Alternate solutions are always welcome. If you have a different, elegant solution to this problem, please add it to this page.

Want a route through all this instead of an archive? The track puts 2,000 problems in a working order, from AMC 10 level to the IMO shortlist.

Source: Omni-MATH, licensed Apache-2.0. Statement and solution reproduced as published; topic and difficulty added by this site.