Maths Olympiad Prep

Library / IMO Shortlist /42 of 365

IMO Shortlist, 2006

Number theory Difficulty 7.9 National Olympiad, round 2 Prove it IMO

Prove that, for every positive integer nn, there exists an integer mm such that 2m+m2^{m}+m is divisible by nn.

Solution

We will prove by induction on dd that, for every positive integer NN, there exist positive integers b0,b1,…,bd−1b_{0}, b_{1}, \ldots, b_{d-1} such that, for each i=0,1,2,…,d−1i=0,1,2, \ldots, d-1, we have bi>Nb_{i}>N and
2bi+bi≡i( mod d). 2^{b_{i}}+b_{i} \equiv i \quad(\bmod d) .
This yields the claim for m=b0m=b_{0}.

The base case d=1d=1 is trivial. Take an a>1a>1 and assume that the statement holds for all d<ad<a. Note that the remainders of 2i2^{i} modulo aa repeat periodically starting with some exponent MM. Let kk be the length of the period; this means that 2M+k′≡2M( mod a)2^{M+k'} \equiv 2^{M}(\bmod a) holds only for those k′k' which are multiples of kk. Note further that the period cannot contain all the aa remainders, since 00 either is missing or is the only number in the period. Thus k<ak<a.

Let d=gcd⁡(a,k)d=\operatorname{gcd}(a, k) and let a′=a/d,k′=k/da'=a / d, k'=k / d. Since 0<k<a0<k<a, we also have 0<d<a0<d<a. By the induction hypothesis, there exist positive integers b0,b1,…,bd−1b_{0}, b_{1}, \ldots, b_{d-1} such that bi>max⁡(2M,N)b_{i}>\max \left(2^{M}, N\right) and
2bi+bi≡i( mod d) for i=0,1,2,…,d−1. \begin{equation*} 2^{b_{i}}+b_{i} \equiv i \quad(\bmod d) \quad \text{ for } \quad i=0,1,2, \ldots, d-1 . \tag{1} \end{equation*}
For each i=0,1,…,d−1i=0,1, \ldots, d-1 consider the sequence
2bi+bi,2bi+k+(bi+k),…,2bi+(a′−1)k+(bi+(a′−1)k). \begin{equation*} 2^{b_{i}}+b_{i}, \quad 2^{b_{i}+k}+\left(b_{i}+k\right), \ldots, \quad 2^{b_{i}+\left(a'-1\right) k}+\left(b_{i}+\left(a'-1\right) k\right) . \tag{2} \end{equation*}
Modulo aa, these numbers are congruent to
2bi+bi,2bi+(bi+k),…,2bi+(bi+(a′−1)k), 2^{b_{i}}+b_{i}, \quad 2^{b_{i}}+\left(b_{i}+k\right), \quad \ldots, \quad 2^{b_{i}}+\left(b_{i}+\left(a'-1\right) k\right),
respectively. The dd sequences contain a′d=aa' d=a numbers altogether. We shall now prove that no two of these numbers are congruent modulo aa.

Suppose that
2bi+(bi+mk)≡2bj+(bj+nk)( mod a) \begin{equation*} 2^{b_{i}}+\left(b_{i}+m k\right) \equiv 2^{b_{j}}+\left(b_{j}+n k\right)(\bmod a) \tag{3} \end{equation*}
for some values of i,j∈{0,1,…,d−1}i, j \in\{0,1, \ldots, d-1\} and m,n∈{0,1,…,a′−1}m, n \in\left\{0,1, \ldots, a'-1\right\}. Since dd is a divisor of aa, we also have
2bi+(bi+mk)≡2bj+(bj+nk)( mod d). 2^{b_{i}}+\left(b_{i}+m k\right) \equiv 2^{b_{j}}+\left(b_{j}+n k\right) \quad(\bmod d) .
Because dd is a divisor of kk and in view of (1), we obtain i≡j( mod d)i \equiv j(\bmod d). As i,j∈{0,1,…,d−1}i, j \in\{0,1, \ldots, d-1\}, this just means that i=ji=j. Substituting this into (3) yields mk≡nk( mod a)m k \equiv n k(\bmod a). Therefore mk′≡nk′( mod a′)m k' \equiv n k'\left(\bmod a'\right); and since a′a' and k′k' are coprime, we get m≡n( mod a′)m \equiv n\left(\bmod a'\right). Hence also m=nm=n.

It follows that the aa numbers that make up the dd sequences (2) satisfy all the requirements; they are certainly all greater than NN because we chose each bi>max⁡(2M,N)b_{i}>\max \left(2^{M}, N\right). So the statement holds for aa, completing the induction.

Want a route through all this instead of an archive? The track puts 2,604 problems in a working order, from Junior Challenge level to the IMO shortlist.

Source: MathNet, licensed CC-BY-4.0. Statement reproduced verbatim; metadata (topic, difficulty) added by this project.