Maths Olympiad Prep

Library / USAMO /102 of 156

USAMO, 2009

Number theory Difficulty 7.5 National Olympiad, round 2 Prove it United States

Let s1,s2,s3,…s_1, s_2, s_3, \dots be an infinite, non-constant sequence of rational numbers, meaning it is not the case that s1=s2=s3=…s_1 = s_2 = s_3 = \dots. Suppose that t1,t2,t3,…t_1, t_2, t_3, \dots is also an infinite, non-constant sequence of rational numbers with the property that (si−sj)(ti−tj)(s_i - s_j)(t_i - t_j) is an integer for all ii and jj. Prove that there exists a rational number rr such that (si−sj)r(s_i - s_j)r and (ti−tj)/r(t_i - t_j)/r are integers for all ii and jj.

Solution

Solution 1 (By Gabriel Carroll). First, we claim there exist i,ji, j such that (si−sj)(ti−tj)≠0(s_i-s_j)(t_i-t_j) \neq 0. Indeed, for any fixed ii, because the sequence s1,s2,…s_1, s_2, \dots is non-constant, there is some jj such that sj≠sis_j \neq s_i. If tj≠tit_j \neq t_i the claim follows, so suppose tj=tit_j = t_i. Because the sequence t1,t2,…t_1, t_2, \dots is nonconstant, there exists kk such that tk≠tit_k \neq t_i. If sk≠sis_k \neq s_i the claim again follows, so suppose sk=sis_k = s_i. Then (sj−sk)(tj−tk)=(sj−si)(ti−tk)≠0(s_j - s_k)(t_j - t_k) = (s_j - s_i)(t_i - t_k) \neq 0, and the claim is proven. We can reorder the pairs (si,ti)(s_i, t_i) relative to each other without affecting either the hypothesis or the conclusion of the problem. So by a suitable reordering, we may assume that (s1−s2)(t1−t2)≠0(s_1 - s_2)(t_1 - t_2) \neq 0.

Second, for any constants aa and bb, we can replace sis_i by si−as_i - a and tit_i by ti−bt_i - b for all ii without affecting either the hypothesis or the conclusion of the problem (since all the differences si−sjs_i - s_j and ti−tjt_i - t_j remain unchanged). In particular, by taking a=s1a = s_1 and b=t1b = t_1, we may assume that s1=t1=0s_1 = t_1 = 0. So we have reduced the problem to the case s1=t1=0s_1 = t_1 = 0, s2≠0s_2 \neq 0, t2≠0t_2 \neq 0.

Call a pair of positive rational numbers (A,B)(A, B) good if ABAB is an integer, and AsjAs_j and BtjBt_j are also integers for all jj.

Third, we show that a good pair exists. We know that for all i≥2i \ge 2, (si−s1)(ti−t1)=siti(s_i - s_1)(t_i - t_1) = s_i t_i is an integer; and for all i,j≥2i, j \ge 2, (si−sj)(ti−tj)=siti−sitj−sjti+sjtj(s_i - s_j)(t_i - t_j) = s_i t_i - s_i t_j - s_j t_i + s_j t_j is an integer, which implies sitj+sjtis_i t_j + s_j t_i is an integer. Write the rational numbers sj,tjs_j, t_j in lowest terms as sj=pj/qjs_j = p_j/q_j and tj=uj/vjt_j = u_j/v_j. We know that, for each jj, sjtj=pjuj/qjvjs_j t_j = p_j u_j / q_j v_j is an integer. Because uju_j is relatively prime to vjv_j, then, pjp_j is divisible by vjv_j, say pj=djvjp_j = d_j v_j for some integer djd_j. We also know that
s2tj+sjt2=p2ujq2vj+pju2qjv2=p2ujqjv2+pju2q2vjq2vjqjv2 s_2 t_j + s_j t_2 = \frac{p_2 u_j}{q_2 v_j} + \frac{p_j u_2}{q_j v_2} = \frac{p_2 u_j q_j v_2 + p_j u_2 q_2 v_j}{q_2 v_j q_j v_2}
is an integer. In particular, qjq_j, being a factor of the denominator, must divide the numerator. But qjq_j divides p2ujqjv2p_2 u_j q_j v_2, so it also divides the other term, pju2q2vj=dju2q2vj2p_j u_2 q_2 v_j = d_j u_2 q_2 v_j^2. Since qjq_j is relatively prime to pj=djvjp_j = d_j v_j, it must divide u2q2u_2 q_2. Moreover, u2q2≠0u_2 q_2 \neq 0, because of our assumption t2≠0t_2 \neq 0. So we have a positive integer A=∣u2q2∣A = |u_2 q_2| such that AsjAs_j is an integer for all jj. Analogously, we can find a positive integer BB such that BtjBt_j is an integer for all jj. This (A,B)(A, B) constitute a good pair, proving existence.

Now we are ready to complete our proof. We know that some good pair exists. We consider a good pair for which the product ABAB is as small as possible. We will show that AB=1AB = 1. Suppose that, for the minimal good pair, AB>1AB > 1; then ABAB has a prime factor pp. If the integer AsiAs_i is divisible by pp for all ii, then we can divide AA by pp and obtain a new good pair (A/p,B)(A/p, B) having a smaller product than before — a contradiction. So for some ii, AsiAs_i is not divisible by pp. Then BtiBt_i must be divisible by pp, because sitis_i t_i is an integer and so (Asi)(Bti)=(AB)(siti)(As_i)(Bt_i) = (AB)(s_i t_i) is an integer divisible by pp. Likewise, there exists some jj such that BtjBt_j is not divisible by pp, but AsjAs_j is. Now write
(AB)(sitj+sjti)−(Asj)(Bti)=(Asi)(Btj). (AB)(s_i t_j + s_j t_i) - (As_j)(Bt_i) = (As_i)(Bt_j).
All the parenthesized factors are integers, and the left-hand side is divisible by pp, but the right-hand side is not. This contradiction completes the proof that the minimal good pair satisfies AB=1AB = 1.

But now take the minimal good pair (A,B)(A, B), and let r=Ar = A. We have that sir=Asis_i r = As_i and ti/r=Btit_i / r = Bt_i are integers for all ii, from which our desired conclusion follows.

Solution 2 (By Lenhard Ng). For pp a prime, define the pp-adic norm ∥⋅∥p\| \cdot \|_p on rational numbers as follows: for r≠0r \neq 0, ∥r∥p\|r\|_p is the unique integer nn for which we can write r=pna/br = p^n a/b with a,ba, b integers not divisible by pp. (By convention, ∥0∥p=+∞\|0\|_p = +\infty.) We will repeatedly use the well-known (or easy to prove) fact that for any rational numbers r1,r2r_1, r_2, we have ∥r1±r2∥p≥min⁡(∥r1∥p,∥r2∥p)\|r_1 \pm r_2\|_p \ge \min(\|r_1\|_p, \|r_2\|_p), with equality whenever ∥r1∥p≠∥r2∥p\|r_1\|_p \neq \|r_2\|_p. The condition of the problem implies that
∥si−sj∥p≥−∥ti−tj∥p(1) \|s_i - s_j\|_p \ge -\|t_i - t_j\|_p \qquad (1)
for all i,ji, j and all primes pp.

We claim in fact that
∥si−sj∥p≥−∥tk−tl∥p \|s_i - s_j\|_p \geq -\|t_k - t_l\|_p
for all i,j,k,li, j, k, l and all prime pp. Suppose otherwise; then there exist i,j,k,l,pi, j, k, l, p for which ∥si−sj∥p<−∥tk−tl∥p\|s_i - s_j\|_p < -\|t_k - t_l\|_p. Since ∥si−sj∥p=∥(si−sk)−(sj−sk)∥p≥min⁡(∥si−sk∥p,∥sj−sk∥p)\|s_i - s_j\|_p = \|(s_i - s_k) - (s_j - s_k)\|_p \geq \min(\|s_i - s_k\|_p, \|s_j - s_k\|_p), at least one of ∥si−sk∥p\|s_i - s_k\|_p and ∥sj−sk∥p\|s_j - s_k\|_p, say the former, is strictly less than −∥tk−tl∥p-\|t_k - t_l\|_p. By (1), it follows that ∥ti−tk∥p>∥tk−tl∥p\|t_i - t_k\|_p > \|t_k - t_l\|_p, and thus ∥ti−tl∥p=∥(ti−tk)+(tk−tl)∥p=∥tk−tl∥p\|t_i - t_l\|_p = \|(t_i - t_k) + (t_k - t_l)\|_p = \|t_k - t_l\|_p. Then by (1) again, ∥si−sl∥p≥−∥tk−tl∥p\|s_i - s_l\|_p \geq -\|t_k - t_l\|_p and ∥sk−sl∥p≥−∥tk−tl∥p\|s_k - s_l\|_p \geq -\|t_k - t_l\|_p, whence ∥si−sk∥p=∥(si−sl)−(sk−sl)∥p≥−∥tk−tl∥p\|s_i - s_k\|_p = \|(s_i - s_l) - (s_k - s_l)\|_p \geq -\|t_k - t_l\|_p, contradicting the assumption that ∥si−sk∥p<−∥tk−tl∥p\|s_i - s_k\|_p < -\|t_k - t_l\|_p. This proves the claim.

Now for each prime pp, define the integer f(p)=min⁡i,j∥si−sj∥pf(p) = \min_{i,j} \|s_i - s_j\|_p. Choose i0,j0,k0,l0i_0, j_0, k_0, l_0 such that si0≠sj0s_{i_0} \neq s_{j_0} and tk0≠tl0t_{k_0} \neq t_{l_0}; then f(p)f(p) exists since it is bounded below by −∥tk0−tl0∥p-\|t_{k_0} - t_{l_0}\|_p (by the claim) and above by ∥si0−sj0∥p\|s_{i_0} - s_{j_0}\|_p. Moreover, if pp does not divide the numerator or denominator of either si0−sj0s_{i_0} - s_{j_0} or tk0−tl0t_{k_0} - t_{l_0}, then ∥si0−sj0∥p=∥tk0−tl0∥p=0\|s_{i_0} - s_{j_0}\|_p = \|t_{k_0} - t_{l_0}\|_p = 0 and thus f(p)=0f(p) = 0. It follows that f(p)=0f(p) = 0 for all but finitely many primes.

We can now define r=∏pp−f(p)r = \prod_p p^{-f(p)}, where the product is over all primes. For any i,ji, j, we have ∥si−sj∥p≥f(p)\|s_i - s_j\|_p \geq f(p) for all pp by construction, and so (si−sj)r(s_i - s_j)r is an integer. On the other hand, for any k,lk, l and any prime pp, ∥tk−tl∥p≥−∥si−sj∥p\|t_k - t_l\|_p \geq -\|s_i - s_j\|_p for all i,ji, j by the claim, and so ∥tk−tl∥p≥−f(p)\|t_k - t_l\|_p \geq -f(p). It follows that (tk−tl)/r(t_k - t_l)/r is an integer for all k,lk, l, whence rr is the desired rational number.

Solution 3 (Based on work by Evan O'Dorney). As in the first solution, we reduce to the case s1=t1=0s_1 = t_1 = 0 and conclude that sitj+sjtis_i t_j + s_j t_i is an integer for i,j≥2i, j \geq 2. For each positive integer n≥2n \geq 2, let SnS_n be the set of rational numbers rr such that sirs_i r and ti/rt_i/r are integers for i=2,3,…,ni = 2, 3, \dots, n. Clearly, we have Sn+1⊆SnS_{n+1} \subseteq S_n. On the other hand, since neither sequence is constant, there exists some nn such that sis_i and tjt_j are nonzero for some indices i,ji, j in {2,3,…,n}\{2, 3, \dots, n\}. If we write si=a/bs_i = a/b and tj=c/dt_j = c/d in lowest terms, then for any r=e/f∈Snr = e/f \in S_n written in lowest terms, ff must divide aa and ee must divide cc. This proves that SnS_n is a finite set, and therefore so is SkS_k for k>nk > n.

Now form the polynomials
P(x)=s2+s3x+⋯+snxn−1andQ(x)=t2+t3x+⋯+tnxn−1 P(x) = s_2 + s_3 x + \dots + s_n x^{n-1} \quad \text{and} \quad Q(x) = t_2 + t_3 x + \dots + t_n x^{n-1}
and note that P(x)Q(x)P(x)Q(x) has integer coefficients (because its coefficients are sums of terms of the form sitj+sjtis_i t_j + s_j t_i and sitis_i t_i). By Gauss' lemma, there exists a rational number rr such that P(x)rP(x)r and Q(x)/rQ(x)/r have integer coefficients, so r∈Snr \in S_n and SnS_n is non-empty.

We now have a decreasing sequence of finite sets S2⊇S3⊇…S_2 \supseteq S_3 \supseteq \dots, each of which is non-empty. Their intersection must then be non-empty (e.g., because there is a least cardinality, and once that is achieved the sets are all equal). For any rr in that intersection, then, we have that sirs_i r and ti/rt_i/r are integers for all ii, giving the desired conclusion.

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.