Maths Olympiad Prep

Track / Stage 8 / 130 of 180 #2470 of 2604

Problem 2470

IMO Shortlist mid-range; USAMO P2/P5
Algebra Difficulty 8.5 Prove it International Mathematical Olympiad · United States

Let P(x)P(x) be a polynomial of degree n>1n > 1 with integer coefficients and let kk be a positive integer. Consider the polynomial
Q(x)=P(P(…(P(x)… )))⏟k P’s Q(x) = \underbrace{P(P(\dots(P(x)\dots)))}_{k \text{ P's}}
Prove that there are at most nn integers tt such that Q(t)=tQ(t) = t.

This one wants a proof. Work it on paper, then read the official solution and mark yourself. Be honest about it: the record is only any use to you if it is.

Next problem →

Official solution

Let Z\mathbb{Z} denote the set of integers. We define
SP={t∣t∈Z and P(t)=t}andSQ={t∣t∈Z and Q(t)=t}. S_P = \{t \mid t \in \mathbb{Z} \text{ and } P(t) = t\} \quad \text{and} \quad S_Q = \{t \mid t \in \mathbb{Z} \text{ and } Q(t) = t\}.
Clearly, SPS_P is a subset of SQS_Q. Also note that there are at most nn elements in SPS_P. This is so because that t∈SPt \in S_P if and only if tt is a root of polynomial P(x)−x=0P(x) - x = 0 of degree nn, which has at most nn roots. If SQ=SPS_Q = S_P, we have nothing to prove. We assume that SPS_P is a proper subset of SQS_Q, and that t∈SQt \in S_Q but t∉SPt \notin S_P.
Consider the sequence {ti}i=0∞\{t_i\}_{i=0}^{\infty} with t0=tt_0 = t, ti+1=P(ti)t_{i+1} = P(t_i) for every nonnegative integer ii. Since t∈SQt \in S_Q, tk=Q(t0)=Q(t)=t=t0t_k = Q(t_0) = Q(t) = t = t_0.
Since the polynomial a−ba - b divides the polynomial am−bma^m - b^m (where mm is a nonnegative integer), it is not difficult to see that the polynomial a−ba - b divides the polynomial P(a)−P(b)P(a) - P(b), where P(x)P(x) is a polynomial with integer coefficients. In our current problem, then, we can conclude that the integer sequence {ti}i=0∞\{t_i\}_{i=0}^{\infty} satisfies the following sequence of divisibility relations
(ti+1−ti)∣(P(ti+1)−P(ti))=ti+2−ti+1 (t_{i+1} - t_i) \mid (P(t_{i+1}) - P(t_i)) = t_{i+2} - t_{i+1}
for every nonnegative integer ii. Since tk+1−tk=t1−t0=P(t)−t≠0t_{k+1} - t_k = t_1 - t_0 = P(t) - t \ne 0, each term in the chain of differences
t1−t0,t2−t1,…,tk−tk−1,tk+1−tkt_1 - t_0, t_2 - t_1, \dots, t_k - t_{k-1}, t_{k+1} - t_k
is a nonzero divisor of the next one, and since tk+1−tk=t1−t0t_{k+1} - t_k = t_1 - t_0, all these differences have equal absolute values. Let ti=max⁡{t0,t1,…,tk}t_i = \max\{t_0, t_1, \dots, t_k\}. Then ti−1−ti=−(ti−ti+1)t_{i-1} - t_i = -(t_i - t_{i+1}), or ti−1=ti+1t_{i-1} = t_{i+1}. It is then not difficult to see that ti+2=tit_{i+2} = t_i for every ii; that is,
t1=P(t0)andt0=P(t1)orP(P(t0))=t0. t_1 = P(t_0) \quad \text{and} \quad t_0 = P(t_1) \quad \text{or} \quad P(P(t_0)) = t_0.

Therefore,
SQ={t∣t∈Z and P(P(t))=t}. S_Q = \{t \mid t \in \mathbb{Z} \text{ and } P(P(t)) = t\}.
Without loss of generality, we may assume that t0<t1t_0 < t_1. If s0s_0 is another element in SQS_Q, let s1=P(s0)s_1 = P(s_0). (It is possible that s0∈SPs_0 \in S_P; that is, s1=s0s_1 = s_0.) We further assume without loss of generality that s0<s1s_0 < s_1 and t0<s0t_0 < s_0; that is, t0<s0≤s1t_0 < s_0 \le s_1 and t0<t1t_0 < t_1. Note that s1−t0s_1 - t_0 divides P(s1)−P(t0)=s0−t1P(s_1) - P(t_0) = s_0 - t_1. We must have t0<s0<s1<t1t_0 < s_0 < s_1 < t_1. Note that s0−t1s_0 - t_1 also divides P(s0)−P(t1)=s1−t0P(s_0) - P(t_1) = s_1 - t_0, it follows that s0−t1=−(s1−t0)s_0 - t_1 = -(s_1 - t_0); that is,
t0+t1=s0+s1=s0+P(s0). t_0 + t_1 = s_0 + s_1 = s_0 + P(s_0).
In other words, s0s_0 is a root of the polynomial P(x)+x=t0+t1P(x) + x = t_0 + t_1. Since P(x)+xP(x) + x has degree nn, there are at most nn (integer) roots (including t0t_0) of P(x)+xP(x) + x. Hence there are at most nn elements in SQS_Q, completing our proof.

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