Ivan Shishkin, Rye (1878)

Problems/General algebraUnreviewed

Polynomial Functional Equation

by Uettechat·
60
Difficulty scaleÉchelle de difficulté

This score reflects both the level of the required concepts and the difficulty of the solution.Ce score tient compte à la fois du niveau des notions nécessaires et de la difficulté de la résolution.

  1. 1–10First steps / middle schoolPremiers pas / collège
  2. 11–25Beginner / high schoolDébutant / lycée
  3. 26–50Intermediate / undergraduateIntermédiaire / licence
  4. 51–70Advanced / graduateAvancé / master
  5. 71–90Expert / specializedExpert / spécialisé
  6. 91–100Research levelNiveau recherche
These levels are approximate guides.Ces niveaux sont des repères approximatifs.
·
English
EnglishFrançais
Unreviewed. This problem has not been reviewed by trusted users yet.

Find all polynomials P∈R[X]P \in \mathbb{R}[X] satisfying the relation:
P(X)P(X+1)=P(X2+X+1)\quad P(X)P(X+1) = P(X^2+X+1)

I solved itMark it doneAdd to my listKeep it in your list

References

  1. Equation fonctionelle de Shapiro / Oral X-ENS Cassini
Details

Download: BibTeXJSON

Hints

3

Hint 1

Open this only if you want a small nudge before looking at the solutions.

Solutions

2
Reveal solutionsAre you sure? Give it a try first.

Solution by Uettechat

Discussions0 useful votes

Let P∈R[X]P \in \mathbb{R}[X] be a polynomial satisfying the equation:
P(X)P(X+1)=P(X2+X+1)P(X)P(X+1) = P(X^2+X+1)

  1. Case where PP is constant:
    Let λ∈R\lambda \in \mathbb{R} such that P=λP = \lambda.
    The equation becomes λ2=λ\lambda^2 = \lambda, which implies λ∈{0,1}\lambda \in \{0, 1\}.
    Thus, the constant solutions are P=0P = 0 and P=1P = 1.

  2. Case where PP is non-constant:
    Let z∈Cz \in \mathbb{C} be a root of PP. Evaluating the functional equation at X=zX = z and at X=z−1X = z - 1 yields:

P(z)P(z+1)=0  ⟹  P(z2+z+1)=0P(z)P(z+1) = 0 \implies P(z^2+z+1) = 0 (1)
P(z−1)P(z)=0  ⟹  P((z−1)2+(z−1)+1)=0  ⟹  P((z−1)2+z)=0P(z-1)P(z) = 0 \implies P((z-1)^2+(z-1)+1) = 0 \implies P((z-1)^2+z) = 0 (2)

Therefore, z2+z+1z^2+z+1 and (z−1)2+z(z-1)^2+z are also roots of PP.

We will now study the root of maximal modulus:
Let ZZ denote the set of (complex) roots of PP, and let λ∈Z\lambda \in Z be a root of maximal modulus.
Applying (1) and (2) to z=λz = \lambda, the numbers λ2+λ+1\lambda^2+\lambda+1 and (λ−1)2+λ(\lambda-1)^2+\lambda are also roots of PP. By the maximality of ∣λ∣\vert{}\lambda\vert{}:
∣λ2+λ+1∣≤∣λ∣and∣(λ−1)2+λ∣≤∣λ∣\vert{}\lambda^2+\lambda+1\vert{} \le \vert{}\lambda\vert{} \quad \text{and} \quad \vert{}(\lambda-1)^2+\lambda\vert{} \le \vert{}\lambda\vert{}By the triangle inequality:
2∣λ∣=∣(λ2+λ+1)−((λ−1)2+λ)∣≤∣λ2+λ+1∣+∣(λ−1)2+λ∣≤2∣λ∣2\vert{}\lambda\vert{} = \vert{}(\lambda^2+\lambda+1) - ((\lambda-1)^2+\lambda)\vert{} \le \vert{}\lambda^2+\lambda+1\vert{} + \vert{}(\lambda-1)^2+\lambda\vert{} \le 2\vert{}\lambda\vert{}Since the bounds are equal, the case of equality in the triangle inequality implies that there exists a real number a≥0a \ge 0 such that:
λ2+λ+1=−a((λ−1)2+λ)\lambda^2+\lambda+1 = -a((\lambda-1)^2+\lambda)Furthermore, the equality case forces both λ2+λ+1\lambda^2+\lambda+1 and (λ−1)2+λ(\lambda-1)^2+\lambda to have modulus exactly ∣λ∣\vert{}\lambda\vert{}. This gives:
∣λ∣=∣−a((λ−1)2+λ)∣=a∣λ∣\vert{}\lambda\vert{} = \vert{}-a((\lambda-1)^2+\lambda)\vert{} = a\vert{}\lambda\vert{}If λ=0\lambda = 0, the initial inequality ∣02+0+1∣≤∣0∣\vert{}0^2+0+1\vert{} \le \vert{}0\vert{} yields 1≤01 \le 0, which is absurd. Thus λ≠0\lambda \neq 0, which implies a=1a = 1. The equation then becomes:
λ2+λ+1=−((λ−1)2+λ)=−λ2+λ−1\lambda^2+\lambda+1 = -((\lambda-1)^2+\lambda) = -\lambda^2+\lambda-1Simplifying gives 2λ2+2=02\lambda^2+2 = 0, which yields λ∈{i,−i}\lambda \in \{i, -i\}.
Thus, necessarily λ∈{i,−i}\lambda \in \{i, -i\}, and in particular ∣λ∣=1\vert{}\lambda\vert{} = 1.

Now let us show that PP has no other complex roots besides ii and −i-i.

Let z∈Zz \in Z be any complex root of PP. We construct the sequence (un)n∈N(u_n)_{n \in \mathbb{N}} defined by:
u0=zandun+1=un2+un+1u_0 = z \quad \text{and} \quad u_{n+1} = u_n^2 + u_n + 1Since P(z)=0  ⟹  P(z2+z+1)=0P(z) = 0 \implies P(z^2+z+1) = 0, an immediate induction shows that un∈Zu_n \in Z for all n∈Nn \in \mathbb{N}.
Because PP is non-constant, the set of roots ZZ is finite. By the Pigeonhole Principle, there exist indices m<nm < n such that um=unu_m = u_n. Therefore, the sequence (un)n∈N(u_n)_{n \in \mathbb{N}} is eventually periodic and enters a cycle of length p≥1p \ge 1.
Let N≥0N \ge 0 be the smallest index such that uNu_N belongs to this cycle, meaning N=min⁡{n∈N∣un+p=un}N = \min \{ n \in \mathbb{N} \mid u_{n+p} = u_n \}.
For any k∈Nk \in \mathbb{N}, write uk=xk+iyku_k = x_k + i y_k with xk,yk∈Rx_k, y_k \in \mathbb{R}. We have:
xk+1=Re(uk+1)=xk2−yk2+xk+1x_{k+1} = \text{Re}(u_{k+1}) = x_k^2 - y_k^2 + x_k + 1Since ∣λ∣=1\vert{}\lambda\vert{} = 1 is the maximal modulus among all roots in ZZ, every root uk∈Zu_k \in Z satisfies ∣uk∣≤1\vert{}u_k\vert{} \le 1. Thus, xk2+yk2≤1x_k^2 + y_k^2 \le 1, which implies −yk2≥xk2−1-y_k^2 \ge x_k^2 - 1. Substituting this yields:
xk+1≥xk2+(xk2−1)+xk+1=2xk2+xkx_{k+1} \ge x_k^2 + (x_k^2 - 1) + x_k + 1 = 2x_k^2 + x_kSumming this inequality over one period of the cycle (from k=Nk = N to N+p−1N + p - 1):
∑k=NN+p−1xk+1≥∑k=NN+p−1(2xk2+xk)\sum_{k=N}^{N+p-1} x_{k+1} \ge \sum_{k=N}^{N+p-1} (2x_k^2 + x_k)Since uN+p=uNu_{N+p} = u_N, the sums ∑k=NN+p−1xk+1\sum_{k=N}^{N+p-1} x_{k+1} and ∑k=NN+p−1xk\sum_{k=N}^{N+p-1} x_k are equal, so subtracting ∑xk\sum x_k from both sides gives:
0≥2∑k=NN+p−1xk2≥00 \ge 2 \sum_{k=N}^{N+p-1} x_k^2 \ge 0This forces xk=0x_k = 0 for all elements uku_k in the cycle (k≥Nk \ge N). For any element uku_k in the cycle (k≥Nk \ge N), since xk=0x_k = 0 and xk+1=0x_{k+1} = 0, the relation xk+1=xk2−yk2+xk+1x_{k+1} = x_k^2 - y_k^2 + x_k + 1 becomes:
0=0−yk2+0+1  ⟹  yk2=1  ⟹  yk∈{−1,1}0 = 0 - y_k^2 + 0 + 1 \implies y_k^2 = 1 \implies y_k \in \{-1, 1\}Hence, uk∈{i,−i}u_k \in \{i, -i\} for all k≥Nk \ge N.

Finally, we prove by backwards induction that uk∈{i,−i}u_k \in \{i, -i\} for all k≥0k \ge 0:
Suppose uk+1∈{i,−i}u_{k+1} \in \{i, -i\}.
Then, uk+1=uk2+uk+1=iu_{k+1} = u_k^2 + u_k + 1 = i or uk+1=uk2+uk+1=−iu_{k+1} = u_k^2 + u_k + 1=-i
We have then uk∈{−1−i,−1+i,−i,i}u_k \in \{-1-i, -1+i, -i, i\}, since ∣uk∣≤1\vert{}u_k\vert{} \leq 1, we have uk∈{−i,i}u_k \in \{-i, i\}

By induction from NN down to 00, we conclude that u0=z∈{i,−i}u_0 = z \in \{i, -i\}.

We have thus shown that every complex root of PP must be ii or −i-i.

Since PP has real coefficients, its non-real complex roots must come in conjugate pairs. Since ii and −i-i are complex conjugates of each other, PP must be of the form (up to a multiplicative constant):
P(X)=c(X2+1)nwith n∈N,c∈RP(X) = c(X^2+1)^n \quad \text{with } n \in \mathbb{N}, c \in \mathbb{R}The leading coefficients of P(X)P(X+1)P(X)P(X+1) and P(X2+X+1)P(X^2+X+1) must match. Identifying them gives c2=c  ⟹  c∈{0,1}c^2 = c \implies c \in \{0, 1\}. Combining this with the constant case gives:
S⊂{0,1}∪{(X2+1)n∣n∈N}S \subset \{0, 1\} \cup \{(X^2+1)^n \mid n \in \mathbb{N}\}

Conversely, if P=0P = 0 or P=1P = 1, the equation is trivially satisfied.
If P(X)=(X2+1)nP(X) = (X^2+1)^n for some n∈Nn \in \mathbb{N}, we verify:
P(X)P(X+1)=(X2+1)n((X+1)2+1)n=((X2+1)(X2+2X+2))nP(X)P(X+1) = (X^2+1)^n ((X+1)^2+1)^n = ((X^2+1)(X^2+2X+2))^nA direct calculation shows that (X2+1)(X2+2X+2)=(X2+X+1)2+1(X^2+1)(X^2+2X+2) = (X^2+X+1)^2+1, so:
P(X)P(X+1)=((X2+X+1)2+1)n=P(X2+X+1)P(X)P(X+1) = ((X^2+X+1)^2+1)^n = P(X^2+X+1)Thus, these polynomials are indeed solutions.

Conclusion:
The set of solutions is
S={0,1}∪{(X2+1)n∣n∈N}S = \{0, 1\} \cup \{(X^2+1)^n \mid n \in \mathbb{N}\}

Solution by Venus2FR

Discussions0 useful votes

soit P un polynome non constant solution. soit aa racine de PP de module maximal. comme P est non constant, en particulier il n’est pas le polynome nul, on en déduit que aa est different de 00, donc inversible (c’est à dire qu’on peut diviser par aa). on vérifie que (X−1)2+(X−1)+1=X2−X+1(X-1)^{2}+(X-1)+1=X^2-X+1 de sorte que l’équation fonctionnelle donne alors : a2+a+1a^{2}+a+1 et a2−a+1a^{2}-a+1 sont deux racines de PP chacune de module inférieur ou égal à module de aa. On pose alors u:=a+1/au:=a+1/a. les deux inégalités précédentes donnent alors (comme aa est inversibl), ∣u−1∣≤1|u-1|\le 1 et ∣u+1∣≤1|u+1|\le 1 et donc uu est dans l’intersection des deux disques du plan complexe de centres respectifs les affixes 1 et -1 et de rayon 1, donc u=0u=0 c’est-à-dire a=ia=i ou a=−ia=-i ainsi si un tel P existe (non contant) il possède i ou -i comme solution mais alors, comme P est un polynome réel, i et -i sont nécessairement solutions de même multiplicité et donc il existe un entier natutel non nul nn et un polynôme réel QQ tel que P=(X2+1)n.QP=(X^{2}+1)^n.Q et i (donc aussi -i) non racine de Q. On vérifie que Q satisfait l’équation fonctionnelle. Si Q est non contant alors d’apres ce qui précède i ou -i est racine de Q, ce qui est exclu. Donc Q est un polynome constant, qui par ailleurs est non nul pusiqu’on a suppose P non constant. Donc P est de la forme P=c(X2+1)nP=c(X^2+1)^n avec c nombre complexe non nul et nn entier naturel non nul. en injectant dans l’équation fonctionnelle on obtient c=1c=1. conclusion partielle : si P est solution non constante alors PP est de la forme (X2+1)n(X^{2}+1)^{n} où nn est entier naturel non nul. il ne reste plus qu’à rechercher les solutions constantes ainsi que celle de la forme ci-dessus... détails à compléter ALTERNATIVE (dans le cas P non constant une fois qu’on a obtenu i ou -i sont racines de module maximal et toute racine de module maximal est i ou -i) on montre que P(1) est nécessairement non nul car dans le cas contraire il y aurait une infinité de zéros, donnés par les termes de la suite définie par récurrence paru0=1,un+1=un2+un+1u_{0}=1, u_{n+1}=u_{n}^{2}+u_{n}+1 qui contient une infinité de termes. en injectant alors 0 dans l’équation fonctionnelle on en déduit que P(0)=1P(0)=1 donc le produit des racines de PP est de module 1 (après avoir obtenu que PP est unitaire) et comme toures les racines sont de module inférieur ou égal à 1, elles sont alors toutes de module 1 et donc égales à i ou -i etc... Détails à compléter

Report

For an unclear, ambiguous, or possibly incorrect statement, please use the Discussion tab on the right. Report content that needs moderator intervention, such as dangerous, clearly non-mathematical, or plagiarized content.