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. 110First steps / middle schoolPremiers pas / collège
  2. 1125Beginner / high schoolDébutant / lycée
  3. 2650Intermediate / undergraduateIntermédiaire / licence
  4. 5170Advanced / graduateAvancé / master
  5. 7190Expert / specializedExpert / spécialisé
  6. 91100Research 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 PR[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

Export references

Hints

3

Hint 1

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

Solutions

1
Reveal solutionsAre you sure? Give it a try first.

Solution by Uettechat

Discussions0 useful votes

Let PR[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 λR\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 zCz \in \mathbb{C} be a root of PP. Evaluating the functional equation at X=zX = z and at X=z1X = 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(z1)P(z)=0    P((z1)2+(z1)+1)=0    P((z1)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 (z1)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 λZ\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 (λ1)2+λ(\lambda-1)^2+\lambda are also roots of PP. By the maximality of λ\vert{}\lambda\vert{}:
λ2+λ+1λand(λ1)2+λλ\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:
2λ=(λ2+λ+1)((λ1)2+λ)λ2+λ+1+(λ1)2+λ2λ2\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 a0a \ge 0 such that:
λ2+λ+1=a((λ1)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 (λ1)2+λ(\lambda-1)^2+\lambda to have modulus exactly λ\vert{}\lambda\vert{}. This gives:
λ=a((λ1)2+λ)=aλ\vert{}\lambda\vert{} = \vert{}-a((\lambda-1)^2+\lambda)\vert{} = a\vert{}\lambda\vert{}If λ=0\lambda = 0, the initial inequality 02+0+10\vert{}0^2+0+1\vert{} \le \vert{}0\vert{} yields 101 \le 0, which is absurd. Thus λ0\lambda \neq 0, which implies a=1a = 1. The equation then becomes:
λ2+λ+1=((λ1)2+λ)=λ2+λ1\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 zZz \in Z be any complex root of PP. We construct the sequence (un)nN(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 unZu_n \in Z for all nNn \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)nN(u_n)_{n \in \mathbb{N}} is eventually periodic and enters a cycle of length p1p \ge 1.
Let N0N \ge 0 be the smallest index such that uNu_N belongs to this cycle, meaning N=min{nNun+p=un}N = \min \{ n \in \mathbb{N} \mid u_{n+p} = u_n \}.
For any kNk \in \mathbb{N}, write uk=xk+iyku_k = x_k + i y_k with xk,ykRx_k, y_k \in \mathbb{R}. We have:
xk+1=Re(uk+1)=xk2yk2+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 ukZu_k \in Z satisfies uk1\vert{}u_k\vert{} \le 1. Thus, xk2+yk21x_k^2 + y_k^2 \le 1, which implies yk2xk21-y_k^2 \ge x_k^2 - 1. Substituting this yields:
xk+1xk2+(xk21)+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+p1N + p - 1):
k=NN+p1xk+1k=NN+p1(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+p1xk+1\sum_{k=N}^{N+p-1} x_{k+1} and k=NN+p1xk\sum_{k=N}^{N+p-1} x_k are equal, so subtracting xk\sum x_k from both sides gives:
02k=NN+p1xk200 \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 (kNk \ge N). For any element uku_k in the cycle (kNk \ge N), since xk=0x_k = 0 and xk+1=0x_{k+1} = 0, the relation xk+1=xk2yk2+xk+1x_{k+1} = x_k^2 - y_k^2 + x_k + 1 becomes:
0=0yk2+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 kNk \ge N.

Finally, we prove by backwards induction that uk{i,i}u_k \in \{i, -i\} for all k0k \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{1i,1+i,i,i}u_k \in \{-1-i, -1+i, -i, i\}, since uk1\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 nN,cRP(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)nnN}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 nNn \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)nnN}S = \{0, 1\} \cup \{(X^2+1)^n \mid n \in \mathbb{N}\}

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.