Determine all surjective functions $ f: \mathbb{Z} \to \mathbb{Z} $ such that $$ f\left(xyz+xf\left(y\right)+yf\left(z\right)+zf\left(x\right)\right)=f\left(x\right)f\left(y\right)f\left(z\right) $$for all $ x,y,z $ in $ \mathbb{Z} $
Problem
Source: 2017 Taiwan TST Round 2, Day 3, Problem 1
Tags: algebra, functional equation
19.04.2017 11:38
do you have all the problems?If you have,could you post them in contedt collections?Thanks
19.04.2017 11:48
We should avoid posting problems from ISL. (Before IMO) So, in my country, the problems are usually posted by their proposers.
19.04.2017 11:52
Are you from Taiwan
19.04.2017 15:12
Maybe the problems will be posted after IMO. Just wait and see if there is any volunteer willing to post Taiwan TSTs.
12.12.2017 09:15
By request, I'm going to post a solution to the reduced version: Find all $f:\mathbb{Z}\to\mathbb{Z}$ surjective such that $f(x)f(y)=f(y+xf(y))$. One can get this reduced version by noticing that $f(0)=1$. It's easy to see that $f(0)=1$. Suppose that $f(a)=2$, then $(x,a):2f(x)=f(2x+a)$. By induction, we can see that $v_2(f(x))\geq v_2(x+a)$. Therefore $v_2(a)\leq v_2(f(0))=1$, and so $a$ is odd. By $2f(x)=f(2x+a)$, we know that $f((2^n-1)a)=2^n \quad\forall n\in\mathbb{N}_0$. Now, suppose that $f(l)=2k+1$ but $l\neq 2ka$, then take $n=v_2(l-2ka)$, we have $((2^n-1)a, l): (2k+1)2^n = f(2^n(2k+1)a-l-2ka-a)$. Taking $v_2$ on the both sides, we get that $n=v_2f(2^n(2k+1)a-l-2ka-a)\geq v_2(2^n(2k+1)a-l-2ka)\geq n+1$, which is a contradiction. Hence we have $f(2ka)=2k+1\forall k\in\mathbb{Z}$. By the identity $2f(x)=f(2x+a)$ it's not hard to get $f(ka)=k+1\quad\forall k\in\mathbb{Z}$. $(1,a^2-a): af(1)=a+1$, and so $a=\pm 1$. If $a=1$, then $f(x)=x+1$; if $a=-1$, then $f(x)=1-x$. It's easy to verify that those two are indeed solutions, and thus are the only two solutions.
03.12.2018 13:41
Here are solutions to this problem. One for the original version and the other for the reduced version.
Attachments:
Problem.pdf (105kb)
21.01.2019 15:01
USJL wrote: By request, I'm going to post a solution to the reduced version: Taking $v_2$ on the both sides, we get that $n=v_2f(2^n(2k+1)a-l-2ka-a)\geq v_2(2^n(2k+1)a-l-2ka)\geq n+1$, which is a contradiction. it seems there is a mistake in your expression. shouldn’t it $n=v_2f(2^n(2k+1)a+l-2ka-a)\geq v_2(2^n(2k+1)a+l-a-2ka)\geq n+1$? but $l-a$ is an odd number and $v_2$ can’t help us.
22.01.2019 11:42
No You are right for $-l\to+l$ but it changes nothing. He uses $v_2(f(x))\ge v_2(x+a)$ (notice the "$+a$" in RHS) And so here $v_2(f(...))\ge v_2(2^n(2k+1)a+l-2ka)$ (the "$a$" in RHS disappears) And since $n=v_2(l-2ka)$ we have $l-2ka=2^nq$ where $q$ is odd And so $v_2(2^n(2k+1)a+l-2ka)=v_2(2^n((2k+1)a+q)\ge n+1$ indeed.
22.01.2019 12:30
USJL wrote: By induction, we can see that $v_2(f(x))\geq v_2(x+a)$. Nice solution! But pco please could you explain me why this is true? Thanks in advance!
22.01.2019 12:39
Just write $x+a=2^nq$ for some odd $q$ Then $f(x)=f(2(2^{n-1}q-a)+a)$ $=2f(2^{n-1}q-a)$ (using $f(2x+a)=2f(x)$ But $f(2^{n-1}q-a)=f(2(2^{n-2}q-a)+a)=2f(2^{n-2}q-a)$ And so $f(x)=2^2f(2^{n-2}q-a)$ And a simple induction gives $f(x)=2^nf(q-a)$ and so $v_2(f(x))\ge n=v_2(x+a)$
28.01.2021 20:16
Very fun problem. Throughout the solution, we maintain that $0$ divides $0$ and no other integer. Let $P(x,y,z)$ denote the original proposition. $P(0,0,0)$ gives $f(0)=0$, $-1$ or $1$. If $f(0)=0$, choose an $x$ satisfying $f(x)=1$, then $P(x,x,0)$ gives $f(x)=1$, contradiction! If $f(0)=-1$, choose $t$ such that $f(t)=1$. Then $P(x,y,0)$ gives $f(x-t)=-f(x)$ $$\implies \ f(x-2t)=-f(x-t)=f(x)$$Since $t \neq 0$, this gives that $f$ is periodic with period $2t$, and hence can't be surjective, contradiction! Therefore $f(0)=1$. Then $P(x,y,0)$ gives $$f(xf(y)+y)=f(x)f(y)$$Call the above $Q(x,y)$. Claim 1: For integers $a,b$, $a \equiv b \pmod{f(b)}$ $\implies$ $f(b) \mid f(a)$. Proof: If $f(b)=0$ it is obvious, else use $Q \left (\dfrac{a-b}{f(b)}, b \right )$. $\blacksquare$ Claim 2: For all integers $y$, all primes which divide $f(y)-1$ also divide $y$. Proof: If $|f(y)-1|=1$, it is vacuously true. Else let $p$ be any prime factor of $f(y)-1$. Choose a $z$ such that $f(z)=p$. Let $S$ be the set of residues $r$ modulo $p$ such that for any integer $x$, $x \equiv r \pmod{p}$ $\implies$ $p \mid f(x)$. $S$ is non-empty, since $z \in S$ by Claim 1. Choose any $t \in S$, and choose any integer $x$ satisfying $x \equiv t-y \pmod p$. Then $xf(y)+y \equiv t \pmod p$ $\implies$ $p \mid f(xf(y)+y)$. So $Q(x,y)$ gives $p \mid f(x)f(y)$ $\implies$ $p \mid f(x)$ $\implies$ $t-y \in S$ since $x$ can be chosen arbitrarily. Using this repeatedly gives $t-ny \in S$ for all positive integers $n$. If $p \nmid y$, then any residue modulo $p$ can be written as $t-ny$ for some positive integer $n$ $\implies$ $S$ contains all residues modulo $p$ $\implies$ $p \mid f(x)$ for all integers $x$, which contradicts surjectivity. Hence $p \mid y$ as required. $\blacksquare$ Claim 3: $f$ is injective over non-zero values. Proof: Let $y_1,y_2$ be integers such that $f(y_1)=f(y_2)=k \neq 0$. Then from $Q(x,y_1)$, $Q(x,y_2)$ and Claim 2 for any arbitrary $x$, all prime factors of $kf(x)-1$ divide both $xk+y_1$ and $xk+y_2$ $\implies$ they all divide $y_1-y_2$. If $y_1 \neq y_2$, then since $f$ is surjective, we can choose an $x$ such that $kf(x)-1$ is divisible by a prime greater than $|y_1-y_2|$, contradiction! Hence $y_1=y_2$ as required. $\blacksquare$ Let $a$ satisfy $f(a)=2$. Clearly $a \neq 0$. Claim 4: $f(x)=\dfrac{x}{a}+1$ if $a \mid x$, and $f(x)=0$ otherwise. Proof: First we prove that $f(x) \neq 0$ $\implies$ $f(x)=\dfrac{x}{a}+1$. Indeed, if $f(x) \neq 0$, comparing $Q(x,a)$ and $Q(a,x)$ we get $$f(2x+a)=f(af(x)+x)=2f(x) \neq 0$$$$\implies \ 2x+a=af(x)+x$$$$\implies \ f(x)=\frac{x}{a}+1$$as required. This immediately gives $f(x)=0$ if $a \nmid x$. Now assume $a \mid x$. If $x = -a$, then $Q(x,a)$ gives $f(x)=0$, consistent with the claim. Else choose $y$ such that $f(y)=\dfrac{x}{a}+1$ $\implies$ $f(y) \neq 0$ $\implies$ $f(y)=\dfrac{y}{a}+1$ $\implies$ $x=y$ as required. $\blacksquare$ If $a = \pm 1$, then by Claim 4 we get the two solutions: $$\boxed{f(x)=1+x \ \ \forall x \in \mathbb{Z}}$$$$\boxed{f(x)=1-x \ \ \forall x \in \mathbb{Z}}$$ Else $f(1)=0$ by Claim 4. Then $P(1,a^2-a)$ gives $$0=f(1)f(a^2-a)=f(f(a^2-a)+a^2-a)=f(a^2)=a+1$$which gives $a=-1$, contradiction! $\blacksquare$