3
$\begingroup$

So here is the Question :-

Find all functions $f$ :- $\mathbb{N}$ $\to$ $\mathbb{N}$ such that :- $$xf(y) + yf(x) = (x + y)f(x^2 + y^2)$$ I tried substituting values for $x$ and $y$, but I couldn't reach to a possible clue to the solution. Any hints or suggestions will be greatly appreciated!

$\endgroup$
5
  • 1
    $\begingroup$ What is $N$? $\mathbb{N}$? If so, do you consider $0$ to be an element of $\mathbb{N}$? $\endgroup$ – Hayden Jun 6 '20 at 4:24
  • 1
    $\begingroup$ N consists the set of natural numbers , 0 is not included . $\endgroup$ – Anonymous Jun 6 '20 at 4:25
  • 1
    $\begingroup$ Plugging in the first numbers gives $f(1)=f(2)=f(5)$. Every constant function is a solution. The question is if that's all; unless I'm wrong, we get by a kind of induction $f(m)=f(1)$ for all $m$ which are sums of two squares, and maybe on other $m$, $f(m)$ is arbitrary. $\endgroup$ – Torsten Schoeneberg Jun 6 '20 at 4:35
  • $\begingroup$ Okay , if you have an idea or so , can you just write it as a solution instead of a comment ? $\endgroup$ – Anonymous Jun 6 '20 at 4:42
  • $\begingroup$ Fair point. Done. $\endgroup$ – Torsten Schoeneberg Jun 6 '20 at 5:00
6
$\begingroup$

Suppose there exists an integer $n$ such that $f(1) <f(n)$. With $x=1$ and $y=n$, we have $$ nf(1)+f(n) = (n+1)f(n^2+1) \implies f(1)<f(g(n))<f(n),$$ where $g(n)=n^2+1$. Repeating the same argument using $f(1)<f(g(n))$, we deduce that $$f(1)<f\left(g^{f(n)-f(1)}(n)\right)<f\left(g^{f(n)-f(1)-1}(n)\right)<\ldots<f(g(n)) <f(n),\tag{1}$$ where $g^m(\cdot)$ denotes the composition of the function $g$ repeated $m$ times.

However, there cannot be more than $f(n)-f(1)-1$ integers between $f(n)$ and $f(1)$. Thus, (1) leads to contradiction because $f(x)\in\mathbb{N},\forall x\in\mathbb{N}$. Hence, $f(1)\nless f(n).$

Using similar arguments, we can show that $f(1)\ngtr f(n)$. Therefore, we conclude that $f(x)=f(1)$ for all values of $x$ which trivially satisfies the given relation.

$\endgroup$
3
$\begingroup$

Setting $x=y=1$ gives $f(1)=f(2)$. From there, try to show via induction that $f(1)=f(n)$ for all $n$ which are sums of two squares.

Remains the question if $f(n)$ can be defined arbitrarily on other $n$. For a counterexample, see comment by user Explorer (thanks!). I do not know if one can conclude further, and whether there are non-constant $f$ which satisfy the criterion.

EDIT: I realise what I had in mind here might not work. One only gets the following: If $f(x)=f(y)$, then $f(x^2+y^2)=f(x)$ as well, and starting with the above, one easily constructs infinitely many numbers $n$ with $f(n)=f(1)$. But to show that is true for all $n$, one needs more work, as in the other answer.

$\endgroup$
1
  • 3
    $\begingroup$ $f(n)$ cannot be arbitrary for other values of $n$. For example, with $x=y=5$, we get $$f(1)=f(5^2+5^2)=f(50)=f(7^2+1)=\frac{f(7)+7f(1)}{7+1}\implies f(7)=f(1).$$ However, $7$ can not be written as sum of two squares. $\endgroup$ – Explorer Jun 6 '20 at 5:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.