Determine all $f : \mathbb R^+ \rightarrow \mathbb R^+$ that satisfies

$f(x + y) = f(x^2 + y^2)$ $\forall x,y \in \mathbb R^+$

My Proof :

I define $g(x) + g(y) = f(x + y)$

Then , we get $g(x^2) - g(x) = c$ ; $c \in \mathbb R^+$

$f(x) = \frac{1 + \sqrt{4c + 1}}{2}$ , which is some constant $k$.

Hence , $f(x) = k$ $\forall k \in \mathbb R^+$.

Is my proof correct?


This doesn't work; there's no reason why $f(x+y)$ can be written as $g(x) + g(y)$, and moreover, $g(x^2) - g(x) = c$ doesn't give you that solution for $f$, because it's not a quadratic in $g(x)$; you'd need $g(x)^2 - g(x) = c$ instead.

The correct solution is as follows: fix $x+y=c$, for some $c \in \mathbb R^+$. Then as we vary $x$ over $(0, c)$, $x^2+y^2$ takes all values between $\frac{c^2}{4}$ and $c^2$ (with an open interval at $c^2$).

So $f(c) = f(a)$ for all $a \in [c^2/4, c^2).$

We can then proceed by "induction" in some sense; $f(1)$ has the same value as $f(x)$ for $x \in [1/4, 1)$, so $f(x)$ is constant on $[1/4, 1]$. We then apply the same method to obtain that $f$ is constant on $[(1/4)^2/4, 1] = [1/64, 1]$, and so on. Similarly, we can extend the range which $f$ is constant to all positive reals greater than $1$, and conclude $f$ is constant. I'll leave this as an exercise.


A bit of geometry:

Consider the first quadrant with $x,y>0$:

1) Let $y+x=c >0;$

$y=-x+c$ is a line with $x,y$ intercepts $c$.

Triangle: $(0,0); P(0,c); Q(c,0)$ has hypothenuse $\overline{PQ} = √2c;$ height from $(0,0)$ to hypothenuse has length $c/√2$.

$y=-x+c$ is tangent to the circle $x^2+y^2=c^2/2$;

The circle passing through $P,Q$ centre $(0,0)$:


For the annulus defined by $c^2/2 \le x^2+y^2 <c^2$ we have:


2) For $d=c/√2$ we find

$f(x^2+y^2)=f(d)=f(c/√2)$ for

$c^2/4 \le x^2+y^2 <c^2/2$.

3) Starting with an arbitrarily large $c>0$ we proceed to annuli with arbitrarily small radii.

Within these annuli the function $f$ is constant.

4) Note: The upper circle due to the $<$ is not included in the annulus.

5) We have disjoint annuli and within each region $f$ is some contant.

6) For $f$ continuos we can conlude $f =c$ (same constant) in the different regions

7) Bonus question: How to proceed without the above assumption?

  • $\begingroup$ I don't think you can take $y+x=0$ since $f$ is defined on the positive reals. $\endgroup$ – Yes it's me Jun 8 '20 at 11:53

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.