6
$\begingroup$

Find all functions $f:\mathbb{R}\rightarrow\mathbb{R}$ such that $$f(x)f\big(f(x)+y\big)=f\big(x^2\big)+f(xy)$$ for all $x,y\in\mathbb R$


Problem by me


Most elegant solution wins!

$\endgroup$
3
$\begingroup$

Not sure this rates as elegant but it is certainly short:

At $x=0,y=-f(0)$ we have $f(0)^2=2f(0)$, a quadratic equation with solutions $0,2$.

Case 1: $f(0)\ne 0$

substitute $x=0$ to get $2f(y+2)=4$ for any $y$, therefore $f\equiv 2$.

Case 2: $f(0)=0$

Case 2a: there is another zero $f(z)=0,z\ne 0$

substitute $x=z$ to get $f(z^2)=f(zy)$ for any $y$, i.e. $f$ must be constant $f\equiv 0$.

Case 2b: $f(0)=0$ is the only zero

Then, letting $y$ equal $0$ and $-f(x)$, respectively, we can make good use of $$ f(x)f(f(x))=f(x^2)=-f(-xf(x)) $$ With $x=1$ this yields $f(f(1))=1$ and $f(-f(1))=-f(1)$. With $x=f(1)$ we now get $f(f(1)^2)=f(1)$ and with $x=-f(1)$, finally, $f(1)=1$ and $f(-1)=-1$.

We now go back to the original equation. Comparing $x=1,y=t$ with $x=-1,y=-t$ we get $f(-z)=-f(z)$ for any $z$. Finally, setting $y=-x$ we get $f(f(x)-x)=0$, i.e. $f(x)=x$ for any $x$.

$\endgroup$
2
  • $\begingroup$ This is similar to my solution. You get the checkmark! $\endgroup$ – Culver Kwan Jul 26 '20 at 3:42
  • $\begingroup$ Hi @CulverKwan, thanks for the entertaining puzzle. I'm new to this section. Do you ever disclose your own solution? I'd like to see whether I missed any shortcuts. $\endgroup$ – Paul Panzer Jul 26 '20 at 3:53
11
$\begingroup$

First, let's get rid of all constant solutions. If $f(x)=c$ is a solution, then the equation gives

$c^2=2c\implies c\in \{0,2\}$. So from now on let's assume $f$ is non-constant.

If some $x$ satisfies $f(x)=0$, then plugging that in,

$f(xy)=-f(x^2)$, and since $f$ is non-constant, $x$ must be zero. Also, putting $x=0$ gives $f(0)f(f(0)+y)=2f(0)$, and again $f(0)\ne 0$ leads to $f$ being constant, so in fact $f(x)=0\iff x=0$.

Plugging $y=0$ now gives

$f(x^2)=f(x)f(f(x))$, so the equation can be written $f(x)f(f(x)+y)=f(x)f(f(x))+f(xy)$. Call this alternate version $(\star)$.

Now we claim that

$f$ is injective. Indeed, suppose $f(a)=f(b)$ with $a\ne b$. As seen already, $a,b$ are non-zero. Now substituting $x\mapsto a$ and $x\mapsto b$ in $(\star)$, we see that all the terms except $f(xy)$ have the same value in both cases. Thus $f(ay)=f(by)$ for all reals $y$. This shows that $f$ takes the same value on any two reals with ratio $a/b$.

Now plugging $x=1$ in the original equation gives $f(1)f(f(1)+y)=f(1)+f(y)$. Using this equation for $y\mapsto ay$ and $y\mapsto by$, we see that in fact $f(f(1)+ay)=f(f(1)+by)$, and given any $c\ne a/b$, one can choose $y$ so that the numbers $f(1)+ay$ and $f(1)+by$ have ratio $c$. Thus given any ratio, one can find two reals with that ratio so that $f$ has the same value on them.

And thus, to prove injectivity,

Note that given any non-zero $x_1,x_2$, we can find $c,d$ so that $c/d=x_1/x_2$ and $f(c)=f(d)$. Using the same kind of argument as on $a,b$, this would give $f(cy)=f(dy)$, so in particular $f(x_1)=f(x_2)$, so in fact $f$ is constant on non-zero reals. Because of $f(x^2)=f(x)f(f(x))$, the only possibility (remembering $f$ is non-constant) is $f(x)=1$ for non-zero $x$ and $0$ for $x=0$; this function, however, doesn't work! So $f$ has to be injective.

Phew! Now we are ready for the finish. Indeed, plug in

$y=-f(x)$; we get $f(x^2)=-f(-xf(x))$. Putting $x\mapsto -x$ gives $f(-xf(x))=f(xf(-x))$, and by injectivity this gives $f(-x)=-f(x)$ (well, as long as $x\ne 0$, but this holds for $x=0$ anyway so we are good). This says $f$ is an odd function, so in fact $f(x^2)=f(xf(x))$. Again, injectivity allows us to "cancel out" the outer $f$'s, giving $f(x)=x$ (again, for non-zero $x$, but $f(0)=0$ is already known).

Thus these are all such $f$:

The constant functions $0$ and $2$, and the identity function.

$\endgroup$
4
  • $\begingroup$ Any more elegant solutions? No need to prove injective. $\endgroup$ – Culver Kwan Jun 19 '20 at 4:26
  • $\begingroup$ This is awesome! $\endgroup$ – CDspace Jun 19 '20 at 21:47
  • $\begingroup$ Does this proof touch on the case when f never takes on the value 0? $\endgroup$ – phenomist Jun 19 '20 at 22:34
  • $\begingroup$ @phenomist the second spoiler block shows that can never happen for non-constant $f$. $\endgroup$ – Ankoganit Jun 20 '20 at 3:01
4
$\begingroup$

Not an answer, but a step toward an answer:

Let $x$ be a number such that $f(x)=0$. Then the definition becomes $0=f(x^2)+f(xy)$. Thus $f(xy)$ is constant as a function of $y$ (for this $x$) and thus either $x=0$ or $f$ is constant. We wind up with two choices: $f\equiv 0$; or $f(x)=0\Rightarrow x=0$ (which latter includes the case that $f$ is never $0$).

(The examples already found by other answerers —

$f\equiv 0$, $f\equiv 2$, and $f(x)=x$

— all meet this criterion.)

$\endgroup$
1
  • $\begingroup$ Good progress, I am using this method! Try finishing it. $\endgroup$ – Culver Kwan Jun 19 '20 at 4:26
3
$\begingroup$

I'm not sure about "all functions", but I've found 2:

$f(x) = 0$

and

$f(x) = 2$

$\endgroup$
3
$\begingroup$

Along with the solutions provided by Steve, I've found that:

$f(x) = x$

Is also a valid solution

$f(x)f(f(x)+y) = f(x)f(x+y) = x(x+y) = x^2+xy$

$f(x^2)+f(xy) = x^2+xy$ (Note this doesn't apply for $f(x) = cx$ where $c\ne1$, as you result with $c^3x^2 + c^2xy = cx^2+cxy$

$\endgroup$

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.