How does one solve the following functional equation when $f:\mathbb{R}\rightarrow\mathbb{R}$


When I assumed it was a polynomial equation, it can be seen through induction that $$f(nx)=nf(x)-\sum _{ i=1 }^{ n-1 }{ (ix^2-1)^2 } $$

for some natural number $n$.

This implies that $$f(n)-nf(1)+\sum _{ i=1 }^{ n-1 }{ (i-1)^2 }=0 $$ is true for all natural $n$, or that for all $n$ $$f(n)-nf(1)+\frac{(n-2)(n-1)(2n-3)}{6}=0$$

Since there are infinite number of solutions to $f(n)-nf(1)+\frac{(n-2)(n-1)(2n-3)}{6}=0$, it can be said that $$f(n)=nf(1)-\frac{(n-2)(n-1)(2n-3)}{6}$$ for all real number $n$.

This implies that $f(n)$ is of the degree $3$, but this is a contradiction since if $f(n)$ had a degree of $3$, the coefficient for $x^2y^2$ would be $0$.

So there appeared to be no polynomial solutions to this function. I further thought that there would be no functions that satisfied this equation.

How does one solve this equation? Any help would be appreciated.

  • $\begingroup$ You can take a look at a generalized version of your question here. :) $\endgroup$ – H. R. Jan 28 '16 at 21:18

$$f(x+y)=f(x)+f(y)-(xy-1)^2\\ f(2)=2f(1)\\ f(3)=f(2)+f(1)-1=3f(1)-1\\ f(2)+f(2)-9=f(4)=f(1)+f(3)-4\\ 4f(1)-9=4f(1)-5$$

  • 1
    $\begingroup$ I get two different expressions from $4=2+2$ and $4=3+1$, not from rearranging earlier lines. $\endgroup$ – Empy2 Jan 28 '16 at 15:08
  • $\begingroup$ Conceptually, first let $y=1$ and solve the equation for $f\vert_{\mathbb N}$, then restrict the original equation to $x,y\in\mathbb N$, which is strong enough to lead to a contradiction. $\endgroup$ – Yai0Phah Jan 28 '16 at 15:17

First put $x=y=0$ into the functional equation to obtain that

$$f(0)=1 \tag{1}$$

Then suppose that $y \ne 0$ so dividing the functional equation by $y$ we can write

$$\frac{f(x+y)-f(x)}{y}=\frac{f(y)-1}{y}-\frac{(xy-1)^2-1}{y} \tag{2}$$

$$\frac{f(x+y)-f(x)}{y}=\frac{f(y)-f(0)}{y-0}-\frac{x^2y^2-2xy}{y} \tag{3}$$

Next, take the limit when $y \to 0$ to obtain

$$f'(x)=f'(0)+2x \tag{4}$$

Where we have assumed that $f(x)$ is differentiable over $\mathbb{R}$. Next, integrating $(4)$ we get

$$f(x)-f(0)=f'(0)x+x^2 \tag{5}$$

Or equivalently

$$f(x)=x^2+f'(0)x+1 \tag{6}$$

But we can observe that $(6)$ will never satisfy the functional equation since $f(x)$ is a second degree polynomial and there are fourth degree terms like $x^2y^2$ in the functional equation . So the only thing that we are left with is that

There is no differentiable $f(x)$ which satisfies the functional equation!

  • $\begingroup$ Are we sure there's a limit when ${y \to 1}$ ? $\endgroup$ – peter.petrov Jan 28 '16 at 14:50
  • $\begingroup$ As the OP said, you need degree at least 4 to produce $x^2y^2$ $\endgroup$ – Empy2 Jan 28 '16 at 14:51
  • $\begingroup$ @peter.petrov: It is a assumption in this solution. I can say that in these types of questions it is a common assumption. :) $\endgroup$ – H. R. Jan 28 '16 at 14:52
  • 2
    $\begingroup$ ignore my earlier comment, looks good. This is a proof that $f$ can't be differentiable, then? $\endgroup$ – hunter Jan 28 '16 at 14:54
  • 1
    $\begingroup$ This answer assumes $f(x)$ is differentiable. You should at least mention it. The original question doesn't mention it. $\endgroup$ – user236182 Jan 28 '16 at 15:01

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.