Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I vaguely remember a question going something like

Let $f$ be a function on $[-1,1]$ with $f$ satisfying (something like) $$f(x^2-1)=(2x)f(x).$$ Show that $f$ is identically zero on $[-1,1]$.

Sorry if I can't give much information. The exact statement has been bugging me for sometime now. I'd like to know what the exact statement is.

Edit: Swapped arguments.

Edit: Replaced $$f(x^2-1)=(2x-1)f(x).$$ with $$f(x^2-1)=2xf(x).$$.

share|cite|improve this question
Well, I'm unsure what it looks like, but I know that it is obvious that f(1) and f(-1) equal zero. I think it might be f(x^2-1)=(2x-1)f(x) instead. – E.Lim Aug 18 '12 at 18:02
Maybe it is f(x^2-1)=2xf(x) instead. Wasn't thinking a lot last night. – E.Lim Aug 19 '12 at 6:14
up vote 3 down vote accepted

The left hand side of $f(x^2-1)=2x f(x)$ is an even function of $x$, so $f$ has to be odd. It follows that we can restrict to the interval $I:=[0,1]$ and have $f(1-x^2)=-2x f(x)$ there.

The map $$T:\quad I\to I,\qquad x\mapsto Tx:= 1-x^2$$ is bijective and defines a discrete time dynamical system on $I$. Two points $x$, $y\in I$ belong to the same orbit iff $x=T^n y$ for some $n\in{\mathbb Z}$.

The point $\tau:={\sqrt{5}-1\over2}$ is a fixed point of $T$, and the set $\{0,1\}$ is an orbit of period $2$. Note that for $x\in\{0,\tau,1\}$ one necessarily has $f(x)=0$. Consider the map $$S:=T^2:\quad x\mapsto 2x^2-x^4\ .$$ From $$Sx \ \cases{<x&$(0< x<\tau)$\cr =x &$(x\in\{0,\tau,1\})$\cr >x&$(\tau<x<1)$\cr}$$ we conclude that $0$ and $1$ are attracting fixed points of $S$ whereas $\tau$ is repelling. This implies that the sets $\{\tau\}$ and $\{0,1\}$ are the only finite orbits of $T$. Put $f(x):=0$ for $x\in\{0,\tau,1\}$. Then choose a point $x_\alpha$ in each infinite orbit $O_\alpha$, put $f(x_\alpha):=1$ (or some arbitrary value), and use the functional equation $f(Tx)=-2x f(x)$ to define $f$ on all of $O_\alpha$. The resulting $f:I\to{\mathbb R}$ will be $\ne0$ at most points of $I$.

This construction shows that we need additional assumptions on $f$ to guarantee $f(x)\equiv0$. The following heuristic argument makes plausible that $f(x)\equiv0$, if we assume that $f$ is differentiable at $0$.

The functional equation $f(Tx)=-2x f(x)$ implies $$f(Sx)=f(2x^2-x^4)=4x(1-x^2) f(x)\ .$$ For $x$ near $0$ this "can be replaced" by $f(2x^2)= 4x f(x)$. We now consider the function $$g(x):={f(x/2)\over x/2}={1\over2}{f(x^2/2)\over x^2/2}={1\over 2}g(x^2)\ .$$ It follows that for all $n\geq1$ we have $$g(x)={1\over2^n}g\bigl(x^{2^n}\bigr)\ .$$ As $n\to\infty$ the right side converges to $0\cdot f'(0)=0$, from which we dare to conclude that $g(x)=0$ for all $x$ sufficiently near $0$, whence $f(x)=0$ for these $x$. Using the fact that the iterates of $S$ push all $x<\tau$ towards $0$ it follows that in fact $f(x)=0$ for $0\leq x<\tau$, and applying $T$ once gives the claim for the interval $]\tau,1[\ $.

share|cite|improve this answer
Wow. I wouldn't have guessed such a condition but I really do think differentiability at 0 is the right one. Thanks. I don't know if this was an actual contest problem. I just assumed it since it involved functional equations. – E.Lim Aug 19 '12 at 17:15

First, you have: $f((\frac{1}{2})^2-1)=f(-\frac{3}{4})=(2\times\frac{1}{2}-1)f(\frac{1}{2})=0$. Then, let's consider $\varphi(x)=x^2-1$ and $(x_n)$ defined by $u_0=-\frac{3}{4}$ and $u_{n+1}=\varphi(u_n)$. It can easily be shown that $\forall n\in\mathbb{N}, f(u_n)=0$ (by recursion).

It can also be shown that $\liminf u_n=-1$ and $\limsup u_n=0$, which means that if $f$ is continuous, then by continuity $f(0)=0$ and $f(-1)=0$.

If $f$ is also monotonic, then $f(x)=0\space\forall x\in[-1,0]$. Now, we have $f(1^2-1)=1\times f(1)$, so $f(1)=0$ as well, so because of the monotony, $f(x)=0 \text{ }\forall x\in [-1,1]$.

Maybe someone can find weaker conditions as well...

share|cite|improve this answer
Hmmm, I think It might have been $$f(x^2-1)=2xf(x).$$ instead. – E.Lim Aug 19 '12 at 6:16
@E.Lim It doesnt change my reasoning much. You have $f(0^2-1)=2\times 0\times f(0)$, so $f(-1)=0$. Then $f((-1)^2-1)=2\times(-1)\times f(-1)$, so $f(0)=0$, then $f(0)=f(1^2-1)=2\times 1\times f(1)$, so $f(1)=0$, and if $f$ is monotonic, it means that $f=0$ on $[-1,1]$. I suspect there might be weaker condition that monotony though... – S4M Aug 19 '12 at 7:58

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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