1

If $x>0,y>0$ and if $f(xy)=f(x)f(y)$, then $f=\, ?$

I tried the problem. And got it as $f(x)^n=f(x^n)$. But answer is $f(x)=x^n$. How?
$f$ is a continuous function.

user26857
  • 52,094
  • Do you know something more about $f$? Is it continuous, differentiable, $C^k$, etc.? – paf Aug 13 '16 at 17:07
  • 1
    Um... $f(x)^n = f(x^n)$ doesn't answer what $f(x)$ is. There are many possible answers to f(x). An obvious answer is $f(x) = x$ but that isn't the only one. It's also clear for any $n$ $f(x) = x^n$ will also work (and $f(x) = x = x^1$ is just a special case). The question is can $f(x)$ be anything else? – fleablood Aug 13 '16 at 17:50
  • 3
    You can find several related links in this post. In particular this one. – Martin Sleziak Aug 14 '16 at 07:34
  • Please, try to make the titles of your questions more informative. E.g., Why does $a<b$ imply $a+c<b+c$? is much more useful for other users than A question about inequality. From How can I ask a good question?: Make your title as descriptive as possible. In many cases one can actually phrase the title as the question, at least in such a way so as to be comprehensible to an expert reader. You can find more tips for choosing a good title here. – Martin Sleziak Aug 14 '16 at 07:37

2 Answers2

3

Trusting that you demand continuity, so that we have $f(x^n)=f(x)^n\;\forall n\in \mathbb R$ (and not just $\mathbb Q$) we can complete the argument as follows:

Suppose that $f(e)=\lambda$. We note that $\lambda ≥0$, since $f(\sqrt e)^2=\lambda$. If $\lambda =0$ then $f(x)$ is identically $0$, which is a valid solution. Otherwise, suppose that $\lambda > 0$. Then for any $x\in \mathbb R$ we write $x=e^{\ln x}$. Then, by what you have already proven, $f(x)=f(e^{\ln x})=f(e)^{\ln x}=\lambda^{\ln x}$. We conclude by writing $$f(x)=\lambda^{\ln x}=e^{(\ln \lambda)( \ln x)}=x^{\ln \lambda}$$

Note: absent continuity, the claim is not generally true. To construct a counterexample, start with a discontinuous function $\Psi(x)$ satisfying $\Psi(x+y)=\Psi(x)+\Psi(y)$. (The usual examples start by thinking of $\mathbb R$ as a vector space over $\mathbb Q$ and letting $\Psi$ denote projection onto one basis line.). Then, to get a discontinuous example of your type, define $f(x)=e^{\Psi(\ln x)}$.

lulu
  • 70,402
  • I think this derivation is wrong as sign(x)x = abs(x) is also multiplicative and continous. The most general statement i could make is So any composition of products of x^n and bsign(x) where n is never 0 when b is not zero and b is always greater equal 0. I'm not sure that covers all the cases though. – worldsmithhelper Mar 28 '21 at 01:03
  • 1
    @worldsmithhelper The problem specifies $x,y>0$ (as does the duplicate problem) – lulu Mar 28 '21 at 01:15
  • Ok, still abs(x) is also a solution. (One you can't distinguish from x). Still you are right, my bad. Is there something you can to generalize the proof to the real number line and discover all the solutions involving compositions of sign? – worldsmithhelper Mar 28 '21 at 01:19
  • @worldsmithhelper I don't see anything very interesting here. Since $f(1)=1=\left(f(-1)\right)^2$ we get $f(-1)=\pm1$ If it is $1$ then $f(-x)=f(x)$ If it is $-1$ then $f(-x)=-f(x)$ and either way the classification follows from the positive case. – lulu Mar 28 '21 at 01:24
  • Would you consider sign(x)*sqrt(abs(x)) interesting over |R ? – worldsmithhelper Mar 28 '21 at 01:25
  • 1
    @worldsmithhelper Not really, no. Everything just comes down to the positive case. – lulu Mar 28 '21 at 01:27
  • Ah i see what you mean, for each positive case there exists two continuations over the complete real numbers. – worldsmithhelper Mar 28 '21 at 01:30
0

4 obvious solutions spring to mind.

$f(x) = 0$ then $f(xy) = 0 = 0*0 =f(x)f(y) $.

$f(x) = 1$ then $f(xy) = 1 = 1*1 = f(x)f(y) $.

$f(x) = x$ then $f(xy) = xy = f(x)f(y)$.

And $f(x) = x^n$ for some $n \in \mathbb R$ then $f(xy) = x^ny^n = f(x)f(y)$.

These are all really the same class of answers. If $n = 1; f(x) = x = x^1$. If $n = 0; f(x) =1 = x^0$. $f(x) = 0$ can be considered an exception. However if we were to allowed $n$ in the extended reals. $n = -\infty; f(x) = 0 = x^{-\infty}$.

But are is this the only class of solutions?

let $b \ne 1; f(b) = c$. Then $c = b^{\log_b c}$. Let $n = \log_b c$. So $f(b) = b^n$.

(I am assuming $c > 0$.[footnote])

Let $x > 0$. Let $k = \log_b x$. So $x = b^k$.

$f(x) = f(b^k) = f(b)^k = b^{nk} = x^n$. And we are done

(I am assuming $f(x^k) = f(x)^k$ for all $k \in \mathbb R$. [footnote 2])

====

[footnote]

If $f(b) = 0$ then $f(1) = f(b)f(1/b) = 0$. S0 $f(x) = f(x)f(1) = 0$ for all $x$ and $f(x) = 0$ is one of the cases we discussed and we can assume we are trying for other cases. So we may assume $f(b) \ne 0$. Furthermore no $x > 0$ is such $f(x) = 0$.

$f(x) = f(x)f(1)$ so $f(1) = 1 > 0$. If $f(b) < 0$ then by intermediate value theorem there exist a value $c$ between $b$ and $x$ where $f(c) = 0$ but we've shown that is not the case. So $f(x) >0$ for all $x$.

[footnote 2]

$f(x^n) = f(x*x*...) = f(x)^n$ and $f({x}^{1/n})^n = f(x)$ so $f({x}^{1/n}) = f(x)^{1/n}$. So $f(x^r)= f(x)^r; \forall r \in \mathbb Q$. As $f$ is continuous, $f(x^y) = f(x)^y$ for all $y \in \mathbb R$.

fleablood
  • 124,253
  • I think this derivation is wrong as sign(x)x = abs(x) is also multiplicative and continous. The most general statement i could make is So any composition of products of x^n and bsign(x) where n is never 0 when b is not zero and always greater equal 0. I'm not sure that covers all the cases though. – worldsmithhelper Mar 28 '21 at 01:00
  • I missed the x,y > 0 part. – worldsmithhelper Mar 28 '21 at 01:31