2
$\begingroup$

Find all functions $f : \mathbb{(0,\infty)}\to\mathbb{R}$ such that $f(x+y) = xf(y)+ yf(x)$ if $f$ is continuous at $x=1$

This problem was looking quite easy at first but the domain of positive reals is posing me a problem. I couldn't plug in zero's for $x$ and $y$. I tried putting $x=y$ but the result $f(2x) = 2xf(x)$ couldn't be used as a recurrence relation $\infty$ times as that would yields $f(0)$ again. I've run out of ideas. Please help.

$\endgroup$
  • $\begingroup$ Setting $y=0$ means that you must have $f(x) = x f(0)$. $\endgroup$ – copper.hat Apr 23 at 17:22
  • $\begingroup$ $y= 0$ is outside the domain. But looking at $x = y = 1$ may prove informative. $\endgroup$ – John Hughes Apr 23 at 17:23
  • $\begingroup$ Also: continuity at $x = 1$ should be a powerful tool, but to use it, you'll need to look at $f(1 + u)$ for small values of $u$. How might you do that? $\endgroup$ – John Hughes Apr 23 at 17:24
  • $\begingroup$ I should read before commenting. $\endgroup$ – copper.hat Apr 23 at 17:26
  • $\begingroup$ I'd say you're entitled, at this point, to the occasional lapse in scanning a question. :) $\endgroup$ – John Hughes Apr 23 at 17:33
5
$\begingroup$

This is one possible continuation of your attempted solution.

By letting $\ y=x=z\ $ you found that $$ f(2z) = 2z f(z). \tag{1} $$ Let $\ y=2z,\ x=z\ $ to get $$ f(3z) = 2z f(z) + z f(2z). \tag{2} $$ Let $\ y=2z,\ x=2z\ $ to get $$ f(4z) = 4z f(2z). \tag{3} $$ Let $\ y=3z,\ x=z\ $ to get $$ f(4z) = 3z f(z) + z f(3z). \tag{4} $$ Now eliminate $\ f(2z),\ f(3z),\ f(4z)\ $ from the four linear equations to get $$ (3-6z+2z^2)f(z) = 0. \tag{5} $$ There are two positive roots of the quadratic which implies that $\ f(z) = 0\ $ for any other values of $\ z.\ $ Use equation $(1)$ to prove $\ f(z) = 0\ $ for the positive roots also.

Notice that no continuity assumption or $\ f(0)\ $ was used to prove $\ f(z)=0\ \forall z.$

$\endgroup$
  • 2
    $\begingroup$ Wow, this seems great! Thank you Somos. $\endgroup$ – Shashwat1337 Apr 24 at 5:15
  • 2
    $\begingroup$ Excellent way to ignore a seemingly important assumption $\endgroup$ – Hagen von Eitzen Apr 24 at 5:33
4
$\begingroup$

If $f$ satisfies the given condition so does $cf$ for any constant $c$. Unless $f(1) = 0$ we might as well assume $f(1) = 1$. Some calculations:

$$f(2) = f(1 + 1) = f(1) + f(1) = 2$$ $$f(3) = f(1 + 2) = 2f(1) + f(2) = 4$$ $$f(4) = f(2 + 2) = 2f(2) + 2 f(2) = 8$$ $$f(4) = f(1 + 3) = 3f(1) + f(3) = 3 + 4 = 7$$ The last two equations are incompatible so we must have $f(1) = 0$.

Doing calculations like the above gives you $f(n) = 0$ and $f(2^{-n}) = 0$ for all natural numbers $n$. The additivity extends this to $f(r) = 0$ for all dyadic rationals $r$. This should give you that $f(x) = 0$ for all positive $x$ if you can exploit the continuity condition.

$\endgroup$

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.