Let $f:\mathbb{R}\rightarrow\mathbb{R}$, $g:\mathbb{R}\rightarrow\mathbb{R}$, and $h:\mathbb{R}\rightarrow\mathbb{R}$ and consider Pexider's equation, $$ f(x) + g(y) = h(x + y) \qquad \qquad (1) $$ where $f$, $g$ and $h$ are unknown. I assume (for simplicity) that $f$, $g$, and $h$ are twice continuously differentiable. Then, we can find a solution by differentiating with respect to $x$ and $y$, $$ h''(x+y)=0 $$ Integrating out, substituting back into $(1)$ and equating coefficients we get, $$ h(z)=c_1z +c_2 + c_3 \qquad f(x)=c_1x +c_2 \qquad g(y)=c_1y +c_3 $$ for arbitrary constants $c_1$, $c_2$ and $c_3$.

I have now found a solution to $(1)$. How do I prove that this solution is unique? It seems immediate, but can't quite convince myself.

  • 1
    $\begingroup$ Well, I am no expert on partial derivatives, but what you have got is $\dfrac{\partial^2(h)}{\partial x \partial y}=0$. Is it enough to conclude what you have concluded? $\endgroup$ – shadow10 Jul 22 '14 at 15:05
  • $\begingroup$ I'm not sure. There are several theorems on uniqueness of solutions of PDEs and ODEs but my suspicion is that uniqueness doesn't even need to use any of this fancy machinery. $\endgroup$ – user103828 Jul 22 '14 at 15:08

Suppose $h$ is twice continuously differentiable. Therefore,

$h''(x+y)=0$ if and only if $h(x+y)=(x+y)k+l$, where $k,l$ are constants (1)

Since $f(x)+g(y)=h(x+y)$, we have $0=\frac{\partial^2}{\partial x \partial y}(L.H.S.) = h''(x+y)$ (2)

Therefore by (1) and (2) we have $f(x)+g(y)=(x+y) k+l$, and since $x,y$ are arbitrary, $f,g$ have to be linear functions.

So, the solution is unique in this sense.

  • $\begingroup$ Yes, uniqueness seems immediate because $h''(x+y)=0 \Leftrightarrow h(x+y)=c_1 (x+y) +c_2$. $\endgroup$ – user103828 Jul 22 '14 at 20:33

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.