I was wondering about real-analytic solutions to the equation
$$f(5 x + x^3) = f(5 x) + f(3 x)$$
If I plug in $x=0$ it follows $f(0)=0$. Also $f(-x) = -f(x)$.
My initial guess was a sum of arcsines but that seems a bad idea ? afterall the $x^3$ part is "inside" and not "outside " ( outside like f(x)^3 ...).
My second guess was a sum of cube roots, but that also seems like a bad idea ?
My third guess was to rewrite it as an integral but I got stuck trying.
My 4 th idea was to use a helping (Abel) equation : $ g(5x + x^3) = g(x) + 1 $ and express $f$ in terms of $g$.
We could also compute the taylor coefficients from the equations but Im not sure if that gives insight ?
Is $f$ a hypergeometric function ??
What are good asymptotics for $f$ ?
I have strong arguments that $f$ is not entire. unless ofcourse $f(x) = 0$. Notice if $a(x)$ is a solution then $C a(x)$ is also one for a real constant $C$.
And how about $C^{oo}$ solutions that are nowhere analytic ? Does that even make sense ?
Would it help to take the derivative on both sides ?
Is there a nonzero elementary solution ?
Is the solution unique up to a constant multiple ?
Is this related to fractals ? How about addition formula's ?
And how would it look on the complex plane ?
It doesn't quite look like you may find a solution which is real-analytic at zero. Otherwise you would have a power-series $f(x) = cx^n + O(x^{n+1})$ with non-zero $c$ which yields $c (5^n - 5^n - 3^n) = 0$. But you could, of course, adjust the constants a bit...