0
$\begingroup$

I am currently doing things related to the Akra–Bazzi theorem. One element in that theorem is the following:

For $n>0$ and sequences of real numbers $a_i, b_i$ of length $n$, where all $a_i>0$ and all $b_i\in (0;1)$, we consider the unique real number $x$ such that $$\sum_{i=1}^n a_i b_i^x = 1$$

My question is: when is $x$ rational/algebraic/transcendental?

For example, for the equation $$\left(\frac{1}{3}\right)^x + \left(\frac{3}{4}\right)^x = 1$$ $x$ is in $(1.1519623;1.1519624)$. It looks like a transcendental number to me, but I have no idea how to show that it is or is not.

$\endgroup$
  • 3
    $\begingroup$ Every number looks like a transcendental number. Also, every number looks like an algebraic irrational, and every number looks like a rational number. $\endgroup$ – Gerry Myerson May 7 '15 at 12:59
  • $\begingroup$ And certainly every number also looks like a computable number. $\endgroup$ – Matemáticos Chibchas May 7 '15 at 13:54
  • $\begingroup$ I have no idea what you mean by that. $\endgroup$ – Manuel Eberl May 7 '15 at 14:40
  • $\begingroup$ Of course the solution of $$ \left(\frac{1}{3}\right)^x + \left(\frac{3}{4}\right)^x = 2 $$ is rational, and the solution of $$ \left(\frac{1}{3}\right)^x + \left(\frac{3}{4}\right)^x = \frac{13}{12} $$ is rational. And of course there are many more like this. So any proof that your number is transcendental seems unlikely to me. See? I can do "seems to me" statements, too. $\endgroup$ – Gerald Edgar May 7 '15 at 14:54
5
$\begingroup$

Schanuel's conjecture implies that if $x$ is an algebraic irrational, $n^x$ for natural numbers $n$ are linearly independent over the rationals: see Will Sawin's answer to this recent question for a proof. It's easy to extend this to positive rationals. Thus Schanuel implies that if the $a_i$ and $b_i$ are rational, $x$ will either be rational or transcendental. So (if you're willing to accept Schanuel) you just have to check for rational solutions, which for something like $(1/3)^x + (3/4)^x = 1$ is not hard.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ That looks pretty interesting, thanks. However, I'm not entirely sure how to prove that the solution is not rational for a case like that. $\endgroup$ – Manuel Eberl May 8 '15 at 6:48

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.