7
$\begingroup$

Does the Gamma function $\Gamma: \mathbb{C} \to \mathbb{C}$ preserve the Kummer ring $\mathbb{Z}[\exp(2\pi\imath/m)]$? And if not, then what about the Gaussian integers $\mathbb{Z}[\imath]$ or the Eisenstein integers $\mathbb{Z}[\exp(2\pi\imath/3)]$?

Is it possible to characterize holomorphic function which do preserve these lattices?

$\endgroup$
2
  • 3
    $\begingroup$ The gamma function has rapid decay in vertical strips, so I don't think it can preserve any lattices. I assume you are asking if it's possible that $\Gamma(L) \subset L$ where $L$ is a lattice. $\endgroup$ – Matt Young Oct 15 '14 at 20:15
  • $\begingroup$ Yes, that's what I'm asking. I know there is a lot to know about $\Gamma$ but I never learned about it's behaviour on the pure imaginary numbers. $\endgroup$ – Vít Tuček Oct 15 '14 at 21:59
13
$\begingroup$

Note that

$$\frac{\Gamma(z) \Gamma(1-z)}{\Gamma(2z) \Gamma(1 - 2z)} = 2 \cos(\pi z),$$

and the RHS is transcendental for any non-rational algebraic number $z$ (by the Gelfond–Schneider theorem). So $\Gamma$ certainly won't preserve any number field $K$. It's most likely true that $\Gamma(z)$ is transcendental for algebraic $z \notin \mathbf{Z}$, but I'm not sure if that's known.

$\endgroup$
3
  • $\begingroup$ I think the same argument works with the simpler expression $\Gamma(z)\Gamma(1-z)$. $\endgroup$ – GH from MO Oct 16 '14 at 15:23
  • 2
    $\begingroup$ But then you get a factor with $\pi$. This expression just gets rid of it and it lets you use Gelfond-Schneider theorem directly. $\endgroup$ – Vít Tuček Oct 17 '14 at 0:29
  • $\begingroup$ @VítTuček: Thank you, that makes sense. $\endgroup$ – GH from MO Oct 20 '14 at 11:54
7
$\begingroup$

I don't know about characterization, but there are lots of such functions. In fact, for any map $g$ of the lattice $L$ into itself, there are continuum-many entire functions $f$ such that $f(z) = g(z)$ for $z \in L$. This is because you can get entire functions that take prescribed values on any subset of $\mathbb C$ without limit points.

$\endgroup$
1
  • $\begingroup$ Thanks! It seems that this extension problem has been dealt with on MO at least 3 times. :-) $\endgroup$ – Vít Tuček Oct 15 '14 at 22:49

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.