2
$\begingroup$

There is a well known fact that if one considers the family of polynomials $$f_n(z) = \sum_{k=1}^n \frac{z^k}{k!}$$ that any sequence of solutions to $f_n(nz_n)=0$ must converge to a point $z$ on the curve $|ze^{1-z}|=1.$ For reference, this curve is called the Szego curve.

My question is about the convergence of the roots. Are the roots proved to be exactly on the curve or near the curve and approaching it?

$\endgroup$
3
  • 1
    $\begingroup$ Page 2 of the following article: math.okstate.edu/~igor/PVNW.pdf might be of some interest. $\endgroup$ Commented Oct 16, 2012 at 19:06
  • $\begingroup$ Thanks! After posting this I realized how easy this question was. I apologize. $\endgroup$ Commented Oct 16, 2012 at 19:34
  • $\begingroup$ k should be from 0 instead of from 1. $$f_n(z) = \sum_{k=0}^n \frac{z^k}{k!}$$ $\endgroup$
    – cvgmt
    Commented May 27 at 3:53

1 Answer 1

7
$\begingroup$

I assume you meant "if $z_n$ satisfies $f(n z_n)=0$ and if the sequence $z_n$ converge to a nonzero limit $z$ then $|z e^{1-z}|=1$ ".

Now, if $n z_n$ is a root of $f_n$ then $z_n$ is algebraic over $\mathbb{Q}$. Assume $z_n$ also lies on your curve. Its real part is a sum of two algebraic numbers, hence is algebraic, and so is $\alpha := 1 - Re(z_n)$. If $\alpha = 0$ then $|z_n| = Re(z_n) = 1$ so that $z_n = 1$, but this would contradict $f(n z_n) = 0$. Thus $\alpha \neq 0$, and the algebraic number $z_n \times \bar z_n = e^{-2 \alpha}$ must be transcendental, a contradiction.

$\endgroup$
1
  • $\begingroup$ This would seem to suggest that no root lies on the curve at all. A much stronger statement. $\endgroup$ Commented Oct 16, 2012 at 19:36

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .