Proof of a conjecture of Pólya on the zeros of successive derivatives of real entire functions (Q869961)

Let \(f\) be a real entire function with only finitely many non-real zeros. In a survey article [Bull. Am. Math. Soc. 49, 178--191 (1943; Zbl 0061.11510)], \textit{G. Polya} has formulated the following conjectures: Conjecture \(A\): If the order of \(f\) is less than 2, then its derivatives, from a certain one onwards, will have no non-real zeros at all. Conjecture \(B\): If the order of \(f\) is greater than 2, then the number of non-real zeros of \(f^{(n)}\) tends to infinity as \(n\to\infty\). Conjecture \(A\) was established by \textit{T. Craven, G. Csordas} and \textit{W. Smith} [Ann. Math. (2) 125, 405--431 (1987; Zbl 0625.30036)] and refined by \textit{H. Ki} and \textit{Y.-O. Kim} [Duke Math. J. 104, No. 1, 45--73 (2000; Zbl 0957.30019)], who proved that the statement holds for all \(f = Ph\), where \(P\) is a real polynomial and \(h\) belongs to the Laguerre-Polya class, that is, the class of entire functions being the locally uniform limit of real polynomials with all zeros real. Only recently, \textit{Langley} [J. Anal. Math. 97, 357--396 (2005)] has shown that, for entire functions of infinite order, \(f^{(n)}\) has infinitely many non-real zeros for all \(n\geq 2\). Finishing a long line of deep investigations, the open case of functions of finite order greater than 2 is completely settled in the paper under review. In fact, the main theorem shows that, for every \(f\) of finite order which is not of the form \(f = Ph\) as above, the number \(N(f^{(n)})\) of non-real zeros of \(f^{(n)}\) satisfies \[ \liminf_{n\to \infty} \frac{N(f^{(n)})}{n}>0\; . \] The proof combines the saddle point method from asymptotic analysis with potential theory and the theory of analytic functions with positive imaginary part in the upper half plane.