Non-real zeros of linear differential polynomials (Q1042656): Difference between revisions

For each non-negative integer \(p\), the class \(V_{2p}\) consists of all entire functions \(f(z)=g(z)\) \(\exp(-az^{2p+2})\), where \(a\geq 0\) and \(g\) is a real entire function (i.e., assuming only real values on the real axis) of genus at most \(2p+1\). The classes \(U_{2p}\) are then given by \(U_0=V_0\) and \(U_{2p}=V_{2p}\setminus V_{2p-2}\) for \(p\geq 1\). Moreover, \(U^*_{2p}\) is the class of entire functions \(f=Ph\), where \(h\in U_{2p}\) and \(P\) is a real polynomial without real zeros. The following result was conjectured by Wiman in 1911 and proved by \textit{T. Sheil-Small} [Ann. Math. (2) 129, No. 1, 179--193 (1989; Zbl 0671.30027)], see also [\textit{S. Edwards} and \textit{S. Hellerstein}, Complex Variables, Theory Appl. 47, No. 1, 25--57 (2002; Zbl 1021.30026)]: if, for some \(p\in \mathbb N\), \(f\in U^*_{2p}\), then \(f^{\prime\prime}\) has at least \(2p\) non-real zeros. In this paper, the author considers the following related problem: if \(f\) is a real entire function with infinitely many non-real zeros, must a linear differential polynomial \(\Psi=f^{(k)}+a_{k-1}f^{(k-1)}+\dots a_0f\) with \(a_j\in\mathbb R\) have non-real zeros and, if so, how many? The question is answered for \(k=2\), in which case it may be assumed with no loss of generality that \(a_1=0\). This is given by the following main result. Let \(f\) be a real entire function with infinitely many non-real zeros, and let \(\omega\) be a positive real number. If \(f\in U^*_{2p}\) for some \(p\in\mathbb N\), then \(f^{\prime\prime}+\omega f\) has at least \(2p\) non-real zeros. If \(f\) has infinite order, then \(f^{\prime\prime}+\omega f\) has infinitely many non-real zeros.