The separation of zeros of solutions of higher order linear differential equations with entire coefficients (Q351145): Difference between revisions

Let \(f_1\) and \(f_2\) be linearly independent entire solutions of \[ w'' + A w =0, \] where \(A\) is an entire function. Suppose that \(f_1\) and \(f_2\) are normalized in such a way that their Wronskian satisfies \[ W(f_1,f_2) =1. \] Then by defining \(E=f_1f_2\) it follows that, for all \(z\in\mathbb{C}\), \[ E(z)=0\qquad \Longrightarrow \qquad E'(z)=\pm1. \] The property above is called the ``Bank-Laine property'' and an entire function satisfying it a ``Bank-Laine function''. \textit{S. M. Elzaidi} [Complex Variables, Theory Appl. 38, No. 3, 201--220 (1999; Zbl 1036.34099)] has shown that the zeros of a finite-order Bank-Laine function cannot be too close to each other. In particular, he has shown that if \(E\) is a finite-order Bank-Laine function with zero sequence \((z_n)\), then there exists \(M>0\) such that \[ |z_m-z_n| \geq \exp\left(-|z_n|^M\right) \] for \(m\not=n\) and \(n\) large. Equivalently, if \(E\) is a finite-order Bank-Laine function, then there exists \(M>0\) such that for all large \(w\) the set \[ V_m = B\big(w,\exp(-|w|^M)\big) \] contains at most one zero of \(E\), where \(B(a,r)\) is the open disc of centre \(a\) and radius \(r\). In the paper under review, the authors prove a counterpart of Elzaidi's result for the higher-order equation \[ w^{(k)} + A_{k-2} w^{(k-2)} + \cdots + A_1 w' + A_0 w =0, \] where \(k\geq 2\) and \(A_0,\ldots,A_{k-2}\) are entire functions. They suppose that the higher-order linear equation above has \(k\) linearly independent entire solutions \(f_1,\ldots,f_k\) such that \(\lambda(f_j)<\infty\) for all \(j\in\{1,\ldots,k\}\), define \(E=f_1\cdots f_k\) and let \[ \max\{\lambda(E),\rho(A_0),\ldots,\rho(A_{k-2})\} < M < \infty. \] Under these assumptions the authors show that for large \(w\in\mathbb{C}\) and \(V_m = B\big(w,\exp(-|w|^M)\big)\) the following conclusions hold. (a) At least one of \(f_1,\ldots,f_k\) has no zeros in \(V_m\). (b) Each of \(f_1,\ldots,f_k\) has at most \(k-1\) zeros in \(V_m\). (c) The product \(E\) has at most \(k(k-1)/2\) zeros in \(V_m\).