Zeros of solutions for higher order linear differential equations with meromorphic coefficients (Q2896288)

Let \(\rho(f)\) denote the order of a meromorphic function \(f\). Estimates of the growth of solutions for the linear differential equations NEWLINE\[NEWLINE f^{(k)}+ A_{k-1}(z)f^{(k-1)}+\dotsb +A_{0}(z)f=0, \eqno(1)NEWLINE\]NEWLINE when \(A_j\), \(j\in \{0, \dotsc, k-1\}\) are polynomials are known at least for five decades (cf. [\textit{H. Wittich}, Neuere Untersuchungen über eindeutige analytische Funktionen. Berlin, Göttingen, Heidelberg: Springer-Verlag (1955; Zbl 0067.05501); \textit{K. Pöschl}, J. Reine Angew. Math. 199, 121--138 (1958; Zbl 0082.07101); ibid. 200, 129--139 (1958; Zbl 0083.31003)]).NEWLINENEWLINE NEWLINEThe maximal growth can be estimated using the following theorem.NEWLINENEWLINETheorem A. Suppose that the coefficients \(A_j\), \(j\in \{0, \dotsc, k-1\}\) are entire. Then \(A_j\), \(j\in \{0, \dotsc, k-1\}\) are polynomials if and only if all solutions \(f\) of (1) are of finite order of growth \(\rho(f)\). More precisely, if \(\rho\geq 1\), then \(\deg A_j \leq (k-j)(\rho-1) \) for all \(j\in \{0, \dotsc, k-1\}\) if and only if \(\rho(f)\leq \rho\).NEWLINENEWLINEFor an entire function \(f\) with the zero sequence \((z_n)\), let \(\lambda(f)=\inf \{ \alpha>0: \sum_n |z_n|^{-\alpha}<\infty\}\) be the convergence exponent of its zeros. It is well-known that \(\lambda(f)\leq \rho(f)\). It can happen that, for all non-trivial solutions of (1), \(\lambda(f)< \rho(f)\) (e.g., when all solutions are the exponent function multiplied by a polynomial).NEWLINENEWLINEUsing a standard substitution of the form \(g=fe^{-\Phi(z)/k}\), where \(\Phi(z)\) is a primitive of \(A_{k-1}\), which does not change zeros, we obtain an equivalent equation NEWLINE\[NEWLINE g^{(k)}+ a_{k-2}(z)g^{(k-2)}+\dotsb +a_{0}(z)g=0, \quad k\geq 2 \eqno(2).NEWLINE\]NEWLINE In 1991, Bank and Langley proved the following theorem.NEWLINENEWLINE Theorem B. Let \(a_j\), \(j\in \{0, \dotsc, k-1\}\) be entire functions of finite order, and let (2) possess a solution base \(g_1, \dotsc, g_n\) such that \(\lambda(g_j)< \infty\), \(j\in \{ 1, \dotsc, n\}\). Then the product \(g_1\dotsm g_n\) is of finite order of growth.NEWLINENEWLINENote that sharp estimates of growth of the coefficients, and consequently, of solutions of (2) in terms of convergence exponents of a solution base has been obtained very recently in a note by \textit{J. Heittokangas} and \textit{J. Rättyä} [Math. Nachr. 284, No. 4, 412--420 (2011; Zbl 1220.34109)].NEWLINENEWLINEThe authors investigate the case when coefficients \(a_j\) are meromorphic.NEWLINENEWLINETheorem 1. Let \(a_j\), \(j\in \{0, \dotsc, k-2\}\) be meromorphic functions of finite order, and let (2) possess a solution base \(g_1, \dotsc, g_n\) such that \(\lambda(g_j)< \infty\), \(j\in \{ 1, \dotsc, n\}\). Then the product \(g_1\dotsm g_n\) is of finite order of growth.NEWLINENEWLINE The result is a generalization of a result by \textit{J. Tu} et al. [J. Math., Wuhan Univ. 27, No. 1, 77--82 (2007; Zbl 1121.34344)]. The proof uses a technique of Nevanlina's theory in an angular domain developed by M. Tsuji.